1 This is gdbint.info, produced by makeinfo version 4.13 from 2 /tmp/android-1596/src/build/../gdb/gdb-7.1.x/gdb/doc/gdbint.texinfo. 3 4 INFO-DIR-SECTION Software development 5 START-INFO-DIR-ENTRY 6 * Gdb-Internals: (gdbint). The GNU debugger's internals. 7 END-INFO-DIR-ENTRY 8 9 Copyright (C) 1990, 1991, 1992, 1993, 1994, 1996, 1998, 1999, 2000, 10 2001, 2002, 2003, 2004, 2005, 2006, 2008, 2009, 2010 Free Software 11 Foundation, Inc. Contributed by Cygnus Solutions. Written by John 12 Gilmore. Second Edition by Stan Shebs. 13 14 Permission is granted to copy, distribute and/or modify this document 15 under the terms of the GNU Free Documentation License, Version 1.1 or 16 any later version published by the Free Software Foundation; with no 17 Invariant Sections, with no Front-Cover Texts, and with no Back-Cover 18 Texts. A copy of the license is included in the section entitled "GNU 19 Free Documentation License". 20 21 This file documents the internals of the GNU debugger GDB. 22 23 Copyright (C) 1990, 1991, 1992, 1993, 1994, 1996, 1998, 1999, 2000, 24 2001, 2002, 2003, 2004, 2005, 2006, 2008, 2009, 2010 Free Software 25 Foundation, Inc. Contributed by Cygnus Solutions. Written by John 26 Gilmore. Second Edition by Stan Shebs. 27 28 Permission is granted to copy, distribute and/or modify this document 29 under the terms of the GNU Free Documentation License, Version 1.1 or 30 any later version published by the Free Software Foundation; with no 31 Invariant Sections, with no Front-Cover Texts, and with no Back-Cover 32 Texts. A copy of the license is included in the section entitled "GNU 33 Free Documentation License". 34 35 36 File: gdbint.info, Node: Top, Next: Summary, Up: (dir) 37 38 Scope of this Document 39 ********************** 40 41 This document documents the internals of the GNU debugger, GDB. It 42 includes description of GDB's key algorithms and operations, as well as 43 the mechanisms that adapt GDB to specific hosts and targets. 44 45 * Menu: 46 47 * Summary:: 48 * Overall Structure:: 49 * Algorithms:: 50 * User Interface:: 51 * libgdb:: 52 * Values:: 53 * Stack Frames:: 54 * Symbol Handling:: 55 * Language Support:: 56 * Host Definition:: 57 * Target Architecture Definition:: 58 * Target Descriptions:: 59 * Target Vector Definition:: 60 * Native Debugging:: 61 * Support Libraries:: 62 * Coding:: 63 * Porting GDB:: 64 * Versions and Branches:: 65 * Start of New Year Procedure:: 66 * Releasing GDB:: 67 * Testsuite:: 68 * Hints:: 69 70 * GDB Observers:: GDB Currently available observers 71 * GNU Free Documentation License:: The license for this documentation 72 * Index:: 73 74 75 File: gdbint.info, Node: Summary, Next: Overall Structure, Prev: Top, Up: Top 76 77 1 Summary 78 ********* 79 80 * Menu: 81 82 * Requirements:: 83 * Contributors:: 84 85 86 File: gdbint.info, Node: Requirements, Next: Contributors, Up: Summary 87 88 1.1 Requirements 89 ================ 90 91 Before diving into the internals, you should understand the formal 92 requirements and other expectations for GDB. Although some of these 93 may seem obvious, there have been proposals for GDB that have run 94 counter to these requirements. 95 96 First of all, GDB is a debugger. It's not designed to be a front 97 panel for embedded systems. It's not a text editor. It's not a shell. 98 It's not a programming environment. 99 100 GDB is an interactive tool. Although a batch mode is available, 101 GDB's primary role is to interact with a human programmer. 102 103 GDB should be responsive to the user. A programmer hot on the trail 104 of a nasty bug, and operating under a looming deadline, is going to be 105 very impatient of everything, including the response time to debugger 106 commands. 107 108 GDB should be relatively permissive, such as for expressions. While 109 the compiler should be picky (or have the option to be made picky), 110 since source code lives for a long time usually, the programmer doing 111 debugging shouldn't be spending time figuring out to mollify the 112 debugger. 113 114 GDB will be called upon to deal with really large programs. 115 Executable sizes of 50 to 100 megabytes occur regularly, and we've 116 heard reports of programs approaching 1 gigabyte in size. 117 118 GDB should be able to run everywhere. No other debugger is 119 available for even half as many configurations as GDB supports. 120 121 122 File: gdbint.info, Node: Contributors, Prev: Requirements, Up: Summary 123 124 1.2 Contributors 125 ================ 126 127 The first edition of this document was written by John Gilmore of 128 Cygnus Solutions. The current second edition was written by Stan Shebs 129 of Cygnus Solutions, who continues to update the manual. 130 131 Over the years, many others have made additions and changes to this 132 document. This section attempts to record the significant contributors 133 to that effort. One of the virtues of free software is that everyone is 134 free to contribute to it; with regret, we cannot actually acknowledge 135 everyone here. 136 137 _Plea:_ This section has only been added relatively recently (four 138 years after publication of the second edition). Additions to this 139 section are particularly welcome. If you or your friends (or 140 enemies, to be evenhanded) have been unfairly omitted from this 141 list, we would like to add your names! 142 143 A document such as this relies on being kept up to date by numerous 144 small updates by contributing engineers as they make changes to the 145 code base. The file `ChangeLog' in the GDB distribution approximates a 146 blow-by-blow account. The most prolific contributors to this important, 147 but low profile task are Andrew Cagney (responsible for over half the 148 entries), Daniel Jacobowitz, Mark Kettenis, Jim Blandy and Eli 149 Zaretskii. 150 151 Eli Zaretskii and Daniel Jacobowitz wrote the sections documenting 152 watchpoints. 153 154 Jeremy Bennett updated the sections on initializing a new 155 architecture and register representation, and added the section on 156 Frame Interpretation. 157 158 159 File: gdbint.info, Node: Overall Structure, Next: Algorithms, Prev: Summary, Up: Top 160 161 2 Overall Structure 162 ******************* 163 164 GDB consists of three major subsystems: user interface, symbol handling 165 (the "symbol side"), and target system handling (the "target side"). 166 167 The user interface consists of several actual interfaces, plus 168 supporting code. 169 170 The symbol side consists of object file readers, debugging info 171 interpreters, symbol table management, source language expression 172 parsing, type and value printing. 173 174 The target side consists of execution control, stack frame analysis, 175 and physical target manipulation. 176 177 The target side/symbol side division is not formal, and there are a 178 number of exceptions. For instance, core file support involves symbolic 179 elements (the basic core file reader is in BFD) and target elements (it 180 supplies the contents of memory and the values of registers). Instead, 181 this division is useful for understanding how the minor subsystems 182 should fit together. 183 184 2.1 The Symbol Side 185 =================== 186 187 The symbolic side of GDB can be thought of as "everything you can do in 188 GDB without having a live program running". For instance, you can look 189 at the types of variables, and evaluate many kinds of expressions. 190 191 2.2 The Target Side 192 =================== 193 194 The target side of GDB is the "bits and bytes manipulator". Although 195 it may make reference to symbolic info here and there, most of the 196 target side will run with only a stripped executable available--or even 197 no executable at all, in remote debugging cases. 198 199 Operations such as disassembly, stack frame crawls, and register 200 display, are able to work with no symbolic info at all. In some cases, 201 such as disassembly, GDB will use symbolic info to present addresses 202 relative to symbols rather than as raw numbers, but it will work either 203 way. 204 205 2.3 Configurations 206 ================== 207 208 "Host" refers to attributes of the system where GDB runs. "Target" 209 refers to the system where the program being debugged executes. In 210 most cases they are the same machine, in which case a third type of 211 "Native" attributes come into play. 212 213 Defines and include files needed to build on the host are host 214 support. Examples are tty support, system defined types, host byte 215 order, host float format. These are all calculated by `autoconf' when 216 the debugger is built. 217 218 Defines and information needed to handle the target format are target 219 dependent. Examples are the stack frame format, instruction set, 220 breakpoint instruction, registers, and how to set up and tear down the 221 stack to call a function. 222 223 Information that is only needed when the host and target are the 224 same, is native dependent. One example is Unix child process support; 225 if the host and target are not the same, calling `fork' to start the 226 target process is a bad idea. The various macros needed for finding the 227 registers in the `upage', running `ptrace', and such are all in the 228 native-dependent files. 229 230 Another example of native-dependent code is support for features that 231 are really part of the target environment, but which require `#include' 232 files that are only available on the host system. Core file handling 233 and `setjmp' handling are two common cases. 234 235 When you want to make GDB work as the traditional native debugger on 236 a system, you will need to supply both target and native information. 237 238 2.4 Source Tree Structure 239 ========================= 240 241 The GDB source directory has a mostly flat structure--there are only a 242 few subdirectories. A file's name usually gives a hint as to what it 243 does; for example, `stabsread.c' reads stabs, `dwarf2read.c' reads 244 DWARF 2, etc. 245 246 Files that are related to some common task have names that share 247 common substrings. For example, `*-thread.c' files deal with debugging 248 threads on various platforms; `*read.c' files deal with reading various 249 kinds of symbol and object files; `inf*.c' files deal with direct 250 control of the "inferior program" (GDB parlance for the program being 251 debugged). 252 253 There are several dozens of files in the `*-tdep.c' family. `tdep' 254 stands for "target-dependent code"--each of these files implements 255 debug support for a specific target architecture (sparc, mips, etc). 256 Usually, only one of these will be used in a specific GDB configuration 257 (sometimes two, closely related). 258 259 Similarly, there are many `*-nat.c' files, each one for native 260 debugging on a specific system (e.g., `sparc-linux-nat.c' is for native 261 debugging of Sparc machines running the Linux kernel). 262 263 The few subdirectories of the source tree are: 264 265 `cli' 266 Code that implements "CLI", the GDB Command-Line Interpreter. 267 *Note Command Interpreter: User Interface. 268 269 `gdbserver' 270 Code for the GDB remote server. 271 272 `gdbtk' 273 Code for Insight, the GDB TK-based GUI front-end. 274 275 `mi' 276 The "GDB/MI", the GDB Machine Interface interpreter. 277 278 `signals' 279 Target signal translation code. 280 281 `tui' 282 Code for "TUI", the GDB Text-mode full-screen User Interface. 283 *Note TUI: User Interface. 284 285 286 File: gdbint.info, Node: Algorithms, Next: User Interface, Prev: Overall Structure, Up: Top 287 288 3 Algorithms 289 ************ 290 291 GDB uses a number of debugging-specific algorithms. They are often not 292 very complicated, but get lost in the thicket of special cases and 293 real-world issues. This chapter describes the basic algorithms and 294 mentions some of the specific target definitions that they use. 295 296 3.1 Prologue Analysis 297 ===================== 298 299 To produce a backtrace and allow the user to manipulate older frames' 300 variables and arguments, GDB needs to find the base addresses of older 301 frames, and discover where those frames' registers have been saved. 302 Since a frame's "callee-saves" registers get saved by younger frames if 303 and when they're reused, a frame's registers may be scattered 304 unpredictably across younger frames. This means that changing the 305 value of a register-allocated variable in an older frame may actually 306 entail writing to a save slot in some younger frame. 307 308 Modern versions of GCC emit Dwarf call frame information ("CFI"), 309 which describes how to find frame base addresses and saved registers. 310 But CFI is not always available, so as a fallback GDB uses a technique 311 called "prologue analysis" to find frame sizes and saved registers. A 312 prologue analyzer disassembles the function's machine code starting 313 from its entry point, and looks for instructions that allocate frame 314 space, save the stack pointer in a frame pointer register, save 315 registers, and so on. Obviously, this can't be done accurately in 316 general, but it's tractable to do well enough to be very helpful. 317 Prologue analysis predates the GNU toolchain's support for CFI; at one 318 time, prologue analysis was the only mechanism GDB used for stack 319 unwinding at all, when the function calling conventions didn't specify 320 a fixed frame layout. 321 322 In the olden days, function prologues were generated by hand-written, 323 target-specific code in GCC, and treated as opaque and untouchable by 324 optimizers. Looking at this code, it was usually straightforward to 325 write a prologue analyzer for GDB that would accurately understand all 326 the prologues GCC would generate. However, over time GCC became more 327 aggressive about instruction scheduling, and began to understand more 328 about the semantics of the prologue instructions themselves; in 329 response, GDB's analyzers became more complex and fragile. Keeping the 330 prologue analyzers working as GCC (and the instruction sets themselves) 331 evolved became a substantial task. 332 333 To try to address this problem, the code in `prologue-value.h' and 334 `prologue-value.c' provides a general framework for writing prologue 335 analyzers that are simpler and more robust than ad-hoc analyzers. When 336 we analyze a prologue using the prologue-value framework, we're really 337 doing "abstract interpretation" or "pseudo-evaluation": running the 338 function's code in simulation, but using conservative approximations of 339 the values registers and memory would hold when the code actually runs. 340 For example, if our function starts with the instruction: 341 342 addi r1, 42 # add 42 to r1 343 we don't know exactly what value will be in `r1' after executing 344 this instruction, but we do know it'll be 42 greater than its original 345 value. 346 347 If we then see an instruction like: 348 349 addi r1, 22 # add 22 to r1 350 we still don't know what `r1's' value is, but again, we can say it 351 is now 64 greater than its original value. 352 353 If the next instruction were: 354 355 mov r2, r1 # set r2 to r1's value 356 then we can say that `r2's' value is now the original value of `r1' 357 plus 64. 358 359 It's common for prologues to save registers on the stack, so we'll 360 need to track the values of stack frame slots, as well as the 361 registers. So after an instruction like this: 362 363 mov (fp+4), r2 364 then we'd know that the stack slot four bytes above the frame pointer 365 holds the original value of `r1' plus 64. 366 367 And so on. 368 369 Of course, this can only go so far before it gets unreasonable. If 370 we wanted to be able to say anything about the value of `r1' after the 371 instruction: 372 373 xor r1, r3 # exclusive-or r1 and r3, place result in r1 374 then things would get pretty complex. But remember, we're just doing 375 a conservative approximation; if exclusive-or instructions aren't 376 relevant to prologues, we can just say `r1''s value is now "unknown". 377 We can ignore things that are too complex, if that loss of information 378 is acceptable for our application. 379 380 So when we say "conservative approximation" here, what we mean is an 381 approximation that is either accurate, or marked "unknown", but never 382 inaccurate. 383 384 Using this framework, a prologue analyzer is simply an interpreter 385 for machine code, but one that uses conservative approximations for the 386 contents of registers and memory instead of actual values. Starting 387 from the function's entry point, you simulate instructions up to the 388 current PC, or an instruction that you don't know how to simulate. Now 389 you can examine the state of the registers and stack slots you've kept 390 track of. 391 392 * To see how large your stack frame is, just check the value of the 393 stack pointer register; if it's the original value of the SP minus 394 a constant, then that constant is the stack frame's size. If the 395 SP's value has been marked as "unknown", then that means the 396 prologue has done something too complex for us to track, and we 397 don't know the frame size. 398 399 * To see where we've saved the previous frame's registers, we just 400 search the values we've tracked -- stack slots, usually, but 401 registers, too, if you want -- for something equal to the 402 register's original value. If the calling conventions suggest a 403 standard place to save a given register, then we can check there 404 first, but really, anything that will get us back the original 405 value will probably work. 406 407 This does take some work. But prologue analyzers aren't 408 quick-and-simple pattern patching to recognize a few fixed prologue 409 forms any more; they're big, hairy functions. Along with inferior 410 function calls, prologue analysis accounts for a substantial portion of 411 the time needed to stabilize a GDB port. So it's worthwhile to look 412 for an approach that will be easier to understand and maintain. In the 413 approach described above: 414 415 * It's easier to see that the analyzer is correct: you just see 416 whether the analyzer properly (albeit conservatively) simulates 417 the effect of each instruction. 