Mercurial > hg > CbC > CbC_gcc
annotate gcc/doc/passes.texi @ 158:494b0b89df80 default tip
...
| author | Shinji KONO <kono@ie.u-ryukyu.ac.jp> |
|---|---|
| date | Mon, 25 May 2020 18:13:55 +0900 |
| parents | 1830386684a0 |
| children |
| rev | line source |
|---|---|
| 111 | 1 @c markers: BUG TODO |
| 0 | 2 |
| 145 | 3 @c Copyright (C) 1988-2020 Free Software Foundation, Inc. |
| 0 | 4 @c This is part of the GCC manual. |
| 5 @c For copying conditions, see the file gcc.texi. | |
| 6 | |
| 7 @node Passes | |
| 8 @chapter Passes and Files of the Compiler | |
| 9 @cindex passes and files of the compiler | |
| 10 @cindex files and passes of the compiler | |
| 11 @cindex compiler passes and files | |
| 111 | 12 @cindex pass dumps |
| 0 | 13 |
| 14 This chapter is dedicated to giving an overview of the optimization and | |
| 15 code generation passes of the compiler. In the process, it describes | |
| 16 some of the language front end interface, though this description is no | |
| 17 where near complete. | |
| 18 | |
| 19 @menu | |
| 20 * Parsing pass:: The language front end turns text into bits. | |
| 21 * Gimplification pass:: The bits are turned into something we can optimize. | |
| 22 * Pass manager:: Sequencing the optimization passes. | |
| 145 | 23 * IPA passes:: Inter-procedural optimizations. |
|
19
58ad6c70ea60
update gcc from 4.4.0 to 4.4.1.
kent@firefly.cr.ie.u-ryukyu.ac.jp
parents:
0
diff
changeset
|
24 * Tree SSA passes:: Optimizations on a high-level representation. |
| 0 | 25 * RTL passes:: Optimizations on a low-level representation. |
| 111 | 26 * Optimization info:: Dumping optimization information from passes. |
| 0 | 27 @end menu |
| 28 | |
| 29 @node Parsing pass | |
| 30 @section Parsing pass | |
| 31 @cindex GENERIC | |
| 32 @findex lang_hooks.parse_file | |
| 33 The language front end is invoked only once, via | |
| 34 @code{lang_hooks.parse_file}, to parse the entire input. The language | |
| 35 front end may use any intermediate language representation deemed | |
| 111 | 36 appropriate. The C front end uses GENERIC trees (@pxref{GENERIC}), plus |
| 0 | 37 a double handful of language specific tree codes defined in |
| 38 @file{c-common.def}. The Fortran front end uses a completely different | |
| 39 private representation. | |
| 40 | |
| 41 @cindex GIMPLE | |
| 42 @cindex gimplification | |
| 43 @cindex gimplifier | |
| 44 @cindex language-independent intermediate representation | |
| 45 @cindex intermediate representation lowering | |
| 46 @cindex lowering, language-dependent intermediate representation | |
| 47 At some point the front end must translate the representation used in the | |
| 48 front end to a representation understood by the language-independent | |
| 49 portions of the compiler. Current practice takes one of two forms. | |
| 111 | 50 The C front end manually invokes the gimplifier (@pxref{GIMPLE}) on each function, |
| 0 | 51 and uses the gimplifier callbacks to convert the language-specific tree |
| 111 | 52 nodes directly to GIMPLE before passing the function off to be compiled. |
| 0 | 53 The Fortran front end converts from a private representation to GENERIC, |
| 54 which is later lowered to GIMPLE when the function is compiled. Which | |
| 55 route to choose probably depends on how well GENERIC (plus extensions) | |
| 56 can be made to match up with the source language and necessary parsing | |
| 57 data structures. | |
| 58 | |
| 59 BUG: Gimplification must occur before nested function lowering, | |
| 60 and nested function lowering must be done by the front end before | |
| 61 passing the data off to cgraph. | |
| 62 | |
| 63 TODO: Cgraph should control nested function lowering. It would | |
| 64 only be invoked when it is certain that the outer-most function | |
| 65 is used. | |
| 66 | |
| 67 TODO: Cgraph needs a gimplify_function callback. It should be | |
| 68 invoked when (1) it is certain that the function is used, (2) | |
| 69 warning flags specified by the user require some amount of | |
| 70 compilation in order to honor, (3) the language indicates that | |
| 71 semantic analysis is not complete until gimplification occurs. | |
| 72 Hum@dots{} this sounds overly complicated. Perhaps we should just | |
| 73 have the front end gimplify always; in most cases it's only one | |
| 74 function call. | |
| 75 | |
| 76 The front end needs to pass all function definitions and top level | |
| 77 declarations off to the middle-end so that they can be compiled and | |
| 78 emitted to the object file. For a simple procedural language, it is | |
| 79 usually most convenient to do this as each top level declaration or | |
| 80 definition is seen. There is also a distinction to be made between | |
| 81 generating functional code and generating complete debug information. | |
| 82 The only thing that is absolutely required for functional code is that | |
| 83 function and data @emph{definitions} be passed to the middle-end. For | |
| 84 complete debug information, function, data and type declarations | |
| 85 should all be passed as well. | |
| 86 | |
| 87 @findex rest_of_decl_compilation | |
| 88 @findex rest_of_type_compilation | |
| 89 @findex cgraph_finalize_function | |
| 90 In any case, the front end needs each complete top-level function or | |
| 91 data declaration, and each data definition should be passed to | |
| 92 @code{rest_of_decl_compilation}. Each complete type definition should | |
| 93 be passed to @code{rest_of_type_compilation}. Each function definition | |
| 94 should be passed to @code{cgraph_finalize_function}. | |
| 95 | |
| 96 TODO: I know rest_of_compilation currently has all sorts of | |
|
19
58ad6c70ea60
update gcc from 4.4.0 to 4.4.1.
kent@firefly.cr.ie.u-ryukyu.ac.jp
parents:
0
diff
changeset
|
97 RTL generation semantics. I plan to move all code generation |
|
58ad6c70ea60
update gcc from 4.4.0 to 4.4.1.
kent@firefly.cr.ie.u-ryukyu.ac.jp
parents:
0
diff
changeset
|
98 bits (both Tree and RTL) to compile_function. Should we hide |
| 0 | 99 cgraph from the front ends and move back to rest_of_compilation |
| 100 as the official interface? Possibly we should rename all three | |
| 101 interfaces such that the names match in some meaningful way and | |
| 102 that is more descriptive than "rest_of". | |
| 103 | |
| 104 The middle-end will, at its option, emit the function and data | |
| 105 definitions immediately or queue them for later processing. | |
| 106 | |
| 107 @node Gimplification pass | |
| 108 @section Gimplification pass | |
| 109 | |
| 110 @cindex gimplification | |
| 111 @cindex GIMPLE | |
| 112 @dfn{Gimplification} is a whimsical term for the process of converting | |
| 113 the intermediate representation of a function into the GIMPLE language | |
| 111 | 114 (@pxref{GIMPLE}). The term stuck, and so words like ``gimplification'', |
| 0 | 115 ``gimplify'', ``gimplifier'' and the like are sprinkled throughout this |
| 116 section of code. | |
| 117 | |
| 118 While a front end may certainly choose to generate GIMPLE directly if | |
| 119 it chooses, this can be a moderately complex process unless the | |
| 120 intermediate language used by the front end is already fairly simple. | |
| 121 Usually it is easier to generate GENERIC trees plus extensions | |
| 122 and let the language-independent gimplifier do most of the work. | |
| 123 | |
| 124 @findex gimplify_function_tree | |
| 125 @findex gimplify_expr | |
| 126 @findex lang_hooks.gimplify_expr | |
| 127 The main entry point to this pass is @code{gimplify_function_tree} | |
| 128 located in @file{gimplify.c}. From here we process the entire | |
| 129 function gimplifying each statement in turn. The main workhorse | |
| 130 for this pass is @code{gimplify_expr}. Approximately everything | |
| 131 passes through here at least once, and it is from here that we | |
| 132 invoke the @code{lang_hooks.gimplify_expr} callback. | |
| 133 | |
| 134 The callback should examine the expression in question and return | |
| 135 @code{GS_UNHANDLED} if the expression is not a language specific | |
| 136 construct that requires attention. Otherwise it should alter the | |
| 137 expression in some way to such that forward progress is made toward | |
| 138 producing valid GIMPLE@. If the callback is certain that the | |
| 139 transformation is complete and the expression is valid GIMPLE, it | |
| 140 should return @code{GS_ALL_DONE}. Otherwise it should return | |
| 141 @code{GS_OK}, which will cause the expression to be processed again. | |
| 142 If the callback encounters an error during the transformation (because | |
| 143 the front end is relying on the gimplification process to finish | |
| 144 semantic checks), it should return @code{GS_ERROR}. | |
| 145 | |
| 146 @node Pass manager | |
| 147 @section Pass manager | |
| 148 | |
| 149 The pass manager is located in @file{passes.c}, @file{tree-optimize.c} | |
| 150 and @file{tree-pass.h}. | |
| 111 | 151 It processes passes as described in @file{passes.def}. |
| 0 | 152 Its job is to run all of the individual passes in the correct order, |
| 153 and take care of standard bookkeeping that applies to every pass. | |
| 154 | |
| 155 The theory of operation is that each pass defines a structure that | |
| 156 represents everything we need to know about that pass---when it | |
| 157 should be run, how it should be run, what intermediate language | |
| 158 form or on-the-side data structures it needs. We register the pass | |
| 159 to be run in some particular order, and the pass manager arranges | |
| 160 for everything to happen in the correct order. | |
| 161 | |
| 162 The actuality doesn't completely live up to the theory at present. | |
| 163 Command-line switches and @code{timevar_id_t} enumerations must still | |
| 164 be defined elsewhere. The pass manager validates constraints but does | |
| 165 not attempt to (re-)generate data structures or lower intermediate | |
| 166 language form based on the requirements of the next pass. Nevertheless, | |
| 167 what is present is useful, and a far sight better than nothing at all. | |
| 168 | |
|
55
77e2b8dfacca
update it from 4.4.3 to 4.5.0
ryoma <e075725@ie.u-ryukyu.ac.jp>
parents:
19
diff
changeset
|
169 Each pass should have a unique name. |
| 0 | 170 Each pass may have its own dump file (for GCC debugging purposes). |
|
55
77e2b8dfacca
update it from 4.4.3 to 4.5.0
ryoma <e075725@ie.u-ryukyu.ac.jp>
parents:
19
diff
changeset
|
171 Passes with a name starting with a star do not dump anything. |
|
77e2b8dfacca
update it from 4.4.3 to 4.5.0
ryoma <e075725@ie.u-ryukyu.ac.jp>
parents:
19
diff
changeset
|
172 Sometimes passes are supposed to share a dump file / option name. |
|
77e2b8dfacca
update it from 4.4.3 to 4.5.0
ryoma <e075725@ie.u-ryukyu.ac.jp>
parents:
19
diff
changeset
|
173 To still give these unique names, you can use a prefix that is delimited |
|
77e2b8dfacca
update it from 4.4.3 to 4.5.0
ryoma <e075725@ie.u-ryukyu.ac.jp>
parents:
19
diff
changeset
|
174 by a space from the part that is used for the dump file / option name. |
|
77e2b8dfacca
update it from 4.4.3 to 4.5.0
ryoma <e075725@ie.u-ryukyu.ac.jp>
parents:
19
diff
changeset
|
175 E.g. When the pass name is "ud dce", the name used for dump file/options |
|
77e2b8dfacca
update it from 4.4.3 to 4.5.0
ryoma <e075725@ie.u-ryukyu.ac.jp>
parents:
19
diff
changeset
|
176 is "dce". |
| 0 | 177 |
| 178 TODO: describe the global variables set up by the pass manager, | |
| 179 and a brief description of how a new pass should use it. | |
|
19
58ad6c70ea60
update gcc from 4.4.0 to 4.4.1.
