SCM Repository
Annotation of /sml/trunk/HISTORY
Parent Directory
| 
Revision Log
Revision 1238 - (view) (download)
| 1 : | dbm | 570 | This is the HISTORY file for the Yale SML/NJ CVS repository. |
| 2 : | |||
| 3 : | An entry should be made for _every_ commit to the repository. | ||
| 4 : | The entries in this file will be used when creating the README | ||
| 5 : | for new versions, so keep that in mind when writing the | ||
| 6 : | description. | ||
| 7 : | |||
| 8 : | The form of an entry should be: | ||
| 9 : | |||
| 10 : | Name: | ||
| 11 : | george | 1003 | Date: yyyy/mm/dd |
| 12 : | dbm | 570 | Tag: <post-commit CVS tag> |
| 13 : | Description: | ||
| 14 : | leunga | 1142 | |
| 15 : | george | 1136 | ---------------------------------------------------------------------- |
| 16 : | blume | 1229 | Name: Matthias Blume |
| 17 : | blume | 1238 | Date: 2002/06/10 16:35:00 EDT |
| 18 : | Tag: blume-20020610-ieeereal | ||
| 19 : | Description: | ||
| 20 : | |||
| 21 : | I back-ported my implementation of IEEEReal.fromString from the | ||
| 22 : | idlbasis-devel branch so that we can test it. | ||
| 23 : | |||
| 24 : | Another small change is that ppDec tries to give more information | ||
| 25 : | than just "<sig>" in the case of functors. However, this code is | ||
| 26 : | broken in some mysterious way if the functor's body's signature | ||
| 27 : | has not been declared by ascription but gets inferred from the | ||
| 28 : | implementation. This needs fixing... | ||
| 29 : | |||
| 30 : | ---------------------------------------------------------------------- | ||
| 31 : | Name: Matthias Blume | ||
| 32 : | blume | 1229 | Date: 2002/05/31 |
| 33 : | Tag: blume-20020531-btrace-mode | ||
| 34 : | Description: | ||
| 35 : | |||
| 36 : | Resurrected SMLofNJ.Internals.BTrace.mode. (It accidentally fell by | ||
| 37 : | the wayside when I switched over to using Controls everywhere.) | ||
| 38 : | |||
| 39 : | ---------------------------------------------------------------------- | ||
| 40 : | george | 1222 | Name: Lal George |
| 41 : | Date: 2002/05/23 12:21:40 EDT | ||
| 42 : | Tag: george-20020523-visual-labels | ||
| 43 : | Description: | ||
| 44 : | |||
| 45 : | Labels are now displayed in the graphical output to make | ||
| 46 : | the fall-through and target blocks obvious. | ||
| 47 : | |||
| 48 : | ---------------------------------------------------------------------- | ||
| 49 : | blume | 1204 | Name: Matthias Blume |
| 50 : | blume | 1220 | Date: 2002/05/22 11:03:00 EDT |
| 51 : | Tag: blume-20020522-shrink | ||
| 52 : | Description: | ||
| 53 : | |||
| 54 : | John tweaked yesterday's fix for 1131 to handle an out-of-memory | ||
| 55 : | situation that comes up when allocating huge arrays. | ||
| 56 : | |||
| 57 : | ---------------------------------------------------------------------- | ||
| 58 : | Name: Matthias Blume | ||
| 59 : | blume | 1215 | Date: 2002/05/21 16:00:00 EDT |
| 60 : | Tag: Release_110_40 | ||
| 61 : | Description: | ||
| 62 : | |||
| 63 : | New working release (110.40). New bootfiles. | ||
| 64 : | |||
| 65 : | [Also: John Reppy fixed GC bug 1131.] | ||
| 66 : | |||
| 67 : | ---------------------------------------------------------------------- | ||
| 68 : | Name: Matthias Blume | ||
| 69 : | blume | 1212 | Date: 2002/05/21 12:35:00 EDT |
| 70 : | Tag: blume-20020521-cmdoc | ||
| 71 : | Description: | ||
| 72 : | |||
| 73 : | CM documentation update. | ||
| 74 : | |||
| 75 : | ---------------------------------------------------------------------- | ||
| 76 : | Name: Matthias Blume | ||
| 77 : | blume | 1211 | Date: 2002/05/21 10:55:00 EDT |
| 78 : | Tag: blume-20020521-misc | ||
| 79 : | Description: | ||
| 80 : | |||
| 81 : | - John tweaked runtime to be silent on heap export (except when | ||
| 82 : | GC messages are on). | ||
| 83 : | - I added a few more things (cross-compiling versions of CMB) to | ||
| 84 : | config/preloads (as suggestions). | ||
| 85 : | |||
| 86 : | ---------------------------------------------------------------------- | ||
| 87 : | Name: Matthias Blume | ||
| 88 : | blume | 1208 | Date: 2002/05/20 22:25:00 EDT |
| 89 : | Tag: blume-20020520-controls | ||
| 90 : | Description: | ||
| 91 : | |||
| 92 : | - Added ControlUtil structure to control-lib.cm. | ||
| 93 : | - Use it throughout. | ||
| 94 : | - Used Controls facility to define MLRISC controls (as opposed to | ||
| 95 : | registering MLRISC control ref cells with Controls after the fact) | ||
| 96 : | - Fixed messed-up controls priorities. | ||
| 97 : | |||
| 98 : | * Removed again all the stuff from config/preloads that one wouldn't | ||
| 99 : | be able to preload at the time the initial heap image is built. | ||
| 100 : | (Many libraries, e.g., CML, do not exist yet at this time. The | ||
| 101 : | only libraries that can be preloaded via config/preloads are those | ||
| 102 : | that come bundled with the bootfiles.) | ||
| 103 : | |||
| 104 : | ---------------------------------------------------------------------- | ||
| 105 : | Name: Matthias Blume | ||
| 106 : | blume | 1204 | Date: 2002/05/20 10:59:00 EDT |
| 107 : | Tag: blume-20020520-preloads | ||
| 108 : | Description: | ||
| 109 : | |||
| 110 : | Added a lot of commented-out suggestions for things to be included | ||
| 111 : | in config/preloads. | ||
| 112 : | |||
| 113 : | ---------------------------------------------------------------------- | ||
| 114 : | leunga | 1203 | Name: Allen Leung |
| 115 : | Date: 2002/05/18 14:20:00 EDT | ||
| 116 : | Tag: leunga-20020518-mdl | ||
| 117 : | Description: | ||
| 118 : | |||
| 119 : | o Made the mdl tool stuff compile and run again. | ||
| 120 : | o I've disabled all the stuff that depends on RTL specifications; they | ||
| 121 : | are all badly broken anyway. | ||
| 122 : | |||
| 123 : | ---------------------------------------------------------------------- | ||
| 124 : | blume | 1201 | Name: Matthias Blume |
| 125 : | Date: 2002/05/17 16:49:00 EDT | ||
| 126 : | Tag: blume-20020517-controls | ||
| 127 : | Description: | ||
| 128 : | |||
| 129 : | 0. John Reppy made several modifications to the SML/NJ library. | ||
| 130 : | In particular, there is a shiny new controls-lib.cm. | ||
| 131 : | |||
| 132 : | 1. Pushed new controls interface through compiler so that everything | ||
| 133 : | compiles again. | ||
| 134 : | |||
| 135 : | 2. Added FormatComb and FORMAT_COMB to the CML version of the | ||
| 136 : | SML/NJ library (so that CML compiles again). | ||
| 137 : | |||
| 138 : | 3. Modified init scripts because XXX_DEFAULT environment variables | ||
| 139 : | are no longer with us. (Boot-time initialization is now done | ||
| 140 : | using the same environment variables that are also used for | ||
| 141 : | startup-time initialization of controls.) | ||
| 142 : | |||
| 143 : | ---------------------------------------------------------------------- | ||
| 144 : | george | 1192 | Name: Lal George |
| 145 : | Date: 2002/05/15 09:20:10 EDT | ||
| 146 : | Tag: george-20020515-pseudo-op-decls | ||
| 147 : | Description: | ||
| 148 : | |||
| 149 : | All pseudo-ops emitted before the first segment declaration | ||
| 150 : | such as TEXT, DATA, and BSS directives are assumed to be global | ||
| 151 : | declarations and are emitted first in the assembly file. This is | ||
| 152 : | useful in a number of situations where one has pseudo-ops that are not | ||
| 153 : | specific to any segment, and also works around the constraint that one | ||
| 154 : | cannot have client pseudo-ops in the TEXT segment. | ||
| 155 : | |||
| 156 : | Because no segment is associated with these declarations it is | ||
| 157 : | an error to allocate any space or objects before the first segment | ||
| 158 : | directive and an exception will be raised. However, we cannot make | ||
| 159 : | this check for client pseudo-ops. | ||
| 160 : | |||
| 161 : | These top level declarations are a field in the CFG graph_info. | ||
| 162 : | In theory you can continue to add to this field after the CFG has been | ||
| 163 : | built -- provided you know what you are doing;-) | ||
| 164 : | |||
| 165 : | ---------------------------------------------------------------------- | ||
| 166 : | blume | 1175 | Name: Matthias Blume |
| 167 : | blume | 1190 | Date: 2002/05/13 16:40:00 EDT |
| 168 : | Tag: blume-20020513-pp-etc | ||
| 169 : | Description: | ||
| 170 : | |||
| 171 : | A few minor bugfixes: | ||
| 172 : | |||
| 173 : | - Stopgap measure for bug recently reported by Elsa Gunter (ppDec). | ||
| 174 : | (Bogus printouts for redefined bindings still occur. Compiler | ||
| 175 : | bug should no longer occur now. We need to redo the prettyprinter | ||
| 176 : | from scratch.) | ||
| 177 : | |||
| 178 : | - CM pathname printer now also adds escape sequences for ( and ) | ||
| 179 : | |||
| 180 : | - commend and docu fixes for ml-nlffi | ||
| 181 : | |||
| 182 : | ---------------------------------------------------------------------- | ||
| 183 : | Name: Matthias Blume | ||
| 184 : | blume | 1189 | Date: 2002/05/10 16:40:00 EDT |
| 185 : | Tag: blume-20020510-erg-textio | ||
| 186 : | Description: | ||
| 187 : | |||
| 188 : | Applied the following bugfix provided by Emden Gansner: | ||
| 189 : | |||
| 190 : | Output is corrupted when outputSubstr is used rather than output. | ||
| 191 : | |||
| 192 : | The problem occurs when a substring | ||
| 193 : | |||
| 194 : | ss = (s, dataStart, dataLen) | ||
| 195 : | |||
| 196 : | where dataStart > 0, fills a stream buffer with avail bytes left. | ||
| 197 : | avail bytes of s, starting at index dataStart, are copied into the | ||
| 198 : | buffer, the buffer is flushed, and then the remaining dataLen-avail | ||
| 199 : | bytes of ss are copied into the beginning of the buffer. Instead of | ||
| 200 : | starting this copy at index dataStart+avail in s, the current code | ||
| 201 : | starts the copy at index avail. | ||
| 202 : | |||
| 203 : | Fix: | ||
| 204 : | In text-io-fn.sml, change line 695 from | ||
| 205 : | val needsFlush = copyVec(v, avail, dataLen-avail, buf, 0) | ||
| 206 : | to | ||
| 207 : | val needsFlush = copyVec(v, dataStart+avail, dataLen-avail, buf, 0) | ||
| 208 : | |||
| 209 : | ---------------------------------------------------------------------- | ||
| 210 : | Name: Matthias Blume | ||
| 211 : | blume | 1186 | Date: 2002/04/12 13:55:00 EDT |
| 212 : | Tag: blume-20020412-assyntax | ||
| 213 : | Description: | ||
| 214 : | |||
| 215 : | 1. Grabbed newer assyntax.h from the XFree86 project. | ||
| 216 : | 2. Fiddled with how to compile X86.prim.asm without warnings. | ||
| 217 : | 3. (Very) Minor cleanup in CM. | ||
| 218 : | |||
| 219 : | ---------------------------------------------------------------------- | ||
| 220 : | Name: Matthias Blume | ||
| 221 : | blume | 1185 | Date: 2002/04/01 (no joke!) 17:07:00 EST |
| 222 : | Tag: blume-20020401-x86div | ||
| 223 : | Description: | ||
| 224 : | |||
| 225 : | Added full support for div/mod/rem/quot on the x86, using the machine | ||
| 226 : | instruction's two results (without clumsily recomputing the remainder) | ||
| 227 : | directly where appropriate. | ||
| 228 : | |||
| 229 : | Some more extensive power-of-two support was added to the x86 instruction | ||
| 230 : | selector (avoiding expensive divs, mods, and muls where they can be | ||
| 231 : | replaced with cheaper shifts and masks). However, this sort of thing | ||
| 232 : | ought to be done earlier, e.g., within the CPS optimizer so that | ||
| 233 : | all architectures benefit from it. | ||
| 234 : | |||
| 235 : | The compiler compiles to a fixed point, but changes might be somewhat | ||
| 236 : | fragile nevertheless. Please, report any strange things that you might | ||
| 237 : | see wrt. div/mod/quot/rem... | ||
| 238 : | |||
| 239 : | ---------------------------------------------------------------------- | ||
| 240 : | Name: Matthias Blume | ||
| 241 : | blume | 1184 | Date: 2002/03/29 17:22:00 |
| 242 : | Tag: blume-20020329-div | ||
| 243 : | Description: | ||
| 244 : | |||
| 245 : | Fixed my broken div/mod logic. Unfortunately, this means that the | ||
| 246 : | inline code for div/mod now has one more comparison than before. | ||
| 247 : | Fast paths (quotient > 0 or remainder = 0) are not affected, though. | ||
| 248 : | The problem was with quotient = 0, because that alone does not tell | ||
| 249 : | us which way the rounding went. One then has to look at whether | ||
| 250 : | remainder and divisor have the same sign... :( | ||
| 251 : | |||
| 252 : | Anyway, I replaced the bootfiles with fresh ones... | ||
| 253 : | |||
| 254 : | ---------------------------------------------------------------------- | ||
| 255 : | Name: Matthias Blume | ||
| 256 : | blume | 1183 | Date: 2002/03/29 14:10:00 EST |
| 257 : | Tag: blume-20020329-inlprims | ||
| 258 : | Description: | ||
| 259 : | |||
| 260 : | NEW BOOTFILES!!! Version number bumped to 110.39.3. | ||
| 261 : | |||
| 262 : | Primops have changed. This means that the bin/boot-file formats have | ||
| 263 : | changed as well. | ||
| 264 : | |||
| 265 : | To make sure that there is no confusion, I made a new version. | ||
| 266 : | |||
| 267 : | |||
| 268 : | CHANGES: | ||
| 269 : | |||
| 270 : | * removed REMT from mltree (remainder should never overflow). | ||
| 271 : | |||
| 272 : | * added primops to deal with divisions of all flavors to the frontend | ||
| 273 : | |||
| 274 : | * handled these primops all the way through so they map to their respective | ||
| 275 : | MLRISC support | ||
| 276 : | |||
| 277 : | * used these primops in the implementation of Int, Int32, Word, Word32 | ||
| 278 : | |||
| 279 : | * removed INLDIV, INLMOD, and INLREM as they are no longer necessary | ||
| 280 : | |||
| 281 : | * parameterized INLMIN, INLMAX, and INLABS by a numkind | ||
| 282 : | |||
| 283 : | * translate.sml now deals with all flavors of INL{MIN,MAX,ABS}, including | ||
| 284 : | floating point | ||
| 285 : | |||
| 286 : | * used INL{MIN,MAX,ABS} in the implementation of Int, Int32, Word, Word32, | ||
| 287 : | and Real (but Real.abs maps to a separate floating-point-only primop) | ||
| 288 : | |||
| 289 : | |||
| 290 : | TODO items: | ||
| 291 : | |||
| 292 : | * Hacked Alpha32 instruction selection, disabling the selection of REMx | ||
| 293 : | instructions because the machine instruction encoder cannot handle | ||
| 294 : | them. (Hppa, PPC, and Sparc instruction selection did not handle | ||
| 295 : | REM in the first place, and REM is supported by the x86 machine coder.) | ||
| 296 : | |||
| 297 : | * Handle DIV and MOD with DIV_TO_NEGINF directly in the x86 instruction | ||
| 298 : | selection phase. (The two can be streamlined because the hardware | ||
| 299 : | delivers both quotient and remainder at the same time anyway.) | ||
| 300 : | |||
| 301 : | * Think about what to do with "valOf(Int32.minInt) div ~1" and friends. | ||
| 302 : | (Currently the behavior is inconsistent both across architectures and | ||
| 303 : | wrt. the draft Basis spec.) | ||
| 304 : | |||
| 305 : | * Word8 should eventually be handled natively, too. | ||
| 306 : | |||
| 307 : | * There seems to be one serious bug in mltree-gen.sml. It appears, though, | ||
| 308 : | as if there currently is no execution path that could trigger it in | ||
| 309 : | SML/NJ. (The assumptions underlying functions arith and promotable do not | ||
| 310 : | hold for things like multiplication and division.) | ||
| 311 : | |||
| 312 : | ---------------------------------------------------------------------- | ||
| 313 : | Name: Matthias Blume | ||
| 314 : | blume | 1181 | Date: 2002/03/27 16:27:00 EST |
| 315 : | Tag: blume-20020327-mlrisc-divisions | ||
| 316 : | Description: | ||
| 317 : | |||
| 318 : | Added support for all four division operations (ML's div, mod, quot, | ||
| 319 : | and rem) to MLRISC. In the course of doing so, I also rationalized | ||
| 320 : | the naming (no more annoying switch-around of DIV and QUOT), by | ||
| 321 : | parameterizing the operation by div_rounding_mode (which can be either | ||
| 322 : | DIV_TO_ZERO or DIV_TO_NEGINF). | ||
| 323 : | |||
| 324 : | The generic MLTreeGen functor takes care of compiling all four | ||
| 325 : | operations down to only round-to-zero div. | ||
| 326 : | |||
| 327 : | Missing pieces: | ||
| 328 : | |||
| 329 : | * Doing something smarter than relying on MLTreeGen on architectures | ||
| 330 : | like, e.g., the x86 where hardware division delivers both quotient and | ||
| 331 : | remainder at the same time. With this, the implementation of the | ||
| 332 : | round-to-neginf operations could be further streamlined. | ||
| 333 : | |||
| 334 : | * Remove inlining support for div/mod/rem from the frontend and replace it | ||
| 335 : | with primops that get carried through to the backend. Do this for all | ||
| 336 : | int and word types. | ||
| 337 : | |||
| 338 : | ---------------------------------------------------------------------- | ||
| 339 : | Name: Matthias Blume | ||
| 340 : | blume | 1180 | Date: 2002/03/25 17:25:00 EST |
| 341 : | Tag: blume-20020325-divmod | ||
| 342 : | Description: | ||
| 343 : | |||
| 344 : | I improved (hopefully without breaking them) the implementation of Int.div, | ||
| 345 : | Int.mod, and Int.rem. For this, the code in translate.sml now takes | ||
| 346 : | advantage of the following observations: | ||
| 347 : | |||
| 348 : | Let q = x quot y r = x rem y | ||
| 349 : | d = x div y m = x mod y | ||
| 350 : | |||
| 351 : | where "quot" is the round-to-zero version of integer division that | ||
| 352 : | hardware usually provides. Then we have: | ||
| 353 : | |||
| 354 : | r = x - q * y where neither the * nor the - will overflow | ||
| 355 : | d = if q >= 0 orelse x = q * y then q else q - 1 | ||
| 356 : | where neither the * nor the - will overflow | ||
| 357 : | m = if q >= 0 orelse r = 0 then r else r + y | ||
| 358 : | where the + will not overflow | ||
| 359 : | |||
| 360 : | This results in substantial simplification of the generated code. | ||
| 361 : | The following table shows the number of CFG nodes and edges generated | ||
| 362 : | for | ||
| 363 : | fun f (x, y) = x OPER y | ||
| 364 : | (* with OPER \in div, mod, quot, rem *) | ||
| 365 : | |||
| 366 : | |||
| 367 : | OPER | nodes(old) | edges(old) | nodes(new) | edges(new) | ||
| 368 : | -------------------------------------------------------- | ||
| 369 : | div | 24 | 39 | 12 | 16 | ||
| 370 : | mod | 41 | 71 | 12 | 16 | ||
| 371 : | quot | 8 | 10 | 8 | 10 | ||
| 372 : | rem | 10 | 14 | 8 | 10 | ||
| 373 : | |||
| 374 : | |||
| 375 : | ---------------------------------------------------------------------- | ||
| 376 : | Name: Matthias Blume | ||
| 377 : | blume | 1179 | Date: 2002/03/25 22:06:00 EST |
| 378 : | Tag: blume-20020325-cprotobug | ||
| 379 : | Description: | ||
| 380 : | |||
| 381 : | Fixed a bug in cproto (c prototype decoder). | ||
| 382 : | |||
| 383 : | ---------------------------------------------------------------------- | ||
| 384 : | Name: Matthias Blume | ||
| 385 : | blume | 1178 | Date: 2002/03/25 16:00:00 EST |
| 386 : | Tag: blume-20020325-raw-primops | ||
| 387 : | Description: | ||
| 388 : | |||
| 389 : | I did some cleanup to Allen's new primop code and | ||
| 390 : | replaced yesterday's bootfiles with new ones. | ||
| 391 : | (But they are stored in the same place.) | ||
| 392 : | |||
| 393 : | ---------------------------------------------------------------------- | ||
| 394 : | Name: Matthias Blume | ||
| 395 : | blume | 1175 | Date: 2002/03/24 22:40:00 EST |
| 396 : | Tag: blume-20020324-bootfiles | ||
| 397 : | Description: | ||
| 398 : | |||
| 399 : | Made the bootfiles that Allen asked for. | ||
| 400 : | |||
| 401 : | ---------------------------------------------------------------------- | ||
| 402 : | leunga | 1174 | Name: Allen Leung |
| 403 : | Date: 2002/03/23 15:50:00 EST | ||
| 404 : | Tag: leunga-20020323-flint-cps-rcc-primops | ||
| 405 : | Description: | ||
| 406 : | |||
| 407 : | 1. Changes to FLINT primops: | ||
| 408 : | |||
| 409 : | (* make a call to a C-function; | ||
| 410 : | * The primop carries C function prototype information and specifies | ||
| 411 : | * which of its (ML-) arguments are floating point. C prototype | ||
| 412 : | * information is for use by the backend, ML information is for | ||
| 413 : | * use by the CPS converter. *) | ||
| 414 : | | RAW_CCALL of { c_proto: CTypes.c_proto, | ||
| 415 : | ml_args: ccall_type list, | ||
| 416 : | ml_res_opt: ccall_type option, | ||
| 417 : | reentrant : bool | ||
| 418 : | } option | ||
| 419 : | (* Allocate uninitialized storage on the heap. | ||
| 420 : | * The record is meant to hold short-lived C objects, i.e., they | ||
| 421 : | * are not ML pointers. With the tag, the representation is | ||
| 422 : | * the same as RECORD with tag tag_raw32 (sz=4), or tag_fblock (sz=8) | ||
| 423 : | *) | ||
| 424 : | | RAW_RECORD of {tag:bool,sz:int} | ||
| 425 : | and ccall_type = CCALL_INT32 | CCALL_REAL64 | CCALL_ML_PTR | ||
| 426 : | |||
| 427 : | 2. These CPS primops are now overloaded: | ||
| 428 : | |||
| 429 : | rawload of {kind:numkind} | ||
| 430 : | rawstore of {kind:numkind} | ||
| 431 : | |||
| 432 : | The one argument form is: | ||
| 433 : | |||
| 434 : | rawload {kind} address | ||
| 435 : | |||
| 436 : | The two argument form is: | ||
| 437 : | |||
| 438 : | rawload {kind} [ml object, byte-offset] | ||
| 439 : | |||
| 440 : | 3. RAW_CCALL/RCC now takes two extra arguments: | ||
| 441 : | |||
| 442 : | a. The first is whether the C call is reentrant, i.e., whether | ||
| 443 : | ML state should be saved and restored. | ||
| 444 : | b. The second argument is a string argument specifying the name of | ||
| 445 : | library and the C function. | ||
| 446 : | |||
| 447 : | These things are currently not handled in the code generator, yet. | ||
| 448 : | |||
| 449 : | 4. In CProto, | ||
| 450 : | |||
| 451 : | An encoding type of "bool" means "ml object" and is mapped into | ||
| 452 : | C prototype of PTR. Note that "bool" is different than "string", | ||
| 453 : | even though "string" is also mapped into PTR, because "bool" | ||
| 454 : | is assigned an CPS type of BOGt, while "string" is assigned INT32t. | ||
| 455 : | |||
| 456 : | 5. Pickler/unpicker | ||
| 457 : | |||
| 458 : | Changed to handle RAW_RECORD and newest RAW_CCALL | ||
| 459 : | |||
| 460 : | 6. MLRiscGen, | ||
| 461 : | |||
| 462 : | 1. Changed to handle the new rawload/rawstore/rawrecord operators. | ||
| 463 : | 2. Code for handling C Calls has been moved to a new module CPSCCalls, | ||
| 464 : | in the file CodeGen/cpscompile/cps-c-calls.sml | ||
| 465 : | |||
| 466 : | 7. Added the conditional move operator | ||
| 467 : | |||
| 468 : | condmove of branch | ||
| 469 : | |||
| 470 : | to cps. Generation of this is still buggy so it is currently | ||
| 471 : | disabled. | ||
| 472 : | |||
| 473 : | ---------------------------------------------------------------------- | ||
| 474 : | george | 1168 | Name: Lal George |
| 475 : | Date: 2002/03/22 14:18:25 EST | ||
| 476 : | george | 1171 | Tag: george-20020322-cps-branch-prob |
| 477 : | george | 1168 | Description: |
| 478 : | |||
| 479 : | george | 1171 | Implemented the Ball-Larus branch prediction-heuristics, and |
| 480 : | george | 1168 | incorporated graphical viewers for control flow graphs. |
| 481 : | |||
| 482 : | george | 1171 | Ball-Larus Heuristics: |
| 483 : | george | 1168 | --------------------- |
| 484 : | See the file compiler/CodeGen/cpscompile/cpsBranchProb.sml. | ||
| 485 : | |||
| 486 : | By design it uses the Dempster-Shafer theory for combining | ||
| 487 : | probabilities. For example, in the function: | ||
| 488 : | |||
| 489 : | fun f(n,acc) = if n = 0 then acc else f(n-1, n*acc) | ||
| 490 : | |||
| 491 : | the ball-larus heuristics predicts that the n=0 is unlikely | ||
| 492 : | (OH-heuristic), and the 'then' branch is unlikely because of the | ||
| 493 : | RH-heuristic -- giving the 'then' branch an even lower combined | ||
| 494 : | george | 1171 | probability using the Dempster-Shafer theory. |
| 495 : | george | 1168 | |
| 496 : | Finally, John Reppy's loop analysis in MLRISC, further lowers the | ||
| 497 : | probability of the 'then' branch because of the loop in the else | ||
| 498 : | branch. | ||
| 499 : | |||
| 500 : | |||
| 501 : | Graphical Viewing: | ||
| 502 : | ------------------ | ||
| 503 : | I merely plugged in Allen's graphical viewers into the compiler. The | ||
| 504 : | additional code is not much. At the top level, saying: | ||
| 505 : | |||
| 506 : | Control.MLRISC.getFlag "cfg-graphical-view" := true; | ||
| 507 : | |||
| 508 : | will display the graphical view of the control flow graph just before | ||
| 509 : | back-patching. daVinci must be in your path for this to work. If | ||
| 510 : | daVinci is not available, then the default viewer can be changed | ||
| 511 : | using: | ||
| 512 : | |||
| 513 : | Control.MLRISC.getString "viewer" | ||
| 514 : | |||
| 515 : | which can be set to "dot" or "vcg" for the corresponding viewers. Of | ||
| 516 : | course, these viewers must be in your path. | ||
| 517 : | |||
| 518 : | The above will display the compilation unit at the level of clusters, | ||
| 519 : | many of which are small, boring, and un-interesting. Also setting: | ||
| 520 : | |||
| 521 : | Control.MLRISC.getInt "cfg-graphical-view_size" | ||
| 522 : | |||
| 523 : | will display clusters that are larger than the value set by the above. | ||
| 524 : | |||
| 525 : | |||
| 526 : | ---------------------------------------------------------------------- | ||
| 527 : | blume | 1159 | Name: Matthias Blume |
| 528 : | Date: 2002/03/21 22:20:00 EST | ||
| 529 : | Tag: blume-20020321-kmp-bugfix | ||
| 530 : | Description: | ||
| 531 : | |||
| 532 : | Changed the interface to the KMP routine in PreString and fixed | ||
| 533 : | a minor bug in one place where it was used. | ||
| 534 : | |||
| 535 : | ---------------------------------------------------------------------- | ||
| 536 : | leunga | 1156 | Name: Allen Leung |
| 537 : | leunga | 1158 | Date: 2002/03/21 20:30:00 EST |
| 538 : | Tag: leunga-20020321-cfg | ||
| 539 : | Description: | ||
| 540 : | |||
| 541 : | Fixed a potential problem in cfg edge splitting. | ||
| 542 : | |||
| 543 : | ---------------------------------------------------------------------- | ||
| 544 : | Name: Allen Leung | ||
| 545 : | leunga | 1156 | Date: 2002/03/21 17:15:00 EST |
| 546 : | Tag: leunga-20020321-x86-fp-cfg | ||
| 547 : | Description: | ||
| 548 : | |||
| 549 : | 1. Recoded the buggy parts of x86-fp. | ||
| 550 : | |||
| 551 : | a. All the block reordering code has been removed. | ||
| 552 : | We now depend on the block placement phases to do this work. | ||
| 553 : | |||
| 554 : | b. Critical edge splitting code has been simplified and moved into the | ||
| 555 : | CFG modules, as where they belong. | ||
| 556 : | |||
| 557 : | Both of these were quite buggy and complex. The code is now much, much | ||
| 558 : | simpler. | ||
| 559 : | |||
| 560 : | 2. X86 backend. | ||
| 561 : | |||
| 562 : | a. Added instructions for 64-bit support. Instruction selection for | ||
| 563 : | 64-bit has not been committed, however, since that | ||
| 564 : | requires changes to MLTREE which haven't been approved by | ||
| 565 : | Lal and John. | ||