418 419 * It's easier to extend the analyzer: you can add support for new 420 instructions, and know that you haven't broken anything that 421 wasn't already broken before. 422 423 * It's orthogonal: to gather new information, you don't need to 424 complicate the code for each instruction. As long as your domain 425 of conservative values is already detailed enough to tell you what 426 you need, then all the existing instruction simulations are 427 already gathering the right data for you. 428 429 430 The file `prologue-value.h' contains detailed comments explaining 431 the framework and how to use it. 432 433 3.2 Breakpoint Handling 434 ======================= 435 436 In general, a breakpoint is a user-designated location in the program 437 where the user wants to regain control if program execution ever reaches 438 that location. 439 440 There are two main ways to implement breakpoints; either as 441 "hardware" breakpoints or as "software" breakpoints. 442 443 Hardware breakpoints are sometimes available as a builtin debugging 444 features with some chips. Typically these work by having dedicated 445 register into which the breakpoint address may be stored. If the PC 446 (shorthand for "program counter") ever matches a value in a breakpoint 447 registers, the CPU raises an exception and reports it to GDB. 448 449 Another possibility is when an emulator is in use; many emulators 450 include circuitry that watches the address lines coming out from the 451 processor, and force it to stop if the address matches a breakpoint's 452 address. 453 454 A third possibility is that the target already has the ability to do 455 breakpoints somehow; for instance, a ROM monitor may do its own 456 software breakpoints. So although these are not literally "hardware 457 breakpoints", from GDB's point of view they work the same; GDB need not 458 do anything more than set the breakpoint and wait for something to 459 happen. 460 461 Since they depend on hardware resources, hardware breakpoints may be 462 limited in number; when the user asks for more, GDB will start trying 463 to set software breakpoints. (On some architectures, notably the 464 32-bit x86 platforms, GDB cannot always know whether there's enough 465 hardware resources to insert all the hardware breakpoints and 466 watchpoints. On those platforms, GDB prints an error message only when 467 the program being debugged is continued.) 468 469 Software breakpoints require GDB to do somewhat more work. The 470 basic theory is that GDB will replace a program instruction with a 471 trap, illegal divide, or some other instruction that will cause an 472 exception, and then when it's encountered, GDB will take the exception 473 and stop the program. When the user says to continue, GDB will restore 474 the original instruction, single-step, re-insert the trap, and continue 475 on. 476 477 Since it literally overwrites the program being tested, the program 478 area must be writable, so this technique won't work on programs in ROM. 479 It can also distort the behavior of programs that examine themselves, 480 although such a situation would be highly unusual. 481 482 Also, the software breakpoint instruction should be the smallest 483 size of instruction, so it doesn't overwrite an instruction that might 484 be a jump target, and cause disaster when the program jumps into the 485 middle of the breakpoint instruction. (Strictly speaking, the 486 breakpoint must be no larger than the smallest interval between 487 instructions that may be jump targets; perhaps there is an architecture 488 where only even-numbered instructions may jumped to.) Note that it's 489 possible for an instruction set not to have any instructions usable for 490 a software breakpoint, although in practice only the ARC has failed to 491 define such an instruction. 492 493 Basic breakpoint object handling is in `breakpoint.c'. However, 494 much of the interesting breakpoint action is in `infrun.c'. 495 496 `target_remove_breakpoint (BP_TGT)' 497 `target_insert_breakpoint (BP_TGT)' 498 Insert or remove a software breakpoint at address 499 `BP_TGT->placed_address'. Returns zero for success, non-zero for 500 failure. On input, BP_TGT contains the address of the breakpoint, 501 and is otherwise initialized to zero. The fields of the `struct 502 bp_target_info' pointed to by BP_TGT are updated to contain other 503 information about the breakpoint on output. The field 504 `placed_address' may be updated if the breakpoint was placed at a 505 related address; the field `shadow_contents' contains the real 506 contents of the bytes where the breakpoint has been inserted, if 507 reading memory would return the breakpoint instead of the 508 underlying memory; the field `shadow_len' is the length of memory 509 cached in `shadow_contents', if any; and the field `placed_size' 510 is optionally set and used by the target, if it could differ from 511 `shadow_len'. 512 513 For example, the remote target `Z0' packet does not require 514 shadowing memory, so `shadow_len' is left at zero. However, the 515 length reported by `gdbarch_breakpoint_from_pc' is cached in 516 `placed_size', so that a matching `z0' packet can be used to 517 remove the breakpoint. 518 519 `target_remove_hw_breakpoint (BP_TGT)' 520 `target_insert_hw_breakpoint (BP_TGT)' 521 Insert or remove a hardware-assisted breakpoint at address 522 `BP_TGT->placed_address'. Returns zero for success, non-zero for 523 failure. See `target_insert_breakpoint' for a description of the 524 `struct bp_target_info' pointed to by BP_TGT; the 525 `shadow_contents' and `shadow_len' members are not used for 526 hardware breakpoints, but `placed_size' may be. 527 528 3.3 Single Stepping 529 =================== 530 531 3.4 Signal Handling 532 =================== 533 534 3.5 Thread Handling 535 =================== 536 537 3.6 Inferior Function Calls 538 =========================== 539 540 3.7 Longjmp Support 541 =================== 542 543 GDB has support for figuring out that the target is doing a `longjmp' 544 and for stopping at the target of the jump, if we are stepping. This 545 is done with a few specialized internal breakpoints, which are visible 546 in the output of the `maint info breakpoint' command. 547 548 To make this work, you need to define a function called 549 `gdbarch_get_longjmp_target', which will examine the `jmp_buf' 550 structure and extract the `longjmp' target address. Since `jmp_buf' is 551 target specific and typically defined in a target header not available 552 to GDB, you will need to determine the offset of the PC manually and 553 return that; many targets define a `jb_pc_offset' field in the tdep 554 structure to save the value once calculated. 555 556 3.8 Watchpoints 557 =============== 558 559 Watchpoints are a special kind of breakpoints (*note breakpoints: 560 Algorithms.) which break when data is accessed rather than when some 561 instruction is executed. When you have data which changes without your 562 knowing what code does that, watchpoints are the silver bullet to hunt 563 down and kill such bugs. 564 565 Watchpoints can be either hardware-assisted or not; the latter type 566 is known as "software watchpoints." GDB always uses hardware-assisted 567 watchpoints if they are available, and falls back on software 568 watchpoints otherwise. Typical situations where GDB will use software 569 watchpoints are: 570 571 * The watched memory region is too large for the underlying hardware 572 watchpoint support. For example, each x86 debug register can 573 watch up to 4 bytes of memory, so trying to watch data structures 574 whose size is more than 16 bytes will cause GDB to use software 575 watchpoints. 576 577 * The value of the expression to be watched depends on data held in 578 registers (as opposed to memory). 579 580 * Too many different watchpoints requested. (On some architectures, 581 this situation is impossible to detect until the debugged program 582 is resumed.) Note that x86 debug registers are used both for 583 hardware breakpoints and for watchpoints, so setting too many 584 hardware breakpoints might cause watchpoint insertion to fail. 585 586 * No hardware-assisted watchpoints provided by the target 587 implementation. 588 589 Software watchpoints are very slow, since GDB needs to single-step 590 the program being debugged and test the value of the watched 591 expression(s) after each instruction. The rest of this section is 592 mostly irrelevant for software watchpoints. 593 594 When the inferior stops, GDB tries to establish, among other 595 possible reasons, whether it stopped due to a watchpoint being hit. It 596 first uses `STOPPED_BY_WATCHPOINT' to see if any watchpoint was hit. 597 If not, all watchpoint checking is skipped. 598 599 Then GDB calls `target_stopped_data_address' exactly once. This 600 method returns the address of the watchpoint which triggered, if the 601 target can determine it. If the triggered address is available, GDB 602 compares the address returned by this method with each watched memory 603 address in each active watchpoint. For data-read and data-access 604 watchpoints, GDB announces every watchpoint that watches the triggered 605 address as being hit. For this reason, data-read and data-access 606 watchpoints _require_ that the triggered address be available; if not, 607 read and access watchpoints will never be considered hit. For 608 data-write watchpoints, if the triggered address is available, GDB 609 considers only those watchpoints which match that address; otherwise, 610 GDB considers all data-write watchpoints. For each data-write 611 watchpoint that GDB considers, it evaluates the expression whose value 612 is being watched, and tests whether the watched value has changed. 613 Watchpoints whose watched values have changed are announced as hit. 614 615 GDB uses several macros and primitives to support hardware 616 watchpoints: 617 618 `TARGET_CAN_USE_HARDWARE_WATCHPOINT (TYPE, COUNT, OTHER)' 619 Return the number of hardware watchpoints of type TYPE that are 620 possible to be set. The value is positive if COUNT watchpoints of 621 this type can be set, zero if setting watchpoints of this type is 622 not supported, and negative if COUNT is more than the maximum 623 number of watchpoints of type TYPE that can be set. OTHER is 624 non-zero if other types of watchpoints are currently enabled (there 625 are architectures which cannot set watchpoints of different types 626 at the same time). 627 628 `TARGET_REGION_OK_FOR_HW_WATCHPOINT (ADDR, LEN)' 629 Return non-zero if hardware watchpoints can be used to watch a 630 region whose address is ADDR and whose length in bytes is LEN. 631 632 `target_insert_watchpoint (ADDR, LEN, TYPE)' 633 `target_remove_watchpoint (ADDR, LEN, TYPE)' 634 Insert or remove a hardware watchpoint starting at ADDR, for LEN 635 bytes. TYPE is the watchpoint type, one of the possible values of 636 the enumerated data type `target_hw_bp_type', defined by 637 `breakpoint.h' as follows: 638 639 enum target_hw_bp_type 640 { 641 hw_write = 0, /* Common (write) HW watchpoint */ 642 hw_read = 1, /* Read HW watchpoint */ 643 hw_access = 2, /* Access (read or write) HW watchpoint */ 644 hw_execute = 3 /* Execute HW breakpoint */ 645 }; 646 647 These two macros should return 0 for success, non-zero for failure. 648 649 `target_stopped_data_address (ADDR_P)' 650 If the inferior has some watchpoint that triggered, place the 651 address associated with the watchpoint at the location pointed to 652 by ADDR_P and return non-zero. Otherwise, return zero. This is 653 required for data-read and data-access watchpoints. It is not 654 required for data-write watchpoints, but GDB uses it to improve 655 handling of those also. 656 657 GDB will only call this method once per watchpoint stop, 658 immediately after calling `STOPPED_BY_WATCHPOINT'. If the 659 target's watchpoint indication is sticky, i.e., stays set after 660 resuming, this method should clear it. For instance, the x86 debug 661 control register has sticky triggered flags. 662 663 `target_watchpoint_addr_within_range (TARGET, ADDR, START, LENGTH)' 664 Check whether ADDR (as returned by `target_stopped_data_address') 665 lies within the hardware-defined watchpoint region described by 666 START and LENGTH. This only needs to be provided if the 667 granularity of a watchpoint is greater than one byte, i.e., if the 668 watchpoint can also trigger on nearby addresses outside of the 669 watched region. 670 671 `HAVE_STEPPABLE_WATCHPOINT' 672 If defined to a non-zero value, it is not necessary to disable a 673 watchpoint to step over it. Like 674 `gdbarch_have_nonsteppable_watchpoint', this is usually set when 675 watchpoints trigger at the instruction which will perform an 676 interesting read or write. It should be set if there is a 677 temporary disable bit which allows the processor to step over the 678 interesting instruction without raising the watchpoint exception 679 again. 680 681 `int gdbarch_have_nonsteppable_watchpoint (GDBARCH)' 682 If it returns a non-zero value, GDB should disable a watchpoint to 683 step the inferior over it. This is usually set when watchpoints 684 trigger at the instruction which will perform an interesting read 685 or write. 686 687 `HAVE_CONTINUABLE_WATCHPOINT' 688 If defined to a non-zero value, it is possible to continue the 689 inferior after a watchpoint has been hit. This is usually set 690 when watchpoints trigger at the instruction following an 691 interesting read or write. 692 693 `CANNOT_STEP_HW_WATCHPOINTS' 694 If this is defined to a non-zero value, GDB will remove all 695 watchpoints before stepping the inferior. 696 697 `STOPPED_BY_WATCHPOINT (WAIT_STATUS)' 698 Return non-zero if stopped by a watchpoint. WAIT_STATUS is of the 699 type `struct target_waitstatus', defined by `target.h'. Normally, 700 this macro is defined to invoke the function pointed to by the 701 `to_stopped_by_watchpoint' member of the structure (of the type 702 `target_ops', defined on `target.h') that describes the 703 target-specific operations; `to_stopped_by_watchpoint' ignores the 704 WAIT_STATUS argument. 705 706 GDB does not require the non-zero value returned by 707 `STOPPED_BY_WATCHPOINT' to be 100% correct, so if a target cannot 708 determine for sure whether the inferior stopped due to a 709 watchpoint, it could return non-zero "just in case". 710 711 3.8.1 Watchpoints and Threads 712 ----------------------------- 713 714 GDB only supports process-wide watchpoints, which trigger in all 715 threads. GDB uses the thread ID to make watchpoints act as if they 716 were thread-specific, but it cannot set hardware watchpoints that only 717 trigger in a specific thread. Therefore, even if the target supports 718 threads, per-thread debug registers, and watchpoints which only affect 719 a single thread, it should set the per-thread debug registers for all 720 threads to the same value. On GNU/Linux native targets, this is 721 accomplished by using `ALL_LWPS' in `target_insert_watchpoint' and 722 `target_remove_watchpoint' and by using `linux_set_new_thread' to 723 register a handler for newly created threads. 724 725 GDB's GNU/Linux support only reports a single event at a time, 726 although multiple events can trigger simultaneously for multi-threaded 727 programs. When multiple events occur, `linux-nat.c' queues subsequent 728 events and returns them the next time the program is resumed. This 729 means that `STOPPED_BY_WATCHPOINT' and `target_stopped_data_address' 730 only need to consult the current thread's state--the thread indicated 731 by `inferior_ptid'. If two threads have hit watchpoints 732 simultaneously, those routines will be called a second time for the 733 second thread. 734 735 3.8.2 x86 Watchpoints 736 --------------------- 737 738 The 32-bit Intel x86 (a.k.a. ia32) processors feature special debug 739 registers designed to facilitate debugging. GDB provides a generic 740 library of functions that x86-based ports can use to implement support 741 for watchpoints and hardware-assisted breakpoints. This subsection 742 documents the x86 watchpoint facilities in GDB. 743 744 (At present, the library functions read and write debug registers 745 directly, and are thus only available for native configurations.) 746 747 To use the generic x86 watchpoint support, a port should do the 748 following: 749 750 * Define the macro `I386_USE_GENERIC_WATCHPOINTS' somewhere in the 751 target-dependent headers. 752 753 * Include the `config/i386/nm-i386.h' header file _after_ defining 754 `I386_USE_GENERIC_WATCHPOINTS'. 755 756 * Add `i386-nat.o' to the value of the Make variable `NATDEPFILES' 757 (*note NATDEPFILES: Native Debugging.). 