| 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: Matthias Blume |
Name: Matthias Blume |
| 17 |
|
Date: 2002/02/13 16:41:00 EST |
| 18 |
|
Tag: blume-20020213-cmlpq |
| 19 |
|
Description: |
| 20 |
|
|
| 21 |
|
Added new priority queue export symbols (which have just been added to |
| 22 |
|
smlnj-lib.cm) to CML's version of smlnj-lib.cm. (Otherwise CML would |
| 23 |
|
not compile and the installer would choke.) |
| 24 |
|
|
| 25 |
|
---------------------------------------------------------------------- |
| 26 |
|
Name: Matthias Blume |
| 27 |
|
Date: 2002/02/13 16:15:00 EST |
| 28 |
|
Tag: blume-20020213-various |
| 29 |
|
Description: |
| 30 |
|
|
| 31 |
|
1. More tweaks to ml-nlffigen: |
| 32 |
|
|
| 33 |
|
- better internal datastructures (resulting in slight speedup) |
| 34 |
|
- "-match" option requires exact match |
| 35 |
|
- "localized" gensym counters (untagged structs/unions nested within |
| 36 |
|
other structs/unions or within typedefs get a fresh counter; their |
| 37 |
|
tag will be prefixed by a concatenation of their parents' tags) |
| 38 |
|
- bug fixes (related to calculation of transitive closure of types |
| 39 |
|
to be included in the output) |
| 40 |
|
|
| 41 |
|
2. Minor Basis updates: |
| 42 |
|
|
| 43 |
|
- added implementations for List.collate and Option.app |
| 44 |
|
|
| 45 |
|
---------------------------------------------------------------------- |
| 46 |
|
Name: Matthias Blume |
| 47 |
|
Date: 2002/02/11 15:55:00 EST |
| 48 |
|
Tag: blume-20020211-gensym |
| 49 |
|
Description: |
| 50 |
|
|
| 51 |
|
Added a "-gensym" option to command line of ml-nlffigen. This can be |
| 52 |
|
used to specify a "stem" -- a string that is inserted in all "gensym'd" |
| 53 |
|
names (ML structure names that correspond to unnamed C structs, unions, |
| 54 |
|
and enums), so that separate runs of ml-nlffigen do not clash. |
| 55 |
|
|
| 56 |
|
---------------------------------------------------------------------- |
| 57 |
|
Name: Matthias Blume |
| 58 |
|
Date: 2002/02/11 12:05:00 EST |
| 59 |
|
Tag: blume-20020211-gensml |
| 60 |
|
Description: |
| 61 |
|
|
| 62 |
|
A quick fix for a problem with GenSML (in the pgraph-util library): |
| 63 |
|
Make generation of toplevel "local" optional. (Strictly speaking, |
| 64 |
|
signature definitions within "local" are not legal SML.) |
| 65 |
|
|
| 66 |
|
Other than that: updates to INSTALL and cm/TODO. |
| 67 |
|
|
| 68 |
|
---------------------------------------------------------------------- |
| 69 |
|
Name: Matthias Blume |
| 70 |
|
Date: 2002/02/08 15:00:00 EST |
| 71 |
|
Tag: blume-20020208-uniquepid |
| 72 |
|
Description: |
| 73 |
|
|
| 74 |
|
0. Version number has been bumped to 110.38.1. NEW BOOTFILES!!! |
| 75 |
|
|
| 76 |
|
1. The installer (config/install.sh) has gotten smarter: |
| 77 |
|
|
| 78 |
|
- Configuration options are a bit easier to specify now |
| 79 |
|
(in config/targets). |
| 80 |
|
- Bug in recognizing .tar.bz2 files fixed. |
| 81 |
|
- Installer automatically resolves dependencies between |
| 82 |
|
configuration options (e.g., if you ask for eXene, you will |
| 83 |
|
also get cml -- regardless whether you asked for it or not). |
| 84 |
|
- Installer can run in "quieter mode" by setting the environment |
| 85 |
|
variable INSTALL_QUIETLY to "true". "Quieter" does not mean |
| 86 |
|
"completely silent", though. |
| 87 |
|
- Build HashCons library as part of smlnj-lib. |
| 88 |
|
|
| 89 |
|
2. A new scheme for assigning persistent identifiers to compilation |
| 90 |
|
units (and, by extension, to types etc.) has been put into place. |
| 91 |
|
This fixes a long-standing bug where types and even dynamic values |
| 92 |
|
can get internally confused, thereby compromising type safety |
| 93 |
|
(abstraction) and dynamic correctness. See |
| 94 |
|
|
| 95 |
|
http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz |
| 96 |
|
|
| 97 |
|
for an example of how things could go wrong until now. |
| 98 |
|
|
| 99 |
|
The downside of the new scheme is that pids are not quite as |
| 100 |
|
persistent as they used to be: CM will generate a fresh pid |
| 101 |
|
for every compilation unit that it thinks it sees for the first |
| 102 |
|
time. That means that if you compile starting from a clean, fresh |
| 103 |
|
source tree at two different times, you end up with different |
| 104 |
|
binaries. |
| 105 |
|
|
| 106 |
|
Cutoff recompilation, however, has not been compromised because |
| 107 |
|
CM keeps pid information in special caches between runs. |
| 108 |
|
|
| 109 |
|
---------------------------------------------------------------------- |
| 110 |
|
Name: Lal George |
| 111 |
|
Date: 2002/02/07 15:34:13 EST 2002 |
| 112 |
|
Tag: <none> |
| 113 |
|
Description: |
| 114 |
|
|
| 115 |
|
Compilers that generate assembly code may produce global labels |
| 116 |
|
whose value is resolved at link time. The various peephole optimization |
| 117 |
|
modules did not take this in account. |
| 118 |
|
|
| 119 |
|
TODO. The Labels.addrOf function should really return an option |
| 120 |
|
type so that clients are forced to deal with this issue, rather |
| 121 |
|
than an exception being raised. |
| 122 |
|
|
| 123 |
|
---------------------------------------------------------------------- |
| 124 |
|
Name: Lal George |
| 125 |
|
Date: 2002/02/06 13:55:02 EST |
| 126 |
|
Tag: george-20020206-ra-breakup |
| 127 |
|
Description: |
| 128 |
|
|
| 129 |
|
1. A bug fix from Allen. |
| 130 |
|
|
| 131 |
|
A typo causes extra fstp %st(0)'s to be generated at compensation |
| 132 |
|
edges, which might cause stack underflow traps at runtime. This |
| 133 |
|
occurs in fft where there are extraneous fstps right before the 'into' |
| 134 |
|
trap instruction (in this case they are harmless since none of the |
| 135 |
|
integers overflow.) |
| 136 |
|
|
| 137 |
|
2. Pulled out various utility modules that were embedded in the modules |
| 138 |
|
of the register allocator. I need these modules for other purposes, but |
| 139 |
|
they are not complete enough to put into a library (just yet). |
| 140 |
|
---------------------------------------------------------------------- |
| 141 |
|
Name: Matthias Blume |
| 142 |
|
Date: 2002/01/31 16:05:00 EST |
| 143 |
|
Tag: blume-20020131-sparc-ccalls |
| 144 |
|
Description: |
| 145 |
|
|
| 146 |
|
1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes) |
| 147 |
|
of extra stack space by mistake. Fixed. |
| 148 |
|
|
| 149 |
|
2. Bug in logic of handling of command-line options in ml-nlffigen fixed. |
| 150 |
|
|
| 151 |
|
---------------------------------------------------------------------- |
| 152 |
|
Name: Allen Leung |
| 153 |
|
Date: 2002/01/30 |
| 154 |
|
Tag: leunga-20020130-nowhere-bug-fix |
| 155 |
|
Description: |
| 156 |
|
|
| 157 |
|
MLRISC bug fixes: |
| 158 |
|
1. Fixed a bindings computation bug in the 'nowhere' program generator tool. |
| 159 |
|
2. MachineInt.fromString was negating its value. |