kent@firefly.cr.ie.u-ryukyu.ac.jp
parents:
0
diff
changeset
|
180 I need to look at what info RTL passes use first@enddots{} |
| 0 | 181 |
| 145 | 182 @node IPA passes |
| 183 @section Inter-procedural optimization passes | |
| 184 @cindex IPA passes | |
| 185 @cindex inter-procedural optimization passes | |
| 186 | |
| 187 The inter-procedural optimization (IPA) passes use call graph | |
| 188 information to perform transformations across function boundaries. | |
| 189 IPA is a critical part of link-time optimization (LTO) and | |
| 190 whole-program (WHOPR) optimization, and these passes are structured | |
| 191 with the needs of LTO and WHOPR in mind by dividing their operations | |
| 192 into stages. For detailed discussion of the LTO/WHOPR IPA pass stages | |
| 193 and interfaces, see @ref{IPA}. | |
| 194 | |
| 195 The following briefly describes the inter-procedural optimization (IPA) | |
| 196 passes, which are split into small IPA passes, regular IPA passes, | |
| 197 and late IPA passes, according to the LTO/WHOPR processing model. | |
| 198 | |
| 199 @menu | |
| 200 * Small IPA passes:: | |
| 201 * Regular IPA passes:: | |
| 202 * Late IPA passes:: | |
| 203 @end menu | |
| 204 | |
| 205 @node Small IPA passes | |
| 206 @subsection Small IPA passes | |
| 207 @cindex small IPA passes | |
| 208 A small IPA pass is a pass derived from @code{simple_ipa_opt_pass}. | |
| 209 As described in @ref{IPA}, it does everything at once and | |
| 210 defines only the @emph{Execute} stage. During this | |
| 211 stage it accesses and modifies the function bodies. | |
| 212 No @code{generate_summary}, @code{read_summary}, or @code{write_summary} | |
| 213 hooks are defined. | |
| 214 | |
| 215 @itemize @bullet | |
| 216 @item IPA free lang data | |
| 217 | |
| 218 This pass frees resources that are used by the front end but are | |
| 219 not needed once it is done. It is located in @file{tree.c} and is described by | |
| 220 @code{pass_ipa_free_lang_data}. | |
| 221 | |
| 222 @item IPA function and variable visibility | |
| 223 | |
| 224 This is a local function pass handling visibilities of all symbols. This | |
| 225 happens before LTO streaming, so @option{-fwhole-program} should be ignored | |
| 226 at this level. It is located in @file{ipa-visibility.c} and is described by | |
| 227 @code{pass_ipa_function_and_variable_visibility}. | |
| 228 | |
| 229 @item IPA remove symbols | |
| 230 | |
| 231 This pass performs reachability analysis and reclaims all unreachable nodes. | |
| 232 It is located in @file{passes.c} and is described by | |
| 233 @code{pass_ipa_remove_symbols}. | |
| 234 | |
| 235 @item IPA OpenACC | |
| 236 | |
| 237 This is a pass group for OpenACC processing. It is located in | |
| 238 @file{tree-ssa-loop.c} and is described by @code{pass_ipa_oacc}. | |
| 239 | |
| 240 @item IPA points-to analysis | |
| 241 | |
| 242 This is a tree-based points-to analysis pass. The idea behind this analyzer | |
| 243 is to generate set constraints from the program, then solve the resulting | |
| 244 constraints in order to generate the points-to sets. It is located in | |
| 245 @file{tree-ssa-structalias.c} and is described by @code{pass_ipa_pta}. | |
| 246 | |
| 247 @item IPA OpenACC kernels | |
| 248 | |
| 249 This is a pass group for processing OpenACC kernels regions. It is a | |
| 250 subpass of the IPA OpenACC pass group that runs on offloaded functions | |
| 251 containing OpenACC kernels loops. It is located in | |
| 252 @file{tree-ssa-loop.c} and is described by | |
| 253 @code{pass_ipa_oacc_kernels}. | |
| 254 | |
| 255 @item Target clone | |
| 256 | |
| 257 This is a pass for parsing functions with multiple target attributes. | |
| 258 It is located in @file{multiple_target.c} and is described by | |
| 259 @code{pass_target_clone}. | |
| 260 | |
| 261 @item IPA auto profile | |
| 262 | |
| 263 This pass uses AutoFDO profiling data to annotate the control flow graph. | |
| 264 It is located in @file{auto-profile.c} and is described by | |
| 265 @code{pass_ipa_auto_profile}. | |
| 266 | |
| 267 @item IPA tree profile | |
| 268 | |
| 269 This pass does profiling for all functions in the call graph. | |
| 270 It calculates branch | |
| 271 probabilities and basic block execution counts. It is located | |
| 272 in @file{tree-profile.c} and is described by @code{pass_ipa_tree_profile}. | |
| 273 | |
| 274 @item IPA free function summary | |
| 275 | |
| 276 This pass is a small IPA pass when argument @code{small_p} is true. | |
| 277 It releases inline function summaries and call summaries. | |
| 278 It is located in @file{ipa-fnsummary.c} and is described by | |
| 279 @code{pass_ipa_free_free_fn_summary}. | |
| 280 | |
| 281 @item IPA increase alignment | |
| 282 | |
| 283 This pass increases the alignment of global arrays to improve | |
| 284 vectorization. It is located in @file{tree-vectorizer.c} | |
| 285 and is described by @code{pass_ipa_increase_alignment}. | |
| 286 | |
| 287 @item IPA transactional memory | |
| 288 | |
| 289 This pass is for transactional memory support. | |
| 290 It is located in @file{trans-mem.c} and is described by | |
| 291 @code{pass_ipa_tm}. | |
| 292 | |
| 293 @item IPA lower emulated TLS | |
| 294 | |
| 295 This pass lowers thread-local storage (TLS) operations | |
| 296 to emulation functions provided by libgcc. | |
| 297 It is located in @file{tree-emutls.c} and is described by | |
| 298 @code{pass_ipa_lower_emutls}. | |
| 299 | |
| 300 @end itemize | |
| 301 | |
| 302 @node Regular IPA passes | |
| 303 @subsection Regular IPA passes | |
| 304 @cindex regular IPA passes | |
| 305 | |
| 306 A regular IPA pass is a pass derived from @code{ipa_opt_pass_d} that | |
| 307 is executed in WHOPR compilation. Regular IPA passes may have summary | |
| 308 hooks implemented in any of the LGEN, WPA or LTRANS stages (@pxref{IPA}). | |
| 309 | |
| 310 @itemize @bullet | |
| 311 @item IPA whole program visibility | |
| 312 | |
| 313 This pass performs various optimizations involving symbol visibility | |
| 314 with @option{-fwhole-program}, including symbol privatization, | |
| 315 discovering local functions, and dismantling comdat groups. It is | |
| 316 located in @file{ipa-visibility.c} and is described by | |
| 317 @code{pass_ipa_whole_program_visibility}. | |
| 318 | |
| 319 @item IPA profile | |
| 320 | |
| 321 The IPA profile pass propagates profiling frequencies across the call | |
| 322 graph. It is located in @file{ipa-profile.c} and is described by | |
| 323 @code{pass_ipa_profile}. | |
