| 14 |
|
|
| 15 |
---------------------------------------------------------------------- |
---------------------------------------------------------------------- |
| 16 |
Name: Matthias Blume |
Name: Matthias Blume |
| 17 |
|
Date: 2001/07/19 15:00:00 EDT |
| 18 |
|
Tag: blume-20010719-libreorg |
| 19 |
|
Description: |
| 20 |
|
|
| 21 |
|
This update puts together a fairly extensive but straightforward change |
| 22 |
|
to the way the libraries that implement the interactive system are |
| 23 |
|
organized: |
| 24 |
|
|
| 25 |
|
The biggest change is the elimination of structure Compiler. As a |
| 26 |
|
replacement for this structure, there is now a CM library |
| 27 |
|
(known as $smlnj/compiler.cm or $smlnj/compiler/current.cm) |
| 28 |
|
that exports all the substructures of the original structure Compiler |
| 29 |
|
directly. So instead of saying Compiler.Foo.bar one now simply |
| 30 |
|
says Foo.bar. (The CM libraries actually export a collection of |
| 31 |
|
structures that is richer than the collection of substructures of |
| 32 |
|
structure Compiler.) |
| 33 |
|
|
| 34 |
|
To make the transition smooth, there is a separate library called |
| 35 |
|
$smlnj/compiler/compiler.cm which puts together and exports the |
| 36 |
|
original structure Compiler (or at least something very close to it). |
| 37 |
|
|
| 38 |
|
There are five members of the original structure Compiler |
| 39 |
|
that are not exported directly but which instead became members |
| 40 |
|
of a new structure Backend (described by signature BACKEND). These are: |
| 41 |
|
structure Profile (: PROFILE), structure Compile (: COMPILE), structure |
| 42 |
|
Interact (: INTERACT), structure Machine (: MACHINE), and val |
| 43 |
|
architecture (: string). |
| 44 |
|
|
| 45 |
|
Structure Compiler.Version has become structure CompilerVersion. |
| 46 |
|
|
| 47 |
|
Cross-compilers for alpha32, hppa, ppc, sparc, and x86 are provided |
| 48 |
|
by $smlnj/compiler/<arch>.cm where <arch> is alpha32, hppa, ppc, sparc, |
| 49 |
|
or x86, respectively. |
| 50 |
|
Each of these exports the same frontend structures that |
| 51 |
|
$smlnj/compiler.cm exports. But they do not have a structure Backend |
| 52 |
|
and instead export some structure <Arch>Backend where <Arch> is Alpha32, |
| 53 |
|
Hppa, PPC, Sparc, or X86, respectively. |
| 54 |
|
|
| 55 |
|
Library $smlnj/compiler/all.cm exports the union of the exports of |
| 56 |
|
$smlnj/compiler/<arch>.cm |
| 57 |
|
|
| 58 |
|
There are no structures <Arch>Compiler anymore, use |
| 59 |
|
$smlnj/compiler/<arch>.cm instead. |
| 60 |
|
|
| 61 |
|
Library host-compiler-0.cm is gone. Instead, the internal library |
| 62 |
|
that instantiates CM is now called cm0.cm. Selection of the host |
| 63 |
|
compiler (backend) is no longer done here but. (Responsibility for it |
| 64 |
|
now lies with $smlnj/compiler/current.cm. This seems to be more |
| 65 |
|
logical.) |
| 66 |
|
|
| 67 |
|
Many individual files have been moved or renamed. Some files have |
| 68 |
|
been split into multiple files, and some "dead" files have been deleted. |
| 69 |
|
|
| 70 |
|
Aside from these changes to library organization, there are also changes |
| 71 |
|
to the way the code itself is organized: |
| 72 |
|
|
| 73 |
|
Structure Binfile has been re-implemented in such a way that it no |
| 74 |
|
longer needs any knowledge of the compiler. It exclusively deals |
| 75 |
|
with the details of binfile layout. It no longer invokes the |
| 76 |
|
compiler (for the purpose of creating new prospective binfile |
| 77 |
|
content), and it no longer has any knowledge of how to interpret |
| 78 |
|
pickles. |
| 79 |
|
|
| 80 |
|
Structure Compile (: COMPILE) has been stripped down to the bare |
| 81 |
|
essentials of compilation. It no longer deals with linking/execution. |
| 82 |
|
The interface has been cleaned up considerably. |