| 160 |
|
|
| 161 |
|
---------------------------------------------------------------------- |
| 162 |
|
Name: Matthias Blume |
| 163 |
|
Date: 2002/01/29 |
| 164 |
|
Tag: blume-20020129-INSTALL |
| 165 |
|
Description: |
| 166 |
|
|
| 167 |
|
- Added somewhat detailed installation instructions (file INSTALL). |
| 168 |
|
- Fixed curl-detection bug in config/install.sh. |
| 169 |
|
- It is now possible to select the URL getter using the URLGETTER |
| 170 |
|
environment variable: |
| 171 |
|
|
| 172 |
|
not set / "unknown" --> automatic detection (script tries wget, |
| 173 |
|
curl, and lynx) |
| 174 |
|
"wget" / "curl" / "lynx" --> use the specified program (script "knows" |
| 175 |
|
how to properly invoke them) |
| 176 |
|
other --> use $URLGETTER directly, it must take |
| 177 |
|
precisely two command-line arguments |
| 178 |
|
(source URL and destination file name) |
| 179 |
|
|
| 180 |
|
---------------------------------------------------------------------- |
| 181 |
|
Name: Matthias Blume |
| 182 |
|
Date: 2002/01/28 |
| 183 |
|
Tag: blume-20020128-sparc-ccalls |
| 184 |
|
Description: |
| 185 |
|
|
| 186 |
|
- Fixed problem with calculation of "used" registers in sparc-c-calls. |
| 187 |
|
- Make use of the allocParam argument in sparc-c-calls. |
| 188 |
|
|
| 189 |
|
---------------------------------------------------------------------- |
| 190 |
|
Name: Matthias Blume |
| 191 |
|
Date: 2002/01/28 |
| 192 |
|
Tag: blume-20020128-allocParam |
| 193 |
|
Description: |
| 194 |
|
|
| 195 |
|
John Reppy: Changes c-calls API to accept client-callback for |
| 196 |
|
allocating extra stack space. |
| 197 |
|
me: Corresponding changes to mlriscGen (using a dummy argument that |
| 198 |
|
does not change the current behavior). |
| 199 |
|
|
| 200 |
|
---------------------------------------------------------------------- |
| 201 |
|
Name: Matthias Blume |
| 202 |
|
Date: 2002/01/28 12:00:00 |
| 203 |
|
Tag: Release_110_38 |
| 204 |
|
Description: |
| 205 |
|
|
| 206 |
|
This time for real!!! |
| 207 |
|
|
| 208 |
|
---------------------------------------------------------------------- |
| 209 |
|
Name: Matthias Blume |
| 210 |
|
Date: 2002/01/28 10:56:00 EST |
| 211 |
|
Tag: blume-20020128-retraction |
| 212 |
|
Description: |
| 213 |
|
|
| 214 |
|
0. Retracted earlier 110.38. (The Release_110_38 tag has been replaced |
| 215 |
|
with blume-Release_110_38-retracted.) |
| 216 |
|
|
| 217 |
|
1. Fixed a problem with incorrect rounding modes in real64.sml. |
| 218 |
|
(Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.) |
| 219 |
|
|
| 220 |
|
2. A bug in ml-nlffigen related to the handling of unnamed structs, unions, |
| 221 |
|
and enums fixed. The naming of corresponding ML identifiers should |
| 222 |
|
now be consistent again. |
| 223 |
|
|
| 224 |
|
---------------------------------------------------------------------- |
| 225 |
|
Name: Allen Leung |
| 226 |
|
Date: 2002/01/27 |
| 227 |
|
Tag: leunga-20020127-nowhere |
| 228 |
|
Description: |
| 229 |
|
|
| 230 |
|
Added a target called nowhere in the configuration scripts. |
| 231 |
|
Enabling this will build the MLRISC 'nowhere' tool (for translating |
| 232 |
|
programs with where-clauses into legal SML code) during installation. |
| 233 |
|
|
| 234 |
|
---------------------------------------------------------------------- |
| 235 |
|
Name: Matthias Blume |
| 236 |
|
Date: 2002/01/25 21:27:00 EST |
| 237 |
|
Tag: blume-Release_110_38-retracted |
| 238 |
|
Description: |
| 239 |
|
|
| 240 |
|
Call it a (working) release! Version is 110.38. Bootfiles are ready. |
| 241 |
|
|
| 242 |
|
README will be added later. |
| 243 |
|
|
| 244 |
|
!!! NOTE: Re-tagged as blume-Release_110_38-retracted. Original tag |
| 245 |
|
(Release_110_38) removed. Reason: Last-minute bug fixes. |
| 246 |
|
|
| 247 |
|
---------------------------------------------------------------------- |
| 248 |
|
Name: Matthias Blume |
| 249 |
|
Date: 2002/01/25 |
| 250 |
|
Tag: blume-20020125-ffi |
| 251 |
|
Description: |
| 252 |
|
|
| 253 |
|
A large number of tweaks and improvements to ml-nlffi-lib and |
| 254 |
|
ml-nlffigen: |
| 255 |
|
|
| 256 |
|
- ML represenation types have been streamlined |
| 257 |
|
- getter and setter functions work with concrete values, not abstract |
| 258 |
|
ones where possible |
| 259 |
|
- ml-nlffigen command line more flexible (see README file there) |
| 260 |
|
- some bugs have been fixed (hopefully) |
| 261 |
|
|
| 262 |
|
---------------------------------------------------------------------- |
| 263 |
|
Name: Lal George |
| 264 |
|
Date: 2002/01/24 |
| 265 |
|
Tag: george-20020124-risc-ra-interface |
| 266 |
|
Description: |
| 267 |
|
|
| 268 |
|
There is a dramatic simplification in the interface to the |
| 269 |
|
register allocator for RISC architectures as a result of making |
| 270 |
|
parallel copy instructions explicit. |
| 271 |
|
|
| 272 |
|
---------------------------------------------------------------------- |
| 273 |
|
Name: Matthias Blume |
| 274 |
|
Date: 2002/01/22 |
| 275 |
|
Tag: blume-20020122-x86-ccalls |
| 276 |
|
Description: |
| 277 |
|
|
| 278 |
|
Bug fix for c-calls on x86 (having to do with how char- and |
| 279 |
|
short-arguments are being handled). |
| 280 |
|
|
| 281 |
|
---------------------------------------------------------------------- |
| 282 |
|
Name: Matthias Blume |
| 283 |
|
Date: 2002/01/21 |
| 284 |
|
Tag: blume-20020121-ff |
| 285 |
|
Description: |
| 286 |
|
|
| 287 |
|
Another day of fiddling with the FFI... |
| 288 |
|
|
| 289 |
|
1. Bug fix/workaround: CKIT does not complain about negative array |
| 290 |
|
dimensions, so ml-nlffigen has to guard itself against this possibility. |
| 291 |
|
(Otherwise a negative dimension would send it into an infinite loop.) |
| 292 |
|
|
| 293 |
|
2. Some of the abstract types (light objects, light pointers, most "base" |
| 294 |
|
types) in structure C are now eqtypes. |
| 295 |
|
|
| 296 |
|
3. Added constructors and test functions for NULL function pointers. |
| 297 |
|
|
| 298 |
|
---------------------------------------------------------------------- |
| 299 |
|
Name: Matthias Blume |
| 300 |
|
Date: 2002/01/18 |
| 301 |
|
Tag: blume-20020118-ready-for-new-release |
| 302 |
|
Description: |
| 303 |
|
|
| 304 |
|
Made config/srcarchiveurl point to a new place. (Will provide boot |
| 305 |
|
files shortly.) |
| 306 |
|
|
| 307 |
|
Maybe we christen this to be 110.38? |
| 308 |
|
|
| 309 |
|
---------------------------------------------------------------------- |
| 310 |
|
Name: Matthias Blume |
| 311 |
|
Date: 2002/01/18 |
| 312 |
|
Tag: blume-20020118-more-ffifiddle |
| 313 |
|
Description: |
| 314 |
|
|
| 315 |
|
Today's FFI fiddling: |
| 316 |
|
|
| 317 |
|
- Provided a structure CGetSet with "convenient" versions of C.Get.* and |
| 318 |
|
C.Set.* that use concrete (MLRep.*) arguments and results instead |
| 319 |
|
of abstract ones. |
| 320 |
|
|
| 321 |
|