| 324 | |
| 325 @item IPA identical code folding | |
| 326 | |
| 327 This is the inter-procedural identical code folding pass. | |
| 328 The goal of this transformation is to discover functions | |
| 329 and read-only variables that have exactly the same semantics. It is | |
| 330 located in @file{ipa-icf.c} and is described by @code{pass_ipa_icf}. | |
| 331 | |
| 332 @item IPA devirtualization | |
| 333 | |
| 334 This pass performs speculative devirtualization based on the type | |
| 335 inheritance graph. When a polymorphic call has only one likely target | |
| 336 in the unit, it is turned into a speculative call. It is located in | |
| 337 @file{ipa-devirt.c} and is described by @code{pass_ipa_devirt}. | |
| 338 | |
| 339 @item IPA constant propagation | |
| 340 | |
| 341 The goal of this pass is to discover functions that are always invoked | |
| 342 with some arguments with the same known constant values and to modify | |
| 343 the functions accordingly. It can also do partial specialization and | |
| 344 type-based devirtualization. It is located in @file{ipa-cp.c} and is | |
| 345 described by @code{pass_ipa_cp}. | |
| 346 | |
| 347 @item IPA scalar replacement of aggregates | |
| 348 | |
| 349 This pass can replace an aggregate parameter with a set of other parameters | |
| 350 representing part of the original, turning those passed by reference | |
| 351 into new ones which pass the value directly. It also removes unused | |
| 352 function return values and unused function parameters. This pass is | |
| 353 located in @file{ipa-sra.c} and is described by @code{pass_ipa_sra}. | |
| 354 | |
| 355 @item IPA constructor/destructor merge | |
| 356 | |
| 357 This pass merges multiple constructors and destructors for static | |
| 358 objects into single functions. It's only run at LTO time unless the | |
| 359 target doesn't support constructors and destructors natively. The | |
| 360 pass is located in @file{ipa.c} and is described by | |
| 361 @code{pass_ipa_cdtor_merge}. | |
| 362 | |
| 363 @item IPA HSA | |
| 364 | |
| 365 This pass is part of the GCC support for HSA (Heterogeneous System | |
| 366 Architecture) accelerators. It is responsible for creation of HSA | |
| 367 clones and emitting HSAIL instructions for them. It is located in | |
| 368 @file{ipa-hsa.c} and is described by @code{pass_ipa_hsa}. | |
| 369 | |
| 370 @item IPA function summary | |
| 371 | |
| 372 This pass provides function analysis for inter-procedural passes. | |
| 373 It collects estimates of function body size, execution time, and frame | |
| 374 size for each function. It also estimates information about function | |
| 375 calls: call statement size, time and how often the parameters change | |
| 376 for each call. It is located in @file{ipa-fnsummary.c} and is | |
| 377 described by @code{pass_ipa_fn_summary}. | |
| 378 | |
| 379 @item IPA inline | |
| 380 | |
| 381 The IPA inline pass handles function inlining with whole-program | |
| 382 knowledge. Small functions that are candidates for inlining are | |
| 383 ordered in increasing badness, bounded by unit growth parameters. | |
| 384 Unreachable functions are removed from the call graph. Functions called | |
| 385 once and not exported from the unit are inlined. This pass is located in | |
| 386 @file{ipa-inline.c} and is described by @code{pass_ipa_inline}. | |
| 387 | |
| 388 @item IPA pure/const analysis | |
| 389 | |
| 390 This pass marks functions as being either const (@code{TREE_READONLY}) or | |
| 391 pure (@code{DECL_PURE_P}). The per-function information is produced | |
| 392 by @code{pure_const_generate_summary}, then the global information is computed | |
| 393 by performing a transitive closure over the call graph. It is located in | |
| 394 @file{ipa-pure-const.c} and is described by @code{pass_ipa_pure_const}. | |
| 395 | |
| 396 @item IPA free function summary | |
| 397 | |
| 398 This pass is a regular IPA pass when argument @code{small_p} is false. | |
| 399 It releases inline function summaries and call summaries. | |
| 400 It is located in @file{ipa-fnsummary.c} and is described by | |
| 401 @code{pass_ipa_free_fn_summary}. | |
| 402 | |
| 403 @item IPA reference | |
| 404 | |
| 405 This pass gathers information about how variables whose scope is | |
| 406 confined to the compilation unit are used. It is located in | |
| 407 @file{ipa-reference.c} and is described by @code{pass_ipa_reference}. | |
| 408 | |
| 409 @item IPA single use | |
| 410 | |
| 411 This pass checks whether variables are used by a single function. | |
| 412 It is located in @file{ipa.c} and is described by | |
| 413 @code{pass_ipa_single_use}. | |
| 414 | |
| 415 @item IPA comdats | |
| 416 | |
| 417 This pass looks for static symbols that are used exclusively | |
| 418 within one comdat group, and moves them into that comdat group. It is | |
| 419 located in @file{ipa-comdats.c} and is described by | |
| 420 @code{pass_ipa_comdats}. | |
| 421 | |
| 422 @end itemize | |
| 423 | |
| 424 @node Late IPA passes | |
| 425 @subsection Late IPA passes | |
| 426 @cindex late IPA passes | |
| 427 | |
| 428 Late IPA passes are simple IPA passes executed after | |
| 429 the regular passes. In WHOPR mode the passes are executed after | |
| 430 partitioning and thus see just parts of the compiled unit. | |
| 431 | |
| 432 @itemize @bullet | |
| 433 @item Materialize all clones | |
| 434 | |
| 435 Once all functions from compilation unit are in memory, produce all clones | |
| 436 and update all calls. It is located in @file{ipa.c} and is described by | |
| 437 @code{pass_materialize_all_clones}. | |
| 438 | |
| 439 @item IPA points-to analysis | |
| 440 | |
| 441 Points-to analysis; this is the same as the points-to-analysis pass | |
| 442 run with the small IPA passes (@pxref{Small IPA passes}). | |
| 443 | |
| 444 @item OpenMP simd clone | |
| 445 | |
| 446 This is the OpenMP constructs' SIMD clone pass. It creates the appropriate | |
| 447 SIMD clones for functions tagged as elemental SIMD functions. | |
| 448 It is located in @file{omp-simd-clone.c} and is described by | |
| 449 @code{pass_omp_simd_clone}. | |
| 450 | |
| 451 @end itemize | |
| 452 | |
|
19
58ad6c70ea60
update gcc from 4.4.0 to 4.4.1.
kent@firefly.cr.ie.u-ryukyu.ac.jp
parents:
0
diff
changeset
|
453 @node Tree SSA passes |
|
58ad6c70ea60
update gcc from 4.4.0 to 4.4.1.
kent@firefly.cr.ie.u-ryukyu.ac.jp
parents:
0
diff
changeset
|
454 @section Tree SSA passes |
| 0 | 455 |
|
19
58ad6c70ea60
update gcc from 4.4.0 to 4.4.1.