| 566 : | |||
| 567 : | b. Added support for FUCOMI and FUCOMIP when generating code for | ||
| 568 : | PentiumPro and above. We only generate these instructions in | ||
| 569 : | the fast-fp mode. | ||
| 570 : | |||
| 571 : | c. Added cases for JP and JNP in X86FreqProps. | ||
| 572 : | |||
| 573 : | 3. CFG | ||
| 574 : | |||
| 575 : | CFG now has a bunch of methods for edge splitting and merging. | ||
| 576 : | |||
| 577 : | 4. Machine description. | ||
| 578 : | |||
| 579 : | John's simplification of MLTREE_BASIS.fcond broke a few machine | ||
| 580 : | description things: | ||
| 581 : | |||
| 582 : | rtl-build.{sig,sml} and hppa.mdl fixed. | ||
| 583 : | |||
| 584 : | NOTE: the machine description stuff in the repository is still broken. | ||
| 585 : | Again, I can't put my fixes in because that involves | ||
| 586 : | changes to MLTREE. | ||
| 587 : | |||
| 588 : | ---------------------------------------------------------------------- | ||
| 589 : | blume | 1145 | Name: Matthias Blume |
| 590 : | blume | 1155 | Date: 2002/03/20 15:55:00 EST |
| 591 : | Tag: blume-20020320-kmp | ||
| 592 : | Description: | ||
| 593 : | |||
| 594 : | Implemented Knuth-Morris-Pratt string matching in PreString and used | ||
| 595 : | it for String.isSubstring, Substring.isSubstring, and | ||
| 596 : | Substring.position. | ||
| 597 : | |||
| 598 : | (Might need some stress-testing. Simple examples worked fine.) | ||
| 599 : | |||
| 600 : | ---------------------------------------------------------------------- | ||
| 601 : | Name: Matthias Blume | ||
| 602 : | blume | 1152 | Date: 2002/03/19 16:37:00 EST |
| 603 : | Tag: blume-20020319-witnesses | ||
| 604 : | Description: | ||
| 605 : | |||
| 606 : | Added a structure C.W and functions convert/Ptr.convert to ml-nlffi-lib. | ||
| 607 : | |||
| 608 : | This implements a generic mechanism for changing constness qualifiers | ||
| 609 : | anywhere within big C types without resorting to outright "casts". | ||
| 610 : | (So far, functions such as C.rw/C.ro or C.Ptr.rw/C.Ptr.ro only let you | ||
| 611 : | modify the constness at the outermost level.) | ||
| 612 : | The implementation of "convert" is based on the idea of "witness" | ||
| 613 : | values -- values that are not used by the operation but whose types | ||
| 614 : | "testify" to their applicability. On the implementation side, "convert" | ||
| 615 : | is simply a projection (returning its second curried argument). With | ||
| 616 : | cross-module inlining, it should not result in any machine code being | ||
| 617 : | generated. | ||
| 618 : | |||
| 619 : | ---------------------------------------------------------------------- | ||
| 620 : | Name: Matthias Blume | ||
| 621 : | blume | 1148 | Date: 2002/03/15 16:40:00 EST |
| 622 : | Tag: blume-20020315-basis | ||
| 623 : | Description: | ||
| 624 : | |||
| 625 : | Provided (preliminary?) implementations for | ||
| 626 : | |||
| 627 : | {String,Substring}.{concatWith,isSuffix,isSubstring} | ||
| 628 : | |||
| 629 : | and | ||
| 630 : | |||
| 631 : | Substring.full | ||
| 632 : | |||
| 633 : | Those are in the Basis spec but they were missing in SML/NJ. | ||
| 634 : | |||
| 635 : | ---------------------------------------------------------------------- | ||
| 636 : | Name: Matthias Blume | ||
| 637 : | blume | 1145 | Date: 2002/03/14 21:30:00 EST |
| 638 : | Tag: blume-20020314-controls | ||
| 639 : | Description: | ||
| 640 : | |||
| 641 : | Controls: | ||
| 642 : | --------- | ||
| 643 : | |||
| 644 : | 1. Factored out the recently-added Controls : CONTROLS stuff and put | ||
| 645 : | it into its own library $/controls-lib.cm. The source tree for | ||
| 646 : | this is under src/smlnj-lib/Controls. | ||
| 647 : | |||
| 648 : | 2. Changed the names of types and functions in this interface, so they | ||
| 649 : | make a bit more "sense": | ||
| 650 : | |||
| 651 : | module -> registry | ||
| 652 : | 'a registry -> 'a group | ||
| 653 : | |||
| 654 : | 3. The interface now deals in ref cells only. The getter/setter interface | ||
| 655 : | is (mostly) gone. | ||
| 656 : | |||
| 657 : | 4. Added a function that lets one register an already-existing ref cell. | ||
| 658 : | |||
| 659 : | 5. Made the corresponding modifications to the rest of the code so that | ||
| 660 : | everything compiles again. | ||
| 661 : | |||
| 662 : | 6. Changed the implementation of Controls.MLRISC back to something closer | ||
| 663 : | to the original. In particular, this module (and therefore MLRISC) | ||
| 664 : | does not depend on Controls. There now is some link-time code in | ||
| 665 : | int-sys.sml that registers the MLRISC controls with the Controls | ||
| 666 : | module. | ||
| 667 : | |||
| 668 : | CM: | ||
| 669 : | --- | ||
| 670 : | |||
| 671 : | * One can now specify the lambda-split aggressiveness in init.cmi. | ||
| 672 : | |||
| 673 : | ---------------------------------------------------------------------- | ||
| 674 : | leunga | 1142 | Name: Allen Leung |
| 675 : | Date: 2002/03/13 17:30:00 EST | ||
| 676 : | Tag: leunga-20020313-x86-fp-unary | ||
| 677 : | Description: | ||
| 678 : | |||
| 679 : | Bug fix for: | ||
| 680 : | |||
| 681 : | > leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml | ||
| 682 : | > Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002 | ||
| 683 : | > - fun f(x,(y,z)) = Real.~ y; | ||
| 684 : | > [autoloading] | ||
| 685 : | > [autoloading done] | ||
| 686 : | > fchsl (%eax), 184(%esp) | ||
| 687 : | > Error: MLRisc bug: X86MCEmitter.emitInstr | ||
| 688 : | > | ||
| 689 : | > uncaught exception Error | ||
| 690 : | > raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19 | ||
| 691 : | |||
| 692 : | The problem was that the code generator did not generate any fp registers | ||
| 693 : | in this case, and the ra didn't know that it needed to run the X86FP phase to | ||
| 694 : | translate the pseudo fp instruction. This only happened with unary fp | ||
| 695 : | operators in certain situations. | ||
| 696 : | |||
| 697 : | ---------------------------------------------------------------------- | ||
| 698 : | blume | 1137 | Name: Matthias Blume |
| 699 : | blume | 1140 | Date: 2002/03/13 14:00:00 EST |
| 700 : | Tag: blume-20020313-overload-etc | ||
| 701 : | Description: | ||
| 702 : | |||
| 703 : | 1. Added _overload as a synonym for overload for backward compatibility. | ||
| 704 : | (Control.overloadKW must be true for either version to be accepted.) | ||
| 705 : | |||
| 706 : | 2. Fixed bug in install script that caused more things to be installed | ||
| 707 : | than what was requested in config/targets. | ||
| 708 : | |||
| 709 : | 3. Made CM aware of the (_)overload construct so that autoloading | ||
| 710 : | works. | ||
| 711 : | |||
| 712 : | ---------------------------------------------------------------------- | ||
| 713 : | Name: Matthias Blume | ||
| 714 : | blume | 1138 | Date: 2002/03/12 22:03:00 EST |
| 715 : | Tag: blume-20020312-url | ||
| 716 : | Description: | ||
| 717 : | |||
| 718 : | Forgot to update BOOT and srcarchiveurl. | ||
| 719 : | |||
| 720 : | ---------------------------------------------------------------------- | ||
| 721 : | Name: Matthias Blume | ||
| 722 : | blume | 1137 | Date: 2002/03/12 17:30:00 EST |
| 723 : | Tag: blume-20020312-version110392 | ||
| 724 : | Description: | ||
| 725 : | |||
| 726 : | Yet another version number bump (because of small changes to the | ||
| 727 : | binfile format). Version number is now 110.39.2. NEW BOOTFILES! | ||
| 728 : | |||
| 729 : | Changes: | ||
| 730 : | |||
| 731 : | The new pid generation scheme described a few weeks ago was overly | ||
| 732 : | complicated. I implemented a new mechanism that is simpler and | ||
| 733 : | provides a bit more "stability": Once CM has seen a compilation | ||
| 734 : | unit, it keeps its identity constant (as long as you do not delete | ||
| 735 : | those crucial CM/GUID/* files). This means that when you change | ||
| 736 : | blume | 1145 | an interface, compile, then go back to the old interface, and |
| 737 : | blume | 1137 | compile again, you arrive at the original pid. |
| 738 : | |||
| 739 : | There now also is a mechanism that instructs CM to use the plain | ||
| 740 : | environment hash as a module's pid (effectively making its GUID | ||
| 741 : | the empty string). For this, "noguid" must be specified as an | ||
| 742 : | option to the .sml file in question within its .cm file. | ||
| 743 : | This is most useful for code that is being generated by tools such | ||
| 744 : | as ml-nlffigen (because during development programmers tend to | ||
| 745 : | erase the tool's entire output directory tree including CM's cached | ||
| 746 : | GUIDs). "noguid" is somewhat dangerous (since it can be used to locally | ||
| 747 : | revert to the old, broken behavior of SML/NJ, but in specific cases | ||
| 748 : | where there is no danger of interface confusion, its use is ok | ||
| 749 : | (I think). | ||
| 750 : | |||
| 751 : | ml-nlffigen by default generates "noguid" annotations. They can be | ||
| 752 : | turned off by specifying -guid in its command line. | ||
| 753 : | |||
| 754 : | ---------------------------------------------------------------------- | ||
| 755 : | george | 1136 | Name: Lal George |
| 756 : | Date: 2002/03/12 12 14:42:36 EST | ||
| 757 : | Tag: george-20020312-frequency-computation | ||
| 758 : | Description: | ||
| 759 : | |||
| 760 : | Integrated jump chaining and static block frequency into the | ||
| 761 : | compiler. More details and numbers later. | ||
| 762 : | leunga | 1025 | |
| 763 : | george | 1003 | ---------------------------------------------------------------------- |
| 764 : | george | 1133 | Name: Lal George |
| 765 : | Date: 2002/03/11 11 22:38:53 EST | ||
| 766 : | Tag: george-20020311-jump-chain-elim | ||
| 767 : | Description: | ||
| 768 : | |||
| 769 : | Tested the jump chain elimination on all architectures (except the | ||
| 770 : | hppa). This is on by default right now and is profitable for the | ||
| 771 : | alpha and x86, however, it may not be profitable for the sparc and ppc | ||
| 772 : | when compiling the compiler. | ||
| 773 : | |||
| 774 : | The gc test will typically jump to a label at the end of the cluster, | ||
| 775 : | where there is another jump to an external cluster containing the actual | ||
| 776 : | code to invoke gc. This is to allow factoring of common gc invocation | ||
| 777 : | sequences. That is to say, we generate: | ||
| 778 : | |||
| 779 : | f: | ||
| 780 : | testgc | ||
| 781 : | ja L1 % jump if above to L1 | ||
| 782 : | |||
| 783 : | L1: | ||
| 784 : | jmp L2 | ||
| 785 : | |||
| 786 : | |||
| 787 : | After jump chain elimination the 'ja L1' instructions is converted to | ||
| 788 : | 'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end | ||
| 789 : | up being implemented in their long form (if L2 is far away) using: | ||
| 790 : | |||
| 791 : | jbe L3 % jump if below or equal to L3 | ||
| 792 : | jmp L2 | ||
| 793 : | L3: | ||
| 794 : | ... | ||
| 795 : | |||
| 796 : | |||
| 797 : | For large compilation units L2 may be far away. | ||
| 798 : | |||
| 799 : | |||
| 800 : | ---------------------------------------------------------------------- | ||
| 801 : | blume | 1132 | Name: Matthias Blume |
| 802 : | Date: 2002/03/11 13:30:00 EST | ||
| 803 : | Tag: blume-20020311-mltreeeval | ||
| 804 : | Description: | ||
| 805 : | |||
| 806 : | A functor parameter was missing. | ||
| 807 : | |||
| 808 : | ---------------------------------------------------------------------- | ||
| 809 : | leunga | 1130 | Name: Allen Leung |
| 810 : | leunga | 1131 | Date: 2002/03/11 10:30:00 EST |
| 811 : | leunga | 1142 | Tag: leunga-20020311-runtime-string0 |
| 812 : | leunga | 1131 | Description: |
| 813 : | |||
| 814 : | The representation of the empty string now points to a | ||
| 815 : | legal null terminated C string instead of unit. It is now possible | ||
| 816 : | to convert an ML string into C string with InlineT.CharVector.getData. | ||
| 817 : | This compiles into one single machine instruction. | ||
| 818 : | |||
| 819 : | ---------------------------------------------------------------------- | ||
| 820 : | Name: Allen Leung | ||
| 821 : | leunga | 1130 | Date: 2002/03/10 23:55:00 EST |
| 822 : | Tag: leunga-20020310-x86-call | ||
| 823 : | Description: | ||
| 824 : | |||
| 825 : | Added machine generation for CALL instruction (relative displacement mode) | ||
| 826 : | |||
| 827 : | ---------------------------------------------------------------------- | ||
| 828 : | blume | 1128 | Name: Matthias Blume |
| 829 : | Date: 2002/03/08 16:05:00 | ||
| 830 : | Tag: blume-20020308-entrypoints | ||
| 831 : | Description: | ||
| 832 : | |||
| 833 : | Version number bumped to 110.39.1. NEW BOOTFILES! | ||
| 834 : | |||
| 835 : | Entrypoints: non-zero offset into a code object where execution should begin. | ||
| 836 : | |||
| 837 : | - Added the notion of an entrypoint to CodeObj. | ||
| 838 : | - Added reading/writing of entrypoint info to Binfile. | ||
| 839 : | - Made runtime system bootloader aware of entrypoints. | ||
| 840 : | - Use the address of the label of the first function given to mlriscGen | ||
| 841 : | as the entrypoint. This address is currently always 0, but it will | ||
| 842 : | not be 0 once we turn on block placement. | ||
| 843 : | - Removed the linkage cluster code (which was The Other Way(tm) of dealing | ||
| 844 : | with entry points) from mlriscGen. | ||
| 845 : | |||
| 846 : | ---------------------------------------------------------------------- | ||
| 847 : | leunga | 1127 | Name: Allen Leung |
| 848 : | Date: 2002/03/07 20:45:00 EST | ||
| 849 : | Tag: leunga-20020307-x86-cmov | ||
| 850 : | Description: | ||
| 851 : | |||
| 852 : | Bug fixes for CMOVcc on x86. | ||
| 853 : | |||
| 854 : | 1. Added machine code generation for CMOVcc | ||
| 855 : | 2. CMOVcc is now generated in preference over SETcc on PentiumPro or above. | ||
| 856 : | 3. CMOVcc cannot have an immediate operand as argument. | ||
| 857 : | |||
| 858 : | ---------------------------------------------------------------------- | ||
| 859 : | blume | 1126 | Name: Matthias Blume |
| 860 : | Date: 2002/03/07 16:15:00 EST | ||
| 861 : | Tag: blume-20020307-controls | ||
| 862 : | Description: | ||
| 863 : | |||
| 864 : | This is a very large but mostly boring patch which makes (almost) | ||
| 865 : | every tuneable compiler knob (i.e., pretty much everything under | ||
| 866 : | Control.* plus a few other things) configurable via both the command | ||
| 867 : | line and environment variables in the style CM did its configuration | ||
| 868 : | until now. | ||
| 869 : | |||
| 870 : | Try starting sml with '-h' (or, if you are brave, '-H') | ||
| 871 : | |||
| 872 : | To this end, I added a structure Controls : CONTROLS to smlnj-lib.cm which | ||
| 873 : | implements the underlying generic mechanism. | ||
| 874 : | |||
| 875 : | The interface to some of the existing such facilities has changed somewhat. | ||
| 876 : | For example, the MLRiscControl module now provides mkFoo instead of getFoo. | ||
| 877 : | (The getFoo interface is still there for backward-compatibility, but its | ||
| 878 : | use is deprecated.) | ||
| 879 : | |||
| 880 : | The ml-build script passes -Cxxx=yyy command-line arguments through so | ||
| 881 : | that one can now twiddle the compiler settings when using this "batch" | ||
| 882 : | compiler. | ||
| 883 : | |||
| 884 : | TODO items: | ||
| 885 : | |||
| 886 : | We should go through and throw out all controls that are no longer | ||
| 887 : | connected to anything. Moreover, we should go through and provide | ||
| 888 : | meaningful (and correct!) documentation strings for those controls | ||
| 889 : | that still are connected. | ||
| 890 : | |||
| 891 : | Currently, multiple calls to Controls.new are accepted (only the first | ||
| 892 : | has any effect). Eventually we should make sure that every control | ||
| 893 : | is being made (via Controls.new) exactly once. Future access can then | ||
| 894 : | be done using Controls.acc. | ||
| 895 : | |||
| 896 : | Finally, it would probably be a good idea to use the getter-setter | ||
| 897 : | interface to controls rather than ref cells. For the time being, both | ||
| 898 : | styles are provided by the Controls module, but getter-setter pairs are | ||
| 899 : | better if thread-safety is of any concern because they can be wrapped. | ||
| 900 : | |||
| 901 : | ***************************************** | ||
| 902 : | |||
| 903 : | One bug fix: The function blockPlacement in three of the MLRISC | ||
| 904 : | backpatch files used to be hard-wired to one of two possibilities at | ||
| 905 : | link time (according to the value of the placementFlag). But (I | ||
| 906 : | think) it should rather sense the flag every time. | ||
| 907 : | |||
| 908 : | ***************************************** | ||
| 909 : | |||
| 910 : | Other assorted changes (by other people who did not supply a HISTORY entry): | ||
| 911 : | |||
| 912 : | 1. the cross-module inliner now works much better (Monnier) | ||
| 913 : | 2. representation of weights, frequencies, and probabilities in MLRISC | ||
| 914 : | changed in preparation of using those for weighted block placement | ||
| 915 : | (Reppy, George) | ||
| 916 : | |||
| 917 : | ---------------------------------------------------------------------- | ||
| 918 : | george | 1116 | Name: Lal George |
| 919 : | george | 1124 | Date: 2002/03/07 14:44:24 EST 2002 |
| 920 : | Tag: george-20020307-weighted-block-placement | ||
| 921 : | |||
| 922 : | Tested the weighted block placement optimization on all architectures | ||
| 923 : | (except the hppa) using AMPL to generate the block and edge frequencies. | ||
| 924 : | Changes were required in the machine properties to correctly | ||
| 925 : | categorize trap instructions. There is an MLRISC flag | ||
| 926 : | "weighted-block-placement" that can be used to enable weighted block | ||
| 927 : | placement, but this will be ineffective without block/edge | ||
| 928 : | frequencies (coming soon). | ||
| 929 : | |||
| 930 : | |||
| 931 : | ---------------------------------------------------------------------- | ||
| 932 : | Name: Lal George | ||
| 933 : | george | 1116 | Date: 2002/03/05 17:24:48 EST |
| 934 : | Tag: george-20020305-linkage-cluster | ||
| 935 : | |||
| 936 : | george | 1124 | In order to support the block placement optimization, a new cluster |
| 937 : | is generated as the very first cluster (called the linkage cluster). | ||
| 938 : | It contains a single jump to the 'real' entry point for the compilation | ||
| 939 : | unit. Block placement has no effect on the linkage cluster itself, but | ||
| 940 : | all the other clusters have full freedom in the manner in which they | ||
| 941 : | reorder blocks or functions. | ||
| 942 : | george | 1116 | |
| 943 : | On the x86 the typical linkage code that is generated is: | ||
| 944 : | ---------------------- | ||
| 945 : | .align 2 | ||
| 946 : | L0: | ||
| 947 : | addl $L1-L0, 72(%esp) | ||
| 948 : | george | 1124 | jmp L1 |
| 949 : | george | 1116 | |
| 950 : | |||
| 951 : | .align 2 | ||
| 952 : | L1: | ||
| 953 : | ---------------------- | ||
| 954 : | |||
| 955 : | 72(%esp) is the memory location for the stdlink register. This | ||
| 956 : | must contain the address of the CPS function being called. In the | ||
| 957 : | george | 1124 | above example, it contains the address of L0; before |
| 958 : | george | 1116 | calling L1 (the real entry point for the compilation unit), it |
| 959 : | must contain the address for L1, and hence | ||
| 960 : | |||
| 961 : | addl $L1-L0, 72(%esp) | ||
| 962 : | |||
| 963 : | george | 1124 | I have tested this on all architectures except the hppa.The increase |
| 964 : | in code size is of course negligible | ||
| 965 : | george | 1116 | |
| 966 : | ---------------------------------------------------------------------- | ||
| 967 : | leunga | 1115 | Name: Allen Leung |
| 968 : | Date: 2002/03/03 13:20:00 EST | ||
| 969 : | Tag: leunga-20020303-mlrisc-tools | ||
| 970 : | |||
| 971 : | Added #[ ... ] expressions to mlrisc tools | ||
| 972 : | |||
| 973 : | ---------------------------------------------------------------------- | ||
| 974 : | blume | 1096 | Name: Matthias Blume |
| 975 : | blume | 1098 | Date: 2002/02/27 12:29:00 EST |
| 976 : | Tag: blume-20020227-cdebug | ||
| 977 : | Description: | ||
| 978 : | |||
| 979 : | - made types in structure C and C_Debug to be equal | ||
| 980 : | - got rid of code duplication (c-int.sml vs. c-int-debug.sml) | ||
| 981 : | - there no longer is a C_Int_Debug (C_Debug is directly derived from C) | ||
| 982 : | |||
| 983 : | ---------------------------------------------------------------------- | ||
| 984 : | Name: Matthias Blume | ||
| 985 : | blume | 1096 | Date: 2002/02/26 12:00:00 EST |
| 986 : | Tag: blume-20020226-ffi | ||
| 987 : | Description: | ||
| 988 : | |||
| 989 : | 1. Fixed a minor bug in CM's "noweb" tool: | ||
| 990 : | If numbering is turned off, then truly don't number (i.e., do not | ||
| 991 : | supply the -L option to noweb). The previous behavior was to supply | ||
| 992 : | -L'' -- which caused noweb to use the "default" line numbering scheme. | ||
| 993 : | Thanks to Chris Richards for pointing this out (and supplying the fix). | ||
| 994 : | |||
| 995 : | 2. Once again, I reworked some aspects of the FFI: | ||
| 996 : | |||
| 997 : | A. The incomplete/complete type business: | ||
| 998 : | |||
| 999 : | - Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are | ||
| 1000 : | gone! | ||
| 1001 : | - ML types representing an incomplete type are now *equal* to | ||
| 1002 : | ML types representing their corresponding complete types (just like | ||
| 1003 : | in C). This is still safe because ml-nlffigen will not generate | ||
| 1004 : | RTTI for incomplete types, nor will it generate functions that | ||
| 1005 : | require access to such RTTI. But when ML code generated from both | ||
| 1006 : | incomplete and complete versions of the C type meet, the ML types | ||
| 1007 : | are trivially interoperable. | ||
| 1008 : | |||
| 1009 : | NOTE: These changes restore the full generality of the translation | ||
| 1010 : | (which was previously lost when I eliminated functorization)! | ||
| 1011 : | |||
| 1012 : | B. Enum types: | ||
| 1013 : | |||
| 1014 : | - Structure C now has a type constructor "enum" that is similar to | ||
| 1015 : | how the "su" constructor works. However, "enum" is not a phantom | ||
| 1016 : | type because each "T enum" has values (and is isomorphic to | ||
| 1017 : | MLRep.Signed.int). | ||
| 1018 : | - There are generic access operations for enum objects (using | ||
| 1019 : | MLRep.Signed.int). | ||
| 1020 : | - ml-nlffigen will generate a structure E_foo for each "enum foo". | ||
| 1021 : | * The structure contains the definition of type "mlrep" (the ML-side | ||
| 1022 : | representation type of the enum). Normally, mlrep is the same | ||
| 1023 : | as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec", | ||
| 1024 : | then mlrep will be defined as a datatype -- thus facilitating | ||
| 1025 : | pattern matching on mlrep values. | ||
| 1026 : | ("-ec" will be suppressed if there are duplicate values in an | ||
| 1027 : | enumeration.) | ||
| 1028 : | * Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep | ||
| 1029 : | will be generated for each C enum constant xxx. | ||
| 1030 : | * Conversion functions m2i and i2m convert between mlrep and | ||
| 1031 : | MLRep.Signed.int. (Without "-ec", these functions are identities.) | ||
| 1032 : | * Coversion functions c and ml convert between mlrep and "tag enum". | ||
| 1033 : | * Access functions (get/set) fetch and store mlrep values. | ||
| 1034 : | - By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed | ||
| 1035 : | enumerations are merged into one single enumeration represented by | ||
| 1036 : | structure E_'. | ||
| 1037 : | |||
| 1038 : | ---------------------------------------------------------------------- | ||
| 1039 : | leunga | 1080 | Name: Allen Leung |
| 1040 : | leunga | 1094 | Date: 2002/02/25 04:45:00 EST |
| 1041 : | Tag: leunga-20020225-cps-spill | ||
| 1042 : | |||
| 1043 : | This is a new implementation of the CPS spill phase. | ||
| 1044 : | The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml | ||
| 1045 : | In case of problems, replace it with the old file spill.sml | ||
| 1046 : | |||
| 1047 : | The current compiler runs into some serious performance problems when | ||
| 1048 : | constructing a large record. This can happen when we try to compile a | ||
| 1049 : | structure with many items. Even a very simple structure like the following | ||
| 1050 : | makes the compiler slow down. | ||
| 1051 : | |||
| 1052 : | structure Foo = struct | ||
| 1053 : | val x_1 = 0w1 : Word32.int | ||
| 1054 : | val x_2 = 0w2 : Word32.int | ||
| 1055 : | val x_3 = 0w3 : Word32.int | ||
| 1056 : | ... | ||
| 1057 : | val x_N = 0wN : Word32.int | ||
| 1058 : | end | ||
| 1059 : | |||
| 1060 : | The following table shows the compile time, from N=1000 to N=4000, | ||
| 1061 : | with the old compiler: | ||
| 1062 : | |||
| 1063 : | N | ||
| 1064 : | 1000 CPS 100 spill 0.04u 0.00s 0.00g | ||
| 1065 : | MLRISC ra 0.06u 0.00s 0.05g | ||
| 1066 : | (spills = 0 reloads = 0) | ||
| 1067 : | TOTAL 0.63u 0.07s 0.21g | ||
| 1068 : | |||
| 1069 : | 1100 CPS 100 spill 8.25u 0.32s 0.64g | ||
| 1070 : | MLRISC ra 5.68u 0.59s 3.93g | ||
| 1071 : | (spills = 0 reloads = 0) | ||
| 1072 : | TOTAL 14.71u 0.99s 4.81g | ||
| 1073 : | |||
| 1074 : | 1500 CPS 100 spill 58.55u 2.34s 1.74g | ||
| 1075 : | MLRISC ra 5.54u 0.65s 3.91g | ||
| 1076 : | (spills = 543 reloads = 1082) | ||
| 1077 : | TOTAL 65.40u 3.13s 6.00g | ||
| 1078 : | |||
| 1079 : | 2000 CPS 100 spill 126.69u 4.84s 3.08g | ||
| 1080 : | MLRISC ra 0.80u 0.10s 0.55g | ||
| 1081 : | (spills = 42 reloads = 84) | ||
| 1082 : | TOTAL 129.42u 5.10s 4.13g | ||
| 1083 : | |||
| 1084 : | 3000 CPS 100 spill 675.59u 19.03s 11.64g | ||
| 1085 : | MLRISC ra 2.69u 0.27s 1.38g | ||
| 1086 : | (spills = 62 reloads = 124) | ||
| 1087 : | TOTAL 682.48u 19.61s 13.99g | ||
| 1088 : | |||
| 1089 : | 4000 CPS 100 spill 2362.82u 56.28s 43.60g | ||
| 1090 : | MLRISC ra 4.96u 0.27s 2.72g | ||
| 1091 : | (spills = 85 reloads = 170) | ||
| 1092 : | TOTAL 2375.26u 57.21s 48.00g | ||
| 1093 : | |||
| 1094 : | As you can see the old cps spill module suffers from some serious | ||
| 1095 : | leunga | 1115 | performance problem. But since I cannot decipher the old code fully, |
| 1096 : | instead of patching the problems up, I'm reimplementing it | ||
| 1097 : | leunga | 1094 | with a different algorithm. The new code is more modular, |
| 1098 : | smaller when compiled, and substantially faster | ||
| 1099 : | (O(n log n) time and O(n) space). Timing of the new spill module: | ||
| 1100 : | |||
| 1101 : | 4000 CPS 100 spill 0.02u 0.00s 0.00g | ||
| 1102 : | MLRISC ra 0.25u 0.02s 0.15g | ||
| 1103 : | (spills=1 reloads=3) | ||
| 1104 : | TOTAL 7.74u 0.34s 1.62g | ||
| 1105 : | |||
| 1106 : | Implementation details: | ||
| 1107 : | |||
| 1108 : | As far as I can tell, the purpose of the CPS spill module is to make sure the | ||
| 1109 : | number of live variables at any program point (the bandwidth) | ||