- Provided word-style bit operations etc. for "int" representation |
| 322 |
|
types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short, |
| 323 |
|
and Long. |
| 324 |
|
|
| 325 |
|
---------------------------------------------------------------------- |
| 326 |
|
Name: Matthias Blume |
| 327 |
|
Date: 2002/01/18 |
| 328 |
|
Tag: blume-20020118-use-x86-fp |
| 329 |
|
Description: |
| 330 |
|
|
| 331 |
|
Now that x86-fast-fp seems to be working, I turned it back on again |
| 332 |
|
by default. (Seems to work fine now, even with the FFI.) |
| 333 |
|
|
| 334 |
|
Other than that, I added some documentation about the FFI to |
| 335 |
|
src/ml-nlffigen/README and updated the FFI test examples in |
| 336 |
|
src/ml-nlffi-lib/Tests/*. |
| 337 |
|
|
| 338 |
|
---------------------------------------------------------------------- |
| 339 |
|
Name: Allen Leung |
| 340 |
|
Date: 2002/01/17 |
| 341 |
|
Tag: leunga-20020117-x86-fast-fp-call |
| 342 |
|
Description: |
| 343 |
|
|
| 344 |
|
1. Fixed a problem with handling return fp values when x86's fast fp |
| 345 |
|
mode is turned on. |
| 346 |
|
|
| 347 |
|
2. Minor pretty printing fix for cellset. Print %st(0) as %st(0) instead |
| 348 |
|
of %f32. |
| 349 |
|
|
| 350 |
|
3. Added a constructor INT32lit to the ast of MLRISC tools. |
| 351 |
|
|
| 352 |
|
---------------------------------------------------------------------- |
| 353 |
|
Name: Matthias Blume |
| 354 |
|
Date: 2002/01/16 |
| 355 |
|
Tag: blume-20020116-ffifiddle |
| 356 |
|
Description: |
| 357 |
|
|
| 358 |
|
More fiddling with the FFI interface: |
| 359 |
|
|
| 360 |
|
- Make constness 'c instead of rw wherever possible. This eliminates |
| 361 |
|
the need for certain explicit coercions. (However, due to ML's |
| 362 |
|
value polymorphism, there will still be many cases where explicit |
| 363 |
|
coercions are necessary. Phantom types are not the whole answer |
| 364 |
|
to modeling a subtyping relationship in ML.) |
| 365 |
|
|
| 366 |
|
- ro/rw coersions for pointers added. (Avoids the detour through */&.) |
| 367 |
|
|
| 368 |
|
- "printf" test example added to src/ml-nlffi-lib/Tests. (Demonstrates |
| 369 |
|
clumsy workaround for varargs problem.) |
| 370 |
|
|
| 371 |
|
---------------------------------------------------------------------- |
| 372 |
|
Name: Lal George |
| 373 |
|
Date: 2002/01/15 |
| 374 |
|
Tag: <none> |
| 375 |
|
Description: |
| 376 |
|
|
| 377 |
|
1. Since COPY instructions are no longer native to the architecture, |
| 378 |
|
a generic functor can be used to implement the expandCopies function. |
| 379 |
|
|
| 380 |
|
2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a |
| 381 |
|
TEXT segment. |
| 382 |
|
|
| 383 |
|
---------------------------------------------------------------------- |
| 384 |
|
Name: Matthias Blume |
| 385 |
|
Date: 2002/01/15 |
| 386 |
|
Tag: blume-20020115-ffiupdates |
| 387 |
|
Description: |
| 388 |
|
|
| 389 |
|
1. Fix for bug resulting in single-precision float values being returned |
| 390 |
|
incorrectly from FFI calls. |
| 391 |
|
|
| 392 |
|
2. Small modifications to C FFI API: |
| 393 |
|
|
| 394 |
|
- memory-allocation routines return straight objects (no options) |
| 395 |
|
and raise an exception in out-of-memory situations |
| 396 |
|
- unsafe extensions to cast between function pointers and pointers |
| 397 |
|
from/to ints |
| 398 |
|
- added structure C_Debug as an alternative to structure C where |
| 399 |
|
pointer-dereferencing (|*| and |*!) always check for null-pointers |
| 400 |
|
- added open_lib' to DynLinkage; open_lib' works like open_lib |
| 401 |
|
but also takes a (possibly empty) list of existing library handles |
| 402 |
|
that the current library depends on |
| 403 |
|
|
| 404 |
|
---------------------------------------------------------------------- |
| 405 |
|
Name: Matthias Blume |
| 406 |
|
Date: 2002/01/10 |
| 407 |
|
Tag: blume-20020110-newffigen |
| 408 |
|
Description: |
| 409 |
|
|
| 410 |
|
1. Updates to portable graph code. |
| 411 |
|
|
| 412 |
|
2. Major update to ml-nlffigen and ml-nlffi-lib. Things are much |
| 413 |
|
more scalable now so that even huge interfaces such as the one |
| 414 |
|
for GTK compile in finite time and space. :-) |
| 415 |
|
See src/ml-nlffigen/README for details on what's new. |
| 416 |
|
|
| 417 |
|
---------------------------------------------------------------------- |
| 418 |
|
Name: Lal George |
| 419 |
|
Date: 2001/01/09 14:31:35 EST 2002 |
| 420 |
|
Tag: george-20011206-rm-native-copy |
| 421 |
|
Description: |
| 422 |
|
|
| 423 |
|
Removed the native COPY and FCOPY instructions |
| 424 |
|
from all the architectures and replaced it with the |
| 425 |
|
explicit COPY instruction from the previous commit. |
| 426 |
|
|
| 427 |
|
It is now possible to simplify many of the optimizations |
| 428 |
|
modules that manipulate copies. This has not been |
| 429 |
|
done in this change. |
| 430 |
|
|
| 431 |
|
---------------------------------------------------------------------- |
| 432 |
|
Name: Lal George |
| 433 |
|
Date: 2001/12/06 16:50:13 EST 2001 |
| 434 |
|
Tag: george-20011206-mlrisc-instruction |
| 435 |
|
Description: |
| 436 |
|
|
| 437 |
|
Changed the representation of instructions from being fully abstract |
| 438 |
|
to being partially concrete. That is to say: |
| 439 |
|
|
| 440 |
|
from |
| 441 |
|
type instruction |
| 442 |
|
|
| 443 |
|
to |
| 444 |
|
type instr (* machine instruction *) |
| 445 |
|
|
| 446 |
|
datatype instruction = |
| 447 |
|
LIVE of {regs: C.cellset, spilled: C.cellset} |
| 448 |
|
| KILL of {regs: C.cellset, spilled: C.cellset} |
| 449 |
|
| COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list} |
| 450 |
|
| ANNOTATION of {i: instruction, a: Annotations.annotation} |
| 451 |
|
| INSTR of instr |
| 452 |
|
|
| 453 |
|
This makes the handling of certain special instructions that appear on |
| 454 |
|
all architectures easier and uniform. |
| 455 |
|
|
| 456 |
|
LIVE and KILL say that a list of registers are live or killed at the |
| 457 |
|
program point where they appear. No spill code is generated when an |
| 458 |
|
element of the 'regs' field is spilled, but the register is moved to |
| 459 |
|
the 'spilled' (which is present, more for debugging than anything else). |
| 460 |
|
|
| 461 |
|
LIVE replaces the (now deprecated) DEFFREG instruction on the alpha. |
| 462 |
|
We used to generate: |
| 463 |
|
|
| 464 |
|
DEFFREG f1 |
| 465 |
|
f1 := f2 + f3 |
| 466 |
|
trapb |
| 467 |
|
|
| 468 |
|
but now generate: |
| 469 |
|
|
| 470 |
|
f1 := f2 + f3 |
| 471 |
|
trapb |
| 472 |
|
LIVE {regs=[f1,f2,f3], spilled=[]} |
| 473 |
|
|
| 474 |
|
Furthermore, the DEFFREG (hack) required that all floating point instruction |
| 475 |
|
use all registers mentioned in the instruction. Therefore f1 := f2 + f3, |
| 476 |
|
defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting |
| 477 |
|
in a cleaner alpha implementation. (Hopefully, intel will not get rid of |