kent@firefly.cr.ie.u-ryukyu.ac.jp
parents:
0
diff
changeset
|
456 The following briefly describes the Tree optimization passes that are |
| 0 | 457 run after gimplification and what source files they are located in. |
| 458 | |
| 459 @itemize @bullet | |
| 460 @item Remove useless statements | |
| 461 | |
| 462 This pass is an extremely simple sweep across the gimple code in which | |
| 463 we identify obviously dead code and remove it. Here we do things like | |
| 464 simplify @code{if} statements with constant conditions, remove | |
| 465 exception handling constructs surrounding code that obviously cannot | |
| 466 throw, remove lexical bindings that contain no variables, and other | |
| 467 assorted simplistic cleanups. The idea is to get rid of the obvious | |
| 468 stuff quickly rather than wait until later when it's more work to get | |
| 469 rid of it. This pass is located in @file{tree-cfg.c} and described by | |
| 470 @code{pass_remove_useless_stmts}. | |
| 471 | |
| 472 @item OpenMP lowering | |
| 473 | |
| 474 If OpenMP generation (@option{-fopenmp}) is enabled, this pass lowers | |
| 475 OpenMP constructs into GIMPLE. | |
| 476 | |
| 477 Lowering of OpenMP constructs involves creating replacement | |
| 478 expressions for local variables that have been mapped using data | |
| 479 sharing clauses, exposing the control flow of most synchronization | |
| 480 directives and adding region markers to facilitate the creation of the | |
| 481 control flow graph. The pass is located in @file{omp-low.c} and is | |
| 482 described by @code{pass_lower_omp}. | |
| 483 | |
| 484 @item OpenMP expansion | |
| 485 | |
| 486 If OpenMP generation (@option{-fopenmp}) is enabled, this pass expands | |
| 487 parallel regions into their own functions to be invoked by the thread | |
| 488 library. The pass is located in @file{omp-low.c} and is described by | |
| 489 @code{pass_expand_omp}. | |
| 490 | |
| 491 @item Lower control flow | |
| 492 | |
| 493 This pass flattens @code{if} statements (@code{COND_EXPR}) | |
| 494 and moves lexical bindings (@code{BIND_EXPR}) out of line. After | |
| 495 this pass, all @code{if} statements will have exactly two @code{goto} | |
| 496 statements in its @code{then} and @code{else} arms. Lexical binding | |
| 497 information for each statement will be found in @code{TREE_BLOCK} rather | |
| 498 than being inferred from its position under a @code{BIND_EXPR}. This | |
| 499 pass is found in @file{gimple-low.c} and is described by | |
| 500 @code{pass_lower_cf}. | |
| 501 | |
| 502 @item Lower exception handling control flow | |
| 503 | |
| 504 This pass decomposes high-level exception handling constructs | |
| 505 (@code{TRY_FINALLY_EXPR} and @code{TRY_CATCH_EXPR}) into a form | |
| 506 that explicitly represents the control flow involved. After this | |
| 507 pass, @code{lookup_stmt_eh_region} will return a non-negative | |
| 508 number for any statement that may have EH control flow semantics; | |
| 509 examine @code{tree_can_throw_internal} or @code{tree_can_throw_external} | |
| 510 for exact semantics. Exact control flow may be extracted from | |
| 511 @code{foreach_reachable_handler}. The EH region nesting tree is defined | |
| 512 in @file{except.h} and built in @file{except.c}. The lowering pass | |
| 513 itself is in @file{tree-eh.c} and is described by @code{pass_lower_eh}. | |
| 514 | |
| 515 @item Build the control flow graph | |
| 516 | |
| 517 This pass decomposes a function into basic blocks and creates all of | |
| 518 the edges that connect them. It is located in @file{tree-cfg.c} and | |
| 519 is described by @code{pass_build_cfg}. | |
| 520 | |
| 521 @item Find all referenced variables | |
| 522 | |
| 523 This pass walks the entire function and collects an array of all | |
| 524 variables referenced in the function, @code{referenced_vars}. The | |
| 525 index at which a variable is found in the array is used as a UID | |
| 526 for the variable within this function. This data is needed by the | |
| 527 SSA rewriting routines. The pass is located in @file{tree-dfa.c} | |
| 528 and is described by @code{pass_referenced_vars}. | |
| 529 | |
| 530 @item Enter static single assignment form | |
| 531 | |
| 532 This pass rewrites the function such that it is in SSA form. After | |
| 533 this pass, all @code{is_gimple_reg} variables will be referenced by | |
| 534 @code{SSA_NAME}, and all occurrences of other variables will be | |
| 535 annotated with @code{VDEFS} and @code{VUSES}; PHI nodes will have | |
| 536 been inserted as necessary for each basic block. This pass is | |
| 537 located in @file{tree-ssa.c} and is described by @code{pass_build_ssa}. | |
| 538 | |
| 539 @item Warn for uninitialized variables | |
| 540 | |
| 541 This pass scans the function for uses of @code{SSA_NAME}s that | |
| 542 are fed by default definition. For non-parameter variables, such | |
| 543 uses are uninitialized. The pass is run twice, before and after | |
| 544 optimization (if turned on). In the first pass we only warn for uses that are | |
| 545 positively uninitialized; in the second pass we warn for uses that | |
| 546 are possibly uninitialized. The pass is located in @file{tree-ssa.c} | |
| 547 and is defined by @code{pass_early_warn_uninitialized} and | |
| 548 @code{pass_late_warn_uninitialized}. | |
| 549 | |
| 550 @item Dead code elimination | |
| 551 | |
| 552 This pass scans the function for statements without side effects whose | |
| 553 result is unused. It does not do memory life analysis, so any value | |
| 554 that is stored in memory is considered used. The pass is run multiple | |
| 555 times throughout the optimization process. It is located in | |
| 556 @file{tree-ssa-dce.c} and is described by @code{pass_dce}. | |
| 557 | |
| 558 @item Dominator optimizations | |
| 559 | |
| 560 This pass performs trivial dominator-based copy and constant propagation, | |
| 561 expression simplification, and jump threading. It is run multiple times | |
|
55
77e2b8dfacca
update it from 4.4.3 to 4.5.0
ryoma <e075725@ie.u-ryukyu.ac.jp>
parents:
19
diff
changeset
|
562 throughout the optimization process. It is located in @file{tree-ssa-dom.c} |
| 0 | 563 and is described by @code{pass_dominator}. |
| 564 | |
| 565 @item Forward propagation of single-use variables | |
| 566 | |
| 567 This pass attempts to remove redundant computation by substituting | |
| 568 variables that are used once into the expression that uses them and | |
| 569 seeing if the result can be simplified. It is located in | |
| 570 @file{tree-ssa-forwprop.c} and is described by @code{pass_forwprop}. | |
| 571 | |
| 572 @item Copy Renaming | |
| 573 | |
| 574 This pass attempts to change the name of compiler temporaries involved in | |
| 575 copy operations such that SSA->normal can coalesce the copy away. When compiler | |
| 576 temporaries are copies of user variables, it also renames the compiler | |
| 577 temporary to the user variable resulting in better use of user symbols. It is | |
| 578 located in @file{tree-ssa-copyrename.c} and is described by | |
| 579 @code{pass_copyrename}. | |
| 580 | |
| 581 @item PHI node optimizations | |
| 582 | |
| 583 This pass recognizes forms of PHI inputs that can be represented as | |
| 584 conditional expressions and rewrites them into straight line code. | |
| 585 It is located in @file{tree-ssa-phiopt.c} and is described by | |
| 586 @code{pass_phiopt}. | |
| 587 | |
| 588 @item May-alias optimization | |
| 589 | |
| 590 This pass performs a flow sensitive SSA-based points-to analysis. | |
| 591 The resulting may-alias, must-alias, and escape analysis information | |
| 592 is used to promote variables from in-memory addressable objects to | |
| 593 non-aliased variables that can be renamed into SSA form. We also | |
| 594 update the @code{VDEF}/@code{VUSE} memory tags for non-renameable | |
| 595 aggregates so that we get fewer false kills. The pass is located | |
| 596 in @file{tree-ssa-alias.c} and is described by @code{pass_may_alias}. | |
| 597 | |
| 598 Interprocedural points-to information is located in | |
| 599 @file{tree-ssa-structalias.c} and described by @code{pass_ipa_pta}. | |
| 600 | |
| 601 @item Profiling | |
| 602 | |
| 111 | 603 This pass instruments the function in order to collect runtime block |
| 0 | 604 and value profiling data. Such data may be fed back into the compiler |
| 605 on a subsequent run so as to allow optimization based on expected | |
| 111 | 606 execution frequencies. The pass is located in @file{tree-profile.c} and |
| 607 is described by @code{pass_ipa_tree_profile}. | |
| 608 | |
| 609 @item Static profile estimation | |
| 610 | |
| 611 This pass implements series of heuristics to guess propababilities | |
| 612 of branches. The resulting predictions are turned into edge profile | |
| 613 by propagating branches across the control flow graphs. | |
| 614 The pass is located in @file{tree-profile.c} and is described by | |
| 615 @code{pass_profile}. | |
| 0 | 616 |
| 617 @item Lower complex arithmetic | |
| 618 | |
| 619 This pass rewrites complex arithmetic operations into their component | |
| 620 scalar arithmetic operations. The pass is located in @file{tree-complex.c} | |
| 621 and is described by @code{pass_lower_complex}. | |
| 622 | |
| 623 @item Scalar replacement of aggregates | |
| 624 | |
| 625 This pass rewrites suitable non-aliased local aggregate variables into | |
| 626 a set of scalar variables. The resulting scalar variables are | |
| 627 rewritten into SSA form, which allows subsequent optimization passes | |
| 628 to do a significantly better job with them. The pass is located in | |
| 629 @file{tree-sra.c} and is described by @code{pass_sra}. | |
| 630 | |
| 631 @item Dead store elimination | |
| 632 | |
| 633 This pass eliminates stores to memory that are subsequently overwritten | |
| 634 by another store, without any intervening loads. The pass is located | |
| 635 in @file{tree-ssa-dse.c} and is described by @code{pass_dse}. | |
| 636 | |
| 637 @item Tail recursion elimination | |
| 638 | |
| 639 This pass transforms tail recursion into a loop. It is located in | |
| 640 @file{tree-tailcall.c} and is described by @code{pass_tail_recursion}. | |
| 641 | |
| 642 @item Forward store motion | |
| 643 | |
| 644 This pass sinks stores and assignments down the flowgraph closer to their | |
| 645 use point. The pass is located in @file{tree-ssa-sink.c} and is | |
| 646 described by @code{pass_sink_code}. | |
| 647 | |
| 648 @item Partial redundancy elimination | |
| 649 | |
| 650 This pass eliminates partially redundant computations, as well as | |
| 651 performing load motion. The pass is located in @file{tree-ssa-pre.c} | |
| 652 and is described by @code{pass_pre}. | |
| 653 | |
| 654 Just before partial redundancy elimination, if | |
| 655 @option{-funsafe-math-optimizations} is on, GCC tries to convert | |
| 656 divisions to multiplications by the reciprocal. The pass is located | |
| 657 in @file{tree-ssa-math-opts.c} and is described by | |
| 658 @code{pass_cse_reciprocal}. | |
| 659 | |
| 660 @item Full redundancy elimination | |
| 661 | |
| 662 This is a simpler form of PRE that only eliminates redundancies that | |
| 111 | 663 occur on all paths. It is located in @file{tree-ssa-pre.c} and |
| 0 | 664 described by @code{pass_fre}. |
| 665 | |
| 666 @item Loop optimization | |
| 667 | |
| 668 The main driver of the pass is placed in @file{tree-ssa-loop.c} | |
| 669 and described by @code{pass_loop}. | |
| 670 | |
| 671 The optimizations performed by this pass are: | |
| 672 | |
| 673 Loop invariant motion. This pass moves only invariants that | |
|
19
58ad6c70ea60
update gcc from 4.4.0 to 4.4.1.