| 1110 : | does not exceed a certain limit, which is determined by the | ||
| 1111 : | size of the spill area. | ||
| 1112 : | |||
| 1113 : | When the bandwidth is too large, we decrease the register pressure by | ||
| 1114 : | packing live variables into spill records. How we achieve this is | ||
| 1115 : | completely different than what we did in the old code. | ||
| 1116 : | |||
| 1117 : | First, there is something about the MLRiscGen code generator | ||
| 1118 : | that we should be aware of: | ||
| 1119 : | |||
| 1120 : | o MLRiscGen performs code motion! | ||
| 1121 : | |||
| 1122 : | In particular, it will move floating point computations and | ||
| 1123 : | address computations involving only the heap pointer to | ||
| 1124 : | their use sites (if there is only a single use). | ||
| 1125 : | What this means is that if we have a CPS record construction | ||
| 1126 : | statement | ||
| 1127 : | |||
| 1128 : | RECORD(k,vl,w,e) | ||
| 1129 : | |||
| 1130 : | we should never count the new record address w as live if w | ||
| 1131 : | has only one use (which is often the case). | ||
| 1132 : | |||
| 1133 : | We should do something similar to floating point, but the transformation | ||
| 1134 : | there is much more complex, so I won't deal with that. | ||
| 1135 : | |||
| 1136 : | Secondly, there are now two new cps primops at our disposal: | ||
| 1137 : | |||
| 1138 : | 1. rawrecord of record_kind option | ||
| 1139 : | This pure operator allocates some uninitialized storage from the heap. | ||
| 1140 : | There are two forms: | ||
| 1141 : | |||
| 1142 : | rawrecord NONE [INT n] allocates a tagless record of length n | ||
| 1143 : | rawrecord (SOME rk) [INT n] allocates a tagged record of length n | ||
| 1144 : | and initializes the tag. | ||
| 1145 : | |||
| 1146 : | 2. rawupdate of cty | ||
| 1147 : | rawupdate cty (v,i,x) | ||
| 1148 : | Assigns to x to the ith component of record v. | ||
| 1149 : | The storelist is not updated. | ||
| 1150 : | |||
| 1151 : | We use these new primops for both spilling and increment record construction. | ||
| 1152 : | |||
| 1153 : | 1. Spilling. | ||
| 1154 : | |||
| 1155 : | This is implemented with a linear scan algorithm (but generalized | ||
| 1156 : | to trees). The algorithm will create a single spill record at the | ||
| 1157 : | beginning of the cps function and use rawupdate to spill to it, | ||
| 1158 : | and SELECT or SELp to reload from it. So both spills and reloads | ||
| 1159 : | are fine-grain operations. In contrast, in the old algorithm | ||
| 1160 : | "spills" have to be bundled together in records. | ||
| 1161 : | |||
| 1162 : | Ideally, we should sink the spill record construction to where | ||
| 1163 : | it is needed. We can even split the spill record into multiple ones | ||
| 1164 : | at the places where they are needed. But CPS is not a good | ||
| 1165 : | representation for global code motion, so I'll keep it simple and | ||
| 1166 : | am not attempting this. | ||
| 1167 : | |||
| 1168 : | 2. Incremental record construction (aka record splitting). | ||
| 1169 : | |||
| 1170 : | Long records with many component values which are simulatenously live | ||
| 1171 : | (recall that single use record addresses are not considered to | ||
| 1172 : | be live) are constructed with rawrecord and rawupdate. | ||
| 1173 : | We allocate space on the heap with rawrecord first, then gradually | ||
| 1174 : | fill it in with rawupdate. This is the technique suggested to me | ||
| 1175 : | by Matthias. | ||
| 1176 : | |||
| 1177 : | Some restrictions on when this is applicable: | ||
| 1178 : | 1. It is not a VECTOR record. The code generator currently does not handle | ||
| 1179 : | this case. VECTOR record uses double indirection like arrays. | ||
| 1180 : | 2. All the record component values are defined in the same "basic block" | ||
| 1181 : | as the record constructor. This is to prevent speculative | ||
| 1182 : | record construction. | ||
| 1183 : | |||
| 1184 : | ---------------------------------------------------------------------- | ||
| 1185 : | Name: Allen Leung | ||
| 1186 : | leunga | 1086 | Date: 2002/02/22 01:02:00 EST |
| 1187 : | Tag: leunga-20020222-mlrisc-tools | ||
| 1188 : | |||
| 1189 : | Minor bug fixes in the parser and rewriter | ||
| 1190 : | |||
| 1191 : | ---------------------------------------------------------------------- | ||
| 1192 : | Name: Allen Leung | ||
| 1193 : | leunga | 1085 | Date: 2002/02/21 20:20:00 EST |
| 1194 : | Tag: leunga-20020221-peephole | ||
| 1195 : | |||
| 1196 : | Regenerated the peephole files. Some contained typos in the specification | ||
| 1197 : | and some didn't compile because of pretty printing bugs in the old version | ||
| 1198 : | of 'nowhere'. | ||
| 1199 : | |||
| 1200 : | ---------------------------------------------------------------------- | ||
| 1201 : | Name: Allen Leung | ||
| 1202 : | leunga | 1079 | Date: 2002/02/19 20:20:00 EST |
| 1203 : | Tag: leunga-20020219-mlrisc-tools | ||
| 1204 : | Description: | ||
| 1205 : | |||
| 1206 : | Minor bug fixes to the mlrisc-tools library: | ||
| 1207 : | |||
| 1208 : | 1. Fixed up parsing colon suffixed keywords | ||
| 1209 : | 2. Added the ability to shut the error messages up | ||
| 1210 : | 3. Reimplemented the pretty printer and fixed up/improved | ||
| 1211 : | the pretty printing of handle and -> types. | ||
| 1212 : | 4. Fixed up generation of literal symbols in the nowhere tool. | ||
| 1213 : | 5. Added some SML keywords to to sml.sty | ||
| 1214 : | |||
| 1215 : | ---------------------------------------------------------------------- | ||
| 1216 : | Name: Matthias Blume | ||
| 1217 : | blume | 1078 | Date: 2002/02/19 16:20:00 EST |
| 1218 : | Tag: blume-20020219-cmffi | ||
| 1219 : | Description: | ||
| 1220 : | |||
| 1221 : | A wild mix of changes, some minor, some major: | ||
| 1222 : | |||
| 1223 : | * All C FFI-related libraries are now anchored under $c: | ||
| 1224 : | $/c.cm --> $c/c.cm | ||
| 1225 : | $/c-int.cm --> $c/internals/c-int.cm | ||
| 1226 : | $/memory.cm --> $c/memory/memory.cm | ||
| 1227 : | |||
| 1228 : | * "make" tool (in CM) now treats its argument pathname slightly | ||
| 1229 : | differently: | ||
| 1230 : | 1. If the native expansion is an absolute name, then before invoking | ||
| 1231 : | the "make" command on it, CM will apply OS.Path.mkRelative | ||
| 1232 : | (with relativeTo = OS.FileSys.getDir()) to it. | ||
| 1233 : | 2. The argument will be passed through to subsequent phases of CM | ||
| 1234 : | processing without "going native". In particular, if the argument | ||
| 1235 : | was an anchored path, then "make" will not lose track of that anchor. | ||
| 1236 : | |||
| 1237 : | * Compiler backends now "know" their respective C calling conventions | ||
| 1238 : | instead of having to be told about it by ml-nlffigen. This relieves | ||
| 1239 : | ml-nlffigen from one of its burdens. | ||
| 1240 : | |||
| 1241 : | * The X86Backend has been split into X86CCallBackend and X86StdCallBackend. | ||
| 1242 : | |||
| 1243 : | * Export C_DEBUG and C_Debug from $c/c.cm. | ||
| 1244 : | |||
| 1245 : | * C type encoding in ml-nlffi-lib has been improved to model the conceptual | ||
| 1246 : | subtyping relationship between incomplete pointers and their complete | ||
| 1247 : | counterparts. For this, ('t, 'c) ptr has been changed to 'o ptr -- | ||
| 1248 : | with the convention of instantiating 'o with ('t, 'c) obj whenever | ||
| 1249 : | the pointer target type is complete. In the incomplete case, 'o | ||
| 1250 : | will be instantiated with some "'c iobj" -- a type obtained by | ||
| 1251 : | using one of the functors PointerToIncompleteType or PointerToCompleteType. | ||
| 1252 : | |||
| 1253 : | Operations that work on both incomplete and complete pointer types are | ||
| 1254 : | typed as taking an 'o ptr while operations that require the target to | ||
| 1255 : | be known are typed as taking some ('t, 'c) obj ptr. | ||
| 1256 : | |||
| 1257 : | voidptr is now a bit "more concrete", namely "type voidptr = void ptr'" | ||
| 1258 : | where void is an eqtype without any values. This makes it possible | ||
| 1259 : | to work on voidptr values using functions meant to operate on light | ||
| 1260 : | incomplete pointers. | ||
| 1261 : | |||
| 1262 : | * As a result of the above, signature POINTER_TO_INCOMPLETE_TYPE has | ||
| 1263 : | been vastly simplified. | ||
| 1264 : | |||
| 1265 : | ---------------------------------------------------------------------- | ||
| 1266 : | Name: Matthias Blume | ||
| 1267 : | blume | 1076 | Date: 2002/02/19 10:48:00 EST |
| 1268 : | Tag: blume-20020219-pqfix | ||
| 1269 : | Description: | ||
| 1270 : | |||
| 1271 : | Applied Chris Okasaki's bug fix for priority queues. | ||
| 1272 : | |||
| 1273 : | ---------------------------------------------------------------------- | ||
| 1274 : | Name: Matthias Blume | ||
| 1275 : | blume | 1073 | Date: 2002/02/15 17:05:00 |
| 1276 : | blume | 1069 | Tag: Release_110_39 |
| 1277 : | Description: | ||
| 1278 : | |||
| 1279 : | blume | 1073 | Last-minute retagging is becoming a tradition... :-( |
| 1280 : | |||
| 1281 : | This is the working release 110.39. | ||
| 1282 : | |||
| 1283 : | ---------------------------------------------------------------------- | ||
| 1284 : | Name: Matthias Blume | ||
| 1285 : | Date: 2002/02/15 16:00:00 EST | ||
| 1286 : | Tag: Release_110_39-orig | ||
| 1287 : | Description: | ||
| 1288 : | |||
| 1289 : | blume | 1069 | Working release 110.39. New bootfiles. |
| 1290 : | |||
| 1291 : | blume | 1073 | (Update: There was a small bug in the installer so it wouldn't work |
| 1292 : | with all shells. So I retagged. -Matthias) | ||
| 1293 : | |||
| 1294 : | blume | 1069 | ---------------------------------------------------------------------- |
| 1295 : | Name: Matthias Blume | ||
| 1296 : | blume | 1068 | Date: 2002/02/15 14:17:00 EST |
| 1297 : | Tag: blume-20020215-showbindings | ||
| 1298 : | Description: | ||
| 1299 : | |||
| 1300 : | Added EnvRef.listBoundSymbols and CM.State.showBindings. Especially | ||
| 1301 : | the latter can be useful for exploring what bindings are available at | ||
| 1302 : | the interactive prompt. (The first function returns only the list | ||
| 1303 : | of symbols that are really bound, the second prints those but also the | ||
| 1304 : | ones that CM's autoloading mechanism knows about.) | ||
| 1305 : | |||
| 1306 : | ---------------------------------------------------------------------- | ||
| 1307 : | Name: Matthias Blume | ||
| 1308 : | blume | 1067 | Date: 2002/02/15 12:08:00 EST |
| 1309 : | Tag: blume-20020215-iptrs | ||
| 1310 : | Description: | ||
| 1311 : | |||
| 1312 : | Two improvements to ml-nlffigen: | ||
| 1313 : | |||
| 1314 : | 1. Write files only if they do not exist or if their current contents | ||
| 1315 : | do not coincide with what's being written. (That is, avoid messing | ||
| 1316 : | with the time stamps unless absolutely necessary.) | ||
| 1317 : | |||
| 1318 : | 2. Implement a "repository" mechanism for generated files related | ||
| 1319 : | to "incomplete pointer types". See the README file for details. | ||
| 1320 : | |||
| 1321 : | ---------------------------------------------------------------------- | ||
| 1322 : | Name: Matthias Blume | ||
| 1323 : | blume | 1066 | Date: 2002/02/14 11:50:00 EST |
| 1324 : | Tag: blume-20020214-quote | ||
| 1325 : | Description: | ||
| 1326 : | |||
| 1327 : | Added a type 't t_' to tag.sml (in ml-nlffi-lib.cm). This is required | ||
| 1328 : | because of the new and improved tag generation scheme. (Thanks to Allen | ||
| 1329 : | Leung for pointing it out.) | ||
| 1330 : | |||
| 1331 : | ---------------------------------------------------------------------- | ||
| 1332 : | george | 1065 | Name: Lal George |
| 1333 : | Date: 2002/02/14 09:55:27 EST 2002 | ||
| 1334 : | Tag: george-20020214-isabelle-bug | ||
| 1335 : | Description: | ||
| 1336 : | |||
| 1337 : | Fixed the MLRISC bug sent by Markus Wenzel regarding the compilation | ||
| 1338 : | of Isabelle on the x86. | ||
| 1339 : | |||
| 1340 : | From Allen: | ||
| 1341 : | ----------- | ||
| 1342 : | I've found the problem: | ||
| 1343 : | |||
| 1344 : | in ra-core.sml, I use the counter "blocked" to keep track of the | ||
| 1345 : | true number of elements in the freeze queue. When the counter goes | ||
| 1346 : | to zero, I skip examining the queue. But I've messed up the | ||
| 1347 : | bookkeeping in combine(): | ||
| 1348 : | |||
| 1349 : | else (); | ||
| 1350 : | case !ucol of | ||
| 1351 : | PSEUDO => (if !cntv > 0 then | ||
| 1352 : | (if !cntu > 0 then blocked := !blocked - 1 else (); | ||
| 1353 : | ^^^^^^^^^^^^^^^^^^^^^^^ | ||
| 1354 : | moveu := mergeMoveList(!movev, !moveu) | ||
| 1355 : | ) | ||
| 1356 : | else (); | ||
| 1357 : | |||
| 1358 : | combine() is called to coalesce two nodes u and v. | ||
| 1359 : | I think I was thinking that if the move counts of u and v are both | ||
| 1360 : | greater than zero then after they are coalesced then one node is | ||
| 1361 : | removed from the freeze queue. Apparently I was thinking that | ||
| 1362 : | both u and v are of low degree, but that's clearly not necessarily true. | ||
| 1363 : | |||
| 1364 : | |||
| 1365 : | 02/12/2002: | ||
| 1366 : | Here's the patch. HOL now compiles. | ||
| 1367 : | |||
| 1368 : | I don't know how this impact on performance (compile | ||
| 1369 : | time or runtime). This bug caused the RA (especially on the x86) | ||
| 1370 : | to go thru the potential spill phase when there are still nodes on the | ||
| 1371 : | freeze queue. | ||
| 1372 : | |||
| 1373 : | |||
| 1374 : | |||
| 1375 : | |||
| 1376 : | ---------------------------------------------------------------------- | ||
| 1377 : | blume | 1058 | Name: Matthias Blume |
| 1378 : | blume | 1064 | Date: 2002/02/13 22:40:00 EST |
| 1379 : | Tag: blume-20020213-fptr-rtti | ||
| 1380 : | Description: | ||
| 1381 : | |||
| 1382 : | Fixed a bug in ml-nlffigen that was introduced with one of the previous | ||
| 1383 : | updates. | ||
| 1384 : | |||
| 1385 : | ---------------------------------------------------------------------- | ||
| 1386 : | Name: Matthias Blume | ||
| 1387 : | blume | 1063 | Date: 2002/02/13 16:41:00 EST |
| 1388 : | Tag: blume-20020213-cmlpq | ||
| 1389 : | Description: | ||
| 1390 : | |||
| 1391 : | Added new priority queue export symbols (which have just been added to | ||
| 1392 : | smlnj-lib.cm) to CML's version of smlnj-lib.cm. (Otherwise CML would | ||
| 1393 : | not compile and the installer would choke.) | ||
| 1394 : | |||
| 1395 : | ---------------------------------------------------------------------- | ||
| 1396 : | Name: Matthias Blume | ||
| 1397 : | blume | 1062 | Date: 2002/02/13 16:15:00 EST |
| 1398 : | Tag: blume-20020213-various | ||
| 1399 : | Description: | ||
| 1400 : | |||
| 1401 : | 1. More tweaks to ml-nlffigen: | ||
| 1402 : | |||
| 1403 : | - better internal datastructures (resulting in slight speedup) | ||
| 1404 : | - "-match" option requires exact match | ||
| 1405 : | - "localized" gensym counters (untagged structs/unions nested within | ||
| 1406 : | other structs/unions or within typedefs get a fresh counter; their | ||
| 1407 : | tag will be prefixed by a concatenation of their parents' tags) | ||
| 1408 : | - bug fixes (related to calculation of transitive closure of types | ||
| 1409 : | to be included in the output) | ||
| 1410 : | |||
| 1411 : | 2. Minor Basis updates: | ||
| 1412 : | |||
| 1413 : | - added implementations for List.collate and Option.app | ||
| 1414 : | |||
| 1415 : | ---------------------------------------------------------------------- | ||
| 1416 : | Name: Matthias Blume | ||
| 1417 : | blume | 1060 | Date: 2002/02/11 15:55:00 EST |
| 1418 : | Tag: blume-20020211-gensym | ||
| 1419 : | Description: | ||
| 1420 : | |||
| 1421 : | Added a "-gensym" option to command line of ml-nlffigen. This can be | ||
| 1422 : | used to specify a "stem" -- a string that is inserted in all "gensym'd" | ||
| 1423 : | names (ML structure names that correspond to unnamed C structs, unions, | ||
| 1424 : | and enums), so that separate runs of ml-nlffigen do not clash. | ||
| 1425 : | |||
| 1426 : | ---------------------------------------------------------------------- | ||
| 1427 : | Name: Matthias Blume | ||
| 1428 : | blume | 1059 | Date: 2002/02/11 12:05:00 EST |
| 1429 : | Tag: blume-20020211-gensml | ||
| 1430 : | Description: | ||
| 1431 : | |||
| 1432 : | A quick fix for a problem with GenSML (in the pgraph-util library): | ||
| 1433 : | Make generation of toplevel "local" optional. (Strictly speaking, | ||
| 1434 : | signature definitions within "local" are not legal SML.) | ||
| 1435 : | |||
| 1436 : | Other than that: updates to INSTALL and cm/TODO. | ||
| 1437 : | |||
| 1438 : | ---------------------------------------------------------------------- | ||
| 1439 : | Name: Matthias Blume | ||
| 1440 : | blume | 1058 | Date: 2002/02/08 15:00:00 EST |
| 1441 : | Tag: blume-20020208-uniquepid | ||
| 1442 : | Description: | ||
| 1443 : | |||
| 1444 : | 0. Version number has been bumped to 110.38.1. NEW BOOTFILES!!! | ||
| 1445 : | |||
| 1446 : | 1. The installer (config/install.sh) has gotten smarter: | ||
| 1447 : | |||
| 1448 : | - Configuration options are a bit easier to specify now | ||
| 1449 : | (in config/targets). | ||
| 1450 : | - Bug in recognizing .tar.bz2 files fixed. | ||
| 1451 : | - Installer automatically resolves dependencies between | ||
| 1452 : | configuration options (e.g., if you ask for eXene, you will | ||
| 1453 : | also get cml -- regardless whether you asked for it or not). | ||
| 1454 : | - Installer can run in "quieter mode" by setting the environment | ||
| 1455 : | variable INSTALL_QUIETLY to "true". "Quieter" does not mean | ||
| 1456 : | "completely silent", though. | ||
| 1457 : | - Build HashCons library as part of smlnj-lib. | ||
| 1458 : | |||
| 1459 : | 2. A new scheme for assigning persistent identifiers to compilation | ||
| 1460 : | units (and, by extension, to types etc.) has been put into place. | ||
| 1461 : | This fixes a long-standing bug where types and even dynamic values | ||
| 1462 : | can get internally confused, thereby compromising type safety | ||
| 1463 : | (abstraction) and dynamic correctness. See | ||
| 1464 : | |||
| 1465 : | http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz | ||
| 1466 : | |||
| 1467 : | for an example of how things could go wrong until now. | ||
| 1468 : | |||
| 1469 : | The downside of the new scheme is that pids are not quite as | ||
| 1470 : | persistent as they used to be: CM will generate a fresh pid | ||
| 1471 : | for every compilation unit that it thinks it sees for the first | ||
| 1472 : | time. That means that if you compile starting from a clean, fresh | ||
| 1473 : | source tree at two different times, you end up with different | ||
| 1474 : | binaries. | ||
| 1475 : | |||
| 1476 : | Cutoff recompilation, however, has not been compromised because | ||
| 1477 : | CM keeps pid information in special caches between runs. | ||
| 1478 : | |||
| 1479 : | ---------------------------------------------------------------------- | ||
| 1480 : | george | 1053 | Name: Lal George |
| 1481 : | george | 1055 | Date: 2002/02/07 15:34:13 EST 2002 |
| 1482 : | Tag: <none> | ||
| 1483 : | Description: | ||
| 1484 : | |||
| 1485 : | Compilers that generate assembly code may produce global labels | ||
| 1486 : | whose value is resolved at link time. The various peephole optimization | ||
| 1487 : | modules did not take this in account. | ||
| 1488 : | |||
| 1489 : | TODO. The Labels.addrOf function should really return an option | ||
| 1490 : | type so that clients are forced to deal with this issue, rather | ||
| 1491 : | than an exception being raised. | ||
| 1492 : | |||
| 1493 : | ---------------------------------------------------------------------- | ||
| 1494 : | Name: Lal George | ||
| 1495 : | george | 1053 | Date: 2002/02/06 13:55:02 EST |
| 1496 : | Tag: george-20020206-ra-breakup | ||
| 1497 : | Description: | ||
| 1498 : | |||
| 1499 : | 1. A bug fix from Allen. | ||
| 1500 : | |||
| 1501 : | A typo causes extra fstp %st(0)'s to be generated at compensation | ||
| 1502 : | edges, which might cause stack underflow traps at runtime. This | ||
| 1503 : | occurs in fft where there are extraneous fstps right before the 'into' | ||
| 1504 : | trap instruction (in this case they are harmless since none of the | ||
| 1505 : | integers overflow.) | ||
| 1506 : | |||
| 1507 : | george | 1054 | 2. Pulled out various utility modules that were embedded in the modules |
| 1508 : | george | 1053 | of the register allocator. I need these modules for other purposes, but |
| 1509 : | george | 1054 | they are not complete enough to put into a library (just yet). |
| 1510 : | george | 1053 | ---------------------------------------------------------------------- |
| 1511 : | blume | 1049 | Name: Matthias Blume |
| 1512 : | Date: 2002/01/31 16:05:00 EST | ||
| 1513 : | Tag: blume-20020131-sparc-ccalls | ||
| 1514 : | Description: | ||
| 1515 : | |||
| 1516 : | 1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes) | ||
| 1517 : | of extra stack space by mistake. Fixed. | ||
| 1518 : | |||
| 1519 : | 2. Bug in logic of handling of command-line options in ml-nlffigen fixed. | ||
| 1520 : | |||
| 1521 : | ---------------------------------------------------------------------- | ||
| 1522 : | leunga | 1048 | Name: Allen Leung |
| 1523 : | Date: 2002/01/30 | ||
| 1524 : | Tag: leunga-20020130-nowhere-bug-fix | ||
| 1525 : | Description: | ||
| 1526 : | |||
| 1527 : | MLRISC bug fixes: | ||
| 1528 : | 1. Fixed a bindings computation bug in the 'nowhere' program generator tool. | ||
| 1529 : | 2. MachineInt.fromString was negating its value. | ||
| 1530 : | |||
| 1531 : | ---------------------------------------------------------------------- | ||
| 1532 : | blume | 1039 | Name: Matthias Blume |
| 1533 : | blume | 1047 | Date: 2002/01/29 |
| 1534 : | Tag: blume-20020129-INSTALL | ||
| 1535 : | Description: | ||
| 1536 : | |||
| 1537 : | - Added somewhat detailed installation instructions (file INSTALL). | ||
| 1538 : | - Fixed curl-detection bug in config/install.sh. | ||
| 1539 : | - It is now possible to select the URL getter using the URLGETTER | ||
| 1540 : | environment variable: | ||
| 1541 : | |||
| 1542 : | not set / "unknown" --> automatic detection (script tries wget, | ||
| 1543 : | curl, and lynx) | ||
| 1544 : | "wget" / "curl" / "lynx" --> use the specified program (script "knows" | ||
| 1545 : | how to properly invoke them) | ||
| 1546 : | other --> use $URLGETTER directly, it must take | ||
| 1547 : | precisely two command-line arguments | ||
| 1548 : | (source URL and destination file name) | ||
| 1549 : | |||
| 1550 : | ---------------------------------------------------------------------- | ||
| 1551 : | Name: Matthias Blume | ||
| 1552 : | blume | 1044 | Date: 2002/01/28 |
| 1553 : | blume | 1045 | Tag: blume-20020128-sparc-ccalls |
| 1554 : | Description: | ||
| 1555 : | |||
| 1556 : | - Fixed problem with calculation of "used" registers in sparc-c-calls. | ||
| 1557 : | - Make use of the allocParam argument in sparc-c-calls. | ||
| 1558 : | |||
| 1559 : | ---------------------------------------------------------------------- | ||
| 1560 : | Name: Matthias Blume | ||
| 1561 : | Date: 2002/01/28 | ||
| 1562 : | blume | 1044 | Tag: blume-20020128-allocParam |
| 1563 : | Description: | ||
| 1564 : | |||
| 1565 : | John Reppy: Changes c-calls API to accept client-callback for | ||
| 1566 : | allocating extra stack space. | ||
| 1567 : | me: Corresponding changes to mlriscGen (using a dummy argument that | ||
| 1568 : | does not change the current behavior). | ||
| 1569 : | |||
| 1570 : | ---------------------------------------------------------------------- | ||
| 1571 : | Name: Matthias Blume | ||
| 1572 : | blume | 1040 | Date: 2002/01/28 12:00:00 |
| 1573 : | Tag: Release_110_38 | ||
| 1574 : | Description: | ||
| 1575 : | |||
| 1576 : | This time for real!!! | ||
| 1577 : | |||
| 1578 : | ---------------------------------------------------------------------- | ||
| 1579 : | Name: Matthias Blume | ||
| 1580 : | blume | 1039 | Date: 2002/01/28 10:56:00 EST |
| 1581 : | Tag: blume-20020128-retraction | ||
| 1582 : | Description: | ||
| 1583 : | |||
| 1584 : | 0. Retracted earlier 110.38. (The Release_110_38 tag has been replaced | ||
| 1585 : | with blume-Release_110_38-retracted.) | ||
| 1586 : | |||
| 1587 : | 1. Fixed a problem with incorrect rounding modes in real64.sml. | ||
| 1588 : | (Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.) | ||
| 1589 : | |||
| 1590 : | 2. A bug in ml-nlffigen related to the handling of unnamed structs, unions, | ||
| 1591 : | and enums fixed. The naming of corresponding ML identifiers should | ||
| 1592 : | now be consistent again. | ||
| 1593 : | |||
| 1594 : | ---------------------------------------------------------------------- | ||
| 1595 : | leunga | 1038 | Name: Allen Leung |
| 1596 : | Date: 2002/01/27 | ||
| 1597 : | Tag: leunga-20020127-nowhere | ||
| 1598 : | Description: | ||
| 1599 : | |||
| 1600 : | Added a target called nowhere in the configuration scripts. | ||
| 1601 : | Enabling this will build the MLRISC 'nowhere' tool (for translating | ||
| 1602 : | programs with where-clauses into legal SML code) during installation. | ||
| 1603 : | |||
| 1604 : | ---------------------------------------------------------------------- | ||
| 1605 : | blume | 1036 | Name: Matthias Blume |
| 1606 : | leunga | 1038 | Date: 2002/01/25 21:27:00 EST |
| 1607 : | blume | 1039 | Tag: blume-Release_110_38-retracted |
| 1608 : | blume | 1037 | Description: |
| 1609 : | |||
| 1610 : | Call it a (working) release! Version is 110.38. Bootfiles are ready. | ||
| 1611 : | |||
| 1612 : | README will be added later. | ||
| 1613 : | |||
| 1614 : | blume | 1039 | !!! NOTE: Re-tagged as blume-Release_110_38-retracted. Original tag |
| 1615 : | (Release_110_38) removed. Reason: Last-minute bug fixes. | ||
| 1616 : | |||
| 1617 : | blume | 1037 | ---------------------------------------------------------------------- |
| 1618 : | Name: Matthias Blume | ||
| 1619 : | blume | 1036 | Date: 2002/01/25 |
| 1620 : | Tag: blume-20020125-ffi | ||
| 1621 : | Description: | ||
| 1622 : | |||
| 1623 : | A large number of tweaks and improvements to ml-nlffi-lib and | ||
| 1624 : | ml-nlffigen: | ||
| 1625 : | |||
| 1626 : | - ML represenation types have been streamlined | ||
| 1627 : | - getter and setter functions work with concrete values, not abstract | ||
| 1628 : | ones where possible | ||
| 1629 : | - ml-nlffigen command line more flexible (see README file there) | ||
| 1630 : | - some bugs have been fixed (hopefully) | ||
| 1631 : | |||
| 1632 : | ---------------------------------------------------------------------- | ||
| 1633 : | george | 1034 | Name: Lal George |
| 1634 : | Date: 2002/01/24 | ||
| 1635 : | Tag: george-20020124-risc-ra-interface | ||
| 1636 : | Description: | ||
| 1637 : | |||
| 1638 : | There is a dramatic simplification in the interface to the | ||
| 1639 : | register allocator for RISC architectures as a result of making | ||
| 1640 : | parallel copy instructions explicit. | ||
| 1641 : | |||