758 759 * Provide implementations for the `I386_DR_LOW_*' macros described 760 below. Typically, each macro should call a target-specific 761 function which does the real work. 762 763 The x86 watchpoint support works by maintaining mirror images of the 764 debug registers. Values are copied between the mirror images and the 765 real debug registers via a set of macros which each target needs to 766 provide: 767 768 `I386_DR_LOW_SET_CONTROL (VAL)' 769 Set the Debug Control (DR7) register to the value VAL. 770 771 `I386_DR_LOW_SET_ADDR (IDX, ADDR)' 772 Put the address ADDR into the debug register number IDX. 773 774 `I386_DR_LOW_RESET_ADDR (IDX)' 775 Reset (i.e. zero out) the address stored in the debug register 776 number IDX. 777 778 `I386_DR_LOW_GET_STATUS' 779 Return the value of the Debug Status (DR6) register. This value is 780 used immediately after it is returned by `I386_DR_LOW_GET_STATUS', 781 so as to support per-thread status register values. 782 783 For each one of the 4 debug registers (whose indices are from 0 to 3) 784 that store addresses, a reference count is maintained by GDB, to allow 785 sharing of debug registers by several watchpoints. This allows users 786 to define several watchpoints that watch the same expression, but with 787 different conditions and/or commands, without wasting debug registers 788 which are in short supply. GDB maintains the reference counts 789 internally, targets don't have to do anything to use this feature. 790 791 The x86 debug registers can each watch a region that is 1, 2, or 4 792 bytes long. The ia32 architecture requires that each watched region be 793 appropriately aligned: 2-byte region on 2-byte boundary, 4-byte region 794 on 4-byte boundary. However, the x86 watchpoint support in GDB can 795 watch unaligned regions and regions larger than 4 bytes (up to 16 796 bytes) by allocating several debug registers to watch a single region. 797 This allocation of several registers per a watched region is also done 798 automatically without target code intervention. 799 800 The generic x86 watchpoint support provides the following API for the 801 GDB's application code: 802 803 `i386_region_ok_for_watchpoint (ADDR, LEN)' 804 The macro `TARGET_REGION_OK_FOR_HW_WATCHPOINT' is set to call this 805 function. It counts the number of debug registers required to 806 watch a given region, and returns a non-zero value if that number 807 is less than 4, the number of debug registers available to x86 808 processors. 809 810 `i386_stopped_data_address (ADDR_P)' 811 The target function `target_stopped_data_address' is set to call 812 this function. This function examines the breakpoint condition 813 bits in the DR6 Debug Status register, as returned by the 814 `I386_DR_LOW_GET_STATUS' macro, and returns the address associated 815 with the first bit that is set in DR6. 816 817 `i386_stopped_by_watchpoint (void)' 818 The macro `STOPPED_BY_WATCHPOINT' is set to call this function. 819 The argument passed to `STOPPED_BY_WATCHPOINT' is ignored. This 820 function examines the breakpoint condition bits in the DR6 Debug 821 Status register, as returned by the `I386_DR_LOW_GET_STATUS' 822 macro, and returns true if any bit is set. Otherwise, false is 823 returned. 824 825 `i386_insert_watchpoint (ADDR, LEN, TYPE)' 826 `i386_remove_watchpoint (ADDR, LEN, TYPE)' 827 Insert or remove a watchpoint. The macros 828 `target_insert_watchpoint' and `target_remove_watchpoint' are set 829 to call these functions. `i386_insert_watchpoint' first looks for 830 a debug register which is already set to watch the same region for 831 the same access types; if found, it just increments the reference 832 count of that debug register, thus implementing debug register 833 sharing between watchpoints. If no such register is found, the 834 function looks for a vacant debug register, sets its mirrored 835 value to ADDR, sets the mirrored value of DR7 Debug Control 836 register as appropriate for the LEN and TYPE parameters, and then 837 passes the new values of the debug register and DR7 to the 838 inferior by calling `I386_DR_LOW_SET_ADDR' and 839 `I386_DR_LOW_SET_CONTROL'. If more than one debug register is 840 required to cover the given region, the above process is repeated 841 for each debug register. 842 843 `i386_remove_watchpoint' does the opposite: it resets the address 844 in the mirrored value of the debug register and its read/write and 845 length bits in the mirrored value of DR7, then passes these new 846 values to the inferior via `I386_DR_LOW_RESET_ADDR' and 847 `I386_DR_LOW_SET_CONTROL'. If a register is shared by several 848 watchpoints, each time a `i386_remove_watchpoint' is called, it 849 decrements the reference count, and only calls 850 `I386_DR_LOW_RESET_ADDR' and `I386_DR_LOW_SET_CONTROL' when the 851 count goes to zero. 852 853 `i386_insert_hw_breakpoint (BP_TGT)' 854 `i386_remove_hw_breakpoint (BP_TGT)' 855 These functions insert and remove hardware-assisted breakpoints. 856 The macros `target_insert_hw_breakpoint' and 857 `target_remove_hw_breakpoint' are set to call these functions. 858 The argument is a `struct bp_target_info *', as described in the 859 documentation for `target_insert_breakpoint'. These functions 860 work like `i386_insert_watchpoint' and `i386_remove_watchpoint', 861 respectively, except that they set up the debug registers to watch 862 instruction execution, and each hardware-assisted breakpoint 863 always requires exactly one debug register. 864 865 `i386_cleanup_dregs (void)' 866 This function clears all the reference counts, addresses, and 867 control bits in the mirror images of the debug registers. It 868 doesn't affect the actual debug registers in the inferior process. 869 870 *Notes:* 871 1. x86 processors support setting watchpoints on I/O reads or writes. 872 However, since no target supports this (as of March 2001), and 873 since `enum target_hw_bp_type' doesn't even have an enumeration 874 for I/O watchpoints, this feature is not yet available to GDB 875 running on x86. 876 877 2. x86 processors can enable watchpoints locally, for the current task 878 only, or globally, for all the tasks. For each debug register, 879 there's a bit in the DR7 Debug Control register that determines 880 whether the associated address is watched locally or globally. The 881 current implementation of x86 watchpoint support in GDB always 882 sets watchpoints to be locally enabled, since global watchpoints 883 might interfere with the underlying OS and are probably 884 unavailable in many platforms. 885 886 3.9 Checkpoints 887 =============== 888 889 In the abstract, a checkpoint is a point in the execution history of 890 the program, which the user may wish to return to at some later time. 891 892 Internally, a checkpoint is a saved copy of the program state, 893 including whatever information is required in order to restore the 894 program to that state at a later time. This can be expected to include 895 the state of registers and memory, and may include external state such 896 as the state of open files and devices. 897 898 There are a number of ways in which checkpoints may be implemented 899 in gdb, e.g. as corefiles, as forked processes, and as some opaque 900 method implemented on the target side. 901 902 A corefile can be used to save an image of target memory and register 903 state, which can in principle be restored later -- but corefiles do not 904 typically include information about external entities such as open 905 files. Currently this method is not implemented in gdb. 906 907 A forked process can save the state of user memory and registers, as 908 well as some subset of external (kernel) state. This method is used to 909 implement checkpoints on Linux, and in principle might be used on other 910 systems. 911 912 Some targets, e.g. simulators, might have their own built-in method 913 for saving checkpoints, and gdb might be able to take advantage of that 914 capability without necessarily knowing any details of how it is done. 915 916 3.10 Observing changes in GDB internals 917 ======================================= 918 919 In order to function properly, several modules need to be notified when 920 some changes occur in the GDB internals. Traditionally, these modules 921 have relied on several paradigms, the most common ones being hooks and 922 gdb-events. Unfortunately, none of these paradigms was versatile 923 enough to become the standard notification mechanism in GDB. The fact 924 that they only supported one "client" was also a strong limitation. 925 926 A new paradigm, based on the Observer pattern of the `Design 927 Patterns' book, has therefore been implemented. The goal was to provide 928 a new interface overcoming the issues with the notification mechanisms 929 previously available. This new interface needed to be strongly typed, 930 easy to extend, and versatile enough to be used as the standard 931 interface when adding new notifications. 932 933 See *note GDB Observers:: for a brief description of the observers 934 currently implemented in GDB. The rationale for the current 935 implementation is also briefly discussed. 936 937 938 File: gdbint.info, Node: User Interface, Next: libgdb, Prev: Algorithms, Up: Top 939 940 4 User Interface 941 **************** 942 943 GDB has several user interfaces, of which the traditional command-line 944 interface is perhaps the most familiar. 945 946 4.1 Command Interpreter 947 ======================= 948 949 The command interpreter in GDB is fairly simple. It is designed to 950 allow for the set of commands to be augmented dynamically, and also has 951 a recursive subcommand capability, where the first argument to a 952 command may itself direct a lookup on a different command list. 953 954 For instance, the `set' command just starts a lookup on the 955 `setlist' command list, while `set thread' recurses to the 956 `set_thread_cmd_list'. 957 958 To add commands in general, use `add_cmd'. `add_com' adds to the 959 main command list, and should be used for those commands. The usual 960 place to add commands is in the `_initialize_XYZ' routines at the ends 961 of most source files. 962 963 To add paired `set' and `show' commands, use `add_setshow_cmd' or 964 `add_setshow_cmd_full'. The former is a slightly simpler interface 965 which is useful when you don't need to further modify the new command 966 structures, while the latter returns the new command structures for 967 manipulation. 968 969 Before removing commands from the command set it is a good idea to 970 deprecate them for some time. Use `deprecate_cmd' on commands or 971 aliases to set the deprecated flag. `deprecate_cmd' takes a `struct 972 cmd_list_element' as it's first argument. You can use the return value 973 from `add_com' or `add_cmd' to deprecate the command immediately after 974 it is created. 975 976 The first time a command is used the user will be warned and offered 977 a replacement (if one exists). Note that the replacement string passed 978 to `deprecate_cmd' should be the full name of the command, i.e., the 979 entire string the user should type at the command line. 980 981 4.2 UI-Independent Output--the `ui_out' Functions 982 ================================================= 983 984 The `ui_out' functions present an abstraction level for the GDB output 985 code. They hide the specifics of different user interfaces supported 986 by GDB, and thus free the programmer from the need to write several 987 versions of the same code, one each for every UI, to produce output. 988 989 4.2.1 Overview and Terminology 990 ------------------------------ 991 992 In general, execution of each GDB command produces some sort of output, 993 and can even generate an input request. 994 995 Output can be generated for the following purposes: 996 997 * to display a _result_ of an operation; 998 999 * to convey _info_ or produce side-effects of a requested operation; 1000 1001 * to provide a _notification_ of an asynchronous event (including 1002 progress indication of a prolonged asynchronous operation); 1003 1004 * to display _error messages_ (including warnings); 1005 1006 * to show _debug data_; 1007 1008 * to _query_ or prompt a user for input (a special case). 1009 1010 This section mainly concentrates on how to build result output, 1011 although some of it also applies to other kinds of output. 1012 1013 Generation of output that displays the results of an operation 1014 involves one or more of the following: 1015 1016 * output of the actual data 1017 1018 * formatting the output as appropriate for console output, to make it 1019 easily readable by humans 1020 1021 * machine oriented formatting-a more terse formatting to allow for 1022 easy parsing by programs which read GDB's output 1023 1024 * annotation, whose purpose is to help legacy GUIs to identify 1025 interesting parts in the output 1026 1027 The `ui_out' routines take care of the first three aspects. 1028 Annotations are provided by separate annotation routines. Note that use 1029 of annotations for an interface between a GUI and GDB is deprecated. 1030 1031 Output can be in the form of a single item, which we call a "field"; 1032 a "list" consisting of identical fields; a "tuple" consisting of 1033 non-identical fields; or a "table", which is a tuple consisting of a 1034 header and a body. In a BNF-like form: 1035 1036 `<table> ==>' 1037 `<header> <body>' 1038 1039 `<header> ==>' 1040 `{ <column> }' 1041 1042 `<column> ==>' 1043 `<width> <alignment> <title>' 1044 1045 `<body> ==>' 1046 `{<row>}' 1047 1048 4.2.2 General Conventions 1049 ------------------------- 1050 1051 Most `ui_out' routines are of type `void', the exceptions are 1052 `ui_out_stream_new' (which returns a pointer to the newly created 1053 object) and the `make_cleanup' routines. 1054 1055 The first parameter is always the `ui_out' vector object, a pointer 1056 to a `struct ui_out'. 1057 1058 The FORMAT parameter is like in `printf' family of functions. When 1059 it is present, there must also be a variable list of arguments 1060 sufficient used to satisfy the `%' specifiers in the supplied format. 1061 1062 When a character string argument is not used in a `ui_out' function 1063 call, a `NULL' pointer has to be supplied instead. 1064 1065 4.2.3 Table, Tuple and List Functions 1066 ------------------------------------- 1067 1068 This section introduces `ui_out' routines for building lists, tuples 1069 and tables. The routines to output the actual data items (fields) are 1070 presented in the next section. 1071 1072 To recap: A "tuple" is a sequence of "fields", each field containing 1073 information about an object; a "list" is a sequence of fields where 1074 each field describes an identical object. 1075 1076 Use the "table" functions when your output consists of a list of 1077 rows (tuples) and the console output should include a heading. Use this 1078 even when you are listing just one object but you still want the header. 1079 1080 Tables can not be nested. Tuples and lists can be nested up to a 1081 maximum of five levels. 1082 1083 The overall structure of the table output code is something like 1084 this: 1085 1086 ui_out_table_begin 1087 ui_out_table_header 1088 ... 1089 ui_out_table_body 1090 ui_out_tuple_begin 1091 ui_out_field_* 1092 ... 1093 ui_out_tuple_end 1094 ... 1095 ui_out_table_end 1096 1097 Here is the description of table-, tuple- and list-related `ui_out' 1098 functions: 1099 1100 -- Function: void ui_out_table_begin (struct ui_out *UIOUT, int 1101 NBROFCOLS, int NR_ROWS, const char *TBLID) 1102 The function `ui_out_table_begin' marks the beginning of the output 1103 of a table. It should always be called before any other `ui_out' 1104 function for a given table. NBROFCOLS is the number of columns in 1105 the table. NR_ROWS is the number of rows in the table. TBLID is 1106 an optional string identifying the table. The string pointed to 1107 by TBLID is copied by the implementation of `ui_out_table_begin', 1108 so the application can free the string if it was `malloc'ed. 1109 1110 The companion function `ui_out_table_end', described below, marks 1111 the end of the table's output. 1112 1113 -- Function: void ui_out_table_header (struct ui_out *UIOUT, int 1114 WIDTH, enum ui_align ALIGNMENT, const char *COLHDR) 1115 `ui_out_table_header' provides the header information for a single 1116 table column. You call this function several times, one each for 1117 every column of the table, after `ui_out_table_begin', but before 1118 `ui_out_table_body'. 1119 1120 The value of WIDTH gives the column width in characters. The 1121 value of ALIGNMENT is one of `left', `center', and `right', and it 1122 specifies how to align the header: left-justify, center, or 1123 right-justify it. COLHDR points to a string that specifies the 1124 column header; the implementation copies that string, so column 1125 header strings in `malloc'ed storage can be freed after the call. 1126 1127 -- Function: void ui_out_table_body (struct ui_out *UIOUT) 1128 This function delimits the table header from the table body. 1129 1130 -- Function: void ui_out_table_end (struct ui_out *UIOUT) 1131 This function signals the end of a table's output. It should be 1132 called after the table body has been produced by the list and 1133 field output functions. 1134 1135 There should be exactly one call to `ui_out_table_end' for each 1136 call to `ui_out_table_begin', otherwise the `ui_out' functions 1137 will signal an internal error. 