kent@firefly.cr.ie.u-ryukyu.ac.jp
parents:
0
diff
changeset
|
674 would be hard to handle on RTL level (function calls, operations that expand to |
| 0 | 675 nontrivial sequences of insns). With @option{-funswitch-loops} it also moves |
| 676 operands of conditions that are invariant out of the loop, so that we can use | |
| 677 just trivial invariantness analysis in loop unswitching. The pass also includes | |
| 678 store motion. The pass is implemented in @file{tree-ssa-loop-im.c}. | |
| 679 | |
| 680 Canonical induction variable creation. This pass creates a simple counter | |
| 681 for number of iterations of the loop and replaces the exit condition of the | |
| 682 loop using it, in case when a complicated analysis is necessary to determine | |
| 683 the number of iterations. Later optimizations then may determine the number | |
| 684 easily. The pass is implemented in @file{tree-ssa-loop-ivcanon.c}. | |
| 685 | |
| 686 Induction variable optimizations. This pass performs standard induction | |
| 687 variable optimizations, including strength reduction, induction variable | |
| 688 merging and induction variable elimination. The pass is implemented in | |
| 689 @file{tree-ssa-loop-ivopts.c}. | |
| 690 | |
| 691 Loop unswitching. This pass moves the conditional jumps that are invariant | |
| 692 out of the loops. To achieve this, a duplicate of the loop is created for | |
| 693 each possible outcome of conditional jump(s). The pass is implemented in | |
| 111 | 694 @file{tree-ssa-loop-unswitch.c}. |
| 695 | |
| 696 Loop splitting. If a loop contains a conditional statement that is | |
| 697 always true for one part of the iteration space and false for the other | |
| 698 this pass splits the loop into two, one dealing with one side the other | |
| 699 only with the other, thereby removing one inner-loop conditional. The | |
| 700 pass is implemented in @file{tree-ssa-loop-split.c}. | |
| 0 | 701 |
| 702 The optimizations also use various utility functions contained in | |
| 703 @file{tree-ssa-loop-manip.c}, @file{cfgloop.c}, @file{cfgloopanal.c} and | |
| 704 @file{cfgloopmanip.c}. | |
| 705 | |
| 706 Vectorization. This pass transforms loops to operate on vector types | |
| 707 instead of scalar types. Data parallelism across loop iterations is exploited | |
| 111 | 708 to group data elements from consecutive iterations into a vector and operate |
| 709 on them in parallel. Depending on available target support the loop is | |
| 0 | 710 conceptually unrolled by a factor @code{VF} (vectorization factor), which is |
| 111 | 711 the number of elements operated upon in parallel in each iteration, and the |
| 0 | 712 @code{VF} copies of each scalar operation are fused to form a vector operation. |
| 713 Additional loop transformations such as peeling and versioning may take place | |
| 111 | 714 to align the number of iterations, and to align the memory accesses in the |
|
55
77e2b8dfacca
update it from 4.4.3 to 4.5.0
ryoma <e075725@ie.u-ryukyu.ac.jp>
parents:
19
diff
changeset
|
715 loop. |
|
77e2b8dfacca
update it from 4.4.3 to 4.5.0
ryoma <e075725@ie.u-ryukyu.ac.jp>
parents:
19
diff
changeset
|
716 The pass is implemented in @file{tree-vectorizer.c} (the main driver), |
| 111 | 717 @file{tree-vect-loop.c} and @file{tree-vect-loop-manip.c} (loop specific parts |
| 718 and general loop utilities), @file{tree-vect-slp} (loop-aware SLP | |
|
55
77e2b8dfacca
update it from 4.4.3 to 4.5.0
ryoma <e075725@ie.u-ryukyu.ac.jp>
parents:
19
diff
changeset
|
719 functionality), @file{tree-vect-stmts.c} and @file{tree-vect-data-refs.c}. |
| 0 | 720 Analysis of data references is in @file{tree-data-ref.c}. |
| 721 | |
|
55
77e2b8dfacca
update it from 4.4.3 to 4.5.0
ryoma <e075725@ie.u-ryukyu.ac.jp>
parents:
19
diff
changeset
|
722 SLP Vectorization. This pass performs vectorization of straight-line code. The |
|
77e2b8dfacca
update it from 4.4.3 to 4.5.0
ryoma <e075725@ie.u-ryukyu.ac.jp>
parents:
19
diff
changeset
|
723 pass is implemented in @file{tree-vectorizer.c} (the main driver), |
| 111 | 724 @file{tree-vect-slp.c}, @file{tree-vect-stmts.c} and |
|
55
77e2b8dfacca
update it from 4.4.3 to 4.5.0
ryoma <e075725@ie.u-ryukyu.ac.jp>
parents:
19
diff
changeset
|
725 @file{tree-vect-data-refs.c}. |
|
77e2b8dfacca
update it from 4.4.3 to 4.5.0
ryoma <e075725@ie.u-ryukyu.ac.jp>
parents:
19
diff
changeset
|
726 |
| 0 | 727 Autoparallelization. This pass splits the loop iteration space to run |
| 728 into several threads. The pass is implemented in @file{tree-parloops.c}. | |
| 729 | |
|
55
77e2b8dfacca
update it from 4.4.3 to 4.5.0
ryoma <e075725@ie.u-ryukyu.ac.jp>
parents:
19
diff
changeset
|
730 Graphite is a loop transformation framework based on the polyhedral |
|
77e2b8dfacca
update it from 4.4.3 to 4.5.0
ryoma <e075725@ie.u-ryukyu.ac.jp>
parents:
19
diff
changeset
|
731 model. Graphite stands for Gimple Represented as Polyhedra. The |
|
77e2b8dfacca
update it from 4.4.3 to 4.5.0
ryoma <e075725@ie.u-ryukyu.ac.jp>
parents:
19
diff
changeset
|
732 internals of this infrastructure are documented in |
|
77e2b8dfacca
update it from 4.4.3 to 4.5.0
ryoma <e075725@ie.u-ryukyu.ac.jp>
parents:
19
diff
changeset
|
733 @w{@uref{http://gcc.gnu.org/wiki/Graphite}}. The passes working on |
|
77e2b8dfacca
update it from 4.4.3 to 4.5.0
ryoma <e075725@ie.u-ryukyu.ac.jp>
parents:
19
diff
changeset
|
734 this representation are implemented in the various @file{graphite-*} |
|
77e2b8dfacca
update it from 4.4.3 to 4.5.0
ryoma <e075725@ie.u-ryukyu.ac.jp>
parents:
19
diff
changeset
|
735 files. |
|
77e2b8dfacca
update it from 4.4.3 to 4.5.0
ryoma <e075725@ie.u-ryukyu.ac.jp>
parents:
19
diff
changeset
|
736 |
| 0 | 737 @item Tree level if-conversion for vectorizer |
| 738 | |
| 739 This pass applies if-conversion to simple loops to help vectorizer. | |
| 740 We identify if convertible loops, if-convert statements and merge | |
| 741 basic blocks in one big block. The idea is to present loop in such | |
| 742 form so that vectorizer can have one to one mapping between statements | |
| 111 | 743 and available vector operations. This pass is located in |
|
67
f6334be47118
update gcc from gcc-4.6-20100522 to gcc-4.6-20110318
nobuyasu <dimolto@cr.ie.u-ryukyu.ac.jp>
parents:
55
diff
changeset
|
744 @file{tree-if-conv.c} and is described by @code{pass_if_conversion}. |
| 0 | 745 |
| 746 @item Conditional constant propagation | |
| 747 | |
| 748 This pass relaxes a lattice of values in order to identify those | |
| 749 that must be constant even in the presence of conditional branches. | |
| 750 The pass is located in @file{tree-ssa-ccp.c} and is described | |
| 751 by @code{pass_ccp}. | |
| 752 | |
| 753 A related pass that works on memory loads and stores, and not just | |
| 754 register values, is located in @file{tree-ssa-ccp.c} and described by | |
| 755 @code{pass_store_ccp}. | |
| 756 | |
| 757 @item Conditional copy propagation | |
| 758 | |
| 759 This is similar to constant propagation but the lattice of values is | |
| 760 the ``copy-of'' relation. It eliminates redundant copies from the | |
| 761 code. The pass is located in @file{tree-ssa-copy.c} and described by | |
| 762 @code{pass_copy_prop}. | |
| 763 | |
| 764 A related pass that works on memory copies, and not just register | |
| 765 copies, is located in @file{tree-ssa-copy.c} and described by | |
| 766 @code{pass_store_copy_prop}. | |
| 767 | |
| 768 @item Value range propagation | |
| 769 | |
| 770 This transformation is similar to constant propagation but | |
| 771 instead of propagating single constant values, it propagates | |
| 772 known value ranges. The implementation is based on Patterson's | |
| 773 range propagation algorithm (Accurate Static Branch Prediction by | |
| 774 Value Range Propagation, J. R. C. Patterson, PLDI '95). In | |
| 775 contrast to Patterson's algorithm, this implementation does not | |