| 478 |
|
this architecture). |
| 479 |
|
|
| 480 |
|
COPYXXX is intended to replace the parallel COPY and FCOPY available on |
| 481 |
|
all the architectures. This will result in further simplification of the |
| 482 |
|
register allocator that must be aware of them for coalescing purposes, and |
| 483 |
|
will also simplify certain aspects of the machine description that provides |
| 484 |
|
callbacks related to parallel copies. |
| 485 |
|
|
| 486 |
|
ANNOTATION should be obvious, and now INSTR represents the honest to God |
| 487 |
|
machine instruction set! |
| 488 |
|
|
| 489 |
|
The <arch>/instructions/<arch>Instr.sml files define certain utility |
| 490 |
|
functions for making porting easier -- essentially converting upper case |
| 491 |
|
to lower case. All machine instructions (of type instr) are in upper case, |
| 492 |
|
and the lower case form generates an MLRISC instruction. For example on |
| 493 |
|
the alpha we have: |
| 494 |
|
|
| 495 |
|
datatype instr = |
| 496 |
|
LDA of {r:cell, b:cell, d:operand} |
| 497 |
|
| ... |
| 498 |
|
|
| 499 |
|
val lda : {r:cell, b:cell, d:operand} -> instruction |
| 500 |
|
... |
| 501 |
|
|
| 502 |
|
where lda is just (INSTR o LDA), etc. |
| 503 |
|
|
| 504 |
|
---------------------------------------------------------------------- |
| 505 |
|
Name: Matthias Blume |
| 506 |
|
Date: 2001/11/22 21:40:00 EST |
| 507 |
|
Tag: Release_110_37 |
| 508 |
|
Description: |
| 509 |
|
|
| 510 |
|
Release 110.37. This time for real. |
| 511 |
|
|
| 512 |
|
---------------------------------------------------------------------- |
| 513 |
|
Name: Matthias Blume |
| 514 |
|
Date: 2001/11/21 16:35:00 EST |
| 515 |
|
Tag: blume-20011121-foot-in-mouth |
| 516 |
|
Description: |
| 517 |
|
|
| 518 |
|
Removed the "Release_110_37" tag because of a serious bug. |
| 519 |
|
This will be re-tagged once the bug is fixed. |
| 520 |
|
|
| 521 |
|
---------------------------------------------------------------------- |
| 522 |
|
Name: Matthias Blume |
| 523 |
|
Date: 2001/11/21 16:14:00 EST |
| 524 |
|
Tag: blume-20011121-forgottenfile |
| 525 |
|
Description: |
| 526 |
|
|
| 527 |
|
Forgot to add a file. (Just a .tex-file -- part of |
| 528 |
|
the CM manual source.) |
| 529 |
|
|
| 530 |
|
---------------------------------------------------------------------- |
| 531 |
|
Name: Matthias Blume |
| 532 |
|
Date: 2001/11/21 16:10:00 EST |
| 533 |
|
Tag: blume-20011121-invalid_110_37 |
| 534 |
|
Description: |
| 535 |
|
|
| 536 |
|
Note: I removed the original tag "Release_110_37" from this commit |
| 537 |
|
because we found a serious bug in all non-x86 backends. |
| 538 |
|
- Matthias |
| 539 |
|
|
| 540 |
|
1. Modifications to the SML/NJ code generator and to the runtime system |
| 541 |
|
so that code object name strings are directly inserted into code |
| 542 |
|
objects at code generation time. The only business the runtime system |
| 543 |
|
has with this is now to read the name strings on occasions. |
| 544 |
|
(The encoding of the name string has also changed somewhat.) |
| 545 |
|
|
| 546 |
|
2. CM now implements a simple "set calculus" for specifying export lists. |
| 547 |
|
In particular, it is now possible to refer to the export lists of |
| 548 |
|
other libraries/groups/sources and form unions as well as differences. |
| 549 |
|
See the latest CM manual for details. |
| 550 |
|
|
| 551 |
|
3. An separate notion of "proxy" libraries has again be eliminated from |
| 552 |
|
CM's model. (Proxy libraries are now simply a special case of using |
| 553 |
|
the export list calculus.) |
| 554 |
|
|
| 555 |
|
4. Some of the existing libraries now take advantage of the new set |
| 556 |
|
calculus. |
| 557 |
|
(Notice that not all libraries have been converted because some |
| 558 |
|
of the existing .cm-files are supposed to be backward compatible |
| 559 |
|
with 110.0.x.) |
| 560 |
|
|
| 561 |
|
5. Some cleanup in stand-alone programs. (Don't use "exnMessage" -- use |
| 562 |
|
"General.exnMessage"! The former relies on a certain hook to be |
| 563 |
|
initialized, and that often does not happen in the stand-alone case.) |
| 564 |
|
|
| 565 |
|
---------------------------------------------------------------------- |
| 566 |
|
Name: Lal George |
| 567 |
|
Date: 2001/11/21 13:56:18 EST |
| 568 |
|
Tag: george-2001121-pseudo-ops |
| 569 |
|
Description: |
| 570 |
|
|
| 571 |
|
Implemented a complete redesign of MLRISC pseudo-ops. Now there |
| 572 |
|
ought to never be any question of incompatabilities with |
| 573 |
|
pseudo-op syntax expected by host assemblers. |
| 574 |
|
|
| 575 |
|
For now, only modules supporting GAS syntax are implemented |
| 576 |
|
but more should follow, such as MASM, and vendor assembler |
| 577 |
|
syntax, e.g. IBM as, Sun as, etc. |
| 578 |
|
|
| 579 |
|
---------------------------------------------------------------------- |
| 580 |
|
Name: Matthias Blume |
| 581 |
|
Date: 2001/11/14 11:52:00 EST |
| 582 |
|
Tag: blume-20011114-srcname |
| 583 |
|
Description: |
| 584 |
|
|
| 585 |
|
1. Routed the name of the current source file to mlriscgen where it |
| 586 |
|
should be directly emitted into the code object. (This last part |
| 587 |
|
is yet to be done.) |
| 588 |
|
|
| 589 |
|
2. Some cleanup of the pgraph code to make it match the proposal that |
| 590 |
|
I put out the other day. (The proposal notwithstanding, things are |
| 591 |
|
still in flux here.) |
| 592 |
|
|
| 593 |
|
---------------------------------------------------------------------- |
| 594 |
|
Name: Lal George |
| 595 |
|
Date: 2001/11/14 09:44:04 EST |
| 596 |
|
Tag: |
| 597 |
|
Description: |
| 598 |
|
|
| 599 |
|
Fix for a backpatching bug reported by Allen. |
| 600 |
|
|
| 601 |
|
Because the boundary between short and long span-dependent |
| 602 |
|
instructions is +/- 128, there are an astounding number of |
| 603 |
|
span-dependent instructions whose size is over estimated. |
| 604 |
|
|
| 605 |
|
Allen came up with the idea of letting the size of span |
| 606 |
|
dependent instructions be non-monotonic, for a maxIter |
| 607 |
|
number of times, after which the size must be monotonically |
| 608 |
|
increasing. |
| 609 |
|
|
| 610 |
|
This table shows the number of span-dependent instructions |
| 611 |
|
whose size was over-estimated as a function of maxIter, for the |
| 612 |
|
file Parse/parse/ml.grm.sml: |
| 613 |
|
|
| 614 |
|
maxIter # of instructions: |
| 615 |
|
10 687 |
| 616 |
|
20 438 |
| 617 |
|
30 198 |
| 618 |
|
40 0 |
| 619 |
|
|
| 620 |
|
In compiling the compiler, there is no significant difference in |
| 621 |
|
compilation speed between maxIter=10 and maxIter=40. Actually, |
| 622 |
|
my measurements showed that maxIter=40 was a tad faster than |
| 623 |
|
maxIter=10! Also 96% of the files in the compiler reach a fix |
| 624 |
|
point within 13 iterations, so fixing maxIter at 40, while high, |
| 625 |
|
is okay. |
| 626 |
|
|
| 627 |
|