1138 1139 The output of the tuples that represent the table rows must follow 1140 the call to `ui_out_table_body' and precede the call to 1141 `ui_out_table_end'. You build a tuple by calling `ui_out_tuple_begin' 1142 and `ui_out_tuple_end', with suitable calls to functions which actually 1143 output fields between them. 1144 1145 -- Function: void ui_out_tuple_begin (struct ui_out *UIOUT, const char 1146 *ID) 1147 This function marks the beginning of a tuple output. ID points to 1148 an optional string that identifies the tuple; it is copied by the 1149 implementation, and so strings in `malloc'ed storage can be freed 1150 after the call. 1151 1152 -- Function: void ui_out_tuple_end (struct ui_out *UIOUT) 1153 This function signals an end of a tuple output. There should be 1154 exactly one call to `ui_out_tuple_end' for each call to 1155 `ui_out_tuple_begin', otherwise an internal GDB error will be 1156 signaled. 1157 1158 -- Function: struct cleanup * make_cleanup_ui_out_tuple_begin_end 1159 (struct ui_out *UIOUT, const char *ID) 1160 This function first opens the tuple and then establishes a cleanup 1161 (*note Cleanups: Coding.) to close the tuple. It provides a 1162 convenient and correct implementation of the non-portable(1) code 1163 sequence: 1164 struct cleanup *old_cleanup; 1165 ui_out_tuple_begin (uiout, "..."); 1166 old_cleanup = make_cleanup ((void(*)(void *)) ui_out_tuple_end, 1167 uiout); 1168 1169 -- Function: void ui_out_list_begin (struct ui_out *UIOUT, const char 1170 *ID) 1171 This function marks the beginning of a list output. ID points to 1172 an optional string that identifies the list; it is copied by the 1173 implementation, and so strings in `malloc'ed storage can be freed 1174 after the call. 1175 1176 -- Function: void ui_out_list_end (struct ui_out *UIOUT) 1177 This function signals an end of a list output. There should be 1178 exactly one call to `ui_out_list_end' for each call to 1179 `ui_out_list_begin', otherwise an internal GDB error will be 1180 signaled. 1181 1182 -- Function: struct cleanup * make_cleanup_ui_out_list_begin_end 1183 (struct ui_out *UIOUT, const char *ID) 1184 Similar to `make_cleanup_ui_out_tuple_begin_end', this function 1185 opens a list and then establishes cleanup (*note Cleanups: Coding.) 1186 that will close the list. 1187 1188 4.2.4 Item Output Functions 1189 --------------------------- 1190 1191 The functions described below produce output for the actual data items, 1192 or fields, which contain information about the object. 1193 1194 Choose the appropriate function accordingly to your particular needs. 1195 1196 -- Function: void ui_out_field_fmt (struct ui_out *UIOUT, char 1197 *FLDNAME, char *FORMAT, ...) 1198 This is the most general output function. It produces the 1199 representation of the data in the variable-length argument list 1200 according to formatting specifications in FORMAT, a `printf'-like 1201 format string. The optional argument FLDNAME supplies the name of 1202 the field. The data items themselves are supplied as additional 1203 arguments after FORMAT. 1204 1205 This generic function should be used only when it is not possible 1206 to use one of the specialized versions (see below). 1207 1208 -- Function: void ui_out_field_int (struct ui_out *UIOUT, const char 1209 *FLDNAME, int VALUE) 1210 This function outputs a value of an `int' variable. It uses the 1211 `"%d"' output conversion specification. FLDNAME specifies the 1212 name of the field. 1213 1214 -- Function: void ui_out_field_fmt_int (struct ui_out *UIOUT, int 1215 WIDTH, enum ui_align ALIGNMENT, const char *FLDNAME, int 1216 VALUE) 1217 This function outputs a value of an `int' variable. It differs 1218 from `ui_out_field_int' in that the caller specifies the desired 1219 WIDTH and ALIGNMENT of the output. FLDNAME specifies the name of 1220 the field. 1221 1222 -- Function: void ui_out_field_core_addr (struct ui_out *UIOUT, const 1223 char *FLDNAME, struct gdbarch *GDBARCH, CORE_ADDR ADDRESS) 1224 This function outputs an address as appropriate for GDBARCH. 1225 1226 -- Function: void ui_out_field_string (struct ui_out *UIOUT, const 1227 char *FLDNAME, const char *STRING) 1228 This function outputs a string using the `"%s"' conversion 1229 specification. 1230 1231 Sometimes, there's a need to compose your output piece by piece using 1232 functions that operate on a stream, such as `value_print' or 1233 `fprintf_symbol_filtered'. These functions accept an argument of the 1234 type `struct ui_file *', a pointer to a `ui_file' object used to store 1235 the data stream used for the output. When you use one of these 1236 functions, you need a way to pass their results stored in a `ui_file' 1237 object to the `ui_out' functions. To this end, you first create a 1238 `ui_stream' object by calling `ui_out_stream_new', pass the `stream' 1239 member of that `ui_stream' object to `value_print' and similar 1240 functions, and finally call `ui_out_field_stream' to output the field 1241 you constructed. When the `ui_stream' object is no longer needed, you 1242 should destroy it and free its memory by calling `ui_out_stream_delete'. 1243 1244 -- Function: struct ui_stream * ui_out_stream_new (struct ui_out 1245 *UIOUT) 1246 This function creates a new `ui_stream' object which uses the same 1247 output methods as the `ui_out' object whose pointer is passed in 1248 UIOUT. It returns a pointer to the newly created `ui_stream' 1249 object. 1250 1251 -- Function: void ui_out_stream_delete (struct ui_stream *STREAMBUF) 1252 This functions destroys a `ui_stream' object specified by 1253 STREAMBUF. 1254 1255 -- Function: void ui_out_field_stream (struct ui_out *UIOUT, const 1256 char *FIELDNAME, struct ui_stream *STREAMBUF) 1257 This function consumes all the data accumulated in 1258 `streambuf->stream' and outputs it like `ui_out_field_string' 1259 does. After a call to `ui_out_field_stream', the accumulated data 1260 no longer exists, but the stream is still valid and may be used 1261 for producing more fields. 1262 1263 *Important:* If there is any chance that your code could bail out 1264 before completing output generation and reaching the point where 1265 `ui_out_stream_delete' is called, it is necessary to set up a cleanup, 1266 to avoid leaking memory and other resources. Here's a skeleton code to 1267 do that: 1268 1269 struct ui_stream *mybuf = ui_out_stream_new (uiout); 1270 struct cleanup *old = make_cleanup (ui_out_stream_delete, mybuf); 1271 ... 1272 do_cleanups (old); 1273 1274 If the function already has the old cleanup chain set (for other 1275 kinds of cleanups), you just have to add your cleanup to it: 1276 1277 mybuf = ui_out_stream_new (uiout); 1278 make_cleanup (ui_out_stream_delete, mybuf); 1279 1280 Note that with cleanups in place, you should not call 1281 `ui_out_stream_delete' directly, or you would attempt to free the same 1282 buffer twice. 1283 1284 4.2.5 Utility Output Functions 1285 ------------------------------ 1286 1287 -- Function: void ui_out_field_skip (struct ui_out *UIOUT, const char 1288 *FLDNAME) 1289 This function skips a field in a table. Use it if you have to 1290 leave an empty field without disrupting the table alignment. The 1291 argument FLDNAME specifies a name for the (missing) filed. 1292 1293 -- Function: void ui_out_text (struct ui_out *UIOUT, const char 1294 *STRING) 1295 This function outputs the text in STRING in a way that makes it 1296 easy to be read by humans. For example, the console 1297 implementation of this method filters the text through a built-in 1298 pager, to prevent it from scrolling off the visible portion of the 1299 screen. 1300 1301 Use this function for printing relatively long chunks of text 1302 around the actual field data: the text it produces is not aligned 1303 according to the table's format. Use `ui_out_field_string' to 1304 output a string field, and use `ui_out_message', described below, 1305 to output short messages. 1306 1307 -- Function: void ui_out_spaces (struct ui_out *UIOUT, int NSPACES) 1308 This function outputs NSPACES spaces. It is handy to align the 1309 text produced by `ui_out_text' with the rest of the table or list. 1310 1311 -- Function: void ui_out_message (struct ui_out *UIOUT, int VERBOSITY, 1312 const char *FORMAT, ...) 1313 This function produces a formatted message, provided that the 1314 current verbosity level is at least as large as given by 1315 VERBOSITY. The current verbosity level is specified by the user 1316 with the `set verbositylevel' command.(2) 1317 1318 -- Function: void ui_out_wrap_hint (struct ui_out *UIOUT, char *INDENT) 1319 This function gives the console output filter (a paging filter) a 1320 hint of where to break lines which are too long. Ignored for all 1321 other output consumers. INDENT, if non-`NULL', is the string to 1322 be printed to indent the wrapped text on the next line; it must 1323 remain accessible until the next call to `ui_out_wrap_hint', or 1324 until an explicit newline is produced by one of the other 1325 functions. If INDENT is `NULL', the wrapped text will not be 1326 indented. 1327 1328 -- Function: void ui_out_flush (struct ui_out *UIOUT) 1329 This function flushes whatever output has been accumulated so far, 1330 if the UI buffers output. 1331 1332 4.2.6 Examples of Use of `ui_out' functions 1333 ------------------------------------------- 1334 1335 This section gives some practical examples of using the `ui_out' 1336 functions to generalize the old console-oriented code in GDB. The 1337 examples all come from functions defined on the `breakpoints.c' file. 1338 1339 This example, from the `breakpoint_1' function, shows how to produce 1340 a table. 1341 1342 The original code was: 1343 1344 if (!found_a_breakpoint++) 1345 { 1346 annotate_breakpoints_headers (); 1347 1348 annotate_field (0); 1349 printf_filtered ("Num "); 1350 annotate_field (1); 1351 printf_filtered ("Type "); 1352 annotate_field (2); 1353 printf_filtered ("Disp "); 1354 annotate_field (3); 1355 printf_filtered ("Enb "); 1356 if (addressprint) 1357 { 1358 annotate_field (4); 1359 printf_filtered ("Address "); 1360 } 1361 annotate_field (5); 1362 printf_filtered ("What\n"); 1363 1364 annotate_breakpoints_table (); 1365 } 1366 1367 Here's the new version: 1368 1369 nr_printable_breakpoints = ...; 1370 1371 if (addressprint) 1372 ui_out_table_begin (ui, 6, nr_printable_breakpoints, "BreakpointTable"); 1373 else 1374 ui_out_table_begin (ui, 5, nr_printable_breakpoints, "BreakpointTable"); 1375 1376 if (nr_printable_breakpoints > 0) 1377 annotate_breakpoints_headers (); 1378 if (nr_printable_breakpoints > 0) 1379 annotate_field (0); 1380 ui_out_table_header (uiout, 3, ui_left, "number", "Num"); /* 1 */ 1381 if (nr_printable_breakpoints > 0) 1382 annotate_field (1); 1383 ui_out_table_header (uiout, 14, ui_left, "type", "Type"); /* 2 */ 1384 if (nr_printable_breakpoints > 0) 1385 annotate_field (2); 1386 ui_out_table_header (uiout, 4, ui_left, "disp", "Disp"); /* 3 */ 1387 if (nr_printable_breakpoints > 0) 1388 annotate_field (3); 1389 ui_out_table_header (uiout, 3, ui_left, "enabled", "Enb"); /* 4 */ 1390 if (addressprint) 1391 { 1392 if (nr_printable_breakpoints > 0) 1393 annotate_field (4); 1394 if (print_address_bits <= 32) 1395 ui_out_table_header (uiout, 10, ui_left, "addr", "Address");/* 5 */ 1396 else 1397 ui_out_table_header (uiout, 18, ui_left, "addr", "Address");/* 5 */ 1398 } 1399 if (nr_printable_breakpoints > 0) 1400 annotate_field (5); 1401 ui_out_table_header (uiout, 40, ui_noalign, "what", "What"); /* 6 */ 1402 ui_out_table_body (uiout); 1403 if (nr_printable_breakpoints > 0) 1404 annotate_breakpoints_table (); 1405 1406 This example, from the `print_one_breakpoint' function, shows how to 1407 produce the actual data for the table whose structure was defined in 1408 the above example. The original code was: 1409 1410 annotate_record (); 1411 annotate_field (0); 1412 printf_filtered ("%-3d ", b->number); 1413 annotate_field (1); 1414 if ((int)b->type > (sizeof(bptypes)/sizeof(bptypes[0])) 1415 || ((int) b->type != bptypes[(int) b->type].type)) 1416 internal_error ("bptypes table does not describe type #%d.", 1417 (int)b->type); 1418 printf_filtered ("%-14s ", bptypes[(int)b->type].description); 1419 annotate_field (2); 1420 printf_filtered ("%-4s ", bpdisps[(int)b->disposition]); 1421 annotate_field (3); 1422 printf_filtered ("%-3c ", bpenables[(int)b->enable]); 1423 ... 1424 1425 This is the new version: 1426 1427 annotate_record (); 1428 ui_out_tuple_begin (uiout, "bkpt"); 1429 annotate_field (0); 1430 ui_out_field_int (uiout, "number", b->number); 1431 annotate_field (1); 1432 if (((int) b->type > (sizeof (bptypes) / sizeof (bptypes[0]))) 1433 || ((int) b->type != bptypes[(int) b->type].type)) 1434 internal_error ("bptypes table does not describe type #%d.", 1435 (int) b->type); 1436 ui_out_field_string (uiout, "type", bptypes[(int)b->type].description); 1437 annotate_field (2); 1438 ui_out_field_string (uiout, "disp", bpdisps[(int)b->disposition]); 1439 annotate_field (3); 1440 ui_out_field_fmt (uiout, "enabled", "%c", bpenables[(int)b->enable]); 1441 ... 1442 1443 This example, also from `print_one_breakpoint', shows how to produce 1444 a complicated output field using the `print_expression' functions which 1445 requires a stream to be passed. It also shows how to automate stream 1446 destruction with cleanups. The original code was: 1447 1448 annotate_field (5); 1449 print_expression (b->exp, gdb_stdout); 1450 1451 The new version is: 1452 1453 struct ui_stream *stb = ui_out_stream_new (uiout); 1454 struct cleanup *old_chain = make_cleanup_ui_out_stream_delete (stb); 1455 ... 1456 annotate_field (5); 1457 print_expression (b->exp, stb->stream); 1458 ui_out_field_stream (uiout, "what", local_stream); 1459 1460 This example, also from `print_one_breakpoint', shows how to use 1461 `ui_out_text' and `ui_out_field_string'. The original code was: 1462 1463 annotate_field (5); 1464 if (b->dll_pathname == NULL) 1465 printf_filtered ("<any library> "); 1466 else 1467 printf_filtered ("library \"%s\" ", b->dll_pathname); 1468 1469 It became: 1470 1471 annotate_field (5); 1472 if (b->dll_pathname == NULL) 1473 { 1474 ui_out_field_string (uiout, "what", "<any library>"); 1475 ui_out_spaces (uiout, 1); 1476 } 1477 else 1478 { 1479 ui_out_text (uiout, "library \""); 1480 ui_out_field_string (uiout, "what", b->dll_pathname); 1481 ui_out_text (uiout, "\" "); 1482 } 1483 1484 The following example from `print_one_breakpoint' shows how to use 1485 `ui_out_field_int' and `ui_out_spaces'. The original code was: 1486 1487 annotate_field (5); 1488 if (b->forked_inferior_pid != 0) 1489 printf_filtered ("process %d ", b->forked_inferior_pid); 1490 1491 It became: 1492 1493 annotate_field (5); 1494 if (b->forked_inferior_pid != 0) 1495 { 1496 ui_out_text (uiout, "process "); 1497 ui_out_field_int (uiout, "what", b->forked_inferior_pid); 1498 ui_out_spaces (uiout, 1); 1499 } 1500 1501 Here's an example of using `ui_out_field_string'. The original code 1502 was: 1503 1504 annotate_field (5); 1505 if (b->exec_pathname != NULL) 1506 printf_filtered ("program \"%s\" ", b->exec_pathname); 1507 1508 It became: 1509 1510 annotate_field (5); 1511 if (b->exec_pathname != NULL) 1512 { 1513 ui_out_text (uiout, "program \""); 1514 ui_out_field_string (uiout, "what", b->exec_pathname); 1515 ui_out_text (uiout, "\" "); 1516 } 1517 1518 Finally, here's an example of printing an address. The original 1519 code: 1520 1521 annotate_field (4); 1522 printf_filtered ("%s ", 1523 hex_string_custom ((unsigned long) b->address, 8)); 1524 1525 It became: 1526 1527 annotate_field (4); 1528 ui_out_field_core_addr (uiout, "Address", b->address); 1529 1530 4.3 Console Printing 1531 ==================== 1532 1533 4.4 TUI 1534 ======= 1535 1536 ---------- Footnotes ---------- 1537 1538 (1) The function cast is not portable ISO C. 1539 1540 (2) As of this writing (April 2001), setting verbosity level is not 1541 yet implemented, and is always returned as zero. So calling 1542 `ui_out_message' with a VERBOSITY argument more than zero will cause 1543 the message to never be printed. 1544 1545 1546 File: gdbint.info, Node: libgdb, Next: Values, Prev: User Interface, Up: Top 1547 1548 5 libgdb 1549 ******** 1550 1551 5.1 libgdb 1.0 1552 ============== 1553 1554 `libgdb' 1.0 was an abortive project of years ago. The theory was to 1555 provide an API to GDB's functionality. 1556 1557 5.2 libgdb 2.0 1558 ============== 1559 1560 `libgdb' 2.0 is an ongoing effort to update GDB so that is better able 1561 to support graphical and other environments. 1562 1563 Since `libgdb' development is on-going, its architecture is still 1564 evolving. The following components have so far been identified: 1565 1566 * Observer - `gdb-events.h'. 1567 1568 * Builder - `ui-out.h' 1569 1570 * Event Loop - `event-loop.h' 1571 1572 * Library - `gdb.h' 1573 1574 The model that ties these components together is described below. 1575 1576 5.3 The `libgdb' Model 1577 ====================== 1578 1579 A client of `libgdb' interacts with the library in two ways. 1580 1581 * As an observer (using `gdb-events') receiving notifications from 1582 `libgdb' of any internal state changes (break point changes, run 1583 state, etc). 1584 1585 * As a client querying `libgdb' (using the `ui-out' builder) to 1586 obtain various status values from GDB. 1587 1588 Since `libgdb' could have multiple clients (e.g., a GUI supporting 1589 the existing GDB CLI), those clients must co-operate when controlling 1590 `libgdb'. In particular, a client must ensure that `libgdb' is idle 1591 (i.e. no other client is using `libgdb') before responding to a 1592 `gdb-event' by making a query. 