| 776 propagate branch probabilities nor it uses more than a single | |
| 777 range per SSA name. This means that the current implementation | |
| 778 cannot be used for branch prediction (though adapting it would | |
| 779 not be difficult). The pass is located in @file{tree-vrp.c} and is | |
| 780 described by @code{pass_vrp}. | |
| 781 | |
| 782 @item Folding built-in functions | |
| 783 | |
| 784 This pass simplifies built-in functions, as applicable, with constant | |
| 785 arguments or with inferable string lengths. It is located in | |
| 786 @file{tree-ssa-ccp.c} and is described by @code{pass_fold_builtins}. | |
| 787 | |
| 788 @item Split critical edges | |
| 789 | |
| 790 This pass identifies critical edges and inserts empty basic blocks | |
| 791 such that the edge is no longer critical. The pass is located in | |
| 792 @file{tree-cfg.c} and is described by @code{pass_split_crit_edges}. | |
| 793 | |
| 794 @item Control dependence dead code elimination | |
| 795 | |
| 796 This pass is a stronger form of dead code elimination that can | |
| 797 eliminate unnecessary control flow statements. It is located | |
| 798 in @file{tree-ssa-dce.c} and is described by @code{pass_cd_dce}. | |
| 799 | |
| 800 @item Tail call elimination | |
| 801 | |
| 802 This pass identifies function calls that may be rewritten into | |
| 803 jumps. No code transformation is actually applied here, but the | |
| 804 data and control flow problem is solved. The code transformation | |
| 805 requires target support, and so is delayed until RTL@. In the | |
| 806 meantime @code{CALL_EXPR_TAILCALL} is set indicating the possibility. | |
| 807 The pass is located in @file{tree-tailcall.c} and is described by | |
| 808 @code{pass_tail_calls}. The RTL transformation is handled by | |
| 809 @code{fixup_tail_calls} in @file{calls.c}. | |
| 810 | |
| 811 @item Warn for function return without value | |
| 812 | |
| 813 For non-void functions, this pass locates return statements that do | |
| 814 not specify a value and issues a warning. Such a statement may have | |
| 815 been injected by falling off the end of the function. This pass is | |
| 816 run last so that we have as much time as possible to prove that the | |
| 817 statement is not reachable. It is located in @file{tree-cfg.c} and | |
| 818 is described by @code{pass_warn_function_return}. | |
| 819 | |
| 820 @item Leave static single assignment form | |
| 821 | |
| 822 This pass rewrites the function such that it is in normal form. At | |
| 823 the same time, we eliminate as many single-use temporaries as possible, | |
| 824 so the intermediate language is no longer GIMPLE, but GENERIC@. The | |
| 825 pass is located in @file{tree-outof-ssa.c} and is described by | |
| 826 @code{pass_del_ssa}. | |
| 827 | |
| 828 @item Merge PHI nodes that feed into one another | |
| 829 | |
| 830 This is part of the CFG cleanup passes. It attempts to join PHI nodes | |
| 831 from a forwarder CFG block into another block with PHI nodes. The | |
| 832 pass is located in @file{tree-cfgcleanup.c} and is described by | |
| 833 @code{pass_merge_phi}. | |
| 834 | |
| 835 @item Return value optimization | |
| 836 | |
| 837 If a function always returns the same local variable, and that local | |
| 838 variable is an aggregate type, then the variable is replaced with the | |
| 839 return value for the function (i.e., the function's DECL_RESULT). This | |
| 840 is equivalent to the C++ named return value optimization applied to | |
| 841 GIMPLE@. The pass is located in @file{tree-nrv.c} and is described by | |
| 842 @code{pass_nrv}. | |
| 843 | |
| 844 @item Return slot optimization | |
| 845 | |
| 846 If a function returns a memory object and is called as @code{var = | |
| 847 foo()}, this pass tries to change the call so that the address of | |
| 848 @code{var} is sent to the caller to avoid an extra memory copy. This | |
| 849 pass is located in @code{tree-nrv.c} and is described by | |
| 850 @code{pass_return_slot}. | |
| 851 | |
| 852 @item Optimize calls to @code{__builtin_object_size} | |
| 853 | |
| 854 This is a propagation pass similar to CCP that tries to remove calls | |
| 855 to @code{__builtin_object_size} when the size of the object can be | |
| 856 computed at compile-time. This pass is located in | |
| 857 @file{tree-object-size.c} and is described by | |
| 858 @code{pass_object_sizes}. | |
| 859 | |
| 860 @item Loop invariant motion | |
| 861 | |
| 862 This pass removes expensive loop-invariant computations out of loops. | |
| 863 The pass is located in @file{tree-ssa-loop.c} and described by | |
| 864 @code{pass_lim}. | |
| 865 | |
| 866 @item Loop nest optimizations | |
| 867 | |
| 868 This is a family of loop transformations that works on loop nests. It | |
| 869 includes loop interchange, scaling, skewing and reversal and they are | |
| 870 all geared to the optimization of data locality in array traversals | |
| 871 and the removal of dependencies that hamper optimizations such as loop | |
| 872 parallelization and vectorization. The pass is located in | |
| 873 @file{tree-loop-linear.c} and described by | |
| 874 @code{pass_linear_transform}. | |
| 875 | |
| 876 @item Removal of empty loops | |
| 877 | |
| 878 This pass removes loops with no code in them. The pass is located in | |
| 879 @file{tree-ssa-loop-ivcanon.c} and described by | |
| 880 @code{pass_empty_loop}. | |
| 881 | |
| 882 @item Unrolling of small loops | |
| 883 | |
| 884 This pass completely unrolls loops with few iterations. The pass | |
| 885 is located in @file{tree-ssa-loop-ivcanon.c} and described by | |
| 886 @code{pass_complete_unroll}. | |
| 887 | |
| 888 @item Predictive commoning | |
| 889 | |
| 890 This pass makes the code reuse the computations from the previous | |
| 891 iterations of the loops, especially loads and stores to memory. | |
| 892 It does so by storing the values of these computations to a bank | |
| 893 of temporary variables that are rotated at the end of loop. To avoid | |
| 894 the need for this rotation, the loop is then unrolled and the copies | |
| 895 of the loop body are rewritten to use the appropriate version of | |
| 896 the temporary variable. This pass is located in @file{tree-predcom.c} | |
| 897 and described by @code{pass_predcom}. | |
| 898 | |
| 899 @item Array prefetching | |
| 900 | |
| 901 This pass issues prefetch instructions for array references inside | |
| 902 loops. The pass is located in @file{tree-ssa-loop-prefetch.c} and | |
| 903 described by @code{pass_loop_prefetch}. | |
| 904 | |
| 905 @item Reassociation | |
| 906 | |
| 907 This pass rewrites arithmetic expressions to enable optimizations that | |
| 908 operate on them, like redundancy elimination and vectorization. The | |
| 909 pass is located in @file{tree-ssa-reassoc.c} and described by | |
| 910 @code{pass_reassoc}. | |
| 911 | |
| 912 @item Optimization of @code{stdarg} functions | |
| 913 | |
| 914 This pass tries to avoid the saving of register arguments into the | |
| 915 stack on entry to @code{stdarg} functions. If the function doesn't | |
| 916 use any @code{va_start} macros, no registers need to be saved. If | |
| 917 @code{va_start} macros are used, the @code{va_list} variables don't | |
| 918 escape the function, it is only necessary to save registers that will | |
| 919 be used in @code{va_arg} macros. For instance, if @code{va_arg} is | |
| 920 only used with integral types in the function, floating point | |
| 921 registers don't need to be saved. This pass is located in | |
| 922 @code{tree-stdarg.c} and described by @code{pass_stdarg}. | |
| 923 | |
| 924 @end itemize | |
| 925 | |
| 926 @node RTL passes | |
| 927 @section RTL passes | |
| 928 | |
|
19
58ad6c70ea60
update gcc from 4.4.0 to 4.4.1.
kent@firefly.cr.ie.u-ryukyu.ac.jp
parents:
0
diff
changeset
|
929 The following briefly describes the RTL generation and optimization |
|
58ad6c70ea60
update gcc from 4.4.0 to 4.4.1.