---------------------------------------------------------------------- |
| 628 |
|
Name: Matthias Blume |
| 629 |
|
Date: 2001/10/31 15:25:00 EST |
| 630 |
|
Tag: blume-20011031-pgraph |
| 631 |
|
Description: |
| 632 |
|
|
| 633 |
|
CKIT: |
| 634 |
|
* Changed the "Function" constructor of type Ast.ctype to carry optional |
| 635 |
|
argument identifiers. |
| 636 |
|
* Changed the return type of TypeUtil.getFunction accordingly. |
| 637 |
|
* Type equality ignores the argument names. |
| 638 |
|
* TypeUtil.composite tries to preserve argument names but gives up quickly |
| 639 |
|
if there is a mismatch. |
| 640 |
|
|
| 641 |
|
installation script: |
| 642 |
|
* attempts to use "curl" if available (unless "wget" is available as well) |
| 643 |
|
|
| 644 |
|
CM: |
| 645 |
|
* has an experimental implementation of "portable graphs" which I will |
| 646 |
|
soon propose as an implementation-independent library format |
| 647 |
|
* there are also new libraries $/pgraph.cm and $/pgraph-util.cm |
| 648 |
|
|
| 649 |
|
NLFFI-LIB: |
| 650 |
|
* some cleanup (all cosmetic) |
| 651 |
|
|
| 652 |
|
NLFFIGEN: |
| 653 |
|
* temporarily disabled the mechanism that suppresses ML output for |
| 654 |
|
C definitions whose identifiers start with an underscore character |
| 655 |
|
* generate val bindings for enum constants |
| 656 |
|
* user can request that only one style (light or heavy) is being used; |
| 657 |
|
default is to use both (command-line arguments: -heavy and -light) |
| 658 |
|
* fixed bug in handling of function types involving incomplete pointers |
| 659 |
|
* generate ML entry points that take record arguments (i.e., using |
| 660 |
|
named arguments) for C functions that have a prototype with named |
| 661 |
|
arguments |
| 662 |
|
(see changes to CKIT) |
| 663 |
|
|
| 664 |
|
---------------------------------------------------------------------- |
| 665 |
|
Name: Allen Leung |
| 666 |
|
Date: 2001/10/27 20:34:00 EDT |
| 667 |
|
Tag: leunga-20011027-x86-fast-fp-call |
| 668 |
|
Description: |
| 669 |
|
|
| 670 |
|
Fixed the bug described in blume-20010920-slowfp. |
| 671 |
|
|
| 672 |
|
The fix involves |
| 673 |
|
1. generating FCOPYs in FSTP in ia32-svid |
| 674 |
|
2. marking a CALL with the appropriate annotation |
| 675 |
|
|
| 676 |
|
---------------------------------------------------------------------- |
| 677 |
|
Name: Matthias Blume |
| 678 |
|
Date: 2001/10/16 11:32:00 EDT |
| 679 |
|
Tag: blume-20011016-netbsd |
| 680 |
|
Description: |
| 681 |
|
|
| 682 |
|
Underscore patch from Chris Richards (fixing problem with compiling |
| 683 |
|
runtime system under recent NetBSD). |
| 684 |
|
|
| 685 |
|
---------------------------------------------------------------------- |
| 686 |
|
Name: Allen Leung |
| 687 |
|
Date: 2001/10/12 17:18:32 EDT 2001 |
| 688 |
|
Tag: leung-20011012-x86-printflowgraph |
| 689 |
|
Description: |
| 690 |
|
|
| 691 |
|
X86RA now uses a valid (instead of dummy) PrintFlowgraph module. |
| 692 |
|
|
| 693 |
|
---------------------------------------------------------------------- |
| 694 |
|
Name: Lal George |
| 695 |
|
Date: 2001/10/11 23:51:34 EDT |
| 696 |
|
Tag: george-20011011-too-many-instrs |
| 697 |
|
Description: |
| 698 |
|
|
| 699 |
|
The representation of a program point never expected to see more |
| 700 |
|
than 65536 instructions in a basic block! |
| 701 |
|
|
| 702 |
|
---------------------------------------------------------------------- |
| 703 |
|
Name: Lal George |
| 704 |
|
Date: 2001/10/09 09:41:37 EDT |
| 705 |
|
Tag: george-20011008-mlrisc-labels |
| 706 |
|
Description: |
| 707 |
|
|
| 708 |
|
Changed the machine description files to support printing of |
| 709 |
|
local and global labels in assembly code, based on host assembler |
| 710 |
|
conventions. |
| 711 |
|
|
| 712 |
|
---------------------------------------------------------------------- |
| 713 |
|
Name: Matthias Blume |
| 714 |
|
Date: 2001/09/25 15:25:00 EDT |
| 715 |
|
Tag: blume-20010925-exninfo |
| 716 |
|
Description: |
| 717 |
|
|
| 718 |
|
I provided a non-hook implementation of exnName (at the toplevel) and |
| 719 |
|
made the "dummy" implementation of exnMessage (at the toplevel) more |
| 720 |
|
useful: if nothing gets "hooked in", then at least you are going to |
| 721 |
|
see the exception name and a message indicating why you don't see more. |
| 722 |
|
|
| 723 |
|
[For the time being, programs that need exnMessage and want to use |
| 724 |
|
ml-build should either use General.exnMessage (strongly recommended) or |
| 725 |
|
refer to structure General at some other point so that CM sees a |
| 726 |
|
static dependency.] |
| 727 |
|
|
| 728 |
|
[Similar remarks go for "print" and "use": If you want to use their |
| 729 |
|
functionality in stand-alone programs generated by ml-build, then use |
| 730 |
|
TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).] |
| 731 |
|
|
| 732 |
|
---------------------------------------------------------------------- |
| 733 |
|
Name: Matthias Blume |
| 734 |
|
Date: 2001/09/20 17:28:00 EDT |
| 735 |
|
Tag: blume-20010920-slowfp |
| 736 |
|
Description: |
| 737 |
|
|
| 738 |
|
Allen says that x86-fast-fp is not safe yet, so I turned it off again... |
| 739 |
|
|
| 740 |
|
---------------------------------------------------------------------- |
| 741 |
|
Name: Matthias Blume |
| 742 |
|
Date: 2001/09/20 17:20:00 EDT |
| 743 |
|
Tag: blume-20010920-canonicalpaths |
| 744 |
|
Description: |
| 745 |
|
|
| 746 |
|
0. Updated the BOOT file (something that I forgot to do earlier). |
| 747 |
|
|
| 748 |
|
1. Small internal change to CM so that it avoids "/../" in filenames |
| 749 |
|
as much as possible (but only where it is safe). |
| 750 |
|
|
| 751 |
|
2. Changed config/_run-sml (resulting in a changed bin/.run-sml) so |
| 752 |
|
that arguments that contain delimiters are passed through correctly. |
| 753 |
|
This change also means that all "special" arguments of the form |
| 754 |
|
@SMLxxx... must come first. |
| 755 |
|
|
| 756 |
|
3. Changed install script to put relative anchor names for tool commands |
| 757 |
|
into pathconfig. |
| 758 |
|
|
| 759 |
|
---------------------------------------------------------------------- |
| 760 |
|
Name: Matthias Blume |
| 761 |
|
Date: 2001/09/18 15:35:00 EDT |
| 762 |
|
Tag: blume-20010918-readme11036 |
| 763 |
|
Description: |
| 764 |
|
|
| 765 |
|
Added README files. |
| 766 |
|
|
| 767 |
|
---------------------------------------------------------------------- |
| 768 |
|
Name: Matthias Blume |
| 769 |
|
Date: 2001/09/18 11:45:00 EDT |
| 770 |
|
Tag: Release_110_36 (retag) |
| 771 |
|
Description: |
| 772 |
|
|
| 773 |
|
Fixed mistake in config/preloads. Retagged as 110.36. |
| 774 |
|
|
| 775 |
|
---------------------------------------------------------------------- |
| 776 |
|
Name: Matthias Blume |
| 777 |
|
Date: 2001/09/18 09:40:00 EDT |
| 778 |
|
Tag: Release_110_36_orig (tag changed) |
| 779 |
|
Description: |
| 780 |