1593 1594 5.4 CLI support 1595 =============== 1596 1597 At present GDB's CLI is very much entangled in with the core of 1598 `libgdb'. Consequently, a client wishing to include the CLI in their 1599 interface needs to carefully co-ordinate its own and the CLI's 1600 requirements. 1601 1602 It is suggested that the client set `libgdb' up to be bi-modal 1603 (alternate between CLI and client query modes). The notes below sketch 1604 out the theory: 1605 1606 * The client registers itself as an observer of `libgdb'. 1607 1608 * The client create and install `cli-out' builder using its own 1609 versions of the `ui-file' `gdb_stderr', `gdb_stdtarg' and 1610 `gdb_stdout' streams. 1611 1612 * The client creates a separate custom `ui-out' builder that is only 1613 used while making direct queries to `libgdb'. 1614 1615 When the client receives input intended for the CLI, it simply 1616 passes it along. Since the `cli-out' builder is installed by default, 1617 all the CLI output in response to that command is routed (pronounced 1618 rooted) through to the client controlled `gdb_stdout' et. al. streams. 1619 At the same time, the client is kept abreast of internal changes by 1620 virtue of being a `libgdb' observer. 1621 1622 The only restriction on the client is that it must wait until 1623 `libgdb' becomes idle before initiating any queries (using the client's 1624 custom builder). 1625 1626 5.5 `libgdb' components 1627 ======================= 1628 1629 Observer - `gdb-events.h' 1630 ------------------------- 1631 1632 `gdb-events' provides the client with a very raw mechanism that can be 1633 used to implement an observer. At present it only allows for one 1634 observer and that observer must, internally, handle the need to delay 1635 the processing of any event notifications until after `libgdb' has 1636 finished the current command. 1637 1638 Builder - `ui-out.h' 1639 -------------------- 1640 1641 `ui-out' provides the infrastructure necessary for a client to create a 1642 builder. That builder is then passed down to `libgdb' when doing any 1643 queries. 1644 1645 Event Loop - `event-loop.h' 1646 --------------------------- 1647 1648 `event-loop', currently non-re-entrant, provides a simple event loop. 1649 A client would need to either plug its self into this loop or, 1650 implement a new event-loop that GDB would use. 1651 1652 The event-loop will eventually be made re-entrant. This is so that 1653 GDB can better handle the problem of some commands blocking instead of 1654 returning. 1655 1656 Library - `gdb.h' 1657 ----------------- 1658 1659 `libgdb' is the most obvious component of this system. It provides the 1660 query interface. Each function is parameterized by a `ui-out' builder. 1661 The result of the query is constructed using that builder before the 1662 query function returns. 1663 1664 1665 File: gdbint.info, Node: Values, Next: Stack Frames, Prev: libgdb, Up: Top 1666 1667 6 Values 1668 ******** 1669 1670 6.1 Values 1671 ========== 1672 1673 GDB uses `struct value', or "values", as an internal abstraction for 1674 the representation of a variety of inferior objects and GDB convenience 1675 objects. 1676 1677 Values have an associated `struct type', that describes a virtual 1678 view of the raw data or object stored in or accessed through the value. 1679 1680 A value is in addition discriminated by its lvalue-ness, given its 1681 `enum lval_type' enumeration type: 1682 1683 ``not_lval'' 1684 This value is not an lval. It can't be assigned to. 1685 1686 ``lval_memory'' 1687 This value represents an object in memory. 1688 1689 ``lval_register'' 1690 This value represents an object that lives in a register. 1691 1692 ``lval_internalvar'' 1693 Represents the value of an internal variable. 1694 1695 ``lval_internalvar_component'' 1696 Represents part of a GDB internal variable. E.g., a structure 1697 field. 1698 1699 ``lval_computed'' 1700 These are "computed" values. They allow creating specialized value 1701 objects for specific purposes, all abstracted away from the core 1702 value support code. The creator of such a value writes specialized 1703 functions to handle the reading and writing to/from the value's 1704 backend data, and optionally, a "copy operator" and a "destructor". 1705 1706 Pointers to these functions are stored in a `struct lval_funcs' 1707 instance (declared in `value.h'), and passed to the 1708 `allocate_computed_value' function, as in the example below. 1709 1710 static void 1711 nil_value_read (struct value *v) 1712 { 1713 /* This callback reads data from some backend, and stores it in V. 1714 In this case, we always read null data. You'll want to fill in 1715 something more interesting. */ 1716 1717 memset (value_contents_all_raw (v), 1718 value_offset (v), 1719 TYPE_LENGTH (value_type (v))); 1720 } 1721 1722 static void 1723 nil_value_write (struct value *v, struct value *fromval) 1724 { 1725 /* Takes the data from FROMVAL and stores it in the backend of V. */ 1726 1727 to_oblivion (value_contents_all_raw (fromval), 1728 value_offset (v), 1729 TYPE_LENGTH (value_type (fromval))); 1730 } 1731 1732 static struct lval_funcs nil_value_funcs = 1733 { 1734 nil_value_read, 1735 nil_value_write 1736 }; 1737 1738 struct value * 1739 make_nil_value (void) 1740 { 1741 struct type *type; 1742 struct value *v; 1743 1744 type = make_nils_type (); 1745 v = allocate_computed_value (type, &nil_value_funcs, NULL); 1746 1747 return v; 1748 } 1749 1750 See the implementation of the `$_siginfo' convenience variable in 1751 `infrun.c' as a real example use of lval_computed. 1752 1753 1754 1755 File: gdbint.info, Node: Stack Frames, Next: Symbol Handling, Prev: Values, Up: Top 1756 1757 7 Stack Frames 1758 ************** 1759 1760 A frame is a construct that GDB uses to keep track of calling and 1761 called functions. 1762 1763 GDB's frame model, a fresh design, was implemented with the need to 1764 support DWARF's Call Frame Information in mind. In fact, the term 1765 "unwind" is taken directly from that specification. Developers wishing 1766 to learn more about unwinders, are encouraged to read the DWARF 1767 specification, available from `http://www.dwarfstd.org'. 1768 1769 GDB's model is that you find a frame's registers by "unwinding" them 1770 from the next younger frame. That is, `get_frame_register' which 1771 returns the value of a register in frame #1 (the next-to-youngest 1772 frame), is implemented by calling frame #0's `frame_register_unwind' 1773 (the youngest frame). But then the obvious question is: how do you 1774 access the registers of the youngest frame itself? 1775 1776 To answer this question, GDB has the "sentinel" frame, the "-1st" 1777 frame. Unwinding registers from the sentinel frame gives you the 1778 current values of the youngest real frame's registers. If F is a 1779 sentinel frame, then `get_frame_type (F) == SENTINEL_FRAME'. 1780 1781 7.1 Selecting an Unwinder 1782 ========================= 1783 1784 The architecture registers a list of frame unwinders (`struct 1785 frame_unwind'), using the functions `frame_unwind_prepend_unwinder' and 1786 `frame_unwind_append_unwinder'. Each unwinder includes a sniffer. 1787 Whenever GDB needs to unwind a frame (to fetch the previous frame's 1788 registers or the current frame's ID), it calls registered sniffers in 1789 order to find one which recognizes the frame. The first time a sniffer 1790 returns non-zero, the corresponding unwinder is assigned to the frame. 1791 1792 7.2 Unwinding the Frame ID 1793 ========================== 1794 1795 Every frame has an associated ID, of type `struct frame_id'. The ID 1796 includes the stack base and function start address for the frame. The 1797 ID persists through the entire life of the frame, including while other 1798 called frames are running; it is used to locate an appropriate `struct 1799 frame_info' from the cache. 1800 1801 Every time the inferior stops, and at various other times, the frame 1802 cache is flushed. Because of this, parts of GDB which need to keep 1803 track of individual frames cannot use pointers to `struct frame_info'. 1804 A frame ID provides a stable reference to a frame, even when the 1805 unwinder must be run again to generate a new `struct frame_info' for 1806 the same frame. 1807 1808 The frame's unwinder's `this_id' method is called to find the ID. 1809 Note that this is different from register unwinding, where the next 1810 frame's `prev_register' is called to unwind this frame's registers. 1811 1812 Both stack base and function address are required to identify the 1813 frame, because a recursive function has the same function address for 1814 two consecutive frames and a leaf function may have the same stack 1815 address as its caller. On some platforms, a third address is part of 1816 the ID to further disambiguate frames--for instance, on IA-64 the 1817 separate register stack address is included in the ID. 1818 1819 An invalid frame ID (`outer_frame_id') returned from the `this_id' 1820 method means to stop unwinding after this frame. 1821 1822 `null_frame_id' is another invalid frame ID which should be used 1823 when there is no frame. For instance, certain breakpoints are attached 1824 to a specific frame, and that frame is identified through its frame ID 1825 (we use this to implement the "finish" command). Using `null_frame_id' 1826 as the frame ID for a given breakpoint means that the breakpoint is not 1827 specific to any frame. The `this_id' method should never return 1828 `null_frame_id'. 1829 1830 7.3 Unwinding Registers 1831 ======================= 1832 1833 Each unwinder includes a `prev_register' method. This method takes a 1834 frame, an associated cache pointer, and a register number. It returns 1835 a `struct value *' describing the requested register, as saved by this 1836 frame. This is the value of the register that is current in this 1837 frame's caller. 1838 1839 The returned value must have the same type as the register. It may 1840 have any lvalue type. In most circumstances one of these routines will 1841 generate the appropriate value: 1842 1843 `frame_unwind_got_optimized' 1844 This register was not saved. 1845 1846 `frame_unwind_got_register' 1847 This register was copied into another register in this frame. This 1848 is also used for unchanged registers; they are "copied" into the 1849 same register. 1850 1851 `frame_unwind_got_memory' 1852 This register was saved in memory. 1853 1854 `frame_unwind_got_constant' 1855 This register was not saved, but the unwinder can compute the 1856 previous value some other way. 1857 1858 `frame_unwind_got_address' 1859 Same as `frame_unwind_got_constant', except that the value is a 1860 target address. This is frequently used for the stack pointer, 1861 which is not explicitly saved but has a known offset from this 1862 frame's stack pointer. For architectures with a flat unified 1863 address space, this is generally the same as 1864 `frame_unwind_got_constant'. 1865 1866 1867 File: gdbint.info, Node: Symbol Handling, Next: Language Support, Prev: Stack Frames, Up: Top 1868 1869 8 Symbol Handling 1870 ***************** 1871 1872 Symbols are a key part of GDB's operation. Symbols include variables, 1873 functions, and types. 1874 1875 Symbol information for a large program can be truly massive, and 1876 reading of symbol information is one of the major performance 1877 bottlenecks in GDB; it can take many minutes to process it all. 1878 Studies have shown that nearly all the time spent is computational, 1879 rather than file reading. 1880 1881 One of the ways for GDB to provide a good user experience is to 1882 start up quickly, taking no more than a few seconds. It is simply not 1883 possible to process all of a program's debugging info in that time, and 1884 so we attempt to handle symbols incrementally. For instance, we create 1885 "partial symbol tables" consisting of only selected symbols, and only 1886 expand them to full symbol tables when necessary. 1887 1888 8.1 Symbol Reading 1889 ================== 1890 1891 GDB reads symbols from "symbol files". The usual symbol file is the 1892 file containing the program which GDB is debugging. GDB can be 1893 directed to use a different file for symbols (with the `symbol-file' 1894 command), and it can also read more symbols via the `add-file' and 1895 `load' commands. In addition, it may bring in more symbols while 1896 loading shared libraries. 1897 1898 Symbol files are initially opened by code in `symfile.c' using the 1899 BFD library (*note Support Libraries::). BFD identifies the type of 1900 the file by examining its header. `find_sym_fns' then uses this 1901 identification to locate a set of symbol-reading functions. 1902 1903 Symbol-reading modules identify themselves to GDB by calling 1904 `add_symtab_fns' during their module initialization. The argument to 1905 `add_symtab_fns' is a `struct sym_fns' which contains the name (or name 1906 prefix) of the symbol format, the length of the prefix, and pointers to 1907 four functions. These functions are called at various times to process 1908 symbol files whose identification matches the specified prefix. 1909 1910 The functions supplied by each module are: 1911 1912 `XYZ_symfile_init(struct sym_fns *sf)' 1913 Called from `symbol_file_add' when we are about to read a new 1914 symbol file. This function should clean up any internal state 1915 (possibly resulting from half-read previous files, for example) 1916 and prepare to read a new symbol file. Note that the symbol file 1917 which we are reading might be a new "main" symbol file, or might 1918 be a secondary symbol file whose symbols are being added to the 1919 existing symbol table. 1920 1921 The argument to `XYZ_symfile_init' is a newly allocated `struct 1922 sym_fns' whose `bfd' field contains the BFD for the new symbol 1923 file being read. Its `private' field has been zeroed, and can be 1924 modified as desired. Typically, a struct of private information 1925 will be `malloc''d, and a pointer to it will be placed in the 1926 `private' field. 1927 1928 There is no result from `XYZ_symfile_init', but it can call 1929 `error' if it detects an unavoidable problem. 1930 1931 `XYZ_new_init()' 1932 Called from `symbol_file_add' when discarding existing symbols. 1933 This function needs only handle the symbol-reading module's 1934 internal state; the symbol table data structures visible to the 1935 rest of GDB will be discarded by `symbol_file_add'. It has no 1936 arguments and no result. It may be called after 1937 `XYZ_symfile_init', if a new symbol table is being read, or may be 1938 called alone if all symbols are simply being discarded. 1939 1940 `XYZ_symfile_read(struct sym_fns *sf, CORE_ADDR addr, int mainline)' 1941 Called from `symbol_file_add' to actually read the symbols from a 1942 symbol-file into a set of psymtabs or symtabs. 1943 1944 `sf' points to the `struct sym_fns' originally passed to 1945 `XYZ_sym_init' for possible initialization. `addr' is the offset 1946 between the file's specified start address and its true address in 1947 memory. `mainline' is 1 if this is the main symbol table being 1948 read, and 0 if a secondary symbol file (e.g., shared library or 1949 dynamically loaded file) is being read. 1950 1951 In addition, if a symbol-reading module creates psymtabs when 1952 XYZ_symfile_read is called, these psymtabs will contain a pointer to a 1953 function `XYZ_psymtab_to_symtab', which can be called from any point in 1954 the GDB symbol-handling code. 1955 1956 `XYZ_psymtab_to_symtab (struct partial_symtab *pst)' 1957 Called from `psymtab_to_symtab' (or the `PSYMTAB_TO_SYMTAB' macro) 1958 if the psymtab has not already been read in and had its 1959 `pst->symtab' pointer set. The argument is the psymtab to be 1960 fleshed-out into a symtab. Upon return, `pst->readin' should have 1961 been set to 1, and `pst->symtab' should contain a pointer to the 1962 new corresponding symtab, or zero if there were no symbols in that 1963 part of the symbol file. 1964 1965 8.2 Partial Symbol Tables 1966 ========================= 1967 1968 GDB has three types of symbol tables: 1969 1970 * Full symbol tables ("symtabs"). These contain the main 1971 information about symbols and addresses. 1972 1973 * Partial symbol tables ("psymtabs"). These contain enough 1974 information to know when to read the corresponding part of the full 1975 symbol table. 1976 1977 * Minimal symbol tables ("msymtabs"). These contain information 1978 gleaned from non-debugging symbols. 1979 1980 This section describes partial symbol tables. 1981 1982 A psymtab is constructed by doing a very quick pass over an 1983 executable file's debugging information. Small amounts of information 1984 are extracted--enough to identify which parts of the symbol table will 1985 need to be re-read and fully digested later, when the user needs the 1986 information. The speed of this pass causes GDB to start up very 1987 quickly. Later, as the detailed rereading occurs, it occurs in small 1988 pieces, at various times, and the delay therefrom is mostly invisible to 1989 the user. 1990 1991 The symbols that show up in a file's psymtab should be, roughly, 1992 those visible to the debugger's user when the program is not running 1993 code from that file. These include external symbols and types, static 1994 symbols and types, and `enum' values declared at file scope. 