| 1642 : | ---------------------------------------------------------------------- | ||
| 1643 : | blume | 1028 | Name: Matthias Blume |
| 1644 : | blume | 1032 | Date: 2002/01/22 |
| 1645 : | Tag: blume-20020122-x86-ccalls | ||
| 1646 : | Description: | ||
| 1647 : | |||
| 1648 : | Bug fix for c-calls on x86 (having to do with how char- and | ||
| 1649 : | short-arguments are being handled). | ||
| 1650 : | |||
| 1651 : | ---------------------------------------------------------------------- | ||
| 1652 : | Name: Matthias Blume | ||
| 1653 : | blume | 1031 | Date: 2002/01/21 |
| 1654 : | Tag: blume-20020121-ff | ||
| 1655 : | Description: | ||
| 1656 : | |||
| 1657 : | Another day of fiddling with the FFI... | ||
| 1658 : | |||
| 1659 : | 1. Bug fix/workaround: CKIT does not complain about negative array | ||
| 1660 : | dimensions, so ml-nlffigen has to guard itself against this possibility. | ||
| 1661 : | (Otherwise a negative dimension would send it into an infinite loop.) | ||
| 1662 : | |||
| 1663 : | 2. Some of the abstract types (light objects, light pointers, most "base" | ||
| 1664 : | types) in structure C are now eqtypes. | ||
| 1665 : | |||
| 1666 : | 3. Added constructors and test functions for NULL function pointers. | ||
| 1667 : | |||
| 1668 : | ---------------------------------------------------------------------- | ||
| 1669 : | Name: Matthias Blume | ||
| 1670 : | blume | 1028 | Date: 2002/01/18 |
| 1671 : | blume | 1030 | Tag: blume-20020118-ready-for-new-release |
| 1672 : | Description: | ||
| 1673 : | |||
| 1674 : | Made config/srcarchiveurl point to a new place. (Will provide boot | ||
| 1675 : | files shortly.) | ||
| 1676 : | |||
| 1677 : | Maybe we christen this to be 110.38? | ||
| 1678 : | |||
| 1679 : | ---------------------------------------------------------------------- | ||
| 1680 : | Name: Matthias Blume | ||
| 1681 : | Date: 2002/01/18 | ||
| 1682 : | blume | 1029 | Tag: blume-20020118-more-ffifiddle |
| 1683 : | Description: | ||
| 1684 : | |||
| 1685 : | Today's FFI fiddling: | ||
| 1686 : | |||
| 1687 : | - Provided a structure CGetSet with "convenient" versions of C.Get.* and | ||
| 1688 : | C.Set.* that use concrete (MLRep.*) arguments and results instead | ||
| 1689 : | of abstract ones. | ||
| 1690 : | |||
| 1691 : | - Provided word-style bit operations etc. for "int" representation | ||
| 1692 : | types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short, | ||
| 1693 : | and Long. | ||
| 1694 : | |||
| 1695 : | ---------------------------------------------------------------------- | ||
| 1696 : | Name: Matthias Blume | ||
| 1697 : | Date: 2002/01/18 | ||
| 1698 : | blume | 1028 | Tag: blume-20020118-use-x86-fp |
| 1699 : | Description: | ||
| 1700 : | |||
| 1701 : | Now that x86-fast-fp seems to be working, I turned it back on again | ||
| 1702 : | by default. (Seems to work fine now, even with the FFI.) | ||
| 1703 : | |||
| 1704 : | Other than that, I added some documentation about the FFI to | ||
| 1705 : | src/ml-nlffigen/README and updated the FFI test examples in | ||
| 1706 : | src/ml-nlffi-lib/Tests/*. | ||
| 1707 : | |||
| 1708 : | ---------------------------------------------------------------------- | ||
| 1709 : | leunga | 1025 | Name: Allen Leung |
| 1710 : | Date: 2002/01/17 | ||
| 1711 : | Tag: leunga-20020117-x86-fast-fp-call | ||
| 1712 : | Description: | ||
| 1713 : | |||
| 1714 : | 1. Fixed a problem with handling return fp values when x86's fast fp | ||
| 1715 : | mode is turned on. | ||
| 1716 : | |||
| 1717 : | 2. Minor pretty printing fix for cellset. Print %st(0) as %st(0) instead | ||
| 1718 : | of %f32. | ||
| 1719 : | |||
| 1720 : | 3. Added a constructor INT32lit to the ast of MLRISC tools. | ||
| 1721 : | |||
| 1722 : | ---------------------------------------------------------------------- | ||
| 1723 : | blume | 1021 | Name: Matthias Blume |
| 1724 : | Date: 2002/01/16 | ||
| 1725 : | Tag: blume-20020116-ffifiddle | ||
| 1726 : | Description: | ||
| 1727 : | |||
| 1728 : | More fiddling with the FFI interface: | ||
| 1729 : | |||
| 1730 : | - Make constness 'c instead of rw wherever possible. This eliminates | ||
| 1731 : | the need for certain explicit coercions. (However, due to ML's | ||
| 1732 : | value polymorphism, there will still be many cases where explicit | ||
| 1733 : | coercions are necessary. Phantom types are not the whole answer | ||
| 1734 : | to modeling a subtyping relationship in ML.) | ||
| 1735 : | |||
| 1736 : | - ro/rw coersions for pointers added. (Avoids the detour through */&.) | ||
| 1737 : | |||
| 1738 : | - "printf" test example added to src/ml-nlffi-lib/Tests. (Demonstrates | ||
| 1739 : | clumsy workaround for varargs problem.) | ||
| 1740 : | |||
| 1741 : | ---------------------------------------------------------------------- | ||
| 1742 : | george | 1016 | Name: Lal George |
| 1743 : | Date: 2002/01/15 | ||
| 1744 : | Tag: <none> | ||
| 1745 : | Description: | ||
| 1746 : | |||
| 1747 : | 1. Since COPY instructions are no longer native to the architecture, | ||
| 1748 : | a generic functor can be used to implement the expandCopies function. | ||
| 1749 : | |||
| 1750 : | 2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a | ||
| 1751 : | TEXT segment. | ||
| 1752 : | |||
| 1753 : | ---------------------------------------------------------------------- | ||
| 1754 : | blume | 1011 | Name: Matthias Blume |
| 1755 : | blume | 1015 | Date: 2002/01/15 |
| 1756 : | Tag: blume-20020115-ffiupdates | ||
| 1757 : | Description: | ||
| 1758 : | |||
| 1759 : | 1. Fix for bug resulting in single-precision float values being returned | ||
| 1760 : | incorrectly from FFI calls. | ||
| 1761 : | |||
| 1762 : | 2. Small modifications to C FFI API: | ||
| 1763 : | |||
| 1764 : | - memory-allocation routines return straight objects (no options) | ||
| 1765 : | and raise an exception in out-of-memory situations | ||
| 1766 : | - unsafe extensions to cast between function pointers and pointers | ||
| 1767 : | from/to ints | ||
| 1768 : | - added structure C_Debug as an alternative to structure C where | ||
| 1769 : | pointer-dereferencing (|*| and |*!) always check for null-pointers | ||
| 1770 : | - added open_lib' to DynLinkage; open_lib' works like open_lib | ||
| 1771 : | but also takes a (possibly empty) list of existing library handles | ||
| 1772 : | that the current library depends on | ||
| 1773 : | |||
| 1774 : | ---------------------------------------------------------------------- | ||
| 1775 : | Name: Matthias Blume | ||
| 1776 : | blume | 1011 | Date: 2002/01/10 |
| 1777 : | Tag: blume-20020110-newffigen | ||
| 1778 : | Description: | ||
| 1779 : | |||
| 1780 : | 1. Updates to portable graph code. | ||
| 1781 : | |||
| 1782 : | 2. Major update to ml-nlffigen and ml-nlffi-lib. Things are much | ||
| 1783 : | more scalable now so that even huge interfaces such as the one | ||
| 1784 : | for GTK compile in finite time and space. :-) | ||
| 1785 : | See src/ml-nlffigen/README for details on what's new. | ||
| 1786 : | |||
| 1787 : | ---------------------------------------------------------------------- | ||
| 1788 : | george | 1003 | Name: Lal George |
| 1789 : | george | 1009 | Date: 2001/01/09 14:31:35 EST 2002 |
| 1790 : | Tag: george-20011206-rm-native-copy | ||
| 1791 : | Description: | ||
| 1792 : | |||
| 1793 : | Removed the native COPY and FCOPY instructions | ||
| 1794 : | from all the architectures and replaced it with the | ||
| 1795 : | explicit COPY instruction from the previous commit. | ||
| 1796 : | |||
| 1797 : | It is now possible to simplify many of the optimizations | ||
| 1798 : | modules that manipulate copies. This has not been | ||
| 1799 : | done in this change. | ||
| 1800 : | |||
| 1801 : | ---------------------------------------------------------------------- | ||
| 1802 : | Name: Lal George | ||
| 1803 : | george | 1003 | Date: 2001/12/06 16:50:13 EST 2001 |
| 1804 : | Tag: george-20011206-mlrisc-instruction | ||
| 1805 : | Description: | ||
| 1806 : | george | 824 | |
| 1807 : | george | 1003 | Changed the representation of instructions from being fully abstract |
| 1808 : | to being partially concrete. That is to say: | ||
| 1809 : | |||
| 1810 : | from | ||
| 1811 : | type instruction | ||
| 1812 : | |||
| 1813 : | to | ||
| 1814 : | type instr (* machine instruction *) | ||
| 1815 : | |||
| 1816 : | datatype instruction = | ||
| 1817 : | LIVE of {regs: C.cellset, spilled: C.cellset} | ||
| 1818 : | | KILL of {regs: C.cellset, spilled: C.cellset} | ||
| 1819 : | | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list} | ||
| 1820 : | | ANNOTATION of {i: instruction, a: Annotations.annotation} | ||
| 1821 : | | INSTR of instr | ||
| 1822 : | |||
| 1823 : | This makes the handling of certain special instructions that appear on | ||
| 1824 : | all architectures easier and uniform. | ||
| 1825 : | |||
| 1826 : | LIVE and KILL say that a list of registers are live or killed at the | ||
| 1827 : | program point where they appear. No spill code is generated when an | ||
| 1828 : | element of the 'regs' field is spilled, but the register is moved to | ||
| 1829 : | the 'spilled' (which is present, more for debugging than anything else). | ||
| 1830 : | |||
| 1831 : | LIVE replaces the (now deprecated) DEFFREG instruction on the alpha. | ||
| 1832 : | We used to generate: | ||
| 1833 : | |||
| 1834 : | DEFFREG f1 | ||
| 1835 : | f1 := f2 + f3 | ||
| 1836 : | trapb | ||
| 1837 : | |||
| 1838 : | but now generate: | ||
| 1839 : | |||
| 1840 : | f1 := f2 + f3 | ||
| 1841 : | trapb | ||
| 1842 : | LIVE {regs=[f1,f2,f3], spilled=[]} | ||
| 1843 : | |||
| 1844 : | Furthermore, the DEFFREG (hack) required that all floating point instruction | ||
| 1845 : | use all registers mentioned in the instruction. Therefore f1 := f2 + f3, | ||
| 1846 : | defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting | ||
| 1847 : | in a cleaner alpha implementation. (Hopefully, intel will not get rid of | ||
| 1848 : | this architecture). | ||
| 1849 : | |||
| 1850 : | COPYXXX is intended to replace the parallel COPY and FCOPY available on | ||
| 1851 : | all the architectures. This will result in further simplification of the | ||
| 1852 : | register allocator that must be aware of them for coalescing purposes, and | ||
| 1853 : | will also simplify certain aspects of the machine description that provides | ||
| 1854 : | callbacks related to parallel copies. | ||
| 1855 : | |||
| 1856 : | ANNOTATION should be obvious, and now INSTR represents the honest to God | ||
| 1857 : | machine instruction set! | ||
| 1858 : | |||
| 1859 : | The <arch>/instructions/<arch>Instr.sml files define certain utility | ||
| 1860 : | functions for making porting easier -- essentially converting upper case | ||
| 1861 : | to lower case. All machine instructions (of type instr) are in upper case, | ||
| 1862 : | and the lower case form generates an MLRISC instruction. For example on | ||
| 1863 : | the alpha we have: | ||
| 1864 : | |||
| 1865 : | datatype instr = | ||
| 1866 : | LDA of {r:cell, b:cell, d:operand} | ||
| 1867 : | | ... | ||
| 1868 : | |||
| 1869 : | val lda : {r:cell, b:cell, d:operand} -> instruction | ||
| 1870 : | ... | ||
| 1871 : | |||
| 1872 : | where lda is just (INSTR o LDA), etc. | ||
| 1873 : | |||
| 1874 : | leunga | 815 | ---------------------------------------------------------------------- |
| 1875 : | blume | 986 | Name: Matthias Blume |
| 1876 : | blume | 993 | Date: 2001/11/22 21:40:00 EST |
| 1877 : | Tag: Release_110_37 | ||
| 1878 : | Description: | ||
| 1879 : | |||
| 1880 : | Release 110.37. This time for real. | ||
| 1881 : | |||
| 1882 : | ---------------------------------------------------------------------- | ||
| 1883 : | Name: Matthias Blume | ||
| 1884 : | blume | 988 | Date: 2001/11/21 16:35:00 EST |
| 1885 : | Tag: blume-20011121-foot-in-mouth | ||
| 1886 : | Description: | ||
| 1887 : | |||
| 1888 : | Removed the "Release_110_37" tag because of a serious bug. | ||
| 1889 : | This will be re-tagged once the bug is fixed. | ||
| 1890 : | |||
| 1891 : | ---------------------------------------------------------------------- | ||
| 1892 : | Name: Matthias Blume | ||
| 1893 : | blume | 987 | Date: 2001/11/21 16:14:00 EST |
| 1894 : | Tag: blume-20011121-forgottenfile | ||
| 1895 : | Description: | ||
| 1896 : | |||
| 1897 : | Forgot to add a file. (Just a .tex-file -- part of | ||
| 1898 : | the CM manual source.) | ||
| 1899 : | |||
| 1900 : | ---------------------------------------------------------------------- | ||
| 1901 : | Name: Matthias Blume | ||
| 1902 : | blume | 986 | Date: 2001/11/21 16:10:00 EST |
| 1903 : | blume | 988 | Tag: blume-20011121-invalid_110_37 |
| 1904 : | blume | 986 | Description: |
| 1905 : | |||
| 1906 : | blume | 988 | Note: I removed the original tag "Release_110_37" from this commit |
| 1907 : | because we found a serious bug in all non-x86 backends. | ||
| 1908 : | - Matthias | ||
| 1909 : | |||
| 1910 : | blume | 986 | 1. Modifications to the SML/NJ code generator and to the runtime system |
| 1911 : | so that code object name strings are directly inserted into code | ||
| 1912 : | objects at code generation time. The only business the runtime system | ||
| 1913 : | has with this is now to read the name strings on occasions. | ||
| 1914 : | (The encoding of the name string has also changed somewhat.) | ||
| 1915 : | |||
| 1916 : | 2. CM now implements a simple "set calculus" for specifying export lists. | ||
| 1917 : | In particular, it is now possible to refer to the export lists of | ||
| 1918 : | other libraries/groups/sources and form unions as well as differences. | ||
| 1919 : | See the latest CM manual for details. | ||
| 1920 : | |||
| 1921 : | 3. An separate notion of "proxy" libraries has again be eliminated from | ||
| 1922 : | CM's model. (Proxy libraries are now simply a special case of using | ||
| 1923 : | the export list calculus.) | ||
| 1924 : | |||
| 1925 : | 4. Some of the existing libraries now take advantage of the new set | ||
| 1926 : | calculus. | ||
| 1927 : | (Notice that not all libraries have been converted because some | ||
| 1928 : | of the existing .cm-files are supposed to be backward compatible | ||
| 1929 : | with 110.0.x.) | ||
| 1930 : | |||
| 1931 : | 5. Some cleanup in stand-alone programs. (Don't use "exnMessage" -- use | ||
| 1932 : | "General.exnMessage"! The former relies on a certain hook to be | ||
| 1933 : | initialized, and that often does not happen in the stand-alone case.) | ||
| 1934 : | |||
| 1935 : | ---------------------------------------------------------------------- | ||
| 1936 : | george | 984 | Name: Lal George |
| 1937 : | Date: 2001/11/21 13:56:18 EST | ||
| 1938 : | Tag: george-2001121-pseudo-ops | ||
| 1939 : | Description: | ||
| 1940 : | |||
| 1941 : | Implemented a complete redesign of MLRISC pseudo-ops. Now there | ||
| 1942 : | ought to never be any question of incompatabilities with | ||
| 1943 : | pseudo-op syntax expected by host assemblers. | ||
| 1944 : | |||
| 1945 : | For now, only modules supporting GAS syntax are implemented | ||
| 1946 : | but more should follow, such as MASM, and vendor assembler | ||
| 1947 : | syntax, e.g. IBM as, Sun as, etc. | ||
| 1948 : | |||
| 1949 : | ---------------------------------------------------------------------- | ||
| 1950 : | blume | 977 | Name: Matthias Blume |
| 1951 : | Date: 2001/11/14 11:52:00 EST | ||
| 1952 : | Tag: blume-20011114-srcname | ||
| 1953 : | Description: | ||
| 1954 : | |||
| 1955 : | 1. Routed the name of the current source file to mlriscgen where it | ||
| 1956 : | should be directly emitted into the code object. (This last part | ||
| 1957 : | is yet to be done.) | ||
| 1958 : | |||
| 1959 : | 2. Some cleanup of the pgraph code to make it match the proposal that | ||
| 1960 : | I put out the other day. (The proposal notwithstanding, things are | ||
| 1961 : | still in flux here.) | ||
| 1962 : | |||
| 1963 : | ---------------------------------------------------------------------- | ||
| 1964 : | george | 976 | Name: Lal George |
| 1965 : | Date: 2001/11/14 09:44:04 EST | ||
| 1966 : | Tag: | ||
| 1967 : | Description: | ||
| 1968 : | |||
| 1969 : | Fix for a backpatching bug reported by Allen. | ||
| 1970 : | |||
| 1971 : | Because the boundary between short and long span-dependent | ||
| 1972 : | instructions is +/- 128, there are an astounding number of | ||
| 1973 : | span-dependent instructions whose size is over estimated. | ||
| 1974 : | |||
| 1975 : | Allen came up with the idea of letting the size of span | ||
| 1976 : | dependent instructions be non-monotonic, for a maxIter | ||
| 1977 : | number of times, after which the size must be monotonically | ||
| 1978 : | increasing. | ||
| 1979 : | |||
| 1980 : | This table shows the number of span-dependent instructions | ||
| 1981 : | whose size was over-estimated as a function of maxIter, for the | ||
| 1982 : | file Parse/parse/ml.grm.sml: | ||
| 1983 : | |||
| 1984 : | maxIter # of instructions: | ||
| 1985 : | 10 687 | ||
| 1986 : | 20 438 | ||
| 1987 : | 30 198 | ||
| 1988 : | 40 0 | ||
| 1989 : | |||
| 1990 : | In compiling the compiler, there is no significant difference in | ||
| 1991 : | compilation speed between maxIter=10 and maxIter=40. Actually, | ||
| 1992 : | my measurements showed that maxIter=40 was a tad faster than | ||
| 1993 : | maxIter=10! Also 96% of the files in the compiler reach a fix | ||
| 1994 : | point within 13 iterations, so fixing maxIter at 40, while high, | ||
| 1995 : | is okay. | ||
| 1996 : | |||
| 1997 : | ---------------------------------------------------------------------- | ||
| 1998 : | blume | 975 | Name: Matthias Blume |
| 1999 : | Date: 2001/10/31 15:25:00 EST | ||
| 2000 : | Tag: blume-20011031-pgraph | ||
| 2001 : | Description: | ||
| 2002 : | |||
| 2003 : | CKIT: | ||
| 2004 : | * Changed the "Function" constructor of type Ast.ctype to carry optional | ||
| 2005 : | argument identifiers. | ||
| 2006 : | * Changed the return type of TypeUtil.getFunction accordingly. | ||
| 2007 : | * Type equality ignores the argument names. | ||
| 2008 : | * TypeUtil.composite tries to preserve argument names but gives up quickly | ||
| 2009 : | if there is a mismatch. | ||
| 2010 : | |||
| 2011 : | installation script: | ||
| 2012 : | * attempts to use "curl" if available (unless "wget" is available as well) | ||
| 2013 : | |||
| 2014 : | CM: | ||
| 2015 : | * has an experimental implementation of "portable graphs" which I will | ||
| 2016 : | soon propose as an implementation-independent library format | ||
| 2017 : | * there are also new libraries $/pgraph.cm and $/pgraph-util.cm | ||
| 2018 : | |||
| 2019 : | NLFFI-LIB: | ||
| 2020 : | * some cleanup (all cosmetic) | ||
| 2021 : | |||
| 2022 : | NLFFIGEN: | ||
| 2023 : | * temporarily disabled the mechanism that suppresses ML output for | ||
| 2024 : | C definitions whose identifiers start with an underscore character | ||
| 2025 : | * generate val bindings for enum constants | ||
| 2026 : | * user can request that only one style (light or heavy) is being used; | ||
| 2027 : | default is to use both (command-line arguments: -heavy and -light) | ||
| 2028 : | * fixed bug in handling of function types involving incomplete pointers | ||
| 2029 : | * generate ML entry points that take record arguments (i.e., using | ||
| 2030 : | named arguments) for C functions that have a prototype with named | ||
| 2031 : | arguments | ||
| 2032 : | (see changes to CKIT) | ||
| 2033 : | |||
| 2034 : | ---------------------------------------------------------------------- | ||
| 2035 : | leunga | 972 | Name: Allen Leung |
| 2036 : | Date: 2001/10/27 20:34:00 EDT | ||
| 2037 : | Tag: leunga-20011027-x86-fast-fp-call | ||
| 2038 : | Description: | ||
| 2039 : | |||
| 2040 : | Fixed the bug described in blume-20010920-slowfp. | ||
| 2041 : | |||
| 2042 : | The fix involves | ||
| 2043 : | 1. generating FCOPYs in FSTP in ia32-svid | ||
| 2044 : | 2. marking a CALL with the appropriate annotation | ||
| 2045 : | |||
| 2046 : | ---------------------------------------------------------------------- | ||
| 2047 : | blume | 961 | Name: Matthias Blume |
| 2048 : | Date: 2001/10/16 11:32:00 EDT | ||
| 2049 : | Tag: blume-20011016-netbsd | ||
| 2050 : | Description: | ||
| 2051 : | |||
| 2052 : | Underscore patch from Chris Richards (fixing problem with compiling | ||
| 2053 : | runtime system under recent NetBSD). | ||
| 2054 : | |||
| 2055 : | ---------------------------------------------------------------------- | ||
| 2056 : | george | 959 | Name: Allen Leung |
| 2057 : | Date: 2001/10/12 17:18:32 EDT 2001 | ||
| 2058 : | Tag: leung-20011012-x86-printflowgraph | ||
| 2059 : | Description: | ||
| 2060 : | |||
| 2061 : | X86RA now uses a valid (instead of dummy) PrintFlowgraph module. | ||
| 2062 : | |||
| 2063 : | ---------------------------------------------------------------------- | ||
| 2064 : | george | 952 | Name: Lal George |
| 2065 : | george | 958 | Date: 2001/10/11 23:51:34 EDT |
| 2066 : | Tag: george-20011011-too-many-instrs | ||
| 2067 : | Description: | ||
| 2068 : | |||
| 2069 : | The representation of a program point never expected to see more | ||
| 2070 : | than 65536 instructions in a basic block! | ||
| 2071 : | |||
| 2072 : | ---------------------------------------------------------------------- | ||
| 2073 : | Name: Lal George | ||
| 2074 : | george | 952 | Date: 2001/10/09 09:41:37 EDT |
| 2075 : | Tag: george-20011008-mlrisc-labels | ||
| 2076 : | Description: | ||
| 2077 : | |||
| 2078 : | Changed the machine description files to support printing of | ||
| 2079 : | local and global labels in assembly code, based on host assembler | ||
| 2080 : | conventions. | ||
| 2081 : | |||
| 2082 : | ---------------------------------------------------------------------- | ||
| 2083 : | blume | 916 | Name: Matthias Blume |
| 2084 : | blume | 939 | Date: 2001/09/25 15:25:00 EDT |
| 2085 : | Tag: blume-20010925-exninfo | ||
| 2086 : | Description: | ||
| 2087 : | |||
| 2088 : | I provided a non-hook implementation of exnName (at the toplevel) and | ||
| 2089 : | made the "dummy" implementation of exnMessage (at the toplevel) more | ||
| 2090 : | useful: if nothing gets "hooked in", then at least you are going to | ||
| 2091 : | see the exception name and a message indicating why you don't see more. | ||
| 2092 : | |||
| 2093 : | [For the time being, programs that need exnMessage and want to use | ||
| 2094 : | ml-build should either use General.exnMessage (strongly recommended) or | ||
| 2095 : | refer to structure General at some other point so that CM sees a | ||
| 2096 : | static dependency.] | ||
| 2097 : | |||
| 2098 : | [Similar remarks go for "print" and "use": If you want to use their | ||
| 2099 : | functionality in stand-alone programs generated by ml-build, then use | ||
| 2100 : | TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).] | ||
| 2101 : | |||
| 2102 : | ---------------------------------------------------------------------- | ||
| 2103 : | Name: Matthias Blume | ||
| 2104 : | blume | 938 | Date: 2001/09/20 17:28:00 EDT |
| 2105 : | Tag: blume-20010920-slowfp | ||
| 2106 : | Description: | ||
| 2107 : | |||
| 2108 : | Allen says that x86-fast-fp is not safe yet, so I turned it off again... | ||
| 2109 : | |||
| 2110 : | ---------------------------------------------------------------------- | ||
| 2111 : | Name: Matthias Blume | ||
| 2112 : | blume | 937 | Date: 2001/09/20 17:20:00 EDT |
| 2113 : | Tag: blume-20010920-canonicalpaths | ||
| 2114 : | Description: | ||
| 2115 : | |||
| 2116 : | 0. Updated the BOOT file (something that I forgot to do earlier). | ||
| 2117 : | |||
| 2118 : | 1. Small internal change to CM so that it avoids "/../" in filenames | ||
| 2119 : | as much as possible (but only where it is safe). | ||
| 2120 : | |||
| 2121 : | 2. Changed config/_run-sml (resulting in a changed bin/.run-sml) so | ||
| 2122 : | that arguments that contain delimiters are passed through correctly. | ||
| 2123 : | This change also means that all "special" arguments of the form | ||
| 2124 : | @SMLxxx... must come first. | ||
| 2125 : | |||
| 2126 : | 3. Changed install script to put relative anchor names for tool commands | ||
| 2127 : | into pathconfig. | ||
| 2128 : | |||
| 2129 : | ---------------------------------------------------------------------- | ||
| 2130 : | Name: Matthias Blume | ||
| 2131 : | blume | 932 | Date: 2001/09/18 15:35:00 EDT |
| 2132 : | Tag: blume-20010918-readme11036 | ||
| 2133 : | Description: | ||
| 2134 : | |||
| 2135 : | Added README files. | ||
| 2136 : | |||
| 2137 : | ---------------------------------------------------------------------- | ||
| 2138 : | Name: Matthias Blume | ||
| 2139 : | blume | 929 | Date: 2001/09/18 11:45:00 EDT |
| 2140 : | Tag: Release_110_36 (retag) | ||
| 2141 : | Description: | ||
| 2142 : | |||
| 2143 : | Fixed mistake in config/preloads. Retagged as 110.36. | ||
| 2144 : | |||
| 2145 : | ---------------------------------------------------------------------- | ||
| 2146 : | Name: Matthias Blume | ||
| 2147 : | blume | 928 | Date: 2001/09/18 09:40:00 EDT |
| 2148 : | blume | 929 | Tag: Release_110_36_orig (tag changed) |
| 2149 : | blume | 928 | Description: |
| 2150 : | |||
| 2151 : | New version (110.36). New bootfiles. | ||
| 2152 : | |||
| 2153 : | ---------------------------------------------------------------------- | ||
| 2154 : | Name: Matthias Blume | ||
| 2155 : | blume | 927 | Date: 2001/09/14 16:15:00 EDT |
| 2156 : | Tag: blume-20010914-x86fastfp | ||
| 2157 : | Description: | ||
| 2158 : | |||
| 2159 : | John committed some changes that Allen made, in particular a (hopefully) | ||
| 2160 : | correctly working version of the x86-fp module. | ||
| 2161 : | |||
| 2162 : | I changed the default setting of the Control.MLRISC.getFlag "x86-fast-fp" | ||
| 2163 : | flag to "true". Everything seems to compile to a fixpoint ok, and | ||
| 2164 : | "mandelbrot" speeds up by about 15%. | ||
| 2165 : | |||
| 2166 : | ---------------------------------------------------------------------- | ||
| 2167 : | Name: Matthias Blume | ||
| 2168 : | blume | 923 | Date: 2001/09/13 11:20:00 EDT |
| 2169 : | Tag: blume-20010913-minimal | ||
| 2170 : | Description: | ||
| 2171 : | |||
| 2172 : | 1. Stefan Monnier's patch to fix a miscompilation problem that | ||
| 2173 : | was brought to light by John Reppy's work on Moby. | ||
| 2174 : | |||
| 2175 : | 2. Implemented a minimal "structure Compiler" that contains just | ||
| 2176 : | "version" and "architecture". The minimal version will be | ||
| 2177 : | available when the full version is not. This is for backward- | ||
| 2178 : | compatibility with code that wants to test Compiler.version. | ||
| 2179 : | |||
| 2180 : | ---------------------------------------------------------------------- | ||
| 2181 : | Name: Matthias Blume | ||
| 2182 : | blume | 920 | Date: 2001/08/28 14:03:00 EDT |
| 2183 : | Tag: blume-20010828-ml-lex | ||
| 2184 : | Description: | ||
| 2185 : | |||
| 2186 : | Fix for bug 1581, received from Neophytos Michael. | ||
| 2187 : | |||
| 2188 : | ---------------------------------------------------------------------- | ||
| 2189 : | Name: Matthias Blume | ||
| 2190 : | blume | 918 | Date: 2001/08/27 11:20:00 EDT |