| 83 |
|
|
| 84 |
|
Utility routines for dealing with linking and execution have been |
| 85 |
|
moved into their own substructures. |
| 86 |
|
|
| 87 |
|
(The ultimate goal of these changes is to provide a light-weight |
| 88 |
|
binfile loader/linker (at least for, e.g., stable libraries) that |
| 89 |
|
does not require CM or the compiler to be present.) |
| 90 |
|
|
| 91 |
|
CM documentation has been updated to reflect the changes to library |
| 92 |
|
organization. |
| 93 |
|
|
| 94 |
|
---------------------------------------------------------------------- |
| 95 |
|
Name: Matthias Blume |
| 96 |
|
Date: 2001/07/10 17:30:00 EDT |
| 97 |
|
Tag: Release_110_34 |
| 98 |
|
Description: |
| 99 |
|
|
| 100 |
|
Minor tweak to 110.34 (re-tagged): |
| 101 |
|
|
| 102 |
|
- README.html file added to CVS repository |
| 103 |
|
- runtime compiles properly under FreeBSD 3.X and 4.X |
| 104 |
|
|
| 105 |
|
---------------------------------------------------------------------- |
| 106 |
|
Name: Matthias Blume |
| 107 |
|
Date: 2001/07/10 17:30:00 EDT |
| 108 |
|
Tag: Release_110_34 |
| 109 |
|
Description: |
| 110 |
|
|
| 111 |
|
New version number (110.34). New bootfiles. |
| 112 |
|
|
| 113 |
|
---------------------------------------------------------------------- |
| 114 |
|
Name: Matthias Blume |
| 115 |
|
Date: 2001/07/09 16:00:00 EDT |
| 116 |
|
Tag: blume-20010709-more-varargs |
| 117 |
|
Description: |
| 118 |
|
|
| 119 |
|
I changed the handling of varargs in ml-nlffigen again: |
| 120 |
|
The ellipsis ... will now simply be ignored (with an accompanying warning). |
| 121 |
|
|
| 122 |
|
The immediate effect is that you can actually call a varargs function |
| 123 |
|
from ML -- but you can't actually supply any arguments beyond the ones |
| 124 |
|
specified explicitly. (For example, you can call printf with its format |
| 125 |
|
string, but you cannot pass additional arguments.) |
| 126 |
|
|
| 127 |
|
This behavior is only marginally more useful than the one before, but |
| 128 |
|
it has the advantage that a function or, more importantly, a function |
| 129 |
|
type never gets dropped on the floor, thus avoiding follow-up problems with |
| 130 |
|
other types that refer to the offending one. |
| 131 |
|
|
| 132 |
|
---------------------------------------------------------------------- |
| 133 |
|
Name: Matthias Blume |
| 134 |
|
Date: 2001/07/09 11:25:00 EDT |
| 135 |
|
Tag: blume-20010709-varargs |
| 136 |
|
Description: |
| 137 |
|
|
| 138 |
|
1. ckit-lib.cm now exports structure Error |
| 139 |
|
2. ml-nlffigen reports occurences of "..." (i.e., varargs function types) |
| 140 |
|
with a warning accompanied by a source location. Moreover, it |
| 141 |
|
merely skips the offending function or type and proceeds with the |
| 142 |
|
rest of its work.u As a result, one can safely feed C code containing |
| 143 |
|
"..." to ml-nlffigen. |
| 144 |
|
3. There are some internal improvements to CM, providing slightly |
| 145 |
|
more general string substitutions in the tools subsystem. |
| 146 |
|
|
| 147 |
|
---------------------------------------------------------------------- |
| 148 |
|
Name: Matthias Blume |
| 149 |
|
Date: 2001/06/27 15:10:00 EDT |
| 150 |
|
Tag: blume-20010627-concur |
| 151 |
|
Description: |
| 152 |
|
|
| 153 |
|
Fixed a small bug in CM's handling of parallel compilation. |
| 154 |
|
(You could observe the bug by Control-C-interrupting an ordinary |
| 155 |
|
CMB.make or CM.stabilize and then attaching some compile servers. |
| 156 |
|
The result was that all of a sudden the previously interrupted |
| 157 |
|
compilation would continue on its own. This was because of |
| 158 |
|
an over-optimization: CM did not bother to clean out certain queues |
| 159 |
|
when no servers were attached "anyway", resulting in the contents |
| 160 |
|