1995 1996 The psymtab also contains the range of instruction addresses that the 1997 full symbol table would represent. 1998 1999 The idea is that there are only two ways for the user (or much of the 2000 code in the debugger) to reference a symbol: 2001 2002 * By its address (e.g., execution stops at some address which is 2003 inside a function in this file). The address will be noticed to 2004 be in the range of this psymtab, and the full symtab will be read 2005 in. `find_pc_function', `find_pc_line', and other `find_pc_...' 2006 functions handle this. 2007 2008 * By its name (e.g., the user asks to print a variable, or set a 2009 breakpoint on a function). Global names and file-scope names will 2010 be found in the psymtab, which will cause the symtab to be pulled 2011 in. Local names will have to be qualified by a global name, or a 2012 file-scope name, in which case we will have already read in the 2013 symtab as we evaluated the qualifier. Or, a local symbol can be 2014 referenced when we are "in" a local scope, in which case the first 2015 case applies. `lookup_symbol' does most of the work here. 2016 2017 The only reason that psymtabs exist is to cause a symtab to be read 2018 in at the right moment. Any symbol that can be elided from a psymtab, 2019 while still causing that to happen, should not appear in it. Since 2020 psymtabs don't have the idea of scope, you can't put local symbols in 2021 them anyway. Psymtabs don't have the idea of the type of a symbol, 2022 either, so types need not appear, unless they will be referenced by 2023 name. 2024 2025 It is a bug for GDB to behave one way when only a psymtab has been 2026 read, and another way if the corresponding symtab has been read in. 2027 Such bugs are typically caused by a psymtab that does not contain all 2028 the visible symbols, or which has the wrong instruction address ranges. 2029 2030 The psymtab for a particular section of a symbol file (objfile) 2031 could be thrown away after the symtab has been read in. The symtab 2032 should always be searched before the psymtab, so the psymtab will never 2033 be used (in a bug-free environment). Currently, psymtabs are allocated 2034 on an obstack, and all the psymbols themselves are allocated in a pair 2035 of large arrays on an obstack, so there is little to be gained by 2036 trying to free them unless you want to do a lot more work. 2037 2038 8.3 Types 2039 ========= 2040 2041 Fundamental Types (e.g., `FT_VOID', `FT_BOOLEAN'). 2042 -------------------------------------------------- 2043 2044 These are the fundamental types that GDB uses internally. Fundamental 2045 types from the various debugging formats (stabs, ELF, etc) are mapped 2046 into one of these. They are basically a union of all fundamental types 2047 that GDB knows about for all the languages that GDB knows about. 2048 2049 Type Codes (e.g., `TYPE_CODE_PTR', `TYPE_CODE_ARRAY'). 2050 ------------------------------------------------------ 2051 2052 Each time GDB builds an internal type, it marks it with one of these 2053 types. The type may be a fundamental type, such as `TYPE_CODE_INT', or 2054 a derived type, such as `TYPE_CODE_PTR' which is a pointer to another 2055 type. Typically, several `FT_*' types map to one `TYPE_CODE_*' type, 2056 and are distinguished by other members of the type struct, such as 2057 whether the type is signed or unsigned, and how many bits it uses. 2058 2059 Builtin Types (e.g., `builtin_type_void', `builtin_type_char'). 2060 --------------------------------------------------------------- 2061 2062 These are instances of type structs that roughly correspond to 2063 fundamental types and are created as global types for GDB to use for 2064 various ugly historical reasons. We eventually want to eliminate 2065 these. Note for example that `builtin_type_int' initialized in 2066 `gdbtypes.c' is basically the same as a `TYPE_CODE_INT' type that is 2067 initialized in `c-lang.c' for an `FT_INTEGER' fundamental type. The 2068 difference is that the `builtin_type' is not associated with any 2069 particular objfile, and only one instance exists, while `c-lang.c' 2070 builds as many `TYPE_CODE_INT' types as needed, with each one 2071 associated with some particular objfile. 2072 2073 8.4 Object File Formats 2074 ======================= 2075 2076 8.4.1 a.out 2077 ----------- 2078 2079 The `a.out' format is the original file format for Unix. It consists 2080 of three sections: `text', `data', and `bss', which are for program 2081 code, initialized data, and uninitialized data, respectively. 2082 2083 The `a.out' format is so simple that it doesn't have any reserved 2084 place for debugging information. (Hey, the original Unix hackers used 2085 `adb', which is a machine-language debugger!) The only debugging 2086 format for `a.out' is stabs, which is encoded as a set of normal 2087 symbols with distinctive attributes. 2088 2089 The basic `a.out' reader is in `dbxread.c'. 2090 2091 8.4.2 COFF 2092 ---------- 2093 2094 The COFF format was introduced with System V Release 3 (SVR3) Unix. 2095 COFF files may have multiple sections, each prefixed by a header. The 2096 number of sections is limited. 2097 2098 The COFF specification includes support for debugging. Although this 2099 was a step forward, the debugging information was woefully limited. 2100 For instance, it was not possible to represent code that came from an 2101 included file. GNU's COFF-using configs often use stabs-type info, 2102 encapsulated in special sections. 2103 2104 The COFF reader is in `coffread.c'. 2105 2106 8.4.3 ECOFF 2107 ----------- 2108 2109 ECOFF is an extended COFF originally introduced for Mips and Alpha 2110 workstations. 2111 2112 The basic ECOFF reader is in `mipsread.c'. 2113 2114 8.4.4 XCOFF 2115 ----------- 2116 2117 The IBM RS/6000 running AIX uses an object file format called XCOFF. 2118 The COFF sections, symbols, and line numbers are used, but debugging 2119 symbols are `dbx'-style stabs whose strings are located in the `.debug' 2120 section (rather than the string table). For more information, see 2121 *note Top: (stabs)Top. 2122 2123 The shared library scheme has a clean interface for figuring out what 2124 shared libraries are in use, but the catch is that everything which 2125 refers to addresses (symbol tables and breakpoints at least) needs to be 2126 relocated for both shared libraries and the main executable. At least 2127 using the standard mechanism this can only be done once the program has 2128 been run (or the core file has been read). 2129 2130 8.4.5 PE 2131 -------- 2132 2133 Windows 95 and NT use the PE ("Portable Executable") format for their 2134 executables. PE is basically COFF with additional headers. 2135 2136 While BFD includes special PE support, GDB needs only the basic COFF 2137 reader. 2138 2139 8.4.6 ELF 2140 --------- 2141 2142 The ELF format came with System V Release 4 (SVR4) Unix. ELF is 2143 similar to COFF in being organized into a number of sections, but it 2144 removes many of COFF's limitations. Debugging info may be either stabs 2145 encapsulated in ELF sections, or more commonly these days, DWARF. 2146 2147 The basic ELF reader is in `elfread.c'. 2148 2149 8.4.7 SOM 2150 --------- 2151 2152 SOM is HP's object file and debug format (not to be confused with IBM's 2153 SOM, which is a cross-language ABI). 2154 2155 The SOM reader is in `somread.c'. 2156 2157 8.5 Debugging File Formats 2158 ========================== 2159 2160 This section describes characteristics of debugging information that 2161 are independent of the object file format. 2162 2163 8.5.1 stabs 2164 ----------- 2165 2166 `stabs' started out as special symbols within the `a.out' format. 2167 Since then, it has been encapsulated into other file formats, such as 2168 COFF and ELF. 2169 2170 While `dbxread.c' does some of the basic stab processing, including 2171 for encapsulated versions, `stabsread.c' does the real work. 2172 2173 8.5.2 COFF 2174 ---------- 2175 2176 The basic COFF definition includes debugging information. The level of 2177 support is minimal and non-extensible, and is not often used. 2178 2179 8.5.3 Mips debug (Third Eye) 2180 ---------------------------- 2181 2182 ECOFF includes a definition of a special debug format. 2183 2184 The file `mdebugread.c' implements reading for this format. 2185 2186 8.5.4 DWARF 2 2187 ------------- 2188 2189 DWARF 2 is an improved but incompatible version of DWARF 1. 2190 2191 The DWARF 2 reader is in `dwarf2read.c'. 2192 2193 8.5.5 Compressed DWARF 2 2194 ------------------------ 2195 2196 Compressed DWARF 2 is not technically a separate debugging format, but 2197 merely DWARF 2 debug information that has been compressed. In this 2198 format, every object-file section holding DWARF 2 debugging information 2199 is compressed and prepended with a header. (The section is also 2200 typically renamed, so a section called `.debug_info' in a DWARF 2 2201 binary would be called `.zdebug_info' in a compressed DWARF 2 binary.) 2202 The header is 12 bytes long: 2203 2204 * 4 bytes: the literal string "ZLIB" 2205 2206 * 8 bytes: the uncompressed size of the section, in big-endian byte 2207 order. 2208 2209 The same reader is used for both compressed an normal DWARF 2 info. 2210 Section decompression is done in `zlib_decompress_section' in 2211 `dwarf2read.c'. 2212 2213 8.5.6 DWARF 3 2214 ------------- 2215 2216 DWARF 3 is an improved version of DWARF 2. 2217 2218 8.5.7 SOM 2219 --------- 2220 2221 Like COFF, the SOM definition includes debugging information. 2222 2223 8.6 Adding a New Symbol Reader to GDB 2224 ===================================== 2225 2226 If you are using an existing object file format (`a.out', COFF, ELF, 2227 etc), there is probably little to be done. 2228 2229 If you need to add a new object file format, you must first add it to 2230 BFD. This is beyond the scope of this document. 2231 2232 You must then arrange for the BFD code to provide access to the 2233 debugging symbols. Generally GDB will have to call swapping routines 2234 from BFD and a few other BFD internal routines to locate the debugging 2235 information. As much as possible, GDB should not depend on the BFD 2236 internal data structures. 2237 2238 For some targets (e.g., COFF), there is a special transfer vector 2239 used to call swapping routines, since the external data structures on 2240 various platforms have different sizes and layouts. Specialized 2241 routines that will only ever be implemented by one object file format 2242 may be called directly. This interface should be described in a file 2243 `bfd/libXYZ.h', which is included by GDB. 2244 2245 8.7 Memory Management for Symbol Files 2246 ====================================== 2247 2248 Most memory associated with a loaded symbol file is stored on its 2249 `objfile_obstack'. This includes symbols, types, namespace data, and 2250 other information produced by the symbol readers. 2251 2252 Because this data lives on the objfile's obstack, it is automatically 2253 released when the objfile is unloaded or reloaded. Therefore one 2254 objfile must not reference symbol or type data from another objfile; 2255 they could be unloaded at different times. 2256 2257 User convenience variables, et cetera, have associated types. 2258 Normally these types live in the associated objfile. However, when the 2259 objfile is unloaded, those types are deep copied to global memory, so 2260 that the values of the user variables and history items are not lost. 2261 2262 2263 File: gdbint.info, Node: Language Support, Next: Host Definition, Prev: Symbol Handling, Up: Top 2264 2265 9 Language Support 2266 ****************** 2267 2268 GDB's language support is mainly driven by the symbol reader, although 2269 it is possible for the user to set the source language manually. 2270 2271 GDB chooses the source language by looking at the extension of the 2272 file recorded in the debug info; `.c' means C, `.f' means Fortran, etc. 2273 It may also use a special-purpose language identifier if the debug 2274 format supports it, like with DWARF. 2275 2276 9.1 Adding a Source Language to GDB 2277 =================================== 2278 2279 To add other languages to GDB's expression parser, follow the following 2280 steps: 2281 2282 _Create the expression parser._ 2283 This should reside in a file `LANG-exp.y'. Routines for building 2284 parsed expressions into a `union exp_element' list are in 2285 `parse.c'. 2286 2287 Since we can't depend upon everyone having Bison, and YACC produces 2288 parsers that define a bunch of global names, the following lines 2289 *must* be included at the top of the YACC parser, to prevent the 2290 various parsers from defining the same global names: 2291 2292 #define yyparse LANG_parse 2293 #define yylex LANG_lex 2294 #define yyerror LANG_error 2295 #define yylval LANG_lval 2296 #define yychar LANG_char 2297 #define yydebug LANG_debug 2298 #define yypact LANG_pact 2299 #define yyr1 LANG_r1 2300 #define yyr2 LANG_r2 2301 #define yydef LANG_def 2302 #define yychk LANG_chk 2303 #define yypgo LANG_pgo 2304 #define yyact LANG_act 2305 #define yyexca LANG_exca 2306 #define yyerrflag LANG_errflag 2307 #define yynerrs LANG_nerrs 2308 2309 At the bottom of your parser, define a `struct language_defn' and 2310 initialize it with the right values for your language. Define an 2311 `initialize_LANG' routine and have it call 2312 `add_language(LANG_language_defn)' to tell the rest of GDB that 2313 your language exists. You'll need some other supporting variables 2314 and functions, which will be used via pointers from your 2315 `LANG_language_defn'. See the declaration of `struct 2316 language_defn' in `language.h', and the other `*-exp.y' files, for 2317 more information. 2318 2319 _Add any evaluation routines, if necessary_ 2320 If you need new opcodes (that represent the operations of the 2321 language), add them to the enumerated type in `expression.h'. Add 2322 support code for these operations in the `evaluate_subexp' function 2323 defined in the file `eval.c'. Add cases for new opcodes in two 2324 functions from `parse.c': `prefixify_subexp' and 2325 `length_of_subexp'. These compute the number of `exp_element's 2326 that a given operation takes up. 2327 2328 _Update some existing code_ 2329 Add an enumerated identifier for your language to the enumerated 2330 type `enum language' in `defs.h'. 2331 2332 Update the routines in `language.c' so your language is included. 2333 These routines include type predicates and such, which (in some 2334 cases) are language dependent. If your language does not appear 2335 in the switch statement, an error is reported. 2336 2337 Also included in `language.c' is the code that updates the variable 2338 `current_language', and the routines that translate the 2339 `language_LANG' enumerated identifier into a printable string. 2340 2341 Update the function `_initialize_language' to include your 2342 language. This function picks the default language upon startup, 2343 so is dependent upon which languages that GDB is built for. 2344 2345 Update `allocate_symtab' in `symfile.c' and/or symbol-reading code 2346 so that the language of each symtab (source file) is set properly. 2347 This is used to determine the language to use at each stack frame 2348 level. Currently, the language is set based upon the extension of 2349 the source file. If the language can be better inferred from the 2350 symbol information, please set the language of the symtab in the 2351 symbol-reading code. 2352 2353 Add helper code to `print_subexp' (in `expprint.c') to handle any 2354 new expression opcodes you have added to `expression.h'. Also, 2355 add the printed representations of your operators to 2356 `op_print_tab'. 2357 2358 _Add a place of call_ 2359 Add a call to `LANG_parse()' and `LANG_error' in `parse_exp_1' 2360 (defined in `parse.c'). 2361 2362 _Edit `Makefile.in'_ 2363 Add dependencies in `Makefile.in'. Make sure you update the macro 2364 variables such as `HFILES' and `OBJS', otherwise your code may not 2365 get linked in, or, worse yet, it may not get `tar'red into the 2366 distribution! 2367 2368 2369 File: gdbint.info, Node: Host Definition, Next: Target Architecture Definition, Prev: Language Support, Up: Top 2370 2371 10 Host Definition 2372 ****************** 2373 2374 With the advent of Autoconf, it's rarely necessary to have host 2375 definition machinery anymore. The following information is provided, 2376 mainly, as an historical reference. 2377 2378 10.1 Adding a New Host 2379 ====================== 2380 2381 GDB's host configuration support normally happens via Autoconf. New 2382 host-specific definitions should not be needed. Older hosts GDB still 2383 use the host-specific definitions and files listed below, but these 2384 mostly exist for historical reasons, and will eventually disappear. 2385 2386 `gdb/config/ARCH/XYZ.mh' 2387 This file is a Makefile fragment that once contained both host and 2388 native configuration information (*note Native Debugging::) for the 2389 machine XYZ. The host configuration information is now handled by 2390 Autoconf. 2391 2392 Host configuration information included definitions for `CC', 2393 `SYSV_DEFINE', `XM_CFLAGS', `XM_ADD_FILES', `XM_CLIBS', 2394 `XM_CDEPS', etc.; see `Makefile.in'. 