| 2191 : | Tag: blume-20010827-readme11035 | ||
| 2192 : | Description: | ||
| 2193 : | |||
| 2194 : | Fleshed out the README file for 110.35. | ||
| 2195 : | |||
| 2196 : | ---------------------------------------------------------------------- | ||
| 2197 : | Name: Matthias Blume | ||
| 2198 : | blume | 916 | Date: 2001/08/24 17:10:00 EDT |
| 2199 : | Tag: Release_110_35 | ||
| 2200 : | Description: | ||
| 2201 : | |||
| 2202 : | New version number (110.35). New bootfiles. | ||
| 2203 : | |||
| 2204 : | ---------------------------------------------------------------------- | ||
| 2205 : | george | 915 | Name: Lal George |
| 2206 : | george | 909 | Date: 2001/08/24 13:47:18 EDT 2001 |
| 2207 : | Tag: george-20010824-MLRISC-graphs | ||
| 2208 : | Description: | ||
| 2209 : | |||
| 2210 : | removed clusters from MLRISC completely and replaced with graphs. | ||
| 2211 : | |||
| 2212 : | ---------------------------------------------------------------------- | ||
| 2213 : | Name: Matthias Blume | ||
| 2214 : | blume | 905 | Date: 2001/08/23 17:50:00 EDT |
| 2215 : | Tag: blume-20010823-toplevel | ||
| 2216 : | Description: | ||
| 2217 : | |||
| 2218 : | - some reorganization of the code that implements various kinds of | ||
| 2219 : | environments in the compiler (static, dynamic, symbolic, combined) | ||
| 2220 : | - re-implemented the EnvRef module so that evalStream works properly | ||
| 2221 : | (if the stream contains references to "use", "CM.make", etc.) | ||
| 2222 : | - cleaned up evalloop.sml and interact.sml (but they need more cleaning) | ||
| 2223 : | |||
| 2224 : | ---------------------------------------------------------------------- | ||
| 2225 : | Name: Matthias Blume | ||
| 2226 : | blume | 904 | Date: 2001/08/20 15:50 EDT |
| 2227 : | Tag: blume20010820-slipup | ||
| 2228 : | Description: | ||
| 2229 : | |||
| 2230 : | I forgot to commit a few files. Here they are... | ||
| 2231 : | |||
| 2232 : | ---------------------------------------------------------------------- | ||
| 2233 : | Name: Matthias Blume | ||
| 2234 : | blume | 903 | Date: 2001/08/20 15:35:00 EDT |
| 2235 : | Tag: blume-20010820-debugprof | ||
| 2236 : | Description: | ||
| 2237 : | |||
| 2238 : | !!!! NEW BOOTFILES !!!! | ||
| 2239 : | |||
| 2240 : | This is another round of reorganizing the compiler sources. This | ||
| 2241 : | time the main goal was to factor out all the "instrumentation" | ||
| 2242 : | passes (for profiling and backtracing) into their own library. | ||
| 2243 : | The difficulty was to do it in such a way that it does not depend | ||
| 2244 : | on elaborate.cm but only on elabdata.cm. | ||
| 2245 : | |||
| 2246 : | Therefore there have been further changes to both elaborate.cm and | ||
| 2247 : | elabdata.cm -- more "generic" things have been moved from the former | ||
| 2248 : | to the latter. As a result, I was forced to split the assignment | ||
| 2249 : | of numbers indicating "primtyc"s into two portions: SML-generic and | ||
| 2250 : | SML/NJ-specific. Since it would have been awkward to maintain, | ||
| 2251 : | I bit the bullet and actually _changed_ the mapping between these | ||
| 2252 : | numbers and primtycs. The bottom line of this is that you need | ||
| 2253 : | a new set of bin- and bootfiles. | ||
| 2254 : | |||
| 2255 : | I have built new bootfiles for all architectures, so doing a fresh | ||
| 2256 : | checkout and config/install.sh should be all you need. | ||
| 2257 : | |||
| 2258 : | The newly created library's name is | ||
| 2259 : | |||
| 2260 : | $smlnj/viscomp/debugprof.cm | ||
| 2261 : | |||
| 2262 : | and its sources live under | ||
| 2263 : | |||
| 2264 : | src/compiler/DebugProf | ||
| 2265 : | |||
| 2266 : | ---------------------------------------------------------------------- | ||
| 2267 : | Name: Matthias Blume | ||
| 2268 : | blume | 902 | Date: 2001/08/15 17:15:00 EDT |
| 2269 : | Tag: blume-20010815-compreorg | ||
| 2270 : | Description: | ||
| 2271 : | |||
| 2272 : | This is a first cut at reorganizing the CM libraries that make up the | ||
| 2273 : | core of the compiler. The idea is to separate out pieces that could | ||
| 2274 : | be used independently by tools, e.g., the parser, the typechecker, etc. | ||
| 2275 : | |||
| 2276 : | The current status is a step in this direction, but it is not quite | ||
| 2277 : | satisfactory yet. Expect more changes in the future. | ||
| 2278 : | |||
| 2279 : | Here is the current (new) organization... | ||
| 2280 : | |||
| 2281 : | What used to be $smlnj/viscomp/core.cm is now divided into | ||
| 2282 : | six CM libraries: | ||
| 2283 : | |||
| 2284 : | $smlnj/viscomp/basics.cm | ||
| 2285 : | /parser.cm | ||
| 2286 : | /elabdata.cm | ||
| 2287 : | /elaborate.cm | ||
| 2288 : | /execute.cm | ||
| 2289 : | /core.cm | ||
| 2290 : | |||
| 2291 : | The CM files for these libraries live under src/system/smlnj/viscomp. | ||
| 2292 : | All these libraries are proxy libraries that contain precisely | ||
| 2293 : | one CM library component. Here are the locations of the components | ||
| 2294 : | (all within the src/compiler tree): | ||
| 2295 : | |||
| 2296 : | Basics/basics.cm | ||
| 2297 : | Parse/parser.cm | ||
| 2298 : | ElabData/elabdata.cm | ||
| 2299 : | Elaborator/elaborate.cm | ||
| 2300 : | Execution/execute.cm | ||
| 2301 : | core.cm | ||
| 2302 : | |||
| 2303 : | [This organization is the same that has been used already | ||
| 2304 : | for a while for the architecture-specific parts of the visible | ||
| 2305 : | compiler and for the old version of core.cm.] | ||
| 2306 : | |||
| 2307 : | As you will notice, many source files have been moved from their | ||
| 2308 : | respective original locations to a new home in one of the above | ||
| 2309 : | subtrees. | ||
| 2310 : | |||
| 2311 : | The division of labor between the new libraries is the following: | ||
| 2312 : | |||
| 2313 : | basics.cm: | ||
| 2314 : | - Simple, basic definitions that pertain to many (or all) of | ||
| 2315 : | the other libraries. | ||
| 2316 : | parser.cm: | ||
| 2317 : | - The SML parser, producing output of type Ast.dec. | ||
| 2318 : | - The type family for Ast is also defined and exported here. | ||
| 2319 : | elabdata.cm: | ||
| 2320 : | - The datatypes that describe input and output of the elaborator. | ||
| 2321 : | This includes types, absyn, and static environments. | ||
| 2322 : | elaborator.cm: | ||
| 2323 : | - The SML/NJ type checker and elaborator. | ||
| 2324 : | This maps an Ast.dec (with a given static environment) to | ||
| 2325 : | an Absyn.dec (with a new static environment). | ||
| 2326 : | - This libraries implements certain modules that used to be | ||
| 2327 : | structures as functors (to remove dependencies on FLINT). | ||
| 2328 : | execute.cm: | ||
| 2329 : | - Everything having to do with executing binary code objects. | ||
| 2330 : | - Dynamic environments. | ||
| 2331 : | core.cm: | ||
| 2332 : | - SML/NJ-specific instantiations of the elaborator and MLRISC. | ||
| 2333 : | - Top-level modules. | ||
| 2334 : | - FLINT (this should eventually become its own library) | ||
| 2335 : | |||
| 2336 : | Notes: | ||
| 2337 : | |||
| 2338 : | I am not 100% happy with the way I separated the elaborator (and its | ||
| 2339 : | data structures) from FLINT. Two instances of the same problem: | ||
| 2340 : | |||
| 2341 : | 1. Data structures contain certain fields that carry FLINT-specific | ||
| 2342 : | information. I hacked around this using exn and the property list | ||
| 2343 : | module from smlnj-lib. But the fact that there are middle-end | ||
| 2344 : | specific fields around at all is a bit annoying. | ||
| 2345 : | |||
| 2346 : | 2. The elaborator calculates certain FLINT-related information. I tried | ||
| 2347 : | to make this as abstract as I could using functorization, but, again, | ||
| 2348 : | the fact that the elaborator has to perform calculations on behalf | ||
| 2349 : | of the middle-end at all is not nice. | ||
| 2350 : | |||
| 2351 : | 3. Having to used exn and property lists is unfortunate because it | ||
| 2352 : | weakens type checking. The other alternative (parameterizing | ||
| 2353 : | nearly *everything*) is not appealing, though. | ||
| 2354 : | |||
| 2355 : | I removed the "rebinding =" warning hack because due to the new organization | ||
| 2356 : | it was awkward to maintain it. As a result, the compiler now issues some of | ||
| 2357 : | these warnings when compiling init.cmi during bootstrap compilation. On | ||
| 2358 : | the plus side, you also get a warning when you do, for example: | ||
| 2359 : | val op = = Int32.+ | ||
| 2360 : | which was not the case up to now. | ||
| 2361 : | |||
| 2362 : | I placed "assign" and "deref" into the _Core structure so that the | ||
| 2363 : | code that deals with the "lazy" keyword can find them there. This | ||
| 2364 : | removes the need for having access to the primitive environment | ||
| 2365 : | during elaboration. | ||
| 2366 : | |||
| 2367 : | ---------------------------------------------------------------------- | ||
| 2368 : | Name: Matthias Blume | ||
| 2369 : | blume | 898 | Date: 2001/08/13 |
| 2370 : | Tag: blume-20010813-closures | ||
| 2371 : | Description: | ||
| 2372 : | |||
| 2373 : | This fix was sent to us by Zhong Shao. It is supposed to improve the | ||
| 2374 : | performance of certain loops by avoiding needless closure allocation. | ||
| 2375 : | |||
| 2376 : | ---------------------------------------------------------------------- | ||
| 2377 : | george | 890 | Name: Lal George |
| 2378 : | george | 896 | Date: 2001/07/31 10:03:23 EDT 2001 |
| 2379 : | Tag: george-20010731-x86-fmalloc | ||
| 2380 : | Description: Fixed bug in x86 calls | ||
| 2381 : | |||
| 2382 : | There was a bug where call instructions would mysteriously | ||
| 2383 : | vanish. The call instruction had to be one that returned | ||
| 2384 : | a floating point value. | ||
| 2385 : | |||
| 2386 : | ---------------------------------------------------------------------- | ||
| 2387 : | Name: Lal George | ||
| 2388 : | george | 890 | Date: 2001/07/19 16:36:29 EDT 2001 |
| 2389 : | Tag: george-20010719-simple-cells | ||
| 2390 : | Description: | ||
| 2391 : | |||
| 2392 : | I have dramatically simplified the interface for CELLS in MLRISC. | ||
| 2393 : | |||
| 2394 : | In summary, the cells interface is broken up into three parts: | ||
| 2395 : | |||
| 2396 : | 1. CellsBasis : CELLS_BASIS | ||
| 2397 : | |||
| 2398 : | CellsBasis is a top level structure and common for all | ||
| 2399 : | architectures. it contains the definitions of basic datatypes | ||
| 2400 : | and utility functions over these types. | ||
| 2401 : | |||
| 2402 : | 2. functor Cells() : CELLS | ||
| 2403 : | |||
| 2404 : | Cells generates an interface for CELLS that incorporates the | ||
| 2405 : | specific resources on the target architecture, such as the | ||
| 2406 : | presence of special register classes, their number and size, | ||
| 2407 : | and various useful substructures. | ||
| 2408 : | |||
| 2409 : | 3. <ARCH>CELLS | ||
| 2410 : | |||
| 2411 : | e.g. SparcCells: SPARCCELLS | ||
| 2412 : | |||
| 2413 : | <ARCH>CELLS usually contains additional bindings for special | ||
| 2414 : | registers on the architecture, such as: | ||
| 2415 : | |||
| 2416 : | val r0 : cell (* register zero *) | ||
| 2417 : | val y : cell (* Y register *) | ||
| 2418 : | val psr : cell (* processor status register *) | ||
| 2419 : | ... | ||
| 2420 : | |||
| 2421 : | The structure returned by applying the Cells functor is opened | ||
| 2422 : | in this interface. | ||
| 2423 : | |||
| 2424 : | The main implication of all this is that the datatypes for cells is | ||
| 2425 : | split between CellsBasis and CELLS -- a fairly simple change for user | ||
| 2426 : | code. | ||
| 2427 : | |||
| 2428 : | In the old scheme the CELLS interface had a definitional binding of | ||
| 2429 : | the form: | ||
| 2430 : | |||
| 2431 : | signature CELLS = sig | ||
| 2432 : | |||
| 2433 : | structure CellsBasis = CellsBasis | ||
| 2434 : | |||
| 2435 : | ... | ||
| 2436 : | |||
| 2437 : | end | ||
| 2438 : | |||
| 2439 : | With all the sharing constraints that goes on in MLRISC, this old | ||
| 2440 : | design quickly leads to errors such as: | ||
| 2441 : | |||
| 2442 : | "structure definition spec inside of sharing ... " | ||
| 2443 : | |||
| 2444 : | |||
| 2445 : | and appears to require an unacceptable amount of sharing and where | ||
| 2446 : | constraint hackery. | ||
| 2447 : | |||
| 2448 : | I think this error message (the interaction of definitional specs and | ||
| 2449 : | sharing) requires more explanation on our web page. | ||
| 2450 : | |||
| 2451 : | ---------------------------------------------------------------------- | ||
| 2452 : | blume | 826 | Name: Matthias Blume |
| 2453 : | blume | 879 | Date: 2001/07/19 15:00:00 EDT |
| 2454 : | Tag: blume-20010719-libreorg | ||
| 2455 : | Description: | ||
| 2456 : | |||
| 2457 : | This update puts together a fairly extensive but straightforward change | ||
| 2458 : | to the way the libraries that implement the interactive system are | ||
| 2459 : | organized: | ||
| 2460 : | |||
| 2461 : | The biggest change is the elimination of structure Compiler. As a | ||
| 2462 : | replacement for this structure, there is now a CM library | ||
| 2463 : | (known as $smlnj/compiler.cm or $smlnj/compiler/current.cm) | ||
| 2464 : | that exports all the substructures of the original structure Compiler | ||
| 2465 : | directly. So instead of saying Compiler.Foo.bar one now simply | ||
| 2466 : | says Foo.bar. (The CM libraries actually export a collection of | ||
| 2467 : | structures that is richer than the collection of substructures of | ||
| 2468 : | structure Compiler.) | ||
| 2469 : | |||
| 2470 : | To make the transition smooth, there is a separate library called | ||
| 2471 : | $smlnj/compiler/compiler.cm which puts together and exports the | ||
| 2472 : | original structure Compiler (or at least something very close to it). | ||
| 2473 : | |||
| 2474 : | There are five members of the original structure Compiler | ||
| 2475 : | that are not exported directly but which instead became members | ||
| 2476 : | of a new structure Backend (described by signature BACKEND). These are: | ||
| 2477 : | structure Profile (: PROFILE), structure Compile (: COMPILE), structure | ||
| 2478 : | Interact (: INTERACT), structure Machine (: MACHINE), and val | ||
| 2479 : | architecture (: string). | ||
| 2480 : | |||
| 2481 : | Structure Compiler.Version has become structure CompilerVersion. | ||
| 2482 : | |||
| 2483 : | Cross-compilers for alpha32, hppa, ppc, sparc, and x86 are provided | ||
| 2484 : | by $smlnj/compiler/<arch>.cm where <arch> is alpha32, hppa, ppc, sparc, | ||
| 2485 : | or x86, respectively. | ||
| 2486 : | Each of these exports the same frontend structures that | ||
| 2487 : | $smlnj/compiler.cm exports. But they do not have a structure Backend | ||
| 2488 : | and instead export some structure <Arch>Backend where <Arch> is Alpha32, | ||
| 2489 : | Hppa, PPC, Sparc, or X86, respectively. | ||
| 2490 : | |||
| 2491 : | Library $smlnj/compiler/all.cm exports the union of the exports of | ||
| 2492 : | $smlnj/compiler/<arch>.cm | ||
| 2493 : | |||
| 2494 : | There are no structures <Arch>Compiler anymore, use | ||
| 2495 : | $smlnj/compiler/<arch>.cm instead. | ||
| 2496 : | |||
| 2497 : | Library host-compiler-0.cm is gone. Instead, the internal library | ||
| 2498 : | that instantiates CM is now called cm0.cm. Selection of the host | ||
| 2499 : | compiler (backend) is no longer done here but. (Responsibility for it | ||
| 2500 : | now lies with $smlnj/compiler/current.cm. This seems to be more | ||
| 2501 : | logical.) | ||
| 2502 : | |||
| 2503 : | Many individual files have been moved or renamed. Some files have | ||
| 2504 : | been split into multiple files, and some "dead" files have been deleted. | ||
| 2505 : | |||
| 2506 : | Aside from these changes to library organization, there are also changes | ||
| 2507 : | to the way the code itself is organized: | ||
| 2508 : | |||
| 2509 : | Structure Binfile has been re-implemented in such a way that it no | ||
| 2510 : | longer needs any knowledge of the compiler. It exclusively deals | ||
| 2511 : | with the details of binfile layout. It no longer invokes the | ||
| 2512 : | compiler (for the purpose of creating new prospective binfile | ||
| 2513 : | content), and it no longer has any knowledge of how to interpret | ||
| 2514 : | pickles. | ||
| 2515 : | |||
| 2516 : | Structure Compile (: COMPILE) has been stripped down to the bare | ||
| 2517 : | essentials of compilation. It no longer deals with linking/execution. | ||
| 2518 : | The interface has been cleaned up considerably. | ||
| 2519 : | |||
| 2520 : | Utility routines for dealing with linking and execution have been | ||
| 2521 : | moved into their own substructures. | ||
| 2522 : | |||
| 2523 : | (The ultimate goal of these changes is to provide a light-weight | ||
| 2524 : | binfile loader/linker (at least for, e.g., stable libraries) that | ||
| 2525 : | does not require CM or the compiler to be present.) | ||
| 2526 : | |||
| 2527 : | CM documentation has been updated to reflect the changes to library | ||
| 2528 : | organization. | ||
| 2529 : | |||
| 2530 : | ---------------------------------------------------------------------- | ||
| 2531 : | Name: Matthias Blume | ||
| 2532 : | blume | 875 | Date: 2001/07/10 17:30:00 EDT |
| 2533 : | Tag: Release_110_34 | ||
| 2534 : | Description: | ||
| 2535 : | |||
| 2536 : | blume | 876 | Minor tweak to 110.34 (re-tagged): |
| 2537 : | |||
| 2538 : | - README.html file added to CVS repository | ||
| 2539 : | - runtime compiles properly under FreeBSD 3.X and 4.X | ||
| 2540 : | |||
| 2541 : | ---------------------------------------------------------------------- | ||
| 2542 : | Name: Matthias Blume | ||
| 2543 : | Date: 2001/07/10 17:30:00 EDT | ||
| 2544 : | Tag: Release_110_34 | ||
| 2545 : | Description: | ||
| 2546 : | |||
| 2547 : | blume | 875 | New version number (110.34). New bootfiles. |
| 2548 : | |||
| 2549 : | ---------------------------------------------------------------------- | ||
| 2550 : | Name: Matthias Blume | ||
| 2551 : | blume | 874 | Date: 2001/07/09 16:00:00 EDT |
| 2552 : | Tag: blume-20010709-more-varargs | ||
| 2553 : | Description: | ||
| 2554 : | |||
| 2555 : | I changed the handling of varargs in ml-nlffigen again: | ||
| 2556 : | The ellipsis ... will now simply be ignored (with an accompanying warning). | ||
| 2557 : | |||
| 2558 : | The immediate effect is that you can actually call a varargs function | ||
| 2559 : | from ML -- but you can't actually supply any arguments beyond the ones | ||
| 2560 : | specified explicitly. (For example, you can call printf with its format | ||
| 2561 : | string, but you cannot pass additional arguments.) | ||
| 2562 : | |||
| 2563 : | This behavior is only marginally more useful than the one before, but | ||
| 2564 : | it has the advantage that a function or, more importantly, a function | ||
| 2565 : | type never gets dropped on the floor, thus avoiding follow-up problems with | ||
| 2566 : | other types that refer to the offending one. | ||
| 2567 : | |||
| 2568 : | ---------------------------------------------------------------------- | ||
| 2569 : | Name: Matthias Blume | ||
| 2570 : | blume | 873 | Date: 2001/07/09 11:25:00 EDT |
| 2571 : | Tag: blume-20010709-varargs | ||
| 2572 : | Description: | ||
| 2573 : | |||
| 2574 : | 1. ckit-lib.cm now exports structure Error | ||
| 2575 : | 2. ml-nlffigen reports occurences of "..." (i.e., varargs function types) | ||
| 2576 : | with a warning accompanied by a source location. Moreover, it | ||
| 2577 : | merely skips the offending function or type and proceeds with the | ||
| 2578 : | rest of its work.u As a result, one can safely feed C code containing | ||
| 2579 : | "..." to ml-nlffigen. | ||
| 2580 : | 3. There are some internal improvements to CM, providing slightly | ||
| 2581 : | more general string substitutions in the tools subsystem. | ||
| 2582 : | |||
| 2583 : | ---------------------------------------------------------------------- | ||
| 2584 : | Name: Matthias Blume | ||
| 2585 : | blume | 854 | Date: 2001/06/27 15:10:00 EDT |
| 2586 : | Tag: blume-20010627-concur | ||
| 2587 : | Description: | ||
| 2588 : | |||
| 2589 : | Fixed a small bug in CM's handling of parallel compilation. | ||
| 2590 : | (You could observe the bug by Control-C-interrupting an ordinary | ||
| 2591 : | CMB.make or CM.stabilize and then attaching some compile servers. | ||
| 2592 : | The result was that all of a sudden the previously interrupted | ||
| 2593 : | compilation would continue on its own. This was because of | ||
| 2594 : | an over-optimization: CM did not bother to clean out certain queues | ||
| 2595 : | when no servers were attached "anyway", resulting in the contents | ||
| 2596 : | of these queues to grab control when new servers did get attached.) | ||
| 2597 : | |||
| 2598 : | There is also another minor update to the CM manual. | ||
| 2599 : | |||
| 2600 : | ---------------------------------------------------------------------- | ||
| 2601 : | Name: Matthias Blume | ||
| 2602 : | blume | 853 | Date: 2001/06/26 16:15:00 EDT |
| 2603 : | Tag: blume-20010626-cmdoc | ||
| 2604 : | Description: | ||
| 2605 : | |||
| 2606 : | Minor typo fixed in CM manual (syntax diagram for libraries). | ||
| 2607 : | |||
| 2608 : | ---------------------------------------------------------------------- | ||
| 2609 : | Name: Matthias Blume | ||
| 2610 : | blume | 852 | Date: 2001/06/25 22:55:00 EDT |
| 2611 : | Tag: blume-20010625-x86pc | ||
| 2612 : | Description: | ||
| 2613 : | |||
| 2614 : | Fixed a nasty bug in the X86 assembly code that caused signal | ||
| 2615 : | handlers to fail (crash) randomly. | ||
| 2616 : | |||
| 2617 : | ---------------------------------------------------------------------- | ||
| 2618 : | Name: Matthias Blume | ||
| 2619 : | blume | 846 | Date: 2001/06/25 12:05:00 EDT |
| 2620 : | Tag: blume-20010625-nlffigen | ||
| 2621 : | Description: | ||
| 2622 : | |||
| 2623 : | This update fixes a number of minor bugs in ml-nlffigen as reported by | ||
| 2624 : | Nick Carter <nbc@andrew.cmu.edu>. | ||
| 2625 : | |||
| 2626 : | 1. Silly but ok typedefs of the form "typedef void myvoid;" are now accepted. | ||
| 2627 : | 2. Default names for generated files are now derived from the name of | ||
| 2628 : | the C file *without its directory*. In particular, this causes generated | ||
| 2629 : | files to be placed locally even if the C file is in some system directory. | ||
| 2630 : | 3. Default names for generated signatures and structures are also derived | ||
| 2631 : | from the C file name without its directory. This avoids silly things | ||
| 2632 : | like "structure GL/GL". | ||
| 2633 : | (Other silly names are still possible because ml-nlffigen does not do | ||
| 2634 : | blume | 847 | a thorough check of whether generated names are legal ML identifiers. |
| 2635 : | When in doubt, use command line arguments to force particular names.) | ||
| 2636 : | blume | 846 | |
| 2637 : | ---------------------------------------------------------------------- | ||
| 2638 : | Name: Matthias Blume | ||
| 2639 : | blume | 845 | Date: 2001/06/21 12:25:00 EDT |
| 2640 : | Tag: blume-20010621-eXene | ||
| 2641 : | Description: | ||
| 2642 : | |||
| 2643 : | eXene now compiles and (sort of) works again. | ||
| 2644 : | |||
| 2645 : | The library name (for version > 110.33) is $/eXene.cm. | ||
| 2646 : | |||
| 2647 : | I also added an new example in src/eXene/examples/nbody. See the | ||
| 2648 : | README file there for details. | ||
| 2649 : | |||
| 2650 : | ---------------------------------------------------------------------- | ||
| 2651 : | Name: Matthias Blume | ||
| 2652 : | blume | 844 | Date: 2001/06/20 16:40:00 EDT |
| 2653 : | Tag: blume-20010620-cml | ||
| 2654 : | Description: | ||
| 2655 : | |||
| 2656 : | CML now compiles and works again. | ||
| 2657 : | |||
| 2658 : | Libraries (for version > 110.33): | ||
| 2659 : | |||
| 2660 : | $cml/cml.cm Main CML library. | ||
| 2661 : | $cml/basis.cm CML's version of $/basis.cm. | ||
| 2662 : | $cml/cml-internal.cm Internal helper library. | ||
| 2663 : | $cml/core-cml.cm Internal helper library. | ||
| 2664 : | $cml-lib/trace-cml.cm Tracing facility. | ||
| 2665 : | $cml-lib/smlnj-lib.cm CML's version of $/smlnj-lib.cm | ||
| 2666 : | |||
| 2667 : | The installer (config/install.sh) has been taught how to properly | ||
| 2668 : | install this stuff. | ||
| 2669 : | |||
| 2670 : | ---------------------------------------------------------------------- | ||
| 2671 : | Name: Matthias Blume | ||
| 2672 : | blume | 843 | Date: 2001/06/19 17:55:00 EDT |
| 2673 : | Tag: blume-20010619-instantiate | ||
| 2674 : | Description: | ||
| 2675 : | |||
| 2676 : | This un-breaks the fix for bug 1432. | ||
| 2677 : | (The bug was originally fixed in 110.9 but I broke it again some | ||
| 2678 : | time after that.) | ||
| 2679 : | |||
| 2680 : | ---------------------------------------------------------------------- | ||
| 2681 : | Name: Matthias Blume | ||
| 2682 : | blume | 842 | Date: 2001/06/19 17:25:00 EDT |
| 2683 : | Tag: blume-20010619-signals | ||
| 2684 : | Description: | ||
| 2685 : | |||
| 2686 : | This should (hopefully) fix the long-standing signal handling bug. | ||
| 2687 : | (The runtime system was constructing a continuation record with an | ||
| 2688 : | incorrect descriptor which would cause the GC to drop data on the floor...) | ||
| 2689 : | |||
| 2690 : | ---------------------------------------------------------------------- | ||
| 2691 : | Name: Matthias Blume | ||
| 2692 : | blume | 840 | Date: 2001/06/15 15:05:00 EDT |
| 2693 : | blume | 841 | Tag: blume-20010615-moresparc |
| 2694 : | Description: | ||
| 2695 : | |||
| 2696 : | Here is a short late-hour update related to Sparc c-calls: | ||
| 2697 : | |||
| 2698 : | -- made handling of double-word arguments a bit smarter | ||
| 2699 : | |||
| 2700 : | -- instruction selection phase tries to collapse certain clumsily | ||
| 2701 : | constructed ML-Trees; typical example: | ||
| 2702 : | |||
| 2703 : | ADD(ty,ADD(_,e,LI d1),LI d2) -> ADD(ty,e,LI(d1+d2)) | ||
| 2704 : | |||
| 2705 : | This currently has no further impact on SML/NJ since mlriscGen does | ||
| 2706 : | not seem to generate such patterns in the first place, and c-calls | ||
| 2707 : | (which did generate them in the beginning) has meanwhile been fixed | ||
| 2708 : | so as to avoid them as well. | ||
| 2709 : | |||
| 2710 : | ---------------------------------------------------------------------- | ||
| 2711 : | Name: Matthias Blume | ||
| 2712 : | Date: 2001/06/15 15:05:00 EDT | ||
| 2713 : | blume | 840 | Tag: blume-20010615-sparc |
| 2714 : | Description: | ||
| 2715 : | |||
| 2716 : | The purpose of this update is to provide an implementation of NLFFI | ||
| 2717 : | on Sparc machines. | ||
| 2718 : | |||
| 2719 : | Here are the changes in detail: | ||
| 2720 : | |||
| 2721 : | * src/MLRISC/sparc/c-calls/sparc-c-calls.sml is a new file containing | ||
| 2722 : | the Sparc implementation of the c-calls API. | ||
| 2723 : | * The Sparc backend of SML/NJ has been modified to uniformely use %fp | ||