of these queues to grab control when new servers did get attached.) |
| 161 |
|
|
| 162 |
|
There is also another minor update to the CM manual. |
| 163 |
|
|
| 164 |
|
---------------------------------------------------------------------- |
| 165 |
|
Name: Matthias Blume |
| 166 |
|
Date: 2001/06/26 16:15:00 EDT |
| 167 |
|
Tag: blume-20010626-cmdoc |
| 168 |
|
Description: |
| 169 |
|
|
| 170 |
|
Minor typo fixed in CM manual (syntax diagram for libraries). |
| 171 |
|
|
| 172 |
|
---------------------------------------------------------------------- |
| 173 |
|
Name: Matthias Blume |
| 174 |
|
Date: 2001/06/25 22:55:00 EDT |
| 175 |
|
Tag: blume-20010625-x86pc |
| 176 |
|
Description: |
| 177 |
|
|
| 178 |
|
Fixed a nasty bug in the X86 assembly code that caused signal |
| 179 |
|
handlers to fail (crash) randomly. |
| 180 |
|
|
| 181 |
|
---------------------------------------------------------------------- |
| 182 |
|
Name: Matthias Blume |
| 183 |
|
Date: 2001/06/25 12:05:00 EDT |
| 184 |
|
Tag: blume-20010625-nlffigen |
| 185 |
|
Description: |
| 186 |
|
|
| 187 |
|
This update fixes a number of minor bugs in ml-nlffigen as reported by |
| 188 |
|
Nick Carter <nbc@andrew.cmu.edu>. |
| 189 |
|
|
| 190 |
|
1. Silly but ok typedefs of the form "typedef void myvoid;" are now accepted. |
| 191 |
|
2. Default names for generated files are now derived from the name of |
| 192 |
|
the C file *without its directory*. In particular, this causes generated |
| 193 |
|
files to be placed locally even if the C file is in some system directory. |
| 194 |
|
3. Default names for generated signatures and structures are also derived |
| 195 |
|
from the C file name without its directory. This avoids silly things |
| 196 |
|
like "structure GL/GL". |
| 197 |
|
(Other silly names are still possible because ml-nlffigen does not do |
| 198 |
|
a thorough check of whether generated names are legal ML identifiers. |
| 199 |
|
When in doubt, use command line arguments to force particular names.) |
| 200 |
|
|
| 201 |
|
---------------------------------------------------------------------- |
| 202 |
|
Name: Matthias Blume |
| 203 |
|
Date: 2001/06/21 12:25:00 EDT |
| 204 |
|
Tag: blume-20010621-eXene |
| 205 |
|
Description: |
| 206 |
|
|
| 207 |
|
eXene now compiles and (sort of) works again. |
| 208 |
|
|
| 209 |
|
The library name (for version > 110.33) is $/eXene.cm. |
| 210 |
|
|
| 211 |
|
I also added an new example in src/eXene/examples/nbody. See the |
| 212 |
|
README file there for details. |
| 213 |
|
|
| 214 |
|
---------------------------------------------------------------------- |
| 215 |
|
Name: Matthias Blume |
| 216 |
|
Date: 2001/06/20 16:40:00 EDT |
| 217 |
|
Tag: blume-20010620-cml |
| 218 |
|
Description: |
| 219 |
|
|
| 220 |
|
CML now compiles and works again. |
| 221 |
|
|
| 222 |
|
Libraries (for version > 110.33): |
| 223 |
|
|
| 224 |
|
$cml/cml.cm Main CML library. |
| 225 |
|
$cml/basis.cm CML's version of $/basis.cm. |
| 226 |
|
$cml/cml-internal.cm Internal helper library. |
| 227 |
|
$cml/core-cml.cm Internal helper library. |
| 228 |
|
$cml-lib/trace-cml.cm Tracing facility. |
| 229 |
|
$cml-lib/smlnj-lib.cm CML's version of $/smlnj-lib.cm |
| 230 |
|
|
| 231 |
|
The installer (config/install.sh) has been taught how to properly |
| 232 |
|
install this stuff. |
| 233 |
|
|
| 234 |
|
---------------------------------------------------------------------- |
| 235 |
|
Name: Matthias Blume |
| 236 |
|
Date: 2001/06/19 17:55:00 EDT |
| 237 |
|
Tag: blume-20010619-instantiate |
| 238 |
|
Description: |
| 239 |
|
|
| 240 |
|
This un-breaks the fix for bug 1432. |
| 241 |
|
(The bug was originally fixed in 110.9 but I broke it again some |
| 242 |
|
time after that.) |
| 243 |
|
|
| 244 |
|
---------------------------------------------------------------------- |
| 245 |
|
Name: Matthias Blume |
| 246 |
|