2395 2396 New host-only configurations do not need this file. 2397 2398 2399 (Files named `gdb/config/ARCH/xm-XYZ.h' were once used to define 2400 host-specific macros, but were no longer needed and have all been 2401 removed.) 2402 2403 Generic Host Support Files 2404 -------------------------- 2405 2406 There are some "generic" versions of routines that can be used by 2407 various systems. 2408 2409 `ser-unix.c' 2410 This contains serial line support for Unix systems. It is 2411 included by default on all Unix-like hosts. 2412 2413 `ser-pipe.c' 2414 This contains serial pipe support for Unix systems. It is 2415 included by default on all Unix-like hosts. 2416 2417 `ser-mingw.c' 2418 This contains serial line support for 32-bit programs running under 2419 Windows using MinGW. 2420 2421 `ser-go32.c' 2422 This contains serial line support for 32-bit programs running 2423 under DOS, using the DJGPP (a.k.a. GO32) execution environment. 2424 2425 `ser-tcp.c' 2426 This contains generic TCP support using sockets. It is included by 2427 default on all Unix-like hosts and with MinGW. 2428 2429 10.2 Host Conditionals 2430 ====================== 2431 2432 When GDB is configured and compiled, various macros are defined or left 2433 undefined, to control compilation based on the attributes of the host 2434 system. While formerly they could be set in host-specific header 2435 files, at present they can be changed only by setting `CFLAGS' when 2436 building, or by editing the source code. 2437 2438 These macros and their meanings (or if the meaning is not documented 2439 here, then one of the source files where they are used is indicated) 2440 are: 2441 2442 `GDBINIT_FILENAME' 2443 The default name of GDB's initialization file (normally 2444 `.gdbinit'). 2445 2446 `SIGWINCH_HANDLER' 2447 If your host defines `SIGWINCH', you can define this to be the name 2448 of a function to be called if `SIGWINCH' is received. 2449 2450 `SIGWINCH_HANDLER_BODY' 2451 Define this to expand into code that will define the function 2452 named by the expansion of `SIGWINCH_HANDLER'. 2453 2454 `CRLF_SOURCE_FILES' 2455 Define this if host files use `\r\n' rather than `\n' as a line 2456 terminator. This will cause source file listings to omit `\r' 2457 characters when printing and it will allow `\r\n' line endings of 2458 files which are "sourced" by gdb. It must be possible to open 2459 files in binary mode using `O_BINARY' or, for fopen, `"rb"'. 2460 2461 `DEFAULT_PROMPT' 2462 The default value of the prompt string (normally `"(gdb) "'). 2463 2464 `DEV_TTY' 2465 The name of the generic TTY device, defaults to `"/dev/tty"'. 2466 2467 `ISATTY' 2468 Substitute for isatty, if not available. 2469 2470 `FOPEN_RB' 2471 Define this if binary files are opened the same way as text files. 2472 2473 `CC_HAS_LONG_LONG' 2474 Define this if the host C compiler supports `long long'. This is 2475 set by the `configure' script. 2476 2477 `PRINTF_HAS_LONG_LONG' 2478 Define this if the host can handle printing of long long integers 2479 via the printf format conversion specifier `ll'. This is set by 2480 the `configure' script. 2481 2482 `LSEEK_NOT_LINEAR' 2483 Define this if `lseek (n)' does not necessarily move to byte number 2484 `n' in the file. This is only used when reading source files. It 2485 is normally faster to define `CRLF_SOURCE_FILES' when possible. 2486 2487 `NORETURN' 2488 If defined, this should be one or more tokens, such as `volatile', 2489 that can be used in both the declaration and definition of 2490 functions to indicate that they never return. The default is 2491 already set correctly if compiling with GCC. This will almost 2492 never need to be defined. 2493 2494 `ATTR_NORETURN' 2495 If defined, this should be one or more tokens, such as 2496 `__attribute__ ((noreturn))', that can be used in the declarations 2497 of functions to indicate that they never return. The default is 2498 already set correctly if compiling with GCC. This will almost 2499 never need to be defined. 2500 2501 `lint' 2502 Define this to help placate `lint' in some situations. 2503 2504 `volatile' 2505 Define this to override the defaults of `__volatile__' or `/**/'. 2506 2507 2508 File: gdbint.info, Node: Target Architecture Definition, Next: Target Descriptions, Prev: Host Definition, Up: Top 2509 2510 11 Target Architecture Definition 2511 ********************************* 2512 2513 GDB's target architecture defines what sort of machine-language 2514 programs GDB can work with, and how it works with them. 2515 2516 The target architecture object is implemented as the C structure 2517 `struct gdbarch *'. The structure, and its methods, are generated 2518 using the Bourne shell script `gdbarch.sh'. 2519 2520 * Menu: 2521 2522 * OS ABI Variant Handling:: 2523 * Initialize New Architecture:: 2524 * Registers and Memory:: 2525 * Pointers and Addresses:: 2526 * Address Classes:: 2527 * Register Representation:: 2528 * Frame Interpretation:: 2529 * Inferior Call Setup:: 2530 * Adding support for debugging core files:: 2531 * Defining Other Architecture Features:: 2532 * Adding a New Target:: 2533 2534 2535 File: gdbint.info, Node: OS ABI Variant Handling, Next: Initialize New Architecture, Up: Target Architecture Definition 2536 2537 11.1 Operating System ABI Variant Handling 2538 ========================================== 2539 2540 GDB provides a mechanism for handling variations in OS ABIs. An OS ABI 2541 variant may have influence over any number of variables in the target 2542 architecture definition. There are two major components in the OS ABI 2543 mechanism: sniffers and handlers. 2544 2545 A "sniffer" examines a file matching a BFD architecture/flavour pair 2546 (the architecture may be wildcarded) in an attempt to determine the OS 2547 ABI of that file. Sniffers with a wildcarded architecture are 2548 considered to be "generic", while sniffers for a specific architecture 2549 are considered to be "specific". A match from a specific sniffer 2550 overrides a match from a generic sniffer. Multiple sniffers for an 2551 architecture/flavour may exist, in order to differentiate between two 2552 different operating systems which use the same basic file format. The 2553 OS ABI framework provides a generic sniffer for ELF-format files which 2554 examines the `EI_OSABI' field of the ELF header, as well as note 2555 sections known to be used by several operating systems. 2556 2557 A "handler" is used to fine-tune the `gdbarch' structure for the 2558 selected OS ABI. There may be only one handler for a given OS ABI for 2559 each BFD architecture. 2560 2561 The following OS ABI variants are defined in `defs.h': 2562 2563 `GDB_OSABI_UNINITIALIZED' 2564 Used for struct gdbarch_info if ABI is still uninitialized. 2565 2566 `GDB_OSABI_UNKNOWN' 2567 The ABI of the inferior is unknown. The default `gdbarch' 2568 settings for the architecture will be used. 2569 2570 `GDB_OSABI_SVR4' 2571 UNIX System V Release 4. 2572 2573 `GDB_OSABI_HURD' 2574 GNU using the Hurd kernel. 2575 2576 `GDB_OSABI_SOLARIS' 2577 Sun Solaris. 2578 2579 `GDB_OSABI_OSF1' 2580 OSF/1, including Digital UNIX and Compaq Tru64 UNIX. 2581 2582 `GDB_OSABI_LINUX' 2583 GNU using the Linux kernel. 2584 2585 `GDB_OSABI_FREEBSD_AOUT' 2586 FreeBSD using the `a.out' executable format. 2587 2588 `GDB_OSABI_FREEBSD_ELF' 2589 FreeBSD using the ELF executable format. 2590 2591 `GDB_OSABI_NETBSD_AOUT' 2592 NetBSD using the `a.out' executable format. 2593 2594 `GDB_OSABI_NETBSD_ELF' 2595 NetBSD using the ELF executable format. 2596 2597 `GDB_OSABI_OPENBSD_ELF' 2598 OpenBSD using the ELF executable format. 2599 2600 `GDB_OSABI_WINCE' 2601 Windows CE. 2602 2603 `GDB_OSABI_GO32' 2604 DJGPP. 2605 2606 `GDB_OSABI_IRIX' 2607 Irix. 2608 2609 `GDB_OSABI_INTERIX' 2610 Interix (Posix layer for MS-Windows systems). 2611 2612 `GDB_OSABI_HPUX_ELF' 2613 HP/UX using the ELF executable format. 2614 2615 `GDB_OSABI_HPUX_SOM' 2616 HP/UX using the SOM executable format. 2617 2618 `GDB_OSABI_QNXNTO' 2619 QNX Neutrino. 2620 2621 `GDB_OSABI_CYGWIN' 2622 Cygwin. 2623 2624 `GDB_OSABI_AIX' 2625 AIX. 2626 2627 2628 Here are the functions that make up the OS ABI framework: 2629 2630 -- Function: const char * gdbarch_osabi_name (enum gdb_osabi OSABI) 2631 Return the name of the OS ABI corresponding to OSABI. 2632 2633 -- Function: void gdbarch_register_osabi (enum bfd_architecture ARCH, 2634 unsigned long MACHINE, enum gdb_osabi OSABI, void 2635 (*INIT_OSABI)(struct gdbarch_info INFO, struct gdbarch 2636 *GDBARCH)) 2637 Register the OS ABI handler specified by INIT_OSABI for the 2638 architecture, machine type and OS ABI specified by ARCH, MACHINE 2639 and OSABI. In most cases, a value of zero for the machine type, 2640 which implies the architecture's default machine type, will 2641 suffice. 2642 2643 -- Function: void gdbarch_register_osabi_sniffer (enum 2644 bfd_architecture ARCH, enum bfd_flavour FLAVOUR, enum 2645 gdb_osabi (*SNIFFER)(bfd *ABFD)) 2646 Register the OS ABI file sniffer specified by SNIFFER for the BFD 2647 architecture/flavour pair specified by ARCH and FLAVOUR. If ARCH 2648 is `bfd_arch_unknown', the sniffer is considered to be generic, 2649 and is allowed to examine FLAVOUR-flavoured files for any 2650 architecture. 2651 2652 -- Function: enum gdb_osabi gdbarch_lookup_osabi (bfd *ABFD) 2653 Examine the file described by ABFD to determine its OS ABI. The 2654 value `GDB_OSABI_UNKNOWN' is returned if the OS ABI cannot be 2655 determined. 2656 2657 -- Function: void gdbarch_init_osabi (struct gdbarch info INFO, struct 2658 gdbarch *GDBARCH, enum gdb_osabi OSABI) 2659 Invoke the OS ABI handler corresponding to OSABI to fine-tune the 2660 `gdbarch' structure specified by GDBARCH. If a handler 2661 corresponding to OSABI has not been registered for GDBARCH's 2662 architecture, a warning will be issued and the debugging session 2663 will continue with the defaults already established for GDBARCH. 2664 2665 -- Function: void generic_elf_osabi_sniff_abi_tag_sections (bfd *ABFD, 2666 asection *SECT, void *OBJ) 2667 Helper routine for ELF file sniffers. Examine the file described 2668 by ABFD and look at ABI tag note sections to determine the OS ABI 2669 from the note. This function should be called via 2670 `bfd_map_over_sections'. 2671 2672 2673 File: gdbint.info, Node: Initialize New Architecture, Next: Registers and Memory, Prev: OS ABI Variant Handling, Up: Target Architecture Definition 2674 2675 11.2 Initializing a New Architecture 2676 ==================================== 2677 2678 * Menu: 2679 2680 * How an Architecture is Represented:: 2681 * Looking Up an Existing Architecture:: 2682 * Creating a New Architecture:: 2683 2684 2685 File: gdbint.info, Node: How an Architecture is Represented, Next: Looking Up an Existing Architecture, Up: Initialize New Architecture 2686 2687 11.2.1 How an Architecture is Represented 2688 ----------------------------------------- 2689 2690 Each `gdbarch' is associated with a single BFD architecture, via a 2691 `bfd_arch_ARCH' in the `bfd_architecture' enumeration. The `gdbarch' 2692 is registered by a call to `register_gdbarch_init', usually from the 2693 file's `_initialize_FILENAME' routine, which will be automatically 2694 called during GDB startup. The arguments are a BFD architecture 2695 constant and an initialization function. 2696 2697 A GDB description for a new architecture, ARCH is created by 2698 defining a global function `_initialize_ARCH_tdep', by convention in 2699 the source file `ARCH-tdep.c'. For example, in the case of the 2700 OpenRISC 1000, this function is called `_initialize_or1k_tdep' and is 2701 found in the file `or1k-tdep.c'. 2702 2703 The resulting object files containing the implementation of the 2704 `_initialize_ARCH_tdep' function are specified in the GDB 2705 `configure.tgt' file, which includes a large case statement pattern 2706 matching against the `--target' option of the `configure' script. The 2707 new `struct gdbarch' is created within the `_initialize_ARCH_tdep' 2708 function by calling `gdbarch_register': 2709 2710 void gdbarch_register (enum bfd_architecture ARCHITECTURE, 2711 gdbarch_init_ftype *INIT_FUNC, 2712 gdbarch_dump_tdep_ftype *TDEP_DUMP_FUNC); 2713 2714 The ARCHITECTURE will identify the unique BFD to be associated with 2715 this `gdbarch'. The INIT_FUNC funciton is called to create and return 2716 the new `struct gdbarch'. The TDEP_DUMP_FUNC function will dump the 2717 target specific details associated with this architecture. 2718 2719 For example the function `_initialize_or1k_tdep' creates its 2720 architecture for 32-bit OpenRISC 1000 architectures by calling: 2721 2722 gdbarch_register (bfd_arch_or32, or1k_gdbarch_init, or1k_dump_tdep); 2723 2724 2725 File: gdbint.info, Node: Looking Up an Existing Architecture, Next: Creating a New Architecture, Prev: How an Architecture is Represented, Up: Initialize New Architecture 2726 2727 11.2.2 Looking Up an Existing Architecture 2728 ------------------------------------------ 2729 2730 The initialization function has this prototype: 2731 2732 static struct gdbarch * 2733 ARCH_gdbarch_init (struct gdbarch_info INFO, 2734 struct gdbarch_list *ARCHES) 2735 2736 The INFO argument contains parameters used to select the correct 2737 architecture, and ARCHES is a list of architectures which have already 2738 been created with the same `bfd_arch_ARCH' value. 2739 2740 The initialization function should first make sure that INFO is 2741 acceptable, and return `NULL' if it is not. Then, it should search 2742 through ARCHES for an exact match to INFO, and return one if found. 2743 Lastly, if no exact match was found, it should create a new 2744 architecture based on INFO and return it. 2745 2746 The lookup is done using `gdbarch_list_lookup_by_info'. It is 2747 passed the list of existing architectures, ARCHES, and the `struct 2748 gdbarch_info', INFO, and returns the first matching architecture it 2749 finds, or `NULL' if none are found. If an architecture is found it can 2750 be returned as the result from the initialization function, otherwise a 2751 new `struct gdbach' will need to be created. 2752 2753 The struct gdbarch_info has the following components: 2754 2755 struct gdbarch_info 2756 { 2757 const struct bfd_arch_info *bfd_arch_info; 2758 int byte_order; 2759 bfd *abfd; 2760 struct gdbarch_tdep_info *tdep_info; 2761 enum gdb_osabi osabi; 2762 const struct target_desc *target_desc; 2763 }; 2764 2765 The `bfd_arch_info' member holds the key details about the 2766 architecture. The `byte_order' member is a value in an enumeration 2767 indicating the endianism. The `abfd' member is a pointer to the full 2768 BFD, the `tdep_info' member is additional custom target specific 2769 information, `osabi' identifies which (if any) of a number of operating 2770 specific ABIs are used by this architecture and the `target_desc' 2771 member is a set of name-value pairs with information about register 2772 usage in this target. 2773 2774 When the `struct gdbarch' initialization function is called, not all 2775 the fields are provided--only those which can be deduced from the BFD. 2776 The `struct gdbarch_info', INFO is used as a look-up key with the list 2777 of existing architectures, ARCHES to see if a suitable architecture 2778 already exists. The TDEP_INFO, OSABI and TARGET_DESC fields may be 2779 added before this lookup to refine the search. 2780 2781 Only information in INFO should be used to choose the new 2782 architecture. Historically, INFO could be sparse, and defaults would 2783 be collected from the first element on ARCHES. However, GDB now fills 2784 in INFO more thoroughly, so new `gdbarch' initialization functions 2785 should not take defaults from ARCHES. 2786 2787 2788 File: gdbint.info, Node: Creating a New Architecture, Prev: Looking Up an Existing Architecture, Up: Initialize New Architecture 2789 2790 11.2.3 Creating a New Architecture 2791 ---------------------------------- 2792 2793 If no architecture is found, then a new architecture must be created, 2794 by calling `gdbarch_alloc' using the supplied `struct gdbarch_info' and 2795 any additional custom target specific information in a `struct 2796 gdbarch_tdep'. The prototype for `gdbarch_alloc' is: 2797 2798 struct gdbarch *gdbarch_alloc (const struct gdbarch_info *INFO, 2799 struct gdbarch_tdep *TDEP); 2800 2801 The newly created struct gdbarch must then be populated. Although 2802 there are default values, in most cases they are not what is required. 2803 2804 For each element, X, there is are a pair of corresponding accessor 2805 functions, one to set the value of that element, `set_gdbarch_X', the 2806 second to either get the value of an element (if it is a variable) or 2807 to apply the element (if it is a function), `gdbarch_X'. Note that 2808 both accessor functions take a pointer to the `struct gdbarch' as first 2809 argument. Populating the new `gdbarch' should use the `set_gdbarch' 2810 functions. 2811 2812 The following sections identify the main elements that should be set 2813 in this way. This is not the complete list, but represents the 2814 functions and elements that must commonly be specified for a new 2815 architecture. Many of the functions and variables are described in the 2816 header file `gdbarch.h'. 2817 2818 This is the main work in defining a new architecture. Implementing 2819 the set of functions to populate the `struct gdbarch'. 