| 2724 : | for accessing the ML frame. Thus, we have a real frame pointer and | ||
| 2725 : | can freely modify %sp without need for an omit-frame-ptr phase. | ||
| 2726 : | The vfp logic in src/compiler/CodeGen/* has been changed to accomodate | ||
| 2727 : | this case. | ||
| 2728 : | * ml-nlffigen has been taught to produce code for different architectures | ||
| 2729 : | and calling conventions. | ||
| 2730 : | * In a way similar to what was done in the x86 case, the Sparc | ||
| 2731 : | backend uses its own specific extension to mltree. (For example, | ||
| 2732 : | it needs to be able to generate UNIMP instructions which are part | ||
| 2733 : | of the calling convention.) | ||
| 2734 : | * ml-nlffi-lib was reorganized to make it more modular (in particular, | ||
| 2735 : | to make it easier to plug in new machine- and os-dependent parts). | ||
| 2736 : | |||
| 2737 : | There are some other fairly unrelated bug fixes and cleanups as well: | ||
| 2738 : | |||
| 2739 : | * I further hacked the .cm files for MLRISC tools (like MDLGen) so | ||
| 2740 : | that they properly share their libraries with existing SML/NJ libraries. | ||
| 2741 : | * I fixed a minor cosmetic bug in CM, supressing certain spurious | ||
| 2742 : | follow-up error messages. | ||
| 2743 : | * Updates to CM/CMB documentation. | ||
| 2744 : | |||
| 2745 : | TODO items: | ||
| 2746 : | |||
| 2747 : | * MLRISC should use a different register as its asmTemp on the Sparc. | ||
| 2748 : | (The current %o2 is a really bad choice because it is part of the | ||
| 2749 : | calling conventions, so things might interfere in unexpected ways.) | ||
| 2750 : | |||
| 2751 : | ---------------------------------------------------------------------- | ||
| 2752 : | Name: Matthias Blume | ||
| 2753 : | blume | 839 | Date: 2001/06/07 |
| 2754 : | Tag: blume-20010607-calls | ||
| 2755 : | Description: | ||
| 2756 : | |||
| 2757 : | A number of internal changes related to C calls and calling conventions: | ||
| 2758 : | |||
| 2759 : | 1. ML-Tree CALL statements now carry a "pops" field. It indicates the | ||
| 2760 : | number of bytes popped implicitly (by the callee). In most cases | ||
| 2761 : | this field is 0 but on x86/win32 it is some non-zero value. This | ||
| 2762 : | is information provided for the benefit of the "omit-frameptr" pass. | ||
| 2763 : | 2. The CALL instruction on the x86 carries a similar "pops" field. | ||
| 2764 : | The instruction selection phase copies its value from the ML-Tree | ||
| 2765 : | CALL statement. | ||
| 2766 : | 3. On all other architectures, the instruction selection phase checks | ||
| 2767 : | whether "pops=0" and complains if not. | ||
| 2768 : | 4. The c-calls implementation for x86 now accepts two calling conventions: | ||
| 2769 : | "ccall" and "stdcall". When "ccall" is selected, the caller cleans | ||
| 2770 : | up after the call and pops is set to 0. For "stdcall", the caller | ||
| 2771 : | does nothing, leaving the cleanup to the callee; pops is set to | ||
| 2772 : | the number of bytes that were pushed onto the stack. | ||
| 2773 : | 5. The cproto decoder (compiler/Semant/types/cproto.sml) now can | ||
| 2774 : | distinguish between "ccall" and "stdcall". | ||
| 2775 : | 6. The UNIMP instruction has been added to the supported Sparc instruction | ||
| 2776 : | set. (This is needed for implementing the official C calling convention | ||
| 2777 : | on this architecture.) | ||
| 2778 : | 7. I fixed some of the .cm files under src/MLRISC/Tools to make them | ||
| 2779 : | work with the latest CM. | ||
| 2780 : | |||
| 2781 : | ---------------------------------------------------------------------- | ||
| 2782 : | Name: Matthias Blume | ||
| 2783 : | blume | 838 | Date: 2001/06/05 15:10:00 EDT |
| 2784 : | Tag: blume-20010605-cm-index | ||
| 2785 : | Description: | ||
| 2786 : | |||
| 2787 : | 0. The "lambdasplit" parameter for class "sml" in CM has been documented. | ||
| 2788 : | |||
| 2789 : | 1. CM can now generate "index files". These are human-readable files | ||
| 2790 : | that list on a per-.cm-file basis each toplevel symbol defined or | ||
| 2791 : | imported. The location of the index file for | ||
| 2792 : | <p>/<d>.cm is <p>/CM/INDEX/<d>.cm. | ||
| 2793 : | To enable index-file generation, set CM.Control.generate_index to true | ||
| 2794 : | or export an environment-symbol: export CM_GENERATE_INDEX=true. | ||
| 2795 : | |||
| 2796 : | The CM manual has been updated accordingly. | ||
| 2797 : | |||
| 2798 : | 2. I made some slight modifications to the c-calls API in MLRISC. | ||
| 2799 : | |||
| 2800 : | a) There is now a callback to support saving/restoring of | ||
| 2801 : | dedicated but caller-save registers around the actual call | ||
| 2802 : | instruction. | ||
| 2803 : | b) One can optionally specify a comment-annotation for the | ||
| 2804 : | call instruction. | ||
| 2805 : | |||
| 2806 : | 3. SML/NJ (mlriscGen.sml) uses this new API for the rawccall primop. | ||
| 2807 : | (For example, the comment annotation shows the C prototype of | ||
| 2808 : | the function being called.) | ||
| 2809 : | |||
| 2810 : | ---------------------------------------------------------------------- | ||
| 2811 : | Name: Matthias Blume | ||
| 2812 : | blume | 837 | Date: 2001/06/01 13:30:00 EDT |
| 2813 : | Tag: blume-20010601-nlffi-cleanup | ||
| 2814 : | Description: | ||
| 2815 : | |||
| 2816 : | This is mostly a cleanup of MLFFI stuff: | ||
| 2817 : | |||
| 2818 : | - some signature files have been put into a more exposed place | ||
| 2819 : | - the ugly 'f type parameter is gone (simplifies types tremendously!) | ||
| 2820 : | - ml-nlffigen changed accordingly | ||
| 2821 : | - tutorial updated | ||
| 2822 : | |||
| 2823 : | Other changes: | ||
| 2824 : | |||
| 2825 : | - author's affiliation in CM manual(s) updated | ||
| 2826 : | - some more recognized keywords added to Allen's sml.sty | ||
| 2827 : | |||
| 2828 : | ---------------------------------------------------------------------- | ||
| 2829 : | Name: Matthias Blume | ||
| 2830 : | blume | 836 | Date: 2001/05/25 15:30:00 EDT |
| 2831 : | Tag: blume-20010525-iptr | ||
| 2832 : | Description: | ||
| 2833 : | |||
| 2834 : | - put the official 110.33-README (as it appears on the ftp server) under | ||
| 2835 : | CVS | ||
| 2836 : | - fixed a small bug related to incomplete pointer types in | ||
| 2837 : | ml-nlffigen | ||
| 2838 : | - small cosmetic change to the ml-nlffi-lib's "arr" type constructor | ||
| 2839 : | (it does not need the 'f type parameter) | ||
| 2840 : | |||
| 2841 : | ---------------------------------------------------------------------- | ||
| 2842 : | Name: Matthias Blume | ||
| 2843 : | blume | 834 | Date: 2001/05/23 14:30:00 EDT |
| 2844 : | Tag: Release_110_33 | ||
| 2845 : | Description: | ||
| 2846 : | |||
| 2847 : | New version number (110.33). New bootfiles. | ||
| 2848 : | |||
| 2849 : | ---------------------------------------------------------------------- | ||
| 2850 : | Name: Matthias Blume | ||
| 2851 : | blume | 833 | Date: 2001/05/22 18:06:00 EDT |
| 2852 : | Tag: blume-20010522-targets | ||
| 2853 : | Description: | ||
| 2854 : | |||
| 2855 : | Made install.sh use file config/targets.customized if it exists, falling | ||
| 2856 : | back to config/targets if it doesn't. This way one can have a customized | ||
| 2857 : | version of the targets file without touching the "real thing", thus | ||
| 2858 : | eliminating the constant fear of accidentally checking something bogus | ||
| 2859 : | back into the CVS repository... (File config/targets.customized must | ||
| 2860 : | not be added to the repository!) | ||
| 2861 : | |||
| 2862 : | ---------------------------------------------------------------------- | ||
| 2863 : | Name: Matthias Blume | ||
| 2864 : | blume | 832 | Date: 2001/05/22 16:30:00 EDT |
| 2865 : | Tag: blume-20010522-minitut | ||
| 2866 : | Description: | ||
| 2867 : | |||
| 2868 : | 1. Bug fix in ml-nlffigen; now (hopefully) correctly handling | ||
| 2869 : | struct returns. | ||
| 2870 : | 2. Added src/ml-nlffi-lib/Doc/mini-tutorial.txt. This is some very | ||
| 2871 : | incomplete, preliminary documentation for NLFFI. | ||
| 2872 : | |||
| 2873 : | ---------------------------------------------------------------------- | ||
| 2874 : | Name: Matthias Blume | ||
| 2875 : | blume | 829 | Date: 2001/05/14 11:30:00 EDT |
| 2876 : | Tag: blume-20010514-script | ||
| 2877 : | Description: | ||
| 2878 : | |||
| 2879 : | Some bugs in install script fixed. | ||
| 2880 : | |||
| 2881 : | In addition to that I also made a slight change to the NLFFI API: | ||
| 2882 : | Functors generated by ml-nlffigen now take the dynamic library as a | ||
| 2883 : | straight functor argument, not as a suspended one. (The original | ||
| 2884 : | functor code used to force the suspension right away anyway, so there | ||
| 2885 : | was nothing gained by this complication of the interface.) | ||
| 2886 : | |||
| 2887 : | ---------------------------------------------------------------------- | ||
| 2888 : | Name: Matthias Blume | ||
| 2889 : | blume | 828 | Date: 2001/05/11 14:35:00 EDT |
| 2890 : | Tag: blume-20010511-ml-nlffi | ||
| 2891 : | Description: | ||
| 2892 : | |||
| 2893 : | I finally took the plunge and added my new FFI code to the main | ||
| 2894 : | repository. For x86-linux it is now ready for prime-time. | ||
| 2895 : | |||
| 2896 : | There are two new subdirectories of "src": | ||
| 2897 : | |||
| 2898 : | - ml-nlffi-lib: | ||
| 2899 : | The utility library for programs using the FFI interface. | ||
| 2900 : | Here is the implementation of $/c.cm and its associated low-level | ||
| 2901 : | partners $/c-int.cm and $/memory.cm. | ||
| 2902 : | - ml-nlffigen: | ||
| 2903 : | A stand-alone program for generating ML glue code from C source | ||
| 2904 : | code. | ||
| 2905 : | |||
| 2906 : | Building ml-nlffigen requires $/ckit-lib.cm. | ||
| 2907 : | |||
| 2908 : | The config/install.sh script has been updates to do the Right Thing | ||
| 2909 : | (hopefully). | ||
| 2910 : | |||
| 2911 : | Notice that the source tree for the C-Kit will not be put under "src" | ||
| 2912 : | but directly under the installation root directory. (This is the | ||
| 2913 : | structure that currently exists on the CVS server when you check out | ||
| 2914 : | module "sml".) Fortunately, config/install.sh knows about this oddity. | ||
| 2915 : | |||
| 2916 : | Bugs: No documentation yet. | ||
| 2917 : | |||
| 2918 : | ---------------------------------------------------------------------- | ||
| 2919 : | Name: Matthias Blume | ||
| 2920 : | blume | 826 | Date: 2001/05/09 16:35:00 EDT |
| 2921 : | Tag: blume-20010509-cpscontract | ||
| 2922 : | Description: | ||
| 2923 : | |||
| 2924 : | Fixed a bug in the accounting code in cpsopt/contract.sml. (The | ||
| 2925 : | wrapper/unwrapper elimination did not decrement usage counts and some | ||
| 2926 : | dead variables got overlooked by the dead-up logic.) | ||
| 2927 : | |||
| 2928 : | ---------------------------------------------------------------------- | ||
| 2929 : | george | 824 | Name: Lal George |
| 2930 : | Date: 2001/05/08 17:26:09 EDT | ||
| 2931 : | Tag: george-20010508-omit-frameptr | ||
| 2932 : | Description: | ||
| 2933 : | |||
| 2934 : | Changes to implement the omit-frame-pointer optimization to support | ||
| 2935 : | raw C calls. For now, there is only support on the Intel x86, but | ||
| 2936 : | other architectures will follow as more experience is gained with this. | ||
| 2937 : | |||
| 2938 : | |||
| 2939 : | ---------------------------------------------------------------------- | ||
| 2940 : | blume | 818 | Name: Matthias Blume |
| 2941 : | blume | 819 | Date: 2001/05/07 14:40:00 EDT |
| 2942 : | Tag: blume-20010507-proxies | ||
| 2943 : | Description: | ||
| 2944 : | |||
| 2945 : | I made into "proxy libraries" all libraries that qualify for such a | ||
| 2946 : | change. (A qualifying library is a library that has another library or | ||
| 2947 : | groups as its sole member and repeats that member's export list | ||
| 2948 : | verbatim. A proxy library avoids this repetition by omitting its export | ||
| 2949 : | list, effectively inheriting the list that its (only) member exports. | ||
| 2950 : | See the CM manual for more explanation.) | ||
| 2951 : | The main effect is that explicit export lists for these libraries | ||
| 2952 : | do not have to be kepts in sync, making maintenance a bit easier. | ||
| 2953 : | |||
| 2954 : | I also added copyright notices to many .cm-files. | ||
| 2955 : | |||
| 2956 : | Last but not least, I made a new set of bootfiles. | ||
| 2957 : | |||
| 2958 : | ---------------------------------------------------------------------- | ||
| 2959 : | Name: Matthias Blume | ||
| 2960 : | blume | 818 | Date: 2001/05/04 17:00:00 EDT |
| 2961 : | Tag: blume-20010504-cm-lsplit | ||
| 2962 : | Description: | ||
| 2963 : | |||
| 2964 : | 0. John merged pending changes to $/smlnj-lib.cm | ||
| 2965 : | |||
| 2966 : | 1. Allen's previous change accidentally backed out of one of Lal's | ||
| 2967 : | earlier changes. I undid this mistake (re-introducing Lal's change). | ||
| 2968 : | |||
| 2969 : | 2. I used the new topOrder' function from graph-scc.sml (from $/smlnj-lib.cm) | ||
| 2970 : | within the compiler where applicable. There is some code simplification | ||
| 2971 : | because of that. | ||
| 2972 : | |||
| 2973 : | 3. The "split" phase (in FLINT) is now part of the default list of phases. | ||
| 2974 : | Compiler.Control.LambdaSplitting.* can be used to globally control the | ||
| 2975 : | lambda-splitting (cross-module-inlining) engine. In addition to that, | ||
| 2976 : | it can now also be controlled on a per-source basis: CM has been taught | ||
| 2977 : | a new tool parameter applicable to ML source files. | ||
| 2978 : | |||
| 2979 : | - To turn lambda-splitting off completely: | ||
| 2980 : | local open Compiler.Control.LambdaSplitting in | ||
| 2981 : | val _ = set Off | ||
| 2982 : | end | ||
| 2983 : | - To make "no lambda-splitting" the global default (but allow per-source | ||
| 2984 : | overriding); this is the initial setting: | ||
| 2985 : | local open Compiler.Control.LambdaSplitting in | ||
| 2986 : | val _ = set (Default NONE) | ||
| 2987 : | end | ||
| 2988 : | - To make "lambda-splitting with aggressiveness a" the global default | ||
| 2989 : | (and allow per-source overriding): | ||
| 2990 : | local open Compiler.Control.LambdaSplitting in | ||
| 2991 : | val _ = set (Default (SOME a)) | ||
| 2992 : | end | ||
| 2993 : | |||
| 2994 : | - To turn lambda-splitting off for a given ML souce file (say: a.sml) | ||
| 2995 : | write (in the respective .cm-file): | ||
| 2996 : | a.sml (lambdasplitting:off) | ||
| 2997 : | - To turn lambda-splitting for a.sml on with minimal aggressiveness: | ||
| 2998 : | a.sml (lambdasplitting:on) | ||
| 2999 : | - To turn lambda-splitting for a.sml on with aggressiveness <a> (where | ||
| 3000 : | <a> is a decimal non-negative integer): | ||
| 3001 : | a.sml (lambdasplitting:<a>) | ||
| 3002 : | - To turn lambda-splitting for a.sml on with maximal aggressiveness: | ||
| 3003 : | a.sml (lambdasplitting:infinity) | ||
| 3004 : | - To use the global default for a.sml: | ||
| 3005 : | a.sml (lambdasplitting:default) | ||
| 3006 : | or simply | ||
| 3007 : | a.sml | ||
| 3008 : | |||
| 3009 : | ---------------------------------------------------------------------- | ||
| 3010 : | leunga | 815 | Name: Allen Leung |
| 3011 : | Date: 2001/05/04 01:57:00 EDT | ||
| 3012 : | Tag: leunga-20010504-sync | ||
| 3013 : | Description: | ||
| 3014 : | leunga | 788 | |
| 3015 : | leunga | 815 | MLRISC features. |
| 3016 : | |||
| 3017 : | 1. Fix to CMPXCHG instructions. | ||
| 3018 : | 2. Changed RA interface to allow annotations in callbacks. | ||
| 3019 : | 3. Added a new method to the stream interface to allow annotations updates. | ||
| 3020 : | |||
| 3021 : | george | 749 | ---------------------------------------------------------------------- |
| 3022 : | blume | 812 | Name: Matthias Blume |
| 3023 : | blume | 814 | Date: 2001/05/01 11:45:00 EDT |
| 3024 : | Tag: blume-20010501-pcedittmp | ||
| 3025 : | Description: | ||
| 3026 : | |||
| 3027 : | Changed install.sh to use the current working directory instead of | ||
| 3028 : | /usr/tmp for a temporary file (pcedittmp). The previous choice | ||
| 3029 : | of /usr/tmp caused trouble with MacOS X because of file premission | ||
| 3030 : | problems. | ||
| 3031 : | |||
| 3032 : | ---------------------------------------------------------------------- | ||
| 3033 : | Name: Matthias Blume | ||
| 3034 : | blume | 812 | Date: 2001/04/20 11:10:00 EDT |
| 3035 : | Tag: blume-20010420-inMLflag | ||
| 3036 : | Description: | ||
| 3037 : | |||
| 3038 : | - added vp_limitPtrMask to vproc-state.h | ||
| 3039 : | (for use by the raw-C-calls mechanism to implement proper interrupt | ||
| 3040 : | handling) | ||
| 3041 : | - made the ML compiler aware of various data-structure offsets so it | ||
| 3042 : | can generate code for accessing the vp_inML flag and vp_limitPtrMask | ||
| 3043 : | - tweaked mlriscGen.sml to have it emit interrupt-handling code for | ||
| 3044 : | raw C-calls | ||
| 3045 : | |||
| 3046 : | ---------------------------------------------------------------------- | ||
| 3047 : | george | 810 | Name: Lal George |
| 3048 : | Date: 2001/04/20 09:15:28 EDT | ||
| 3049 : | Tag: george-20010420-macosX | ||
| 3050 : | Description: | ||
| 3051 : | |||
| 3052 : | - Changes to port to Mac OS X; Darwin. | ||
| 3053 : | |||
| 3054 : | - In the process I found that sqrt was broken on the PPC, because the | ||
| 3055 : | fsqrt instruction is not implemented. | ||
| 3056 : | |||
| 3057 : | ---------------------------------------------------------------------- | ||
| 3058 : | blume | 799 | Name: Matthias Blume |
| 3059 : | blume | 808 | Date: 2001/04/18 12:45:00 EDT |
| 3060 : | Tag: blume-20010418-ccalls | ||
| 3061 : | Description: | ||
| 3062 : | |||
| 3063 : | - fixed two off-by-4 errors in the x86-specific c-calls implementation | ||
| 3064 : | (this bug prevented structure arguments containing pointers from being | ||
| 3065 : | passed correctly) | ||
| 3066 : | - changed the raw-C-call code in mlriscGen.sml in such a way that | ||
| 3067 : | structure arguments are represented as a pointer to the beginning | ||
| 3068 : | of the structure (instead of having a series of synthesized arguments, | ||
| 3069 : | one for each structure member) | ||
| 3070 : | |||
| 3071 : | - made makeml script's verbosity level configurable via environment | ||
| 3072 : | variable (MAKEML_VERBOSITY) | ||
| 3073 : | |||
| 3074 : | blume | 813 | - eliminated placeholder implementations for f32l, w16s, i16s, and f32s |
| 3075 : | blume | 808 | in rawmem-x86.sml; we are now using the real thing |
| 3076 : | |||
| 3077 : | ---------------------------------------------------------------------- | ||
| 3078 : | Name: Matthias Blume | ||
| 3079 : | blume | 806 | Date: 2001/03/22 16:25:00 EST |
| 3080 : | Tag: blume-20010322-bootfiles | ||
| 3081 : | Description: | ||
| 3082 : | |||
| 3083 : | Created a new set of bootfiles (for your automatic installation convenience). | ||
| 3084 : | |||
| 3085 : | ---------------------------------------------------------------------- | ||
| 3086 : | Name: Matthias Blume | ||
| 3087 : | blume | 805 | Date: 2001/03/22 15:10:00 EST |
| 3088 : | Tag: blume-20010322-rawmem-parcm | ||
| 3089 : | Description: | ||
| 3090 : | |||
| 3091 : | 1. All "raw memory access" primitives for the new FFI are implemented now | ||
| 3092 : | (at least on the x86). | ||
| 3093 : | 2. Some further cleanup of CM's parallel make mechanism. | ||
| 3094 : | |||
| 3095 : | ---------------------------------------------------------------------- | ||
| 3096 : | Name: Matthias Blume | ||
| 3097 : | blume | 801 | Date: 2001/03/19 17:53:00 EST |
| 3098 : | Tag: blume-20010319-parallel | ||
| 3099 : | Description: | ||
| 3100 : | |||
| 3101 : | Parallel make (using compile servers) now works again. | ||
| 3102 : | |||
| 3103 : | To this end, CM.stabilize and CMB.make have been modified to work in | ||
| 3104 : | two passes when compile servers are attached: | ||
| 3105 : | 1. Compile everything, do not perform stabilization; this pass | ||
| 3106 : | uses compile servers | ||
| 3107 : | 2. Stabilize everything; this pass does not use compile servers | ||
| 3108 : | If there are no compile servers, the two passes are combined into one | ||
| 3109 : | (as before). Splitting the passes increases the inherent parallelism | ||
| 3110 : | in the dependency graph because the entire graph including all | ||
| 3111 : | libraries is available at the same time. This, in turn, improves | ||
| 3112 : | server utilization. The downside is that the master process will | ||
| 3113 : | have to do some extra work after compilation is done (because for | ||
| 3114 : | technical reasons it must re-read all the binfiles during stabilization). | ||
| 3115 : | |||
| 3116 : | ---------------------------------------------------------------------- | ||
| 3117 : | Name: Matthias Blume | ||
| 3118 : | blume | 800 | Date: 2001/03/16 12:22:00 EST |
| 3119 : | Tag: blume-20010316-bootfiles | ||
| 3120 : | Description: | ||
| 3121 : | |||
| 3122 : | Created a new set of bootfiles (for your automatic installation convenience). | ||
| 3123 : | |||
| 3124 : | ---------------------------------------------------------------------- | ||
| 3125 : | Name: Matthias Blume | ||
| 3126 : | blume | 799 | Date: 2001/03/16 11:00:00 EST |
| 3127 : | Tag: blume-20010316-MLTREE-fixup | ||
| 3128 : | Description: | ||
| 3129 : | |||
| 3130 : | This is a minor fixup for an (untagged) earlier commit by Allen. | ||
| 3131 : | (A file was missing). | ||
| 3132 : | |||
| 3133 : | ---------------------------------------------------------------------- | ||
| 3134 : | leunga | 788 | Name: Allen Leung |
| 3135 : | leunga | 796 | Date: Mon Mar 5 18:54:57 EST 2001 |
| 3136 : | Tag: leunga-20010305-cut-support | ||
| 3137 : | |||
| 3138 : | 1. New support for alternative control-flow in MLTREE. | ||
| 3139 : | Currently we support | ||
| 3140 : | |||
| 3141 : | FLOW_TO(CALL ...., [k1,...,kn]) | ||
| 3142 : | |||
| 3143 : | This is needed for 'cuts to' in C-- and try/handle-like constructs | ||
| 3144 : | in Moby | ||
| 3145 : | |||
| 3146 : | New assembler flag "asm-show-cutsto" to turn on control-flow debugging. | ||
| 3147 : | |||
| 3148 : | 2. Register Allocator | ||
| 3149 : | |||
| 3150 : | Changes in interface [from Fermin, John] | ||
| 3151 : | |||
| 3152 : | 3. Alpha 8-bit SLL support [Fermin] | ||
| 3153 : | |||
| 3154 : | 4. All architectures | ||
| 3155 : | |||
| 3156 : | A new module (ClusterExpandCopies) for expanding parallel copies. | ||
| 3157 : | |||
| 3158 : | ---------------------------------------------------------------------- | ||
| 3159 : | Name: Allen Leung | ||
| 3160 : | leunga | 788 | Date: 2001/02/27 23:07:00 EST |
| 3161 : | Tag: leunga-20010227-minor-stuff | ||
| 3162 : | |||
| 3163 : | 1. Alpha bug fix for CMOVNE | ||
| 3164 : | 2. Handle mltree COND(..,FCMP ...,...) | ||
| 3165 : | 3. Bug fix in simplifier | ||
| 3166 : | |||
| 3167 : | ---------------------------------------------------------------------- | ||
| 3168 : | blume | 777 | Name: Matthias Blume |
| 3169 : | blume | 781 | Date: 2001/01/30 17:50:00 EST |
| 3170 : | Tag: blume-20010130-sync | ||
| 3171 : | Description: | ||
| 3172 : | |||
| 3173 : | This is just a minor update to sync my devel branch with the main brach. | ||
| 3174 : | The only visible change is the addition of some README files. | ||
| 3175 : | |||
| 3176 : | ---------------------------------------------------------------------- | ||
| 3177 : | Name: Matthias Blume | ||
| 3178 : | blume | 778 | Date: 2001/01/12 23:30:00 JST |
| 3179 : | Tag: blume-20010112-bootfiles | ||
| 3180 : | Description: | ||
| 3181 : | |||
| 3182 : | Made a new set of bootfiles that goes with the current state of the | ||
| 3183 : | repository. | ||
| 3184 : | |||
| 3185 : | ---------------------------------------------------------------------- | ||
| 3186 : | Name: Matthias Blume | ||
| 3187 : | blume | 777 | Date: 2001/01/12 21:20:00 JST |
| 3188 : | Tag: blume-20010112-sync | ||
| 3189 : | Description: | ||
| 3190 : | |||
| 3191 : | I am just flushing out some minor changes that had accumulated in | ||
| 3192 : | my private branch in order to sync with the main tree. (This is | ||
| 3193 : | mainly because I had CVS trouble when trying to merge _into_ my | ||
| 3194 : | private branch.) | ||
| 3195 : | |||
| 3196 : | Most people should be completely unaffected by this. | ||
| 3197 : | |||
| 3198 : | ---------------------------------------------------------------------- | ||
| 3199 : | leunga | 775 | Name: Allen Leung |
| 3200 : | Date: Thu Jan 11 21:03:00 EST 2001 | ||
| 3201 : | Tag: leunga-20010111-labexp=mltree | ||
| 3202 : | Description: | ||
| 3203 : | |||
| 3204 : | 1. Removed the type LabelExp and replace it by MLTree. | ||
| 3205 : | 2. Rewritten mltree-simplify with the pattern matcher tool. | ||
| 3206 : | 3. There were some bugs in alpha code generator which would break | ||
| 3207 : | 64-bit code generation. | ||
| 3208 : | 4. Redo the tools to generate code with the | ||
| 3209 : | 5. The CM files in MLRISC (and in src/system/smlnj/MLRISC) | ||
| 3210 : | are now generated by perl scripts. | ||
| 3211 : | |||
| 3212 : | ---------------------------------------------------------------------- | ||
| 3213 : | blume | 771 | Name: Matthias Blume |
| 3214 : | blume | 774 | Date: 2001/01/10 21:55:00 JST |
| 3215 : | Tag: blume-20010110-rcc | ||
| 3216 : | Description: | ||
| 3217 : | |||
| 3218 : | The RCC stuff now seems to work (but only on the x86). | ||
| 3219 : | This required hacking of the c-calls interface (and -implementation) in | ||
| 3220 : | MLRISC. | ||
| 3221 : | |||
| 3222 : | Normal compiler users should be unaffected. | ||
| 3223 : | |||
| 3224 : | ---------------------------------------------------------------------- | ||
| 3225 : | Name: Matthias Blume | ||
| 3226 : | blume | 773 | Date: 2001/01/09 01:20:00 JST |
| 3227 : | Tag: blume-20010109-rcc | ||
| 3228 : | Description: | ||
| 3229 : | |||
| 3230 : | This is a fairly big patch, flushing out a large number of pending | ||
| 3231 : | changes that I made to my development copy over the last couple of days. | ||
| 3232 : | |||
| 3233 : | Of practical relevance at this moment is a workaround for a pickling | ||
| 3234 : | bug that Allen ran into the other day. The cause of the bug itself is | ||
| 3235 : | still unknown and it might be hard to fix it properly, but the | ||
| 3236 : | workaround has some merits of its own (namely somewhat reducing pickling | ||
| 3237 : | overhead for certain libraries). Therefore, I think this solution should | ||
| 3238 : | be satisfactory at this time. | ||
| 3239 : | |||
| 3240 : | The rest of the changes (i.e., the vast majority) has to do with my | ||
| 3241 : | ongoing efforts of providing direct support for C function calls from | ||
| 3242 : | ML. At the moment there is a new primop "RAW_CCALL", typing magic | ||
| 3243 : | in types/cproto.sml (invoked from FLINT/trans/translate.sml), a new | ||
| 3244 : | case in the FLINT CPS datatype (RCC), changes to cps/convert.sml to | ||