Date: 2001/06/19 17:25:00 EDT |
| 247 |
|
Tag: blume-20010619-signals |
| 248 |
|
Description: |
| 249 |
|
|
| 250 |
|
This should (hopefully) fix the long-standing signal handling bug. |
| 251 |
|
(The runtime system was constructing a continuation record with an |
| 252 |
|
incorrect descriptor which would cause the GC to drop data on the floor...) |
| 253 |
|
|
| 254 |
|
---------------------------------------------------------------------- |
| 255 |
|
Name: Matthias Blume |
| 256 |
|
Date: 2001/06/15 15:05:00 EDT |
| 257 |
|
Tag: blume-20010615-moresparc |
| 258 |
|
Description: |
| 259 |
|
|
| 260 |
|
Here is a short late-hour update related to Sparc c-calls: |
| 261 |
|
|
| 262 |
|
-- made handling of double-word arguments a bit smarter |
| 263 |
|
|
| 264 |
|
-- instruction selection phase tries to collapse certain clumsily |
| 265 |
|
constructed ML-Trees; typical example: |
| 266 |
|
|
| 267 |
|
ADD(ty,ADD(_,e,LI d1),LI d2) -> ADD(ty,e,LI(d1+d2)) |
| 268 |
|
|
| 269 |
|
This currently has no further impact on SML/NJ since mlriscGen does |
| 270 |
|
not seem to generate such patterns in the first place, and c-calls |
| 271 |
|
(which did generate them in the beginning) has meanwhile been fixed |
| 272 |
|
so as to avoid them as well. |
| 273 |
|
|
| 274 |
|
---------------------------------------------------------------------- |
| 275 |
|
Name: Matthias Blume |
| 276 |
|
Date: 2001/06/15 15:05:00 EDT |
| 277 |
|
Tag: blume-20010615-sparc |
| 278 |
|
Description: |
| 279 |
|
|
| 280 |
|
The purpose of this update is to provide an implementation of NLFFI |
| 281 |
|
on Sparc machines. |
| 282 |
|
|
| 283 |
|
Here are the changes in detail: |
| 284 |
|
|
| 285 |
|
* src/MLRISC/sparc/c-calls/sparc-c-calls.sml is a new file containing |
| 286 |
|
the Sparc implementation of the c-calls API. |
| 287 |
|
* The Sparc backend of SML/NJ has been modified to uniformely use %fp |
| 288 |
|
for accessing the ML frame. Thus, we have a real frame pointer and |
| 289 |
|
can freely modify %sp without need for an omit-frame-ptr phase. |
| 290 |
|
The vfp logic in src/compiler/CodeGen/* has been changed to accomodate |
| 291 |
|
this case. |
| 292 |
|
* ml-nlffigen has been taught to produce code for different architectures |
| 293 |
|
and calling conventions. |
| 294 |
|
* In a way similar to what was done in the x86 case, the Sparc |
| 295 |
|
backend uses its own specific extension to mltree. (For example, |
| 296 |
|
it needs to be able to generate UNIMP instructions which are part |
| 297 |
|
of the calling convention.) |
| 298 |
|
* ml-nlffi-lib was reorganized to make it more modular (in particular, |
| 299 |
|
to make it easier to plug in new machine- and os-dependent parts). |
| 300 |
|
|
| 301 |
|
There are some other fairly unrelated bug fixes and cleanups as well: |
| 302 |
|
|
| 303 |
|
* I further hacked the .cm files for MLRISC tools (like MDLGen) so |
| 304 |
|
that they properly share their libraries with existing SML/NJ libraries. |
| 305 |
|
* I fixed a minor cosmetic bug in CM, supressing certain spurious |
| 306 |
|
follow-up error messages. |
| 307 |
|
* Updates to CM/CMB documentation. |
| 308 |
|
|
| 309 |
|
TODO items: |
| 310 |
|
|
| 311 |
|
* MLRISC should use a different register as its asmTemp on the Sparc. |
| 312 |
|
(The current %o2 is a really bad choice because it is part of the |
| 313 |
|
calling conventions, so things might interfere in unexpected ways.) |
| 314 |
|
|
| 315 |
|
---------------------------------------------------------------------- |
| 316 |
|
Name: Matthias Blume |
| 317 |
|
Date: 2001/06/07 |
| 318 |
|
Tag: blume-20010607-calls |
| 319 |
|
Description: |
| 320 |
|
|
| 321 |
|
A number of internal changes related to C calls and calling conventions: |
| 322 |
|
|
| 323 |
|
1. ML-Tree CALL statements now carry a "pops" field. It indicates the |
| 324 |
|