|
|
| 781 |
|
New version (110.36). New bootfiles. |
| 782 |
|
|
| 783 |
|
---------------------------------------------------------------------- |
| 784 |
|
Name: Matthias Blume |
| 785 |
|
Date: 2001/09/14 16:15:00 EDT |
| 786 |
|
Tag: blume-20010914-x86fastfp |
| 787 |
|
Description: |
| 788 |
|
|
| 789 |
|
John committed some changes that Allen made, in particular a (hopefully) |
| 790 |
|
correctly working version of the x86-fp module. |
| 791 |
|
|
| 792 |
|
I changed the default setting of the Control.MLRISC.getFlag "x86-fast-fp" |
| 793 |
|
flag to "true". Everything seems to compile to a fixpoint ok, and |
| 794 |
|
"mandelbrot" speeds up by about 15%. |
| 795 |
|
|
| 796 |
|
---------------------------------------------------------------------- |
| 797 |
|
Name: Matthias Blume |
| 798 |
|
Date: 2001/09/13 11:20:00 EDT |
| 799 |
|
Tag: blume-20010913-minimal |
| 800 |
|
Description: |
| 801 |
|
|
| 802 |
|
1. Stefan Monnier's patch to fix a miscompilation problem that |
| 803 |
|
was brought to light by John Reppy's work on Moby. |
| 804 |
|
|
| 805 |
|
2. Implemented a minimal "structure Compiler" that contains just |
| 806 |
|
"version" and "architecture". The minimal version will be |
| 807 |
|
available when the full version is not. This is for backward- |
| 808 |
|
compatibility with code that wants to test Compiler.version. |
| 809 |
|
|
| 810 |
|
---------------------------------------------------------------------- |
| 811 |
|
Name: Matthias Blume |
| 812 |
|
Date: 2001/08/28 14:03:00 EDT |
| 813 |
|
Tag: blume-20010828-ml-lex |
| 814 |
|
Description: |
| 815 |
|
|
| 816 |
|
Fix for bug 1581, received from Neophytos Michael. |
| 817 |
|
|
| 818 |
|
---------------------------------------------------------------------- |
| 819 |
|
Name: Matthias Blume |
| 820 |
|
Date: 2001/08/27 11:20:00 EDT |
| 821 |
|
Tag: blume-20010827-readme11035 |
| 822 |
|
Description: |
| 823 |
|
|
| 824 |
|
Fleshed out the README file for 110.35. |
| 825 |
|
|
| 826 |
|
---------------------------------------------------------------------- |
| 827 |
|
Name: Matthias Blume |
| 828 |
|
Date: 2001/08/24 17:10:00 EDT |
| 829 |
|
Tag: Release_110_35 |
| 830 |
|
Description: |
| 831 |
|
|
| 832 |
|
New version number (110.35). New bootfiles. |
| 833 |
|
|
| 834 |
|
---------------------------------------------------------------------- |
| 835 |
|
Name: Lal George |
| 836 |
|
Date: 2001/08/24 13:47:18 EDT 2001 |
| 837 |
|
Tag: george-20010824-MLRISC-graphs |
| 838 |
|
Description: |
| 839 |
|
|
| 840 |
|
removed clusters from MLRISC completely and replaced with graphs. |
| 841 |
|
|
| 842 |
|
---------------------------------------------------------------------- |
| 843 |
|
Name: Matthias Blume |
| 844 |
|
Date: 2001/08/23 17:50:00 EDT |
| 845 |
|
Tag: blume-20010823-toplevel |
| 846 |
|
Description: |
| 847 |
|
|
| 848 |
|
- some reorganization of the code that implements various kinds of |
| 849 |
|
environments in the compiler (static, dynamic, symbolic, combined) |
| 850 |
|
- re-implemented the EnvRef module so that evalStream works properly |
| 851 |
|
(if the stream contains references to "use", "CM.make", etc.) |
| 852 |
|
- cleaned up evalloop.sml and interact.sml (but they need more cleaning) |
| 853 |
|
|
| 854 |
|
---------------------------------------------------------------------- |
| 855 |
|
Name: Matthias Blume |
| 856 |
|
Date: 2001/08/20 15:50 EDT |
| 857 |
|
Tag: blume20010820-slipup |
| 858 |
|
Description: |
| 859 |
|
|
| 860 |
|
I forgot to commit a few files. Here they are... |
| 861 |
|
|
| 862 |
|
---------------------------------------------------------------------- |
| 863 |
|
Name: Matthias Blume |
| 864 |
|
Date: 2001/08/20 15:35:00 EDT |
| 865 |
|
Tag: blume-20010820-debugprof |
| 866 |
|
Description: |
| 867 |
|
|
| 868 |
|
!!!! NEW BOOTFILES !!!! |
| 869 |
|
|
| 870 |
|
This is another round of reorganizing the compiler sources. This |
| 871 |
|
time the main goal was to factor out all the "instrumentation" |
| 872 |
|
passes (for profiling and backtracing) into their own library. |
| 873 |
|
The difficulty was to do it in such a way that it does not depend |
| 874 |
|
on elaborate.cm but only on elabdata.cm. |
| 875 |
|
|
| 876 |
|
Therefore there have been further changes to both elaborate.cm and |
| 877 |
|
elabdata.cm -- more "generic" things have been moved from the former |
| 878 |
|
to the latter. As a result, I was forced to split the assignment |
| 879 |
|
of numbers indicating "primtyc"s into two portions: SML-generic and |
| 880 |
|
SML/NJ-specific. Since it would have been awkward to maintain, |
| 881 |
|
I bit the bullet and actually _changed_ the mapping between these |
| 882 |
|
numbers and primtycs. The bottom line of this is that you need |
| 883 |
|
a new set of bin- and bootfiles. |
| 884 |
|
|
| 885 |
|
I have built new bootfiles for all architectures, so doing a fresh |
| 886 |
|
checkout and config/install.sh should be all you need. |
| 887 |
|
|
| 888 |
|
The newly created library's name is |
| 889 |
|
|
| 890 |
|
$smlnj/viscomp/debugprof.cm |
| 891 |
|
|
| 892 |
|
and its sources live under |
| 893 |
|
|
| 894 |
|
src/compiler/DebugProf |
| 895 |
|
|
| 896 |
|
---------------------------------------------------------------------- |
| 897 |
|
Name: Matthias Blume |
| 898 |
|
Date: 2001/08/15 17:15:00 EDT |
| 899 |
|
Tag: blume-20010815-compreorg |
| 900 |
|
Description: |
| 901 |
|
|
| 902 |
|
This is a first cut at reorganizing the CM libraries that make up the |
| 903 |
|
core of the compiler. The idea is to separate out pieces that could |
| 904 |
|
be used independently by tools, e.g., the parser, the typechecker, etc. |
| 905 |
|
|
| 906 |
|
The current status is a step in this direction, but it is not quite |
| 907 |
|
satisfactory yet. Expect more changes in the future. |
| 908 |
|
|
| 909 |
|
Here is the current (new) organization... |
| 910 |
|
|
| 911 |
|
What used to be $smlnj/viscomp/core.cm is now divided into |
| 912 |
|
six CM libraries: |
| 913 |
|
|
| 914 |
|
$smlnj/viscomp/basics.cm |
| 915 |
|
/parser.cm |
| 916 |
|
/elabdata.cm |
| 917 |
|
/elaborate.cm |
| 918 |
|
/execute.cm |
| 919 |
|
/core.cm |
| 920 |
|
|
| 921 |
|
The CM files for these libraries live under src/system/smlnj/viscomp. |
| 922 |
|
All these libraries are proxy libraries that contain precisely |
| 923 |
|
one CM library component. Here are the locations of the components |
| 924 |
|
(all within the src/compiler tree): |
| 925 |
|
|
| 926 |
|
Basics/basics.cm |
| 927 |
|
Parse/parser.cm |
| 928 |
|
ElabData/elabdata.cm |
| 929 |
|
Elaborator/elaborate.cm |
| 930 |
|
Execution/execute.cm |
| 931 |
|
core.cm |
| 932 |
|
|
| 933 |
|
[This organization is the same that has been used already |
| 934 |
|
for a while for the architecture-specific parts of the visible |
| 935 |
|
compiler and for the old version of core.cm.] |
| 936 |
|
|
| 937 |
|
As you will notice, many source files have been moved from their |