| 3245 : | translate uses of RAW_CCALL into RCC, and changes to mlriscGen.sml to | ||
| 3246 : | handle RCC. | ||
| 3247 : | |||
| 3248 : | The last part (the changes to mlriscGen.sml) are still known to be | ||
| 3249 : | wrong on the x86 and not implemented on all other architectures. But | ||
| 3250 : | the infrastructure is in place. I had to change a few functor | ||
| 3251 : | signatures in the backend to be able to route the CCalls interface | ||
| 3252 : | from MLRISC there, and I had to specialize the mltree type (on the | ||
| 3253 : | x86) to include the necessary extensions. (The extensions themselves | ||
| 3254 : | were already there and redy to go in MLRISC/x86). | ||
| 3255 : | |||
| 3256 : | Everything should be very happy as soon as someone helps me with | ||
| 3257 : | mlriscGen.sml... | ||
| 3258 : | |||
| 3259 : | In any case, nothing of this should matter to anyone as long as the | ||
| 3260 : | new primop is not being used (which is going to be the case unless you | ||
| 3261 : | find it where I hid it :). The rest of the compiler is completely | ||
| 3262 : | unaffected. | ||
| 3263 : | |||
| 3264 : | ---------------------------------------------------------------------- | ||
| 3265 : | Name: Matthias Blume | ||
| 3266 : | blume | 772 | Date: 2001/01/05 00:30:00 JST |
| 3267 : | Tag: blume-20010105-primops | ||
| 3268 : | Description: | ||
| 3269 : | |||
| 3270 : | Added some experimental support for work that I am doing right now. | ||
| 3271 : | These changes mostly concern added primops, but there is also a new | ||
| 3272 : | experimental C library in the runtime system (but currently not enabled | ||
| 3273 : | anywhere except on Linux/X86). | ||
| 3274 : | |||
| 3275 : | In the course of adding primops (and playing with them), I discovered that | ||
| 3276 : | Zhong's INL_PRIM hack (no type info for certain primops) was, in fact, badly | ||
| 3277 : | broken. (Zhong was very right he labeled this stuff as "major gross hack".) | ||
| 3278 : | To recover, I made type information in INL_PRIM mandatory and changed | ||
| 3279 : | prim.sml as well as built-in.sml accordingly. The InLine structure now | ||
| 3280 : | has complete, correct type information (i.e., no bottom types). | ||
| 3281 : | |||
| 3282 : | Since all these changes mean that we need new binfiles, I also bumped the | ||
| 3283 : | version number to 110.32.1. | ||
| 3284 : | |||
| 3285 : | ---------------------------------------------------------------------- | ||
| 3286 : | Name: Matthias Blume | ||
| 3287 : | blume | 771 | Date: 2000/12/30 22:10:00 JST |
| 3288 : | Tag: blume-20001230-various | ||
| 3289 : | Description: | ||
| 3290 : | |||
| 3291 : | Added proxy libraries for MLRISC and let MLRISC libraries refer | ||
| 3292 : | to each other using path anchors. (See CM manual for explanation.) | ||
| 3293 : | |||
| 3294 : | Updated CM documentation. | ||
| 3295 : | |||
| 3296 : | Fixed some bugs in CM. | ||
| 3297 : | |||
| 3298 : | Implemented "proxy" libraries (= syntactic sugar for CM). | ||
| 3299 : | |||
| 3300 : | Added "-quiet" option to makeml and changed runtime system accordingly. | ||
| 3301 : | |||
| 3302 : | Added cleanup handler for exportML to reset timers and compiler stats. | ||
| 3303 : | |||
| 3304 : | ---------------------------------------------------------------------- | ||
| 3305 : | george | 761 | Name: Lal George |
| 3306 : | Date: 2000/12/22 22:22:58 EST 2000 | ||
| 3307 : | Tag: Release_110_32 | ||
| 3308 : | Description: | ||
| 3309 : | |||
| 3310 : | Infinite precision used throughout MLRISC. | ||
| 3311 : | see MLRISC/mltree/machine-int.sig | ||
| 3312 : | |||
| 3313 : | ---------------------------------------------------------------------- | ||
| 3314 : | blume | 756 | Name: Matthias Blume |
| 3315 : | blume | 760 | Date: 2000/12/22 23:16:00 JST |
| 3316 : | Tag: blume-20001222-warn | ||
| 3317 : | Description: | ||
| 3318 : | blume | 759 | |
| 3319 : | blume | 760 | Corrected wording and formatting of some CM warning message which I |
| 3320 : | broke in my previous patch. | ||
| 3321 : | blume | 759 | |
| 3322 : | ---------------------------------------------------------------------- | ||
| 3323 : | Name: Matthias Blume | ||
| 3324 : | blume | 763 | Date: 2000/12/22 21:20:00 JST |
| 3325 : | Tag: blume-20001222-anchorenv | ||
| 3326 : | Description: | ||
| 3327 : | |||
| 3328 : | Fixed CM's handling of anchor environments in connection with CMB.make. | ||
| 3329 : | |||
| 3330 : | ---------------------------------------------------------------------- | ||
| 3331 : | Name: Matthias Blume | ||
| 3332 : | blume | 758 | Date: 2000/12/22 13:15:00 JST |
| 3333 : | Tag: blume-20001222-cleanup | ||
| 3334 : | Description: | ||
| 3335 : | |||
| 3336 : | Removed src/cm/ffi which does not (and did not) belong here. | ||
| 3337 : | |||
| 3338 : | ---------------------------------------------------------------------- | ||
| 3339 : | Name: Matthias Blume | ||
| 3340 : | blume | 757 | Date: 2000/12/21 23:55:00 JST |
| 3341 : | Tag: blume-20001221-exn | ||
| 3342 : | Description: | ||
| 3343 : | |||
| 3344 : | Probably most important: CM no longer silently swallows all exceptions | ||
| 3345 : | in the compiler. | ||
| 3346 : | Plus: some other minor CM changes. For example, CM now reports some | ||
| 3347 : | sizes for generated binfiles (code, data, envpickle, lambdapickle). | ||
| 3348 : | |||
| 3349 : | ---------------------------------------------------------------------- | ||
| 3350 : | Name: Matthias Blume | ||
| 3351 : | blume | 756 | Date: 2000/12/15 00:01:05 JST |
| 3352 : | Tag: blume-20001215-dirtool | ||
| 3353 : | Description: | ||
| 3354 : | |||
| 3355 : | - "dir" tool added. | ||
| 3356 : | - improvements and cleanup to Tools structure | ||
| 3357 : | - documentation updates | ||
| 3358 : | |||
| 3359 : | ---------------------------------------------------------------------- | ||
| 3360 : | leunga | 752 | Name: Allen Leung |
| 3361 : | leunga | 755 | Date: Thu Dec 14 03:45:24 EST 2000 |
| 3362 : | Description: | ||
| 3363 : | Tag: leunga-20001214-int-inf | ||
| 3364 : | Description: | ||
| 3365 : | |||
| 3366 : | In IntInf, added these standard functions, which are missing from our | ||
| 3367 : | implementation: | ||
| 3368 : | |||
| 3369 : | andb : int * int -> int | ||
| 3370 : | xorb : int * int -> int | ||
| 3371 : | orb : int * int -> int | ||
| 3372 : | notb : int -> int | ||
| 3373 : | << : int * word -> int | ||
| 3374 : | ~>> : int * word -> int | ||
| 3375 : | |||
| 3376 : | Not tested, I hope they are correct. | ||
| 3377 : | |||
| 3378 : | ---------------------------------------------------------------------- | ||
| 3379 : | Name: Allen Leung | ||
| 3380 : | leunga | 752 | Date: Fri Dec 8 19:23:26 EST 2000 |
| 3381 : | Description: | ||
| 3382 : | Tag: leunga-20001208-nowhere | ||
| 3383 : | Description: | ||
| 3384 : | |||
| 3385 : | Slight improvements to the 'nowhere' tool to handle OR-patterns, | ||
| 3386 : | to generate better error messages etc. Plus a brief manual. | ||
| 3387 : | |||
| 3388 : | ---------------------------------------------------------------------- | ||
| 3389 : | george | 749 | Name: Lal George |
| 3390 : | Date: 2000/12/08 09:54:02 EST 2000 | ||
| 3391 : | Tag: Release_110_31 | ||
| 3392 : | Description: | ||
| 3393 : | george | 705 | |
| 3394 : | george | 749 | - Version 110.31 |
| 3395 : | leunga | 591 | ---------------------------------------------------------------------- |
| 3396 : | leunga | 744 | Name: Allen Leung |
| 3397 : | Date: Thu Dec 7 22:01:04 EST 2000 | ||
| 3398 : | Tag: leunga-20001207-cell-monster-hack | ||
| 3399 : | Description: | ||
| 3400 : | |||
| 3401 : | Major MLRISC internal changes. Affect all clients. | ||
| 3402 : | Summary: | ||
| 3403 : | |||
| 3404 : | 1. Type CELLS.cell = int is now replaced by a datatype. | ||
| 3405 : | As a result, the old regmap is now gone. Almost all interfaces | ||
| 3406 : | in MLRISC change as a consequence. | ||
| 3407 : | |||
| 3408 : | 2. A new brand version of machine description tool (v3.0) that generates | ||
| 3409 : | modules expecting the new interface. The old version is removed. | ||
| 3410 : | |||
| 3411 : | 3. The RA interface has been further abstracted into two new functors. | ||
| 3412 : | RISC_RA and X86RA. These functors have much simpler interfaces. | ||
| 3413 : | [See also directory MLRISC/demo.] | ||
| 3414 : | |||
| 3415 : | 4. Some other new source->source code generation tools are available: | ||
| 3416 : | |||
| 3417 : | a. MLRISC/Tools/RewriteGen -- generate rewriters from rules. | ||
| 3418 : | b. MLRISC/Tools/WhereGen -- expands conditional pattern matching rules. | ||
| 3419 : | I use this tool to generate the peephole optimizers---with the new | ||
| 3420 : | cell type changes, peephole rules are becoming difficult to write | ||
| 3421 : | without conditional pattern matching. | ||
| 3422 : | |||
| 3423 : | 5. More Intmap -> IntHashTable change. Previous changes by Matthias didn't | ||
| 3424 : | cover the entire MLRISC source tree so many things broke. | ||
| 3425 : | |||
| 3426 : | 6. CM files have been moved to the subdirectory MLRISC/cm. | ||
| 3427 : | They are moved because there are a lot of them and they clutter up the | ||
| 3428 : | root dir. | ||
| 3429 : | |||
| 3430 : | 7. More detailed documentation to come... | ||
| 3431 : | |||
| 3432 : | NOTE: To rebuild from 110.30 (ftp distribution), you'll have to do | ||
| 3433 : | a makeml -rebuild first. This is because of other other | ||
| 3434 : | changes that Matthias has made (see below). | ||
| 3435 : | |||
| 3436 : | |||
| 3437 : | ---------------------------------------------------------------------- | ||
| 3438 : | blume | 733 | Name: Matthias Blume |
| 3439 : | blume | 742 | Date: 2000/11/30 23:12:00 JST |
| 3440 : | Tag: blume-20001130-filereorg | ||
| 3441 : | Description: | ||
| 3442 : | |||
| 3443 : | Some manual updates and some file reorganizations in CM. | ||
| 3444 : | |||
| 3445 : | ---------------------------------------------------------------------- | ||
| 3446 : | Name: Matthias Blume | ||
| 3447 : | blume | 737 | Date: 2000/11/24 17:45:00 JST |
| 3448 : | Tag: blume-20001124-link | ||
| 3449 : | Description: | ||
| 3450 : | |||
| 3451 : | Drastically improved link traversal code for the case that the dynamic | ||
| 3452 : | value was already loaded at bootstrap time. As a result, CM and CMB | ||
| 3453 : | now both load blazingly fast -- even on a very slow machine. Also, | ||
| 3454 : | memory consumption has been further reduced by this. | ||
| 3455 : | |||
| 3456 : | Warning: The format of the PIDMAP file has changed. THerefore, to | ||
| 3457 : | bootstrap you have to do this: | ||
| 3458 : | |||
| 3459 : | 1. Run CMB.make | ||
| 3460 : | 2. Make a symbolic link for the boot directory: | ||
| 3461 : | ln -s sml.boot.ARCH-OS xxx | ||
| 3462 : | 3. "Rebuild" the boot directory: | ||
| 3463 : | ./makeml -boot xxx -rebuild sml ; rm xxx | ||
| 3464 : | 4. Boot normally: | ||
| 3465 : | ./makeml | ||
| 3466 : | |||
| 3467 : | ---------------------------------------------------------------------- | ||
| 3468 : | Name: Matthias Blume | ||
| 3469 : | blume | 735 | Date: 2000/11/21 21:20:00 JST |
| 3470 : | Tag: blume-20001121-tools | ||
| 3471 : | Description: | ||
| 3472 : | |||
| 3473 : | Continued hacking on autoloading problem -- with success this time. | ||
| 3474 : | Also changed tool-plugin mechanism. See new CM manual. | ||
| 3475 : | |||
| 3476 : | ---------------------------------------------------------------------- | ||
| 3477 : | Name: Matthias Blume | ||
| 3478 : | blume | 734 | Date: 2000/11/19 14:30:00 JST |
| 3479 : | Tag: blume-20001119-autoload | ||
| 3480 : | Description: | ||
| 3481 : | |||
| 3482 : | Some hacking to make autoloading faster. Success for CMB, no success | ||
| 3483 : | so far for CM. There is a reduced structure CM' that autoloads faster. | ||
| 3484 : | (This is a temporary, non-documented hack to be eliminated again when | ||
| 3485 : | the general problem is solved.) | ||
| 3486 : | |||
| 3487 : | ---------------------------------------------------------------------- | ||
| 3488 : | Name: Matthias Blume | ||
| 3489 : | blume | 733 | Date: 2000/11/17 14:10:00 JST |
| 3490 : | Tag: blume-20001117-pickle-lib | ||
| 3491 : | Description: | ||
| 3492 : | |||
| 3493 : | 1. Eliminated comp-lib.cm | ||
| 3494 : | 2. Made pickle-lib.cm | ||
| 3495 : | 3. Eliminated all uses of intset.sml (from comp-lib.cm) | ||
| 3496 : | 4. Replaced all uses of intmap.{sig,sml} (from comp-lib.cm) with | ||
| 3497 : | equivalent constructs from smlnj-lib.cm (INtHashTable). | ||
| 3498 : | 5. Point 4. also goes for those uses of intmap.* in MLRISC. | ||
| 3499 : | Duplicated intmap modules thrown out. | ||
| 3500 : | 6. Hunted down all duplicated SCC code and replaced it with | ||
| 3501 : | equivalent stuff (GraphSCCFn from smlnj-lib.cm). | ||
| 3502 : | 7. Rewrote Feedback module. | ||
| 3503 : | 8. Moved sortedlist.sml into viscomp-lib.cm. Eventually it | ||
| 3504 : | should be thrown out and equivalent modules from smlnj-lib.cm | ||
| 3505 : | should be used (IntRedBlackSet, IntListSet, ...). | ||
| 3506 : | |||
| 3507 : | Confirmed that compiler compiles to fixpoint. | ||
| 3508 : | |||
| 3509 : | ---------------------------------------------------------------------- | ||
| 3510 : | leunga | 731 | Name: Allen Leung |
| 3511 : | Date: 2000/11/10 18:00:00 | ||
| 3512 : | Tag: leunga-20001110-new-x86-fp | ||
| 3513 : | |||
| 3514 : | A new x86 floating point code generator has been added. | ||
| 3515 : | By default this is turned off. To turn this on, do: | ||
| 3516 : | |||
| 3517 : | CM.autoload "$smlnj/compiler.cm"; | ||
| 3518 : | Compiler.Control.MLRISC.getFlag "x86-fast-fp" := true; | ||
| 3519 : | |||
| 3520 : | Changes: | ||
| 3521 : | |||
| 3522 : | 1. Changed FTAN to FPTAN so that the assembly output is correct. | ||
| 3523 : | 2. Changed the extension callback for FTANGENT to generate: | ||
| 3524 : | |||
| 3525 : | fptan | ||
| 3526 : | fstp %st(0) | ||
| 3527 : | instead of | ||
| 3528 : | fptan | ||
| 3529 : | fstpl ftempmem | ||
| 3530 : | |||
| 3531 : | 3. Numerous assembly fixes for x86. | ||
| 3532 : | |||
| 3533 : | 5. Cleaned up the machine code output module x86/x86MC.sml and added | ||
| 3534 : | support for a whole bunch of instructions and addressing modes: | ||
| 3535 : | |||
| 3536 : | fadd/fsub/fsubr/fmul/fdiv/fdivr %st, %st(n) | ||
| 3537 : | faddp/fsubp/fsubrp/fmulp/fdivp/fdivrp %st, %st(n) | ||
| 3538 : | fadd/fsub/fsubr/fmul/fdiv/fdivr %st(n), %st | ||
| 3539 : | fiadd/fisub/fisubr/fimul/fidiv/fidivr mem | ||
| 3540 : | fxch %st(n) | ||
| 3541 : | fld %st(n) | ||
| 3542 : | fst %st(n) | ||
| 3543 : | fst mem | ||
| 3544 : | fstp %st(n) | ||
| 3545 : | fucom %st(n) | ||
| 3546 : | fucomp %st(n) | ||
| 3547 : | |||
| 3548 : | All these are now generated when the fast fp mode is turned on. | ||
| 3549 : | |||
| 3550 : | 6. Removed the dedicated registers %st(0), ..., %st(7) from X86CpsRegs | ||
| 3551 : | |||
| 3552 : | ---------------------------------------------------------------------- | ||
| 3553 : | blume | 729 | Name: Matthias Blume |
| 3554 : | Date: 2000/11/09 11:20:00 JST | ||
| 3555 : | Tag: blume-20001109-scc | ||
| 3556 : | Description: | ||
| 3557 : | |||
| 3558 : | Eliminated some code duplication: | ||
| 3559 : | |||
| 3560 : | 1. Added "where" clause to GraphSCCFn in SML/NJ Library. | ||
| 3561 : | (Otherwise the functor is useless.) | ||
| 3562 : | 2. Used GraphSCCFn where SCCUtilFun was used previously. | ||
| 3563 : | 3. Got rid of SCCUtilFun (in comp-lib.cm). | ||
| 3564 : | |||
| 3565 : | ---------------------------------------------------------------------- | ||
| 3566 : | george | 721 | Name: Lal George |
| 3567 : | Date: 2000/11/06 09:02:21 EST 2000 | ||
| 3568 : | Tag: Release_110_30 | ||
| 3569 : | Description: | ||
| 3570 : | |||
| 3571 : | - Version 110.30 | ||
| 3572 : | ---------------------------------------------------------------------- | ||
| 3573 : | blume | 715 | Name: Matthias Blume |
| 3574 : | blume | 716 | Date: 2000/11/04 14:45:00 |
| 3575 : | Tag: blume-20001104-mlbuild | ||
| 3576 : | Description: | ||
| 3577 : | |||
| 3578 : | - Made ml-build faster on startup. | ||
| 3579 : | - Documentation fixes. | ||
| 3580 : | |||
| 3581 : | ---------------------------------------------------------------------- | ||
| 3582 : | Name: Matthias Blume | ||
| 3583 : | blume | 715 | Date: 2000/11/02 17:00:00 JST |
| 3584 : | Tag: blume-20001102-condcomp | ||
| 3585 : | Description: | ||
| 3586 : | |||
| 3587 : | - Small tweaks to pickler -- new BOOTFILES! | ||
| 3588 : | - Version bumped to 110.29.2. | ||
| 3589 : | - Added conditional compilation facility to init.cmi (see comment there). | ||
| 3590 : | ---------------------------------------------------------------------- | ||
| 3591 : | leunga | 713 | Name: Allen Leung |
| 3592 : | Date: 2000/10/23 19:31:00 | ||
| 3593 : | Tag: leunga-20001023-demo-ra | ||
| 3594 : | |||
| 3595 : | 1. Minor RA changes that improves spilling on x86 (affects Moby and C-- only) | ||
| 3596 : | 2. Test programs for the graph library updated | ||
| 3597 : | 3. Some new MLRISC demo programs added | ||
| 3598 : | |||
| 3599 : | ---------------------------------------------------------------------- | ||
| 3600 : | blume | 710 | Name: Matthias Blume |
| 3601 : | blume | 711 | Date: 2000/08/31 22:15:00 JST |
| 3602 : | Tag: blume-20001017-errmsg | ||
| 3603 : | Description: | ||
| 3604 : | |||
| 3605 : | More error message grief: Where there used to be no messages, there | ||
| 3606 : | now were some that had bogus error regions. Fixed. | ||
| 3607 : | |||
| 3608 : | ---------------------------------------------------------------------- | ||
| 3609 : | Name: Matthias Blume | ||
| 3610 : | blume | 710 | Date: 2000/08/31 17:30:00 JST |
| 3611 : | Tag: blume-20001017-v110p29p1 | ||
| 3612 : | Description: | ||
| 3613 : | |||
| 3614 : | I made a version 110.29.1 with new bootfiles. | ||
| 3615 : | |||
| 3616 : | Changes: Modified pickler/unpickler for faster and leaner unpickling. | ||
| 3617 : | CM documentation changes and a small bugfix in CM's error reporting. | ||
| 3618 : | |||
| 3619 : | ---------------------------------------------------------------------- | ||
| 3620 : | george | 705 | Name: Lal George |
| 3621 : | Date: 2000/09/27 14:42:35 EDT | ||
| 3622 : | Tag: george-20000927-nodestatus | ||
| 3623 : | Description: | ||
| 3624 : | |||
| 3625 : | Changed the type of the nodestatus, so that: | ||
| 3626 : | |||
| 3627 : | SPILLED(~1) is now SPILLED | ||
| 3628 : | SPILLED(m) where m>=0 is now MEMREG(m) | ||
| 3629 : | SPILLED(s) where s<~1 is now SPILL_LOC(~s) | ||
| 3630 : | |||
| 3631 : | ---------------------------------------------------------------------- | ||
| 3632 : | blume | 697 | Name: Matthias Blume |
| 3633 : | blume | 703 | Date: 2000/09/07 14:45:00 JST |
| 3634 : | Tag: blume-20000907-cmerrmsg | ||
| 3635 : | Description: | ||
| 3636 : | |||
| 3637 : | Small tweak to CM to avoid getting ML syntax error messages twice. | ||
| 3638 : | |||
| 3639 : | ---------------------------------------------------------------------- | ||
| 3640 : | Name: Matthias Blume | ||
| 3641 : | blume | 701 | Date: 2000/08/31 18:00:00 JST |
| 3642 : | Tag: blume-20000831-cvsbootfiles | ||
| 3643 : | Description: | ||
| 3644 : | |||
| 3645 : | New URL for boot files (because the 110.29 files on the BL server do | ||
| 3646 : | now work correctly with my updated install scripts for yacc and lex). | ||
| 3647 : | |||
| 3648 : | ---------------------------------------------------------------------- | ||
| 3649 : | Name: Matthias Blume | ||
| 3650 : | blume | 697 | Date: 2000/08/08 12:33:00 JST |
| 3651 : | Tag: blume-20000808-manual | ||
| 3652 : | Description: | ||
| 3653 : | |||
| 3654 : | Tiny update to CM manual. | ||
| 3655 : | |||
| 3656 : | ---------------------------------------------------------------------- | ||
| 3657 : | leunga | 695 | Name: Allen Leung |
| 3658 : | Date: 2000/08/7 19:31:00 | ||
| 3659 : | Tag: leunga-20000807-a-whole-bunch-of-stuff | ||
| 3660 : | |||
| 3661 : | Moby, C--, SSA, x86, machine descriptions etc. Should only affect C-- | ||
| 3662 : | and Mobdy. | ||
| 3663 : | |||
| 3664 : | 1. x86 | ||
| 3665 : | |||
| 3666 : | a. Fixes to peephole module by John and Dan. | ||
| 3667 : | b. Assembly fix to SETcc by Allen. | ||
| 3668 : | c. Fix to c-call by John. | ||
| 3669 : | d. Fix to spilling by John. (This one deals with the missing FSTPT case) | ||
| 3670 : | e. Instruction selection optimization to SETcc as suggested by John. | ||
| 3671 : | |||
| 3672 : | For example, | ||
| 3673 : | |||
| 3674 : | MV(32, x, COND(32, CMP(32, LT, a, b), LI 1, LI 0)) | ||
| 3675 : | |||
| 3676 : | should generate: | ||
| 3677 : | |||
| 3678 : | MOVL a, x | ||
| 3679 : | SUBL b, x | ||
| 3680 : | SHRL 31, x | ||
| 3681 : | |||
| 3682 : | 2. IR stuff | ||
| 3683 : | |||
| 3684 : | A bunch of new DJ-graph related algorithms added. These | ||
| 3685 : | speed up SSA construction. | ||
| 3686 : | |||
| 3687 : | 3. SSA + Scheduling | ||
| 3688 : | |||
| 3689 : | Added code for SSA and scheduling to the repository | ||
| 3690 : | |||
| 3691 : | ---------------------------------------------------------------------- | ||
| 3692 : | george | 694 | Name: Lal George |
| 3693 : | Date: 2000/07/27 11:53:14 EDT | ||
| 3694 : | |||
| 3695 : | Tag: lal-20000727-linux-ppc | ||
| 3696 : | Description: | ||
| 3697 : | |||
| 3698 : | Made changes to support Linux PPC. | ||
| 3699 : | p.s. I have confirmation that the 110.29 boot files work fine. | ||
| 3700 : | |||
| 3701 : | ---------------------------------------------------------------------- | ||
| 3702 : | blume | 684 | Name: Matthias Blume |
| 3703 : | blume | 692 | Date: 2000/07/27 17:40:00 JST |
| 3704 : | Tag: blume-20000727-scripts | ||
| 3705 : | Description: | ||
| 3706 : | |||
| 3707 : | !!!! WARNING !!!! | ||
| 3708 : | You must recompile the runtime system! | ||
| 3709 : | !!!! WARNING !!!! | ||
| 3710 : | |||
| 3711 : | This is basically another round of script-enhancements: | ||
| 3712 : | |||
| 3713 : | 1. sml, ml-build, and ml-makedepend accept options -D and -U to define | ||
| 3714 : | and undefine CM preprocessor symbols. | ||
| 3715 : | |||
| 3716 : | 2. ml-build avoids generating a new heap image if it finds that the | ||
| 3717 : | existing one is still ok. (The condition is that no ML file had to | ||
| 3718 : | be recompiled and all ML files are found to be older that the heap | ||
| 3719 : | file.) | ||
| 3720 : | |||
| 3721 : | To make this work smoothly, I also hacked the runtime system as | ||
| 3722 : | well as SMLofNJ.SysInfo to get access to the heap image suffix | ||
| 3723 : | (.sparc-solaris, ...) that is currently being used. | ||
| 3724 : | |||
| 3725 : | Moreover, the signature of CM.mk_standalone has changed. See the | ||
| 3726 : | CM manual. | ||
| 3727 : | |||
| 3728 : | 3. ml-makedepend accepts additional options -n, -a, and -o. (See the | ||
| 3729 : | CM manual for details.) | ||
| 3730 : | |||
| 3731 : | 4. More CM manual updates: | ||
| 3732 : | - all of the above has been documented. | ||
| 3733 : | - there is now a section describing the (CM-related) command line | ||
| 3734 : | arguments that are accepted by the "sml" command | ||
| 3735 : | |||
| 3736 : | ---------------------------------------------------------------------- | ||
| 3737 : | Name: Matthias Blume | ||
| 3738 : | blume | 691 | Date: 2000/07/25 16:20:00 JST |
| 3739 : | Tag: blume-20000725-makedepend | ||
| 3740 : | Description: | ||
| 3741 : | |||
| 3742 : | Added a script called ml-makedepend. This can be used in makefiles | ||
| 3743 : | for Unix' make in a way very similar to the "makedepend" command for | ||
| 3744 : | C. | ||
| 3745 : | |||
| 3746 : | The script internally uses function CM.sources. | ||
| 3747 : | |||
| 3748 : | Synopsis: | ||
| 3749 : | |||
| 3750 : | ml-makedepend [-f makefile] cmfile targetname | ||
| 3751 : | |||
| 3752 : | The default for the makefile is "makefile" (or "Makefile" should | ||
| 3753 : | "makefile" not exist). | ||
| 3754 : | |||
| 3755 : | ml-makedepend adds a cmfile/targetname-specific section to this | ||
| 3756 : | makefile (after removing the previous version of this section). The | ||
| 3757 : | section contains a single dependency specification with targetname on | ||
| 3758 : | the LHS (targetname is an arbitrary name), and a list of files derived | ||
| 3759 : | from the cmfile on the RHS. Some of the files on the RHS are | ||
| 3760 : | ARCH/OPSYS-specific. Therefore, ml-makedepend inserts references to | ||
| 3761 : | "make" variables $(ARCH) and $(OPSYS) in place of the corresponding | ||
| 3762 : | path names. The makefile writer is responsible for making sure that | ||
| 3763 : | these variables have correct at the time "make" is invoked. | ||
| 3764 : | |||
| 3765 : | ---------------------------------------------------------------------- | ||
| 3766 : | Name: Matthias Blume | ||
| 3767 : | blume | 690 | Date: 2000/07/22 23:30:00 JST |
| 3768 : | Tag: blume-20000722-urlupdate | ||
| 3769 : | Description: | ||
| 3770 : | |||
| 3771 : | Changed BOOT and config/srcarchiveurl to point to BL server: | ||
| 3772 : | |||
| 3773 : | ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.29/ | ||
| 3774 : | |||
| 3775 : | ---------------------------------------------------------------------- | ||
| 3776 : | Name: Matthias Blume | ||
| 3777 : | blume | 686 | Date: 2000/07/18 18:00:00 JST |
| 3778 : | Tag: blume-20000718-Version_110_29 | ||
| 3779 : | Description: | ||
| 3780 : | |||
| 3781 : | 1. Updated src/compiler/TopLevel/main/version.sml to version 110.29 | ||
| 3782 : | |||
| 3783 : | 2. Updated config/version to 110.29 | ||
| 3784 : | |||
| 3785 : | 3. Updated config/srcarchiveurl | ||
| 3786 : | |||
| 3787 : | 3. New boot files! | ||
| 3788 : | ftp://ftp.cs.princeton.edu/pub/people/blume/sml/110.29-autofetch | ||
| 3789 : | |||
| 3790 : | ---------------------------------------------------------------------- | ||
| 3791 : | Name: Matthias Blume | ||
| 3792 : | blume | 684 | Date: 2000/07/11 13:58:00 JST |
| 3793 : | Tag: blume-20000711-doctypo | ||
| 3794 : | Description: | ||
| 3795 : | |||
| 3796 : | Fixed a few typos in CM manual. | ||
| 3797 : | |||
| 3798 : | ---------------------------------------------------------------------- | ||
| 3799 : | leunga | 683 | Name: Allen Leung |
| 3800 : | Date: 2000/06/15 00:38:00 | ||
| 3801 : | Tag: leunga-20000704-sparc-x86 | ||
| 3802 : | |||
| 3803 : | 1. x86 peephole improvement sp += k; sp -= k => nop [from John] | ||
| 3804 : | 2. fix to x86 RET bug [found by Dan Grossman] | ||
| 3805 : | 3. sparc assembly bug fix for ticc instructions [found by Fermin] | ||
| 3806 : | |||
| 3807 : | Affects c-- and moby only | ||
| 3808 : | |||
| 3809 : | ---------------------------------------------------------------------- | ||
| 3810 : | blume | 680 | Name: Matthias Blume |
| 3811 : | blume | 682 | Date: 2000/07/04 15:26:00 |
| 3812 : | Tag: blume-20000704-trigger | ||
| 3813 : | Description: | ||
| 3814 : | |||
| 3815 : | 1. Improvements to CM manual. | ||
| 3816 : | 2. SMLofNJ.Internals.BTrace.trigger reinstated as an alternative way | ||
| 3817 : | of getting a back-trace. The function, when called, raises an | ||