number of bytes popped implicitly (by the callee). In most cases |
| 325 |
|
this field is 0 but on x86/win32 it is some non-zero value. This |
| 326 |
|
is information provided for the benefit of the "omit-frameptr" pass. |
| 327 |
|
2. The CALL instruction on the x86 carries a similar "pops" field. |
| 328 |
|
The instruction selection phase copies its value from the ML-Tree |
| 329 |
|
CALL statement. |
| 330 |
|
3. On all other architectures, the instruction selection phase checks |
| 331 |
|
whether "pops=0" and complains if not. |
| 332 |
|
4. The c-calls implementation for x86 now accepts two calling conventions: |
| 333 |
|
"ccall" and "stdcall". When "ccall" is selected, the caller cleans |
| 334 |
|
up after the call and pops is set to 0. For "stdcall", the caller |
| 335 |
|
does nothing, leaving the cleanup to the callee; pops is set to |
| 336 |
|
the number of bytes that were pushed onto the stack. |
| 337 |
|
5. The cproto decoder (compiler/Semant/types/cproto.sml) now can |
| 338 |
|
distinguish between "ccall" and "stdcall". |
| 339 |
|
6. The UNIMP instruction has been added to the supported Sparc instruction |
| 340 |
|
set. (This is needed for implementing the official C calling convention |
| 341 |
|
on this architecture.) |
| 342 |
|
7. I fixed some of the .cm files under src/MLRISC/Tools to make them |
| 343 |
|
work with the latest CM. |
| 344 |
|
|
| 345 |
|
---------------------------------------------------------------------- |
| 346 |
|
Name: Matthias Blume |
| 347 |
|
Date: 2001/06/05 15:10:00 EDT |
| 348 |
|
Tag: blume-20010605-cm-index |
| 349 |
|
Description: |
| 350 |
|
|
| 351 |
|
0. The "lambdasplit" parameter for class "sml" in CM has been documented. |
| 352 |
|
|
| 353 |
|
1. CM can now generate "index files". These are human-readable files |
| 354 |
|
that list on a per-.cm-file basis each toplevel symbol defined or |
| 355 |
|
imported. The location of the index file for |
| 356 |
|
<p>/<d>.cm is <p>/CM/INDEX/<d>.cm. |
| 357 |
|
To enable index-file generation, set CM.Control.generate_index to true |
| 358 |
|
or export an environment-symbol: export CM_GENERATE_INDEX=true. |
| 359 |
|
|
| 360 |
|
The CM manual has been updated accordingly. |
| 361 |
|
|
| 362 |
|
2. I made some slight modifications to the c-calls API in MLRISC. |
| 363 |
|
|
| 364 |
|
a) There is now a callback to support saving/restoring of |
| 365 |
|
dedicated but caller-save registers around the actual call |
| 366 |
|
instruction. |
| 367 |
|
b) One can optionally specify a comment-annotation for the |
| 368 |
|
call instruction. |
| 369 |
|
|
| 370 |
|
3. SML/NJ (mlriscGen.sml) uses this new API for the rawccall primop. |
| 371 |
|
(For example, the comment annotation shows the C prototype of |
| 372 |
|
the function being called.) |
| 373 |
|
|
| 374 |
|
---------------------------------------------------------------------- |
| 375 |
|
Name: Matthias Blume |
| 376 |
|
Date: 2001/06/01 13:30:00 EDT |
| 377 |
|
Tag: blume-20010601-nlffi-cleanup |
| 378 |
|
Description: |
| 379 |
|
|
| 380 |
|
This is mostly a cleanup of MLFFI stuff: |
| 381 |
|
|
| 382 |
|
- some signature files have been put into a more exposed place |
| 383 |
|
- the ugly 'f type parameter is gone (simplifies types tremendously!) |
| 384 |
|
- ml-nlffigen changed accordingly |
| 385 |
|
- tutorial updated |
| 386 |
|
|
| 387 |
|
Other changes: |
| 388 |
|
|
| 389 |
|
- author's affiliation in CM manual(s) updated |
| 390 |
|
- some more recognized keywords added to Allen's sml.sty |
| 391 |
|
|
| 392 |
|
---------------------------------------------------------------------- |
| 393 |
|
Name: Matthias Blume |
| 394 |
Date: 2001/05/25 15:30:00 EDT |
Date: 2001/05/25 15:30:00 EDT |
| 395 |
Tag: blume-20010525-iptr |
Tag: blume-20010525-iptr |
| 396 |
Description: |
Description: |
Parent Directory
| 
Revision Log
| 
Patch