kent@firefly.cr.ie.u-ryukyu.ac.jp
parents:
0
diff
changeset
|
930 passes that are run after the Tree optimization passes. |
| 0 | 931 |
| 932 @itemize @bullet | |
| 933 @item RTL generation | |
| 934 | |
| 935 @c Avoiding overfull is tricky here. | |
| 936 The source files for RTL generation include | |
| 937 @file{stmt.c}, | |
| 938 @file{calls.c}, | |
| 939 @file{expr.c}, | |
| 940 @file{explow.c}, | |
| 941 @file{expmed.c}, | |
| 942 @file{function.c}, | |
| 943 @file{optabs.c} | |
| 944 and @file{emit-rtl.c}. | |
| 945 Also, the file | |
| 946 @file{insn-emit.c}, generated from the machine description by the | |
| 947 program @code{genemit}, is used in this pass. The header file | |
| 948 @file{expr.h} is used for communication within this pass. | |
| 949 | |
| 950 @findex genflags | |
| 951 @findex gencodes | |
| 952 The header files @file{insn-flags.h} and @file{insn-codes.h}, | |
| 953 generated from the machine description by the programs @code{genflags} | |
| 954 and @code{gencodes}, tell this pass which standard names are available | |
| 955 for use and which patterns correspond to them. | |
| 956 | |
|
19
58ad6c70ea60
update gcc from 4.4.0 to 4.4.1.
kent@firefly.cr.ie.u-ryukyu.ac.jp
parents:
0
diff
changeset
|
957 @item Generation of exception landing pads |
| 0 | 958 |
| 959 This pass generates the glue that handles communication between the | |
| 960 exception handling library routines and the exception handlers within | |
| 961 the function. Entry points in the function that are invoked by the | |
| 962 exception handling library are called @dfn{landing pads}. The code | |
|
19
58ad6c70ea60
update gcc from 4.4.0 to 4.4.1.
kent@firefly.cr.ie.u-ryukyu.ac.jp
parents:
0
diff
changeset
|
963 for this pass is located in @file{except.c}. |
| 0 | 964 |
|
19
58ad6c70ea60
update gcc from 4.4.0 to 4.4.1.
kent@firefly.cr.ie.u-ryukyu.ac.jp
parents:
0
diff
changeset
|
965 @item Control flow graph cleanup |
| 0 | 966 |
| 967 This pass removes unreachable code, simplifies jumps to next, jumps to | |
| 968 jump, jumps across jumps, etc. The pass is run multiple times. | |
| 969 For historical reasons, it is occasionally referred to as the ``jump | |
| 970 optimization pass''. The bulk of the code for this pass is in | |
| 971 @file{cfgcleanup.c}, and there are support routines in @file{cfgrtl.c} | |
| 972 and @file{jump.c}. | |
| 973 | |
| 974 @item Forward propagation of single-def values | |
| 975 | |
| 976 This pass attempts to remove redundant computation by substituting | |
| 977 variables that come from a single definition, and | |
| 978 seeing if the result can be simplified. It performs copy propagation | |
| 979 and addressing mode selection. The pass is run twice, with values | |
|
19
58ad6c70ea60
update gcc from 4.4.0 to 4.4.1.
kent@firefly.cr.ie.u-ryukyu.ac.jp
parents:
0
diff
changeset
|
980 being propagated into loops only on the second run. The code is |
|
58ad6c70ea60
update gcc from 4.4.0 to 4.4.1.
kent@firefly.cr.ie.u-ryukyu.ac.jp
parents:
0
diff
changeset
|
981 located in @file{fwprop.c}. |
| 0 | 982 |
| 983 @item Common subexpression elimination | |
| 984 | |
| 985 This pass removes redundant computation within basic blocks, and | |
| 986 optimizes addressing modes based on cost. The pass is run twice. | |
|
19
58ad6c70ea60
update gcc from 4.4.0 to 4.4.1.
kent@firefly.cr.ie.u-ryukyu.ac.jp
parents:
0
diff
changeset
|
987 The code for this pass is located in @file{cse.c}. |
| 0 | 988 |
|
19
58ad6c70ea60
update gcc from 4.4.0 to 4.4.1.
kent@firefly.cr.ie.u-ryukyu.ac.jp
parents:
0
diff
changeset
|
989 @item Global common subexpression elimination |
| 0 | 990 |
| 991 This pass performs two | |
| 992 different types of GCSE depending on whether you are optimizing for | |
| 993 size or not (LCM based GCSE tends to increase code size for a gain in | |
| 994 speed, while Morel-Renvoise based GCSE does not). | |
| 995 When optimizing for size, GCSE is done using Morel-Renvoise Partial | |
| 996 Redundancy Elimination, with the exception that it does not try to move | |
| 997 invariants out of loops---that is left to the loop optimization pass. | |
| 998 If MR PRE GCSE is done, code hoisting (aka unification) is also done, as | |
| 999 well as load motion. | |
| 1000 If you are optimizing for speed, LCM (lazy code motion) based GCSE is | |
| 1001 done. LCM is based on the work of Knoop, Ruthing, and Steffen. LCM | |
| 1002 based GCSE also does loop invariant code motion. We also perform load | |
| 1003 and store motion when optimizing for speed. | |
| 1004 Regardless of which type of GCSE is used, the GCSE pass also performs | |
| 1005 global constant and copy propagation. | |
| 1006 The source file for this pass is @file{gcse.c}, and the LCM routines | |
| 1007 are in @file{lcm.c}. | |
| 1008 | |
| 1009 @item Loop optimization | |
| 1010 | |
| 1011 This pass performs several loop related optimizations. | |
| 1012 The source files @file{cfgloopanal.c} and @file{cfgloopmanip.c} contain | |
| 1013 generic loop analysis and manipulation code. Initialization and finalization | |
| 1014 of loop structures is handled by @file{loop-init.c}. | |
| 1015 A loop invariant motion pass is implemented in @file{loop-invariant.c}. | |
| 111 | 1016 Basic block level optimizations---unrolling, and peeling loops--- |
| 1017 are implemented in @file{loop-unroll.c}. | |
| 0 | 1018 Replacing of the exit condition of loops by special machine-dependent |
| 1019 instructions is handled by @file{loop-doloop.c}. | |
| 1020 | |
| 1021 @item Jump bypassing | |
| 1022 | |
| 1023 This pass is an aggressive form of GCSE that transforms the control | |
| 1024 flow graph of a function by propagating constants into conditional | |
| 1025 branch instructions. The source file for this pass is @file{gcse.c}. | |
| 1026 | |
| 1027 @item If conversion | |
| 1028 | |
| 1029 This pass attempts to replace conditional branches and surrounding | |
| 1030 assignments with arithmetic, boolean value producing comparison | |
| 1031 instructions, and conditional move instructions. In the very last | |
| 111 | 1032 invocation after reload/LRA, it will generate predicated instructions |
|
19
58ad6c70ea60
update gcc from 4.4.0 to 4.4.1.
kent@firefly.cr.ie.u-ryukyu.ac.jp
parents:
0
diff
changeset
|
1033 when supported by the target. The code is located in @file{ifcvt.c}. |
| 0 | 1034 |
| 1035 @item Web construction | |
| 1036 | |
| 1037 This pass splits independent uses of each pseudo-register. This can | |
| 1038 improve effect of the other transformation, such as CSE or register | |
|
19
58ad6c70ea60
update gcc from 4.4.0 to 4.4.1.
kent@firefly.cr.ie.u-ryukyu.ac.jp
parents:
0
diff
changeset
|
1039 allocation. The code for this pass is located in @file{web.c}. |
| 0 | 1040 |
| 1041 @item Instruction combination | |
| 1042 | |
| 1043 This pass attempts to combine groups of two or three instructions that | |
| 1044 are related by data flow into single instructions. It combines the | |
| 1045 RTL expressions for the instructions by substitution, simplifies the | |
| 1046 result using algebra, and then attempts to match the result against | |
|
19
58ad6c70ea60
update gcc from 4.4.0 to 4.4.1.
kent@firefly.cr.ie.u-ryukyu.ac.jp
parents:
0
diff
changeset
|
1047 the machine description. The code is located in @file{combine.c}. |
| 0 | 1048 |
|
19
58ad6c70ea60
update gcc from 4.4.0 to 4.4.1.
kent@firefly.cr.ie.u-ryukyu.ac.jp
parents:
0
diff
changeset
|
1049 @item Mode switching optimization |
| 0 | 1050 |
| 1051 This pass looks for instructions that require the processor to be in a | |
| 1052 specific ``mode'' and minimizes the number of mode changes required to | |
| 1053 satisfy all users. What these modes are, and what they apply to are | |
|
19
58ad6c70ea60
update gcc from 4.4.0 to 4.4.1.
kent@firefly.cr.ie.u-ryukyu.ac.jp
parents:
0
diff
changeset
|
1054 completely target-specific. The code for this pass is located in |
|
58ad6c70ea60
update gcc from 4.4.0 to 4.4.1.
kent@firefly.cr.ie.u-ryukyu.ac.jp
parents:
0
diff
changeset
|
1055 @file{mode-switching.c}. |
| 0 | 1056 |
| 1057 @cindex modulo scheduling | |
| 1058 @cindex sms, swing, software pipelining | |
| 1059 @item Modulo scheduling | |
| 1060 | |
| 1061 This pass looks at innermost loops and reorders their instructions | |
| 1062 by overlapping different iterations. Modulo scheduling is performed | |
|
19
58ad6c70ea60
update gcc from 4.4.0 to 4.4.1.
kent@firefly.cr.ie.u-ryukyu.ac.jp
parents:
0
diff
changeset
|
1063 immediately before instruction scheduling. The code for this pass is |
|
58ad6c70ea60
update gcc from 4.4.0 to 4.4.1.