| 938 |
|
respective original locations to a new home in one of the above |
| 939 |
|
subtrees. |
| 940 |
|
|
| 941 |
|
The division of labor between the new libraries is the following: |
| 942 |
|
|
| 943 |
|
basics.cm: |
| 944 |
|
- Simple, basic definitions that pertain to many (or all) of |
| 945 |
|
the other libraries. |
| 946 |
|
parser.cm: |
| 947 |
|
- The SML parser, producing output of type Ast.dec. |
| 948 |
|
- The type family for Ast is also defined and exported here. |
| 949 |
|
elabdata.cm: |
| 950 |
|
- The datatypes that describe input and output of the elaborator. |
| 951 |
|
This includes types, absyn, and static environments. |
| 952 |
|
elaborator.cm: |
| 953 |
|
- The SML/NJ type checker and elaborator. |
| 954 |
|
This maps an Ast.dec (with a given static environment) to |
| 955 |
|
an Absyn.dec (with a new static environment). |
| 956 |
|
- This libraries implements certain modules that used to be |
| 957 |
|
structures as functors (to remove dependencies on FLINT). |
| 958 |
|
execute.cm: |
| 959 |
|
- Everything having to do with executing binary code objects. |
| 960 |
|
- Dynamic environments. |
| 961 |
|
core.cm: |
| 962 |
|
- SML/NJ-specific instantiations of the elaborator and MLRISC. |
| 963 |
|
- Top-level modules. |
| 964 |
|
- FLINT (this should eventually become its own library) |
| 965 |
|
|
| 966 |
|
Notes: |
| 967 |
|
|
| 968 |
|
I am not 100% happy with the way I separated the elaborator (and its |
| 969 |
|
data structures) from FLINT. Two instances of the same problem: |
| 970 |
|
|
| 971 |
|
1. Data structures contain certain fields that carry FLINT-specific |
| 972 |
|
information. I hacked around this using exn and the property list |
| 973 |
|
module from smlnj-lib. But the fact that there are middle-end |
| 974 |
|
specific fields around at all is a bit annoying. |
| 975 |
|
|
| 976 |
|
2. The elaborator calculates certain FLINT-related information. I tried |
| 977 |
|
to make this as abstract as I could using functorization, but, again, |
| 978 |
|
the fact that the elaborator has to perform calculations on behalf |
| 979 |
|
of the middle-end at all is not nice. |
| 980 |
|
|
| 981 |
|
3. Having to used exn and property lists is unfortunate because it |
| 982 |
|
weakens type checking. The other alternative (parameterizing |
| 983 |
|
nearly *everything*) is not appealing, though. |
| 984 |
|
|
| 985 |
|
I removed the "rebinding =" warning hack because due to the new organization |
| 986 |
|
it was awkward to maintain it. As a result, the compiler now issues some of |
| 987 |
|
these warnings when compiling init.cmi during bootstrap compilation. On |
| 988 |
|
the plus side, you also get a warning when you do, for example: |
| 989 |
|
val op = = Int32.+ |
| 990 |
|
which was not the case up to now. |
| 991 |
|
|
| 992 |
|
I placed "assign" and "deref" into the _Core structure so that the |
| 993 |
|
code that deals with the "lazy" keyword can find them there. This |
| 994 |
|
removes the need for having access to the primitive environment |
| 995 |
|
during elaboration. |
| 996 |
|
|
| 997 |
|
---------------------------------------------------------------------- |
| 998 |
|
Name: Matthias Blume |
| 999 |
|
Date: 2001/08/13 |
| 1000 |
|
Tag: blume-20010813-closures |
| 1001 |
|
Description: |
| 1002 |
|
|
| 1003 |
|
This fix was sent to us by Zhong Shao. It is supposed to improve the |
| 1004 |
|
performance of certain loops by avoiding needless closure allocation. |
| 1005 |
|
|
| 1006 |
|
---------------------------------------------------------------------- |
| 1007 |
|
Name: Lal George |
| 1008 |
|
Date: 2001/07/31 10:03:23 EDT 2001 |
| 1009 |
|
Tag: george-20010731-x86-fmalloc |
| 1010 |
|
Description: Fixed bug in x86 calls |
| 1011 |
|
|
| 1012 |
|
There was a bug where call instructions would mysteriously |
| 1013 |
|
vanish. The call instruction had to be one that returned |
| 1014 |
|
a floating point value. |
| 1015 |
|
|
| 1016 |
|
---------------------------------------------------------------------- |
| 1017 |
|
Name: Lal George |
| 1018 |
|
Date: 2001/07/19 16:36:29 EDT 2001 |
| 1019 |
|
Tag: george-20010719-simple-cells |
| 1020 |
|
Description: |
| 1021 |
|
|
| 1022 |
|
I have dramatically simplified the interface for CELLS in MLRISC. |
| 1023 |
|
|
| 1024 |
|
In summary, the cells interface is broken up into three parts: |
| 1025 |
|
|
| 1026 |
|
1. CellsBasis : CELLS_BASIS |
| 1027 |
|
|
| 1028 |
|
CellsBasis is a top level structure and common for all |
| 1029 |
|
architectures. it contains the definitions of basic datatypes |
| 1030 |
|
and utility functions over these types. |
| 1031 |
|
|
| 1032 |
|
2. functor Cells() : CELLS |
| 1033 |
|
|
| 1034 |
|
Cells generates an interface for CELLS that incorporates the |
| 1035 |
|
specific resources on the target architecture, such as the |
| 1036 |
|
presence of special register classes, their number and size, |
| 1037 |
|
and various useful substructures. |
| 1038 |
|
|
| 1039 |
|
3. <ARCH>CELLS |
| 1040 |
|
|
| 1041 |
|
e.g. SparcCells: SPARCCELLS |
| 1042 |
|
|
| 1043 |
|
<ARCH>CELLS usually contains additional bindings for special |
| 1044 |
|
registers on the architecture, such as: |
| 1045 |
|
|
| 1046 |
|
val r0 : cell (* register zero *) |
| 1047 |
|
val y : cell (* Y register *) |
| 1048 |
|
val psr : cell (* processor status register *) |
| 1049 |
|
... |
| 1050 |
|
|
| 1051 |
|
The structure returned by applying the Cells functor is opened |
| 1052 |
|
in this interface. |
| 1053 |
|
|
| 1054 |
|
The main implication of all this is that the datatypes for cells is |
| 1055 |
|
split between CellsBasis and CELLS -- a fairly simple change for user |
| 1056 |
|
code. |
| 1057 |
|
|
| 1058 |
|
In the old scheme the CELLS interface had a definitional binding of |
| 1059 |
|
the form: |
| 1060 |
|
|
| 1061 |
|
signature CELLS = sig |
| 1062 |
|
|
| 1063 |
|
structure CellsBasis = CellsBasis |
| 1064 |
|
|
| 1065 |
|
... |
| 1066 |
|
|
| 1067 |
|
end |
| 1068 |
|
|
| 1069 |
|
With all the sharing constraints that goes on in MLRISC, this old |
| 1070 |
|
design quickly leads to errors such as: |
| 1071 |
|
|
| 1072 |
|
"structure definition spec inside of sharing ... " |
| 1073 |
|
|
| 1074 |
|
|
| 1075 |
|
and appears to require an unacceptable amount of sharing and where |
| 1076 |
|
constraint hackery. |
| 1077 |
|
|
| 1078 |
|
I think this error message (the interaction of definitional specs and |
| 1079 |
|
sharing) requires more explanation on our web page. |
| 1080 |
|
|
| 1081 |
|
---------------------------------------------------------------------- |
| 1082 |
|
Name: Matthias Blume |
| 1083 |
|
Date: 2001/07/19 15:00:00 EDT |
| 1084 |
|
Tag: blume-20010719-libreorg |
| 1085 |
|
Description: |
| 1086 |
|
|
| 1087 |
|
This update puts together a fairly extensive but straightforward change |
| 1088 |
|
to the way the libraries that implement the interactive system are |
| 1089 |