| 3818 : | internal exception which explicitly carries the full back-trace history, | ||
| 3819 : | so it is unaffected by any intervening handle-raise pairs ("trivial" | ||
| 3820 : | or not). The interactive loop will print that history once it arrives | ||
| 3821 : | at top level. | ||
| 3822 : | Short of having all exceptions implicitly carry the full history, the | ||
| 3823 : | recommended way of using this facility is: | ||
| 3824 : | - compile your program with instrumentation "on" | ||
| 3825 : | - run it, when it raises an exception, look at the history | ||
| 3826 : | - if the history is "cut off" because of some handler, go and modify | ||
| 3827 : | your program so that it explicitly calls BTrace.trigger | ||
| 3828 : | - recompile (still instrumented), and rerun; look at the full history | ||
| 3829 : | |||
| 3830 : | ---------------------------------------------------------------------- | ||
| 3831 : | Name: Matthias Blume | ||
| 3832 : | Date: 2000/07/03 15:36:00 JST | ||
| 3833 : | blume | 680 | Tag: blume-20000702-manual |
| 3834 : | Description: | ||
| 3835 : | |||
| 3836 : | Small corrections and updates to CM manual. | ||
| 3837 : | |||
| 3838 : | ---------------------------------------------------------------------- | ||
| 3839 : | blume | 682 | Name: Matthias Blume |
| 3840 : | blume | 679 | Date: 2000/06/29 16:04:00 JST |
| 3841 : | Tag: blume-20000629-yacctool | ||
| 3842 : | Description: | ||
| 3843 : | |||
| 3844 : | Changes: | ||
| 3845 : | |||
| 3846 : | 1. Class "mlyacc" now takes separate arguments to pass options to | ||
| 3847 : | generated .sml- and .sig-files independently. | ||
| 3848 : | 2. Corresponding CM manual updates. | ||
| 3849 : | 3. BTrace module now also reports call sites. (However, for loop clusters | ||
| 3850 : | it only shows from where the cluster was entered.) There are associated | ||
| 3851 : | modifications to core.sml, internals.{sig,sml}, btrace.sml, and btimp.sml. | ||
| 3852 : | |||
| 3853 : | ---------------------------------------------------------------------- | ||
| 3854 : | blume | 675 | Name: Matthias Blume |
| 3855 : | blume | 678 | Date: 2000/06/27 16:51:00 JST |
| 3856 : | Tag: blume-20000627-noweb | ||
| 3857 : | Description: | ||
| 3858 : | |||
| 3859 : | Changes: | ||
| 3860 : | |||
| 3861 : | 1. Implemented "subdir" and "witness" options for noweb tool. | ||
| 3862 : | This caused some slight internal changes in CM's tool implementation. | ||
| 3863 : | 2. Fixed bug in "tool plugin" mechanism. This is essentially cleaning | ||
| 3864 : | some remaining issues from earlier path anchor changes. | ||
| 3865 : | 3. Updated CM manual accordingly. | ||
| 3866 : | |||
| 3867 : | 4. Changed implementation of back-tracing so that I now consider it | ||
| 3868 : | ready for prime-time. | ||
| 3869 : | |||
| 3870 : | In particular, you don't have to explicitly trigger the back-trace | ||
| 3871 : | anymore. Instead, if you are running BTrace-instrumented code and | ||
| 3872 : | there is an uncaught exception (regardless of whether or not it was | ||
| 3873 : | raised in instrumented code), the top-level evalloop will print | ||
| 3874 : | the back-trace. | ||
| 3875 : | |||
| 3876 : | Features: | ||
| 3877 : | |||
| 3878 : | - Instrumented and uninstrumented code work together seemlessly. | ||
| 3879 : | (Of course, uninstrumented code is never mentioned in actual | ||
| 3880 : | back-traces.) | ||
| 3881 : | |||
| 3882 : | - Asymptotic time- and space-complexity of instrumented code is | ||
| 3883 : | equal to that of uninstrumented code. (This means that | ||
| 3884 : | tail-recursion is preserved by the instrumentation phase.) | ||
| 3885 : | |||
| 3886 : | - Modules whose code has been instrumented in different sessions | ||
| 3887 : | work together without problem. | ||
| 3888 : | |||
| 3889 : | - There is no penalty whatsoever on uninstrumented code. | ||
| 3890 : | |||
| 3891 : | - There is no penalty on "raise" expressions, even in | ||
| 3892 : | instrumented code. | ||
| 3893 : | |||
| 3894 : | A potential bug (or perhaps it is a feature, too): | ||
| 3895 : | |||
| 3896 : | A back-trace reaches no further than the outermost instrumented | ||
| 3897 : | non-trivial "raise". Here, a "trivial" raise is one that is the | ||
| 3898 : | sole RHS of a "handle" rule. Thus, back-traces reach trough | ||
| 3899 : | |||
| 3900 : | <exp> handle e => raise e | ||
| 3901 : | |||
| 3902 : | and even | ||
| 3903 : | |||
| 3904 : | <exp> handle Foo => raise Bar | ||
| 3905 : | |||
| 3906 : | and, of course, through | ||
| 3907 : | |||
| 3908 : | <exp> handle Foo => ... | ||
| 3909 : | |||
| 3910 : | if the exception was not Foo. | ||
| 3911 : | |||
| 3912 : | Back-traces always reach right through any un-instrumented code | ||
| 3913 : | including any of its "handle" expressions, trivial or not. | ||
| 3914 : | |||
| 3915 : | To try this out, do the following: | ||
| 3916 : | |||
| 3917 : | - Erase all existing binfiles for your program. | ||
| 3918 : | (You may keep binfiles for those modules where you think you | ||
| 3919 : | definitely don't need back-tracing.) | ||
| 3920 : | - Turn on back-trace instrumentation: | ||
| 3921 : | SMLofNJ.Internals.BTrace.mode (SOME true); | ||
| 3922 : | - Recompile your program. (I.e., run "CM.make" or "use".) | ||
| 3923 : | - You may now turn instrumentation off again (if you want): | ||
| 3924 : | SMLofNJ.Internals.BTrace.mode (SOME false); | ||
| 3925 : | - Run your program as usual. If it raises an exception that | ||
| 3926 : | reaches the interactive toplevel, then a back-trace will | ||
| 3927 : | automatically be printed. After that, the toplevel loop | ||
| 3928 : | will print the exception history as usual. | ||
| 3929 : | |||
| 3930 : | ---------------------------------------------------------------------- | ||
| 3931 : | Name: Matthias Blume | ||
| 3932 : | blume | 677 | Date: 2000/06/26 09:56:46 JST |
| 3933 : | Tag: blume-20000626-setup | ||
| 3934 : | Description: | ||
| 3935 : | |||
| 3936 : | CM: - setup-parameter to "sml" added; this can be used to run arbitrary | ||
| 3937 : | ML code before and after compiling a file (e.g., to set compiler | ||
| 3938 : | flags) | ||
| 3939 : | |||
| 3940 : | Compiler: - improved btrace API (in core.sml, internals.{sig,sml}) | ||
| 3941 : | - associated changes to btrace.sml (BTrace instrumentation pass) | ||
| 3942 : | - cleaner implementation of btimp.sml (BTrace tracing and report | ||
| 3943 : | module) | ||
| 3944 : | |||
| 3945 : | CM manual: * new path encoding documented | ||
| 3946 : | * description of setup-parameter to "sml" added | ||
| 3947 : | |||
| 3948 : | The biggest user-visible change to back-tracing is that it is no | ||
| 3949 : | longer necessary to compile all traced modules within the same | ||
| 3950 : | session. (This was a real limitation.) | ||
| 3951 : | |||
| 3952 : | ---------------------------------------------------------------------- | ||
| 3953 : | Name: Matthias Blume | ||
| 3954 : | blume | 676 | Date: 2000/06/24 12:40:00 JST |
| 3955 : | Tag: blume-20000624-startup | ||
| 3956 : | Description: | ||
| 3957 : | |||
| 3958 : | Fixes startup slowdown problem. (I was calling SrcPath.sync a _tad_ | ||
| 3959 : | bit too often -- to put it mildly. :) | ||
| 3960 : | |||
| 3961 : | ---------------------------------------------------------------------- | ||
| 3962 : | Name: Matthias Blume | ||
| 3963 : | blume | 675 | Date: 2000/06/23 18:20:00 JST |
| 3964 : | Tag: blume-20000623-btrace | ||
| 3965 : | Description: | ||
| 3966 : | |||
| 3967 : | This updates adds a backtrace facility to aid programmers in debugging | ||
| 3968 : | their programs. This involves the following changes: | ||
| 3969 : | |||
| 3970 : | 1. Module system/smlnj/init/core.sml (structure _Core) now has hooks for | ||
| 3971 : | keeping track of the current call stack. When programs are compiled | ||
| 3972 : | in a special mode, the compiler will insert calls to these hooks | ||
| 3973 : | into the user program. | ||
| 3974 : | "Hook" means that it is possible for different implementations of | ||
| 3975 : | back-tracing to register themselves (at different times). | ||
| 3976 : | |||
| 3977 : | 2. compiler/MiscUtil/profile/btrace.sml implements the annotation phase | ||
| 3978 : | as an Absyn.dec->Absyn.dec rewrite. Normally this phase is turned off. | ||
| 3979 : | It can be turned on using this call: | ||
| 3980 : | SMLofNJ.Internals.BTrace.mode (SOME true); | ||
| 3981 : | Turning it off again: | ||
| 3982 : | SMLofNJ.Internals.BTrace.mode (SOME false); | ||
| 3983 : | Querying the current status: | ||
| 3984 : | SMLofNJ.Internals.BTrace.mode NONE; | ||
| 3985 : | Annotated programs are about twice as big as normal ones, and they | ||
| 3986 : | run a factor of 2 to 4 slower with a dummy back-trace plugin (one | ||
| 3987 : | where all hooks do nothing). The slowdown with a plugin that is | ||
| 3988 : | actually useful (such as the one supplied by default) is even greater, | ||
| 3989 : | but in the case of the default plugin it is still only an constant | ||
| 3990 : | factor (amortized). | ||
| 3991 : | |||
| 3992 : | 3. system/Basis/Implementation/NJ/internals.{sig,sml} have been augmented | ||
| 3993 : | with a sub-structure BTrace for controlling back-tracing. In particular, | ||
| 3994 : | the above-mentioned function "mode" controls whether the annotation | ||
| 3995 : | phase is invoked by the compiler. Another important function is | ||
| 3996 : | "trigger": when called it aborts the current execution and causes | ||
| 3997 : | the top-level loop to print a full back-trace. | ||
| 3998 : | |||
| 3999 : | 4. compiler/MiscUtil/profile/btimp.sml is the current default plugin | ||
| 4000 : | for back-tracing. It keeps track of the dynamic call stack and in | ||
| 4001 : | addition to that it keeps a partial history at each "level" of that | ||
| 4002 : | stack. For example, if a tail-calls b, b tail-calls c, and c tail-calls | ||
| 4003 : | d and b (at separate times, dynamically), then the report will show: | ||
| 4004 : | |||
| 4005 : | GOTO d | ||
| 4006 : | /c | ||
| 4007 : | GOTO \b | ||
| 4008 : | CALL a | ||
| 4009 : | |||
| 4010 : | This shows that there was an initial non-tail call of a, then a | ||
| 4011 : | tail-call to b or c, looping behavior in a cluster of functions that | ||
| 4012 : | consist of b and c, and then a goto from that cluster (i.e., either from | ||
| 4013 : | b or from c) to d. | ||
| 4014 : | |||
| 4015 : | Note that (depending on the user program) the amount of information | ||
| 4016 : | that the back-trace module has to keep track of at each level is bounded | ||
| 4017 : | by a constant. Thus, the whole implementation has the same asymptotical | ||
| 4018 : | complexity as the original program (both in space and in time). | ||
| 4019 : | |||
| 4020 : | 5. compiler/TopLevel/interact/evalloop.sml has been modified to | ||
| 4021 : | handle the special exception SMLofNJ.Internals.BTrace.BTrace | ||
| 4022 : | which is raised by the "trigger" function mentioned above. | ||
| 4023 : | |||
| 4024 : | Notes on usage: | ||
| 4025 : | |||
| 4026 : | - Annotated code works well together with unannotated code: | ||
| 4027 : | Unannotated calls simply do not show up at all in the backtrace. | ||
| 4028 : | |||
| 4029 : | - It is not a good idea to let modules that were annotated during | ||
| 4030 : | different sessions run at the same time. This is because the compiler | ||
| 4031 : | chooses small integers to identify individual functions, and there | ||
| 4032 : | will be clashes if different modules were compiled in separate sessions. | ||
| 4033 : | (Nothing will crash, and you will even be told about the clashes, but | ||
| 4034 : | back-trace information will in general not be useful.) | ||
| 4035 : | |||
| 4036 : | - Back-tracing can be confused by callcc and capture. | ||
| 4037 : | |||
| 4038 : | - The only way of getting a back-trace right now is to explicitly | ||
| 4039 : | invoke the "trigger" function from your user program. Eventually, we | ||
| 4040 : | should make every exception carry back-trace information (if | ||
| 4041 : | available). But since this creates more overhead at "raise"-time | ||
| 4042 : | (similar to the current exnHistory overhead), I have not yet | ||
| 4043 : | implemented this. (The implementation will be rather easy.) With | ||
| 4044 : | exceptions carrying back-trace information, this facility will be even | ||
| 4045 : | more useful because users don't need to modify their programs... | ||
| 4046 : | |||
| 4047 : | - While it is possible to compile the compiler with back-trace | ||
| 4048 : | annotations turned on (I did it to get some confidence in | ||
| 4049 : | correctness), you must make absolutely sure that core.sml and | ||
| 4050 : | btimp.sml are compiled WITHOUT annotation! (core.sml cannot actually | ||
| 4051 : | be compiled with annotation because there is no core access yet, but | ||
| 4052 : | if you compile btimp.sml with annotation, then the system will go into | ||
| 4053 : | an infinite recursion and crash.) | ||
| 4054 : | Since CM currently does not know about BTrace, the only way to turn | ||
| 4055 : | annotations on and off for different modules of the compiler is to | ||
| 4056 : | interrupt CMB.make, change the settings, and re-invoke it. Of course, | ||
| 4057 : | this is awkward and clumsy. | ||
| 4058 : | |||
| 4059 : | Sample sessions: | ||
| 4060 : | |||
| 4061 : | Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000 | ||
| 4062 : | - SMLofNJ.Internals.BTrace.mode (SOME true); | ||
| 4063 : | [autoloading] | ||
| 4064 : | [autoloading done] | ||
| 4065 : | val it = false : bool | ||
| 4066 : | - structure X = struct | ||
| 4067 : | - fun main n = let | ||
| 4068 : | - fun a (x, 0) = d x | ||
| 4069 : | - | a (x, n) = b (x, n - 1) | ||
| 4070 : | - and b (x, n) = c (x, n) | ||
| 4071 : | - and c (x, n) = a (x, n) | ||
| 4072 : | - and d x = e (x, 3) | ||
| 4073 : | - and e (x, 0) = f x | ||
| 4074 : | - | e (x, n) = e (x, n - 1) | ||
| 4075 : | - and f 0 = SMLofNJ.Internals.BTrace.trigger () | ||
| 4076 : | - | f n = n * g (n - 1) | ||
| 4077 : | - and g n = a (n, 3) | ||
| 4078 : | - in | ||
| 4079 : | - f n | ||
| 4080 : | - end | ||
| 4081 : | - end; | ||
| 4082 : | structure X : sig val main : int -> int end | ||
| 4083 : | - X.main 3; | ||
| 4084 : | *** BACK-TRACE *** | ||
| 4085 : | GOTO stdIn:4.2-13.20: X.main[2].f | ||
| 4086 : | GOTO-( stdIn:4.2-13.20: X.main[2].e | ||
| 4087 : | GOTO stdIn:4.2-13.20: X.main[2].d | ||
| 4088 : | / stdIn:4.2-13.20: X.main[2].a | ||
| 4089 : | | stdIn:4.2-13.20: X.main[2].b | ||
| 4090 : | GOTO-\ stdIn:4.2-13.20: X.main[2].c | ||
| 4091 : | CALL stdIn:4.2-13.20: X.main[2].g | ||
| 4092 : | GOTO stdIn:4.2-13.20: X.main[2].f | ||
| 4093 : | GOTO-( stdIn:4.2-13.20: X.main[2].e | ||
| 4094 : | GOTO stdIn:4.2-13.20: X.main[2].d | ||
| 4095 : | / stdIn:4.2-13.20: X.main[2].a | ||
| 4096 : | | stdIn:4.2-13.20: X.main[2].b | ||
| 4097 : | GOTO-\ stdIn:4.2-13.20: X.main[2].c | ||
| 4098 : | CALL stdIn:4.2-13.20: X.main[2].g | ||
| 4099 : | GOTO stdIn:4.2-13.20: X.main[2].f | ||
| 4100 : | GOTO-( stdIn:4.2-13.20: X.main[2].e | ||
| 4101 : | GOTO stdIn:4.2-13.20: X.main[2].d | ||
| 4102 : | / stdIn:4.2-13.20: X.main[2].a | ||
| 4103 : | | stdIn:4.2-13.20: X.main[2].b | ||
| 4104 : | GOTO-\ stdIn:4.2-13.20: X.main[2].c | ||
| 4105 : | CALL stdIn:4.2-13.20: X.main[2].g | ||
| 4106 : | GOTO stdIn:4.2-13.20: X.main[2].f | ||
| 4107 : | CALL stdIn:2.15-17.4: X.main[2] | ||
| 4108 : | - | ||
| 4109 : | |||
| 4110 : | (Note that because of a FLINt bug the above code currently does not | ||
| 4111 : | compile without BTrace turned on.) | ||
| 4112 : | |||
| 4113 : | Here is another example, using my modified Tiger compiler: | ||
| 4114 : | |||
| 4115 : | Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000 | ||
| 4116 : | - SMLofNJ.Internals.BTrace.mode (SOME true); | ||
| 4117 : | [autoloading] | ||
| 4118 : | [autoloading done] | ||
| 4119 : | val it = false : bool | ||
| 4120 : | - CM.make "sources.cm"; | ||
| 4121 : | [autoloading] | ||
| 4122 : | ... | ||
| 4123 : | [autoloading done] | ||
| 4124 : | [scanning sources.cm] | ||
| 4125 : | [parsing (sources.cm):parse.sml] | ||
| 4126 : | [creating directory CM/SKEL ...] | ||
| 4127 : | [parsing (sources.cm):tiger.lex.sml] | ||
| 4128 : | ... | ||
| 4129 : | [wrote CM/sparc-unix/semant.sml] | ||
| 4130 : | [compiling (sources.cm):main.sml] | ||
| 4131 : | [wrote CM/sparc-unix/main.sml] | ||
| 4132 : | [New bindings added.] | ||
| 4133 : | val it = true : bool | ||
| 4134 : | - Main.compile ("../testcases/merge.tig", "foo.out"); | ||
| 4135 : | *** BACK-TRACE *** | ||
| 4136 : | CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trvar | ||
| 4137 : | CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp | ||
| 4138 : | CALL lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2] | ||
| 4139 : | GOTO lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2] | ||
| 4140 : | CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp | ||
| 4141 : | CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp | ||
| 4142 : | CALL lib/semant.sml:488.3-505.6: SemantFun[2].transDec.trdec[2].transBody[2] | ||
| 4143 : | / lib/semant.sml:411.65-543.8: SemantFun[2].transDec | ||
| 4144 : | CALL-\ lib/semant.sml:413.2-540.9: SemantFun[2].transDec.trdec[2] | ||
| 4145 : | CALL lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp | ||
| 4146 : | CALL lib/semant.sml:8.52-558.4: SemantFun[2].transProg[2] | ||
| 4147 : | CALL main.sml:1.18-118.4: Main.compile[2] | ||
| 4148 : | - | ||
| 4149 : | |||
| 4150 : | ---------------------------------------------------------------------- | ||
| 4151 : | blume | 673 | Name: Matthias Blumen |
| 4152 : | Date: 2000/06/21 18:00:00 JST | ||
| 4153 : | Tag: blume-20000621-manual | ||
| 4154 : | Description: | ||
| 4155 : | |||
| 4156 : | CM manual update: Path environments documented. | ||
| 4157 : | |||
| 4158 : | ---------------------------------------------------------------------- | ||
| 4159 : | blume | 666 | Name: Matthias Blume |
| 4160 : | blume | 672 | Date: 2000/06/19 13:40:00 |
| 4161 : | Tag: blume-20000619-manual | ||
| 4162 : | Description: | ||
| 4163 : | |||
| 4164 : | CM manual and system/README update. This only covers the fact that | ||
| 4165 : | there are no more implicit anchors. (Path environments and the "bind" | ||
| 4166 : | option to "cm" have yet to be documented.) | ||
| 4167 : | |||
| 4168 : | ---------------------------------------------------------------------- | ||
| 4169 : | Name: Matthias Blume | ||
| 4170 : | blume | 671 | Date: 2000/06/19 11:05:00 JST |
| 4171 : | Tag: blume-20000619-chdir-bugfix | ||
| 4172 : | Description: | ||
| 4173 : | |||
| 4174 : | Fixed a bug in new SrcPath module that sometimes led to a bad chDir call. | ||
| 4175 : | |||
| 4176 : | ---------------------------------------------------------------------- | ||
| 4177 : | Name: Matthias Blume | ||
| 4178 : | blume | 670 | Date: 2000/06/18 22:00:10 JST |
| 4179 : | Tag: blume-20000618-implicit-anchors-really-gone | ||
| 4180 : | Description: | ||
| 4181 : | |||
| 4182 : | I updates the previous HISTORY entry where I forgot to mention that | ||
| 4183 : | implicit anchors are no longer with us. | ||
| 4184 : | |||
| 4185 : | The current update also gets rid of the (now useless) controller | ||
| 4186 : | CM.Control.implicit_anchors. | ||
| 4187 : | |||
| 4188 : | ---------------------------------------------------------------------- | ||
| 4189 : | Name: Matthias Blume | ||
| 4190 : | blume | 666 | Date: 2000/06/16 17:30:00 JST |
| 4191 : | Tag: blume-20000616-anchorenv | ||
| 4192 : | Description: | ||
| 4193 : | |||
| 4194 : | This patch implements the long anticipated (just kidding :) "anchor | ||
| 4195 : | environment" mechanism. In the course of doing this, I also | ||
| 4196 : | re-implemented CM's internal "SrcPath" module from scratch. The new | ||
| 4197 : | one should be more robust in certain boundary cases. In any case, it | ||
| 4198 : | is a lot cleaner than its predecessor (IMHO). | ||
| 4199 : | |||
| 4200 : | This time, although there is yet another boot file format change, I | ||
| 4201 : | kept the unpickler backward-compatible. As a result, no new bootfiles | ||
| 4202 : | are necessary and bootstrapping is straightforward. (You cannot read | ||
| 4203 : | new bootfiles into an old system, but the other way around is no | ||
| 4204 : | problem.) | ||
| 4205 : | |||
| 4206 : | Visible changes: | ||
| 4207 : | |||
| 4208 : | blume | 670 | ** 0. Implicit path anchors (without the leading $-symbol) are no |
| 4209 : | longer recognized at all. This means that such path names are not | ||
| 4210 : | illegal either. For example, the name basis.cm simply refers to a | ||
| 4211 : | local file called "basis.cm" (i.e, the name is an ordinary path | ||
| 4212 : | relative to .cm-files directory). Or, to put it differently, only | ||
| 4213 : | names that start with $ are anchored paths. | ||
| 4214 : | blume | 666 | |
| 4215 : | blume | 670 | ** 1. The $<singlearc> abbreviation for $/<singlearc> has finally |
| 4216 : | vanished. | ||
| 4217 : | |||
| 4218 : | John (Reppy) had critizised this as soon as I originally proposed and | ||
| 4219 : | blume | 666 | implemented it, but at that time I did not really deeply believe |
| 4220 : | him. :) Now I came full-circle because I need the $<singlearc> syntax | ||
| 4221 : | in another place where it cannot be seen as an abbreviation for | ||
| 4222 : | $/<singlearc>. To avoid the confusion, $<singlearc> now means what it | ||
| 4223 : | seems to mean (i.e., it "expands" into the corresponding anchor | ||
| 4224 : | value). | ||
| 4225 : | |||
| 4226 : | However, when paths are used as members in CM description files, it | ||
| 4227 : | continues to be true that there must be at least another arc after the | ||
| 4228 : | anchor. This is now enforced separately during semantic analysis | ||
| 4229 : | blume | 670 | (i.e., from a lexical/syntactical point of view, the notation is ok.) |
| 4230 : | blume | 666 | |
| 4231 : | ** 2. The "cm" class now accepts an option "bind". The option's value | ||
| 4232 : | is a sub-option list of precisely two items -- one labeled "anchor" | ||
| 4233 : | and the other one labeled "value". As you might expect, "anchor" is | ||
| 4234 : | used to specify an anchor name to be bound, and "value" specifies what | ||
| 4235 : | the anchor is being bound to. | ||
| 4236 : | |||
| 4237 : | The value must be a directory name and can be given in either standard | ||
| 4238 : | syntax (including the possibility that it is itself an anchored path) | ||
| 4239 : | or native syntax. | ||
| 4240 : | |||
| 4241 : | Examples: | ||
| 4242 : | |||
| 4243 : | foo.cm (bind:(anchor:bar value:$mystuff/bar)) | ||
| 4244 : | lib.cm (bind:(anchor:a value:"H:\\x\\y\\z")) (* only works under windows *) | ||
| 4245 : | |||
| 4246 : | and so on. | ||
| 4247 : | |||
| 4248 : | The meaning of this is that the .cm-file will be processed with an | ||
| 4249 : | augmented anchor environment where the given anchor(s) is/are bound to | ||
| 4250 : | the given values(s). | ||
| 4251 : | |||
| 4252 : | The rationale for having this feature is this: Suppose you are trying | ||
| 4253 : | to use two different (already stable) libraries a.cm and b.cm (that | ||
| 4254 : | you perhaps didn't write yourself). Further, suppose each of these | ||
| 4255 : | two libraries internally uses its own auxiliary library $aux/lib.cm. | ||
| 4256 : | Normally you would now have a problem because the anchor "lib" can not | ||
| 4257 : | be bound to more than one value globally. Therefore, the project that | ||
| 4258 : | uses both a.cm and b.cm must locally redirect the anchor to some other | ||
| 4259 : | place: | ||
| 4260 : | |||
| 4261 : | a.cm (bind:(anchor:lib value:/usr/lib/smlnj/a-stuff)) | ||
| 4262 : | b.cm (bind:(anchor:lib value:/usr/lib/smlnj/b-stuff)) | ||
| 4263 : | |||
| 4264 : | This hard-wires $lib/aux.cm to /usr/lib/smlnj/a-stuff/aux.cm or | ||
| 4265 : | /usr/lib/smlnj/b-stuff/aux.cm, respectively. | ||
| 4266 : | |||
| 4267 : | Hard-wiring path names is a bit inflexible (and CM will verbosely warn | ||
| 4268 : | you when you do so at the time of CM.stabilize). Therefore, you can | ||
| 4269 : | also use an anchored path as the value: | ||
| 4270 : | |||
| 4271 : | a.cm (bind:(anchor:lib value:$a-lib)) | ||
| 4272 : | b.cm (bind:(anchor:lib value:$b-lib)) | ||
| 4273 : | |||
| 4274 : | Now you can globally configure (using the usual CM.Anchor.anchor or | ||
| 4275 : | pathconfig machinery) bindings for "a-lib" and "b-lib". Since "lib" | ||
| 4276 : | itself is always locally bound, setting it globally is no longer | ||
| 4277 : | meaningful or necessary (but it does not hurt either). In fact, "lib" | ||
| 4278 : | can still be used as a global anchor for separate purposes. As a | ||
| 4279 : | matter of fact, one can locally define "lib" in terms of a global | ||
| 4280 : | "lib": | ||
| 4281 : | |||
| 4282 : | a.cm (bind:(anchor:lib value:$lib/a)) | ||
| 4283 : | b.cm (bind:(anchor:lib value:$lib/b)) | ||
| 4284 : | |||
| 4285 : | ** 3: The encoding of path names has changed. This affects the way | ||
| 4286 : | path names are shown in CM's progress report and also the internal | ||
| 4287 : | protocol encoding used for parallel make. | ||
| 4288 : | |||
| 4289 : | The encoding now uses one or more ':'-separated segments. Each | ||
| 4290 : | segments corresponds to a file that has been specified relative to the | ||
| 4291 : | file given by its preceding segment. The first segment is either | ||
| 4292 : | relative to the CWD, absolute, or anchored. Each segment itself is | ||
| 4293 : | basically a Unix pathname; all segments but the first are relative. | ||
| 4294 : | |||
| 4295 : | Example: | ||
| 4296 : | |||
| 4297 : | $foo/bar/baz.cm:a/b/c.sml | ||
| 4298 : | |||
| 4299 : | This path denotes the file bar/a/b/c.sml relative to the directory | ||
| 4300 : | denoted by anchor "foo". Notice that the encoding also includes | ||
| 4301 : | baz.cm which is the .cm-file that listed a/b/c.sml. As usual, such | ||
| 4302 : | paths are resolved relative to the .cm-files directory, so baz.cm must | ||
| 4303 : | be ignored to get the "real" pathname. | ||
| 4304 : | |||
| 4305 : | To make this fact more obvious, CM puts the names of such "virtual | ||
| 4306 : | arcs" into parentheses when they appear in progress reports. (No | ||
| 4307 : | parentheses will appear in the internal protocol encoding.) Thus, | ||
| 4308 : | what you really see is: | ||
| 4309 : | |||
| 4310 : | $foo/bar/(baz.cm):a/b/c.sml | ||
| 4311 : | |||
| 4312 : | I find this notation to be much more informative than before. | ||
| 4313 : | |||
| 4314 : | Another new feature of the encoding is that special characters | ||
| 4315 : | including parentheses, colons, (back)slashes, and white space are | ||
| 4316 : | written as \ddd (where ddd is the decimal encoding of the character). | ||
| 4317 : | |||
| 4318 : | blume | 670 | *** The CM manual still needs to be updated. |
| 4319 : | |||
| 4320 : | blume | 666 | ---------------------------------------------------------------------- |
| 4321 : | leunga | 660 | Name: Allen Leung |
| 4322 : | Date: 2000/0 |