kent@firefly.cr.ie.u-ryukyu.ac.jp
parents:
0
diff
changeset
|
1064 located in @file{modulo-sched.c}. |
| 0 | 1065 |
| 1066 @item Instruction scheduling | |
| 1067 | |
| 1068 This pass looks for instructions whose output will not be available by | |
| 1069 the time that it is used in subsequent instructions. Memory loads and | |
| 1070 floating point instructions often have this behavior on RISC machines. | |
| 1071 It re-orders instructions within a basic block to try to separate the | |
| 1072 definition and use of items that otherwise would cause pipeline | |
| 1073 stalls. This pass is performed twice, before and after register | |
|
19
58ad6c70ea60
update gcc from 4.4.0 to 4.4.1.
kent@firefly.cr.ie.u-ryukyu.ac.jp
parents:
0
diff
changeset
|
1074 allocation. The code for this pass is located in @file{haifa-sched.c}, |
| 0 | 1075 @file{sched-deps.c}, @file{sched-ebb.c}, @file{sched-rgn.c} and |
| 1076 @file{sched-vis.c}. | |
| 1077 | |
| 1078 @item Register allocation | |
| 1079 | |
| 1080 These passes make sure that all occurrences of pseudo registers are | |
| 1081 eliminated, either by allocating them to a hard register, replacing | |
| 1082 them by an equivalent expression (e.g.@: a constant) or by placing | |
| 1083 them on the stack. This is done in several subpasses: | |
| 1084 | |
| 1085 @itemize @bullet | |
| 1086 @item | |
| 1087 The integrated register allocator (@acronym{IRA}). It is called | |
| 1088 integrated because coalescing, register live range splitting, and hard | |
| 1089 register preferencing are done on-the-fly during coloring. It also | |
| 111 | 1090 has better integration with the reload/LRA pass. Pseudo-registers spilled |
| 1091 by the allocator or the reload/LRA have still a chance to get | |
| 1092 hard-registers if the reload/LRA evicts some pseudo-registers from | |
| 0 | 1093 hard-registers. The allocator helps to choose better pseudos for |
| 1094 spilling based on their live ranges and to coalesce stack slots | |
| 1095 allocated for the spilled pseudo-registers. IRA is a regional | |
| 1096 register allocator which is transformed into Chaitin-Briggs allocator | |
| 1097 if there is one region. By default, IRA chooses regions using | |
| 1098 register pressure but the user can force it to use one region or | |
| 1099 regions corresponding to all loops. | |
| 1100 | |
| 1101 Source files of the allocator are @file{ira.c}, @file{ira-build.c}, | |
| 1102 @file{ira-costs.c}, @file{ira-conflicts.c}, @file{ira-color.c}, | |
| 1103 @file{ira-emit.c}, @file{ira-lives}, plus header files @file{ira.h} | |
| 1104 and @file{ira-int.h} used for the communication between the allocator | |
| 1105 and the rest of the compiler and between the IRA files. | |
| 1106 | |
| 1107 @cindex reloading | |
| 1108 @item | |
| 1109 Reloading. This pass renumbers pseudo registers with the hardware | |
| 1110 registers numbers they were allocated. Pseudo registers that did not | |
| 1111 get hard registers are replaced with stack slots. Then it finds | |
| 1112 instructions that are invalid because a value has failed to end up in | |
| 1113 a register, or has ended up in a register of the wrong kind. It fixes | |
| 1114 up these instructions by reloading the problematical values | |
| 1115 temporarily into registers. Additional instructions are generated to | |
| 1116 do the copying. | |
| 1117 | |
| 1118 The reload pass also optionally eliminates the frame pointer and inserts | |
| 1119 instructions to save and restore call-clobbered registers around calls. | |
| 1120 | |
| 1121 Source files are @file{reload.c} and @file{reload1.c}, plus the header | |
| 1122 @file{reload.h} used for communication between them. | |
| 111 | 1123 |
| 1124 @cindex Local Register Allocator (LRA) | |
| 1125 @item | |
| 1126 This pass is a modern replacement of the reload pass. Source files | |
| 1127 are @file{lra.c}, @file{lra-assign.c}, @file{lra-coalesce.c}, | |
| 1128 @file{lra-constraints.c}, @file{lra-eliminations.c}, | |
| 1129 @file{lra-lives.c}, @file{lra-remat.c}, @file{lra-spills.c}, the | |
| 1130 header @file{lra-int.h} used for communication between them, and the | |
| 1131 header @file{lra.h} used for communication between LRA and the rest of | |
| 1132 compiler. | |
| 1133 | |
| 1134 Unlike the reload pass, intermediate LRA decisions are reflected in | |
| 1135 RTL as much as possible. This reduces the number of target-dependent | |
| 1136 macros and hooks, leaving instruction constraints as the primary | |
| 1137 source of control. | |
| 1138 | |
| 1139 LRA is run on targets for which TARGET_LRA_P returns true. | |
| 0 | 1140 @end itemize |
| 1141 | |
| 1142 @item Basic block reordering | |
| 1143 | |
| 1144 This pass implements profile guided code positioning. If profile | |
| 1145 information is not available, various types of static analysis are | |
| 1146 performed to make the predictions normally coming from the profile | |
| 1147 feedback (IE execution frequency, branch probability, etc). It is | |
| 1148 implemented in the file @file{bb-reorder.c}, and the various | |
| 1149 prediction routines are in @file{predict.c}. | |
| 1150 | |
| 1151 @item Variable tracking | |
| 1152 | |
| 1153 This pass computes where the variables are stored at each | |
| 1154 position in code and generates notes describing the variable locations | |
| 1155 to RTL code. The location lists are then generated according to these | |
| 1156 notes to debug information if the debugging information format supports | |
|
19
58ad6c70ea60
update gcc from 4.4.0 to 4.4.1.
kent@firefly.cr.ie.u-ryukyu.ac.jp
parents:
0
diff
changeset
|
1157 location lists. The code is located in @file{var-tracking.c}. |
| 0 | 1158 |
| 1159 @item Delayed branch scheduling | |
| 1160 | |
| 1161 This optional pass attempts to find instructions that can go into the | |
|
19
58ad6c70ea60
update gcc from 4.4.0 to 4.4.1.
kent@firefly.cr.ie.u-ryukyu.ac.jp
parents:
0
diff
changeset
|
1162 delay slots of other instructions, usually jumps and calls. The code |
|
58ad6c70ea60
update gcc from 4.4.0 to 4.4.1.
kent@firefly.cr.ie.u-ryukyu.ac.jp
parents:
0
diff
changeset
|
1163 for this pass is located in @file{reorg.c}. |
| 0 | 1164 |
| 1165 @item Branch shortening | |
| 1166 | |
| 1167 On many RISC machines, branch instructions have a limited range. | |
| 1168 Thus, longer sequences of instructions must be used for long branches. | |
| 1169 In this pass, the compiler figures out what how far each instruction | |
| 1170 will be from each other instruction, and therefore whether the usual | |
| 1171 instructions, or the longer sequences, must be used for each branch. | |
|
19
58ad6c70ea60
update gcc from 4.4.0 to 4.4.1.
kent@firefly.cr.ie.u-ryukyu.ac.jp
parents:
0
diff
changeset
|
1172 The code for this pass is located in @file{final.c}. |
| 0 | 1173 |
| 1174 @item Register-to-stack conversion | |
| 1175 | |
| 1176 Conversion from usage of some hard registers to usage of a register | |
| 1177 stack may be done at this point. Currently, this is supported only | |
|
19
58ad6c70ea60
update gcc from 4.4.0 to 4.4.1.
kent@firefly.cr.ie.u-ryukyu.ac.jp
parents:
0
diff
changeset
|
1178 for the floating-point registers of the Intel 80387 coprocessor. The |
|
58ad6c70ea60
update gcc from 4.4.0 to 4.4.1.
kent@firefly.cr.ie.u-ryukyu.ac.jp
parents:
0
diff
changeset
|
1179 code for this pass is located in @file{reg-stack.c}. |
| 0 | 1180 |
| 1181 @item Final | |
| 1182 | |
| 1183 This pass outputs the assembler code for the function. The source files | |
| 1184 are @file{final.c} plus @file{insn-output.c}; the latter is generated | |
| 1185 automatically from the machine description by the tool @file{genoutput}. | |
| 1186 The header file @file{conditions.h} is used for communication between | |
| 111 | 1187 these files. |
| 0 | 1188 |
| 1189 @item Debugging information output | |
| 1190 | |
| 1191 This is run after final because it must output the stack slot offsets | |
| 1192 for pseudo registers that did not get hard registers. Source files | |
| 131 | 1193 are @file{dbxout.c} for DBX symbol table format, @file{dwarfout.c} for |
| 1194 DWARF symbol table format, files @file{dwarf2out.c} and @file{dwarf2asm.c} | |
| 1195 for DWARF2 symbol table format, and @file{vmsdbgout.c} for VMS debug | |
| 1196 symbol table format. | |
| 0 | 1197 |
| 1198 @end itemize | |
| 111 | 1199 |
| 1200 @node Optimization info | |
| 1201 @section Optimization info | |
| 1202 @include optinfo.texi |