|
organized: |
| 1090 |
|
|
| 1091 |
|
The biggest change is the elimination of structure Compiler. As a |
| 1092 |
|
replacement for this structure, there is now a CM library |
| 1093 |
|
(known as $smlnj/compiler.cm or $smlnj/compiler/current.cm) |
| 1094 |
|
that exports all the substructures of the original structure Compiler |
| 1095 |
|
directly. So instead of saying Compiler.Foo.bar one now simply |
| 1096 |
|
says Foo.bar. (The CM libraries actually export a collection of |
| 1097 |
|
structures that is richer than the collection of substructures of |
| 1098 |
|
structure Compiler.) |
| 1099 |
|
|
| 1100 |
|
To make the transition smooth, there is a separate library called |
| 1101 |
|
$smlnj/compiler/compiler.cm which puts together and exports the |
| 1102 |
|
original structure Compiler (or at least something very close to it). |
| 1103 |
|
|
| 1104 |
|
There are five members of the original structure Compiler |
| 1105 |
|
that are not exported directly but which instead became members |
| 1106 |
|
of a new structure Backend (described by signature BACKEND). These are: |
| 1107 |
|
structure Profile (: PROFILE), structure Compile (: COMPILE), structure |
| 1108 |
|
Interact (: INTERACT), structure Machine (: MACHINE), and val |
| 1109 |
|
architecture (: string). |
| 1110 |
|
|
| 1111 |
|
Structure Compiler.Version has become structure CompilerVersion. |
| 1112 |
|
|
| 1113 |
|
Cross-compilers for alpha32, hppa, ppc, sparc, and x86 are provided |
| 1114 |
|
by $smlnj/compiler/<arch>.cm where <arch> is alpha32, hppa, ppc, sparc, |
| 1115 |
|
or x86, respectively. |
| 1116 |
|
Each of these exports the same frontend structures that |
| 1117 |
|
$smlnj/compiler.cm exports. But they do not have a structure Backend |
| 1118 |
|
and instead export some structure <Arch>Backend where <Arch> is Alpha32, |
| 1119 |
|
Hppa, PPC, Sparc, or X86, respectively. |
| 1120 |
|
|
| 1121 |
|
Library $smlnj/compiler/all.cm exports the union of the exports of |
| 1122 |
|
$smlnj/compiler/<arch>.cm |
| 1123 |
|
|
| 1124 |
|
There are no structures <Arch>Compiler anymore, use |
| 1125 |
|
$smlnj/compiler/<arch>.cm instead. |
| 1126 |
|
|
| 1127 |
|
Library host-compiler-0.cm is gone. Instead, the internal library |
| 1128 |
|
that instantiates CM is now called cm0.cm. Selection of the host |
| 1129 |
|
compiler (backend) is no longer done here but. (Responsibility for it |
| 1130 |
|
now lies with $smlnj/compiler/current.cm. This seems to be more |
| 1131 |
|
logical.) |
| 1132 |
|
|
| 1133 |
|
Many individual files have been moved or renamed. Some files have |
| 1134 |
|
been split into multiple files, and some "dead" files have been deleted. |
| 1135 |
|
|
| 1136 |
|
Aside from these changes to library organization, there are also changes |
| 1137 |
|
to the way the code itself is organized: |
| 1138 |
|
|
| 1139 |
|
Structure Binfile has been re-implemented in such a way that it no |
| 1140 |
|
longer needs any knowledge of the compiler. It exclusively deals |
| 1141 |
|
with the details of binfile layout. It no longer invokes the |
| 1142 |
|
compiler (for the purpose of creating new prospective binfile |
| 1143 |
|
content), and it no longer has any knowledge of how to interpret |
| 1144 |
|
pickles. |
| 1145 |
|
|
| 1146 |
|
Structure Compile (: COMPILE) has been stripped down to the bare |
| 1147 |
|
essentials of compilation. It no longer deals with linking/execution. |
| 1148 |
|
The interface has been cleaned up considerably. |
| 1149 |
|
|
| 1150 |
|
Utility routines for dealing with linking and execution have been |
| 1151 |
|
moved into their own substructures. |
| 1152 |
|
|
| 1153 |
|
(The ultimate goal of these changes is to provide a light-weight |
| 1154 |
|
binfile loader/linker (at least for, e.g., stable libraries) that |
| 1155 |
|
does not require CM or the compiler to be present.) |
| 1156 |
|
|
| 1157 |
|
CM documentation has been updated to reflect the changes to library |
| 1158 |
|
organization. |
| 1159 |
|
|
| 1160 |
|
---------------------------------------------------------------------- |
| 1161 |
|
Name: Matthias Blume |
| 1162 |
|
Date: 2001/07/10 17:30:00 EDT |
| 1163 |
|
Tag: Release_110_34 |
| 1164 |
|
Description: |
| 1165 |
|
|
| 1166 |
|
Minor tweak to 110.34 (re-tagged): |
| 1167 |
|
|
| 1168 |
|
- README.html file added to CVS repository |
| 1169 |
|
- runtime compiles properly under FreeBSD 3.X and 4.X |
| 1170 |
|
|
| 1171 |
|
---------------------------------------------------------------------- |
| 1172 |
|
Name: Matthias Blume |
| 1173 |
|
Date: 2001/07/10 17:30:00 EDT |
| 1174 |
|
Tag: Release_110_34 |
| 1175 |
|
Description: |
| 1176 |
|
|
| 1177 |
|
New version number (110.34). New bootfiles. |
| 1178 |
|
|
| 1179 |
|
---------------------------------------------------------------------- |
| 1180 |
|
Name: Matthias Blume |
| 1181 |
|
Date: 2001/07/09 16:00:00 EDT |
| 1182 |
|
Tag: blume-20010709-more-varargs |
| 1183 |
|
Description: |
| 1184 |
|
|
| 1185 |
|
I changed the handling of varargs in ml-nlffigen again: |
| 1186 |
|
The ellipsis ... will now simply be ignored (with an accompanying warning). |
| 1187 |
|
|
| 1188 |
|
The immediate effect is that you can actually call a varargs function |
| 1189 |
|
from ML -- but you can't actually supply any arguments beyond the ones |
| 1190 |
|
specified explicitly. (For example, you can call printf with its format |
| 1191 |
|
string, but you cannot pass additional arguments.) |
| 1192 |
|
|
| 1193 |
|
This behavior is only marginally more useful than the one before, but |
| 1194 |
|
it has the advantage that a function or, more importantly, a function |
| 1195 |
|
type never gets dropped on the floor, thus avoiding follow-up problems with |
| 1196 |
|
other types that refer to the offending one. |
| 1197 |
|
|
| 1198 |
|
---------------------------------------------------------------------- |
| 1199 |
|
Name: Matthias Blume |
| 1200 |
Date: 2001/07/09 11:25:00 EDT |
Date: 2001/07/09 11:25:00 EDT |
| 1201 |
Tag: blume-20010709-varargs |
Tag: blume-20010709-varargs |
| 1202 |
Description: |
Description: |
| 3734 |
elaborator). There were a lot of changes during my "linkpath" trials |
elaborator). There were a lot of changes during my "linkpath" trials |
| 3735 |
that could have been reverted to their original state but weren't. |
that could have been reverted to their original state but weren't. |
| 3736 |
Please, don't be too harsh on me for messing with this code a bit more |
Please, don't be too harsh on me for messing with this code a bit more |
| 3737 |
than what was strictly necessary... (I _did_ resist the tempation |
than what was strictly necessary... (I _did_ resist the temptation |
| 3738 |
of doing any "global reformatting" to avoid an untimely death at |
of doing any "global reformatting" to avoid an untimely death at |
| 3739 |
Dave's hands. :) |
Dave's hands. :) |
| 3740 |
|
|
Parent Directory
| 
Revision Log
| 
Patch