2820 2821 `struct gdbarch_tdep' is not defined within GDB--it is up to the 2822 user to define this struct if it is needed to hold custom target 2823 information that is not covered by the standard `struct gdbarch'. For 2824 example with the OpenRISC 1000 architecture it is used to hold the 2825 number of matchpoints available in the target (along with other 2826 information). 2827 2828 If there is no additional target specific information, it can be set 2829 to `NULL'. 2830 2831 2832 File: gdbint.info, Node: Registers and Memory, Next: Pointers and Addresses, Prev: Initialize New Architecture, Up: Target Architecture Definition 2833 2834 11.3 Registers and Memory 2835 ========================= 2836 2837 GDB's model of the target machine is rather simple. GDB assumes the 2838 machine includes a bank of registers and a block of memory. Each 2839 register may have a different size. 2840 2841 GDB does not have a magical way to match up with the compiler's idea 2842 of which registers are which; however, it is critical that they do 2843 match up accurately. The only way to make this work is to get accurate 2844 information about the order that the compiler uses, and to reflect that 2845 in the `gdbarch_register_name' and related functions. 2846 2847 GDB can handle big-endian, little-endian, and bi-endian 2848 architectures. 2849 2850 2851 File: gdbint.info, Node: Pointers and Addresses, Next: Address Classes, Prev: Registers and Memory, Up: Target Architecture Definition 2852 2853 11.4 Pointers Are Not Always Addresses 2854 ====================================== 2855 2856 On almost all 32-bit architectures, the representation of a pointer is 2857 indistinguishable from the representation of some fixed-length number 2858 whose value is the byte address of the object pointed to. On such 2859 machines, the words "pointer" and "address" can be used interchangeably. 2860 However, architectures with smaller word sizes are often cramped for 2861 address space, so they may choose a pointer representation that breaks 2862 this identity, and allows a larger code address space. 2863 2864 For example, the Renesas D10V is a 16-bit VLIW processor whose 2865 instructions are 32 bits long(1). If the D10V used ordinary byte 2866 addresses to refer to code locations, then the processor would only be 2867 able to address 64kb of instructions. However, since instructions must 2868 be aligned on four-byte boundaries, the low two bits of any valid 2869 instruction's byte address are always zero--byte addresses waste two 2870 bits. So instead of byte addresses, the D10V uses word addresses--byte 2871 addresses shifted right two bits--to refer to code. Thus, the D10V can 2872 use 16-bit words to address 256kb of code space. 2873 2874 However, this means that code pointers and data pointers have 2875 different forms on the D10V. The 16-bit word `0xC020' refers to byte 2876 address `0xC020' when used as a data address, but refers to byte address 2877 `0x30080' when used as a code address. 2878 2879 (The D10V also uses separate code and data address spaces, which also 2880 affects the correspondence between pointers and addresses, but we're 2881 going to ignore that here; this example is already too long.) 2882 2883 To cope with architectures like this--the D10V is not the only 2884 one!--GDB tries to distinguish between "addresses", which are byte 2885 numbers, and "pointers", which are the target's representation of an 2886 address of a particular type of data. In the example above, `0xC020' 2887 is the pointer, which refers to one of the addresses `0xC020' or 2888 `0x30080', depending on the type imposed upon it. GDB provides 2889 functions for turning a pointer into an address and vice versa, in the 2890 appropriate way for the current architecture. 2891 2892 Unfortunately, since addresses and pointers are identical on almost 2893 all processors, this distinction tends to bit-rot pretty quickly. Thus, 2894 each time you port GDB to an architecture which does distinguish 2895 between pointers and addresses, you'll probably need to clean up some 2896 architecture-independent code. 2897 2898 Here are functions which convert between pointers and addresses: 2899 2900 -- Function: CORE_ADDR extract_typed_address (void *BUF, struct type 2901 *TYPE) 2902 Treat the bytes at BUF as a pointer or reference of type TYPE, and 2903 return the address it represents, in a manner appropriate for the 2904 current architecture. This yields an address GDB can use to read 2905 target memory, disassemble, etc. Note that BUF refers to a buffer 2906 in GDB's memory, not the inferior's. 2907 2908 For example, if the current architecture is the Intel x86, this 2909 function extracts a little-endian integer of the appropriate 2910 length from BUF and returns it. However, if the current 2911 architecture is the D10V, this function will return a 16-bit 2912 integer extracted from BUF, multiplied by four if TYPE is a 2913 pointer to a function. 2914 2915 If TYPE is not a pointer or reference type, then this function 2916 will signal an internal error. 2917 2918 -- Function: CORE_ADDR store_typed_address (void *BUF, struct type 2919 *TYPE, CORE_ADDR ADDR) 2920 Store the address ADDR in BUF, in the proper format for a pointer 2921 of type TYPE in the current architecture. Note that BUF refers to 2922 a buffer in GDB's memory, not the inferior's. 2923 2924 For example, if the current architecture is the Intel x86, this 2925 function stores ADDR unmodified as a little-endian integer of the 2926 appropriate length in BUF. However, if the current architecture 2927 is the D10V, this function divides ADDR by four if TYPE is a 2928 pointer to a function, and then stores it in BUF. 2929 2930 If TYPE is not a pointer or reference type, then this function 2931 will signal an internal error. 2932 2933 -- Function: CORE_ADDR value_as_address (struct value *VAL) 2934 Assuming that VAL is a pointer, return the address it represents, 2935 as appropriate for the current architecture. 2936 2937 This function actually works on integral values, as well as 2938 pointers. For pointers, it performs architecture-specific 2939 conversions as described above for `extract_typed_address'. 2940 2941 -- Function: CORE_ADDR value_from_pointer (struct type *TYPE, 2942 CORE_ADDR ADDR) 2943 Create and return a value representing a pointer of type TYPE to 2944 the address ADDR, as appropriate for the current architecture. 2945 This function performs architecture-specific conversions as 2946 described above for `store_typed_address'. 2947 2948 Here are two functions which architectures can define to indicate the 2949 relationship between pointers and addresses. These have default 2950 definitions, appropriate for architectures on which all pointers are 2951 simple unsigned byte addresses. 2952 2953 -- Function: CORE_ADDR gdbarch_pointer_to_address (struct gdbarch 2954 *GDBARCH, struct type *TYPE, char *BUF) 2955 Assume that BUF holds a pointer of type TYPE, in the appropriate 2956 format for the current architecture. Return the byte address the 2957 pointer refers to. 2958 2959 This function may safely assume that TYPE is either a pointer or a 2960 C++ reference type. 2961 2962 -- Function: void gdbarch_address_to_pointer (struct gdbarch *GDBARCH, 2963 struct type *TYPE, char *BUF, CORE_ADDR ADDR) 2964 Store in BUF a pointer of type TYPE representing the address ADDR, 2965 in the appropriate format for the current architecture. 2966 2967 This function may safely assume that TYPE is either a pointer or a 2968 C++ reference type. 2969 2970 ---------- Footnotes ---------- 2971 2972 (1) Some D10V instructions are actually pairs of 16-bit 2973 sub-instructions. However, since you can't jump into the middle of 2974 such a pair, code addresses can only refer to full 32 bit instructions, 2975 which is what matters in this explanation. 2976 2977 2978 File: gdbint.info, Node: Address Classes, Next: Register Representation, Prev: Pointers and Addresses, Up: Target Architecture Definition 2979 2980 11.5 Address Classes 2981 ==================== 2982 2983 Sometimes information about different kinds of addresses is available 2984 via the debug information. For example, some programming environments 2985 define addresses of several different sizes. If the debug information 2986 distinguishes these kinds of address classes through either the size 2987 info (e.g, `DW_AT_byte_size' in DWARF 2) or through an explicit address 2988 class attribute (e.g, `DW_AT_address_class' in DWARF 2), the following 2989 macros should be defined in order to disambiguate these types within 2990 GDB as well as provide the added information to a GDB user when 2991 printing type expressions. 2992 2993 -- Function: int gdbarch_address_class_type_flags (struct gdbarch 2994 *GDBARCH, int BYTE_SIZE, int DWARF2_ADDR_CLASS) 2995 Returns the type flags needed to construct a pointer type whose 2996 size is BYTE_SIZE and whose address class is DWARF2_ADDR_CLASS. 2997 This function is normally called from within a symbol reader. See 2998 `dwarf2read.c'. 2999 3000 -- Function: char * gdbarch_address_class_type_flags_to_name (struct 3001 gdbarch *GDBARCH, int TYPE_FLAGS) 3002 Given the type flags representing an address class qualifier, 3003 return its name. 3004 3005 -- Function: int gdbarch_address_class_name_to_type_flags (struct 3006 gdbarch *GDBARCH, int NAME, int *TYPE_FLAGS_PTR) 3007 Given an address qualifier name, set the `int' referenced by 3008 TYPE_FLAGS_PTR to the type flags for that address class qualifier. 3009 3010 Since the need for address classes is rather rare, none of the 3011 address class functions are defined by default. Predicate functions 3012 are provided to detect when they are defined. 3013 3014 Consider a hypothetical architecture in which addresses are normally 3015 32-bits wide, but 16-bit addresses are also supported. Furthermore, 3016 suppose that the DWARF 2 information for this architecture simply uses 3017 a `DW_AT_byte_size' value of 2 to indicate the use of one of these 3018 "short" pointers. The following functions could be defined to 3019 implement the address class functions: 3020 3021 somearch_address_class_type_flags (int byte_size, 3022 int dwarf2_addr_class) 3023 { 3024 if (byte_size == 2) 3025 return TYPE_FLAG_ADDRESS_CLASS_1; 3026 else 3027 return 0; 3028 } 3029 3030 static char * 3031 somearch_address_class_type_flags_to_name (int type_flags) 3032 { 3033 if (type_flags & TYPE_FLAG_ADDRESS_CLASS_1) 3034 return "short"; 3035 else 3036 return NULL; 3037 } 3038 3039 int 3040 somearch_address_class_name_to_type_flags (char *name, 3041 int *type_flags_ptr) 3042 { 3043 if (strcmp (name, "short") == 0) 3044 { 3045 *type_flags_ptr = TYPE_FLAG_ADDRESS_CLASS_1; 3046 return 1; 3047 } 3048 else 3049 return 0; 3050 } 3051 3052 The qualifier `@short' is used in GDB's type expressions to indicate 3053 the presence of one of these "short" pointers. For example if the 3054 debug information indicates that `short_ptr_var' is one of these short 3055 pointers, GDB might show the following behavior: 3056 3057 (gdb) ptype short_ptr_var 3058 type = int * @short 3059 3060 3061 File: gdbint.info, Node: Register Representation, Next: Frame Interpretation, Prev: Address Classes, Up: Target Architecture Definition 3062 3063 11.6 Register Representation 3064 ============================ 3065 3066 * Menu: 3067 3068 * Raw and Cooked Registers:: 3069 * Register Architecture Functions & Variables:: 3070 * Register Information Functions:: 3071 * Register and Memory Data:: 3072 * Register Caching:: 3073 3074 3075 File: gdbint.info, Node: Raw and Cooked Registers, Next: Register Architecture Functions & Variables, Up: Register Representation 3076 3077 11.6.1 Raw and Cooked Registers 3078 ------------------------------- 3079 3080 GDB considers registers to be a set with members numbered linearly from 3081 0 upwards. The first part of that set corresponds to real physical 3082 registers, the second part to any "pseudo-registers". Pseudo-registers 3083 have no independent physical existence, but are useful representations 3084 of information within the architecture. For example the OpenRISC 1000 3085 architecture has up to 32 general purpose registers, which are 3086 typically represented as 32-bit (or 64-bit) integers. However the GPRs 3087 are also used as operands to the floating point operations, and it 3088 could be convenient to define a set of pseudo-registers, to show the 3089 GPRs represented as floating point values. 3090 3091 For any architecture, the implementer will decide on a mapping from 3092 hardware to GDB register numbers. The registers corresponding to real 3093 hardware are referred to as "raw" registers, the remaining registers are 3094 "pseudo-registers". The total register set (raw and pseudo) is called 3095 the "cooked" register set. 3096 3097 3098 File: gdbint.info, Node: Register Architecture Functions & Variables, Next: Register Information Functions, Prev: Raw and Cooked Registers, Up: Register Representation 3099 3100 11.6.2 Functions and Variables Specifying the Register Architecture 3101 ------------------------------------------------------------------- 3102 3103 These `struct gdbarch' functions and variables specify the number and 3104 type of registers in the architecture. 3105 3106 -- Architecture Function: CORE_ADDR read_pc (struct regcache *REGCACHE) 3107 3108 -- Architecture Function: void write_pc (struct regcache *REGCACHE, 3109 CORE_ADDR VAL) 3110 Read or write the program counter. The default value of both 3111 functions is `NULL' (no function available). If the program 3112 counter is just an ordinary register, it can be specified in 3113 `struct gdbarch' instead (see `pc_regnum' below) and it will be 3114 read or written using the standard routines to access registers. 3115 This function need only be specified if the program counter is not 3116 an ordinary register. 3117 3118 Any register information can be obtained using the supplied 3119 register cache, REGCACHE. *Note Register Caching: Register 3120 Caching. 3121 3122 3123 -- Architecture Function: void pseudo_register_read (struct gdbarch 3124 *GDBARCH, struct regcache *REGCACHE, int REGNUM, const 3125 gdb_byte *BUF) 3126 3127 -- Architecture Function: void pseudo_register_write (struct gdbarch 3128 *GDBARCH, struct regcache *REGCACHE, int REGNUM, const 3129 gdb_byte *BUF) 3130 These functions should be defined if there are any 3131 pseudo-registers. The default value is `NULL'. REGNUM is the 3132 number of the register to read or write (which will be a "cooked" 3133 register number) and BUF is the buffer where the value read will be 3134 placed, or from which the value to be written will be taken. The 3135 value in the buffer may be converted to or from a signed or 3136 unsigned integral value using one of the utility functions (*note 3137 Using Different Register and Memory Data Representations: Register 3138 and Memory Data.). 3139 3140 The access should be for the specified architecture, GDBARCH. Any 3141 register information can be obtained using the supplied register 3142 cache, REGCACHE. *Note Register Caching: Register Caching. 3143 3144 3145 -- Architecture Variable: int sp_regnum 3146 This specifies the register holding the stack pointer, which may 3147 be a raw or pseudo-register. It defaults to -1 (not defined), but 3148 it is an error for it not to be defined. 3149 3150 The value of the stack pointer register can be accessed withing 3151 GDB as the variable `$sp'. 3152 3153 3154 -- Architecture Variable: int pc_regnum 3155 This specifies the register holding the program counter, which may 3156 be a raw or pseudo-register. It defaults to -1 (not defined). If 3157 `pc_regnum' is not defined, then the functions `read_pc' and 3158 `write_pc' (see above) must be defined. 3159 3160 The value of the program counter (whether defined as a register, or 3161 through `read_pc' and `write_pc') can be accessed withing GDB as 3162 the variable `$pc'. 3163 3164 3165 -- Architecture Variable: int ps_regnum 3166 This specifies the register holding the processor status (often 3167 called the status register), which may be a raw or 3168 pseudo-register. It defaults to -1 (not defined). 3169 3170 If defined, the value of this register can be accessed withing GDB 3171 as the variable `$ps'. 3172 3173 3174 -- Architecture Variable: int fp0_regnum 3175 This specifies the first floating point register. It defaults to 3176 0. `fp0_regnum' is not needed unless the target offers support 3177 for floating point. 3178 3179 3180 3181 File: gdbint.info, Node: Register Information Functions, Next: Register and Memory Data, Prev: Register Architecture Functions & Variables, Up: Register Representation 3182 3183 11.6.3 Functions Giving Register Information 3184 -------------------------------------------- 3185 3186 These functions return information about registers. 3187 3188 -- Architecture Function: const char * register_name (struct gdbarch 3189 *GDBARCH, int REGNUM) 3190 This function should convert a register number (raw or pseudo) to a 3191 register name (as a C `const char *'). This is used both to 3192 determine the name of a register for output and to work out the 3193 meaning of any register names used as input. The function may 3194 also return `NULL', to indicate that REGNUM is not a valid 3195 register. 3196 3197 For example with the OpenRISC 1000, GDB registers 0-31 are the 3198 General Purpose Registers, register 32 is the program counter and 3199 register 33 is the supervision register (i.e. the processor status 3200 register), which map to the strings `"gpr00"' through `"gpr31"', 3201 `"pc"' and `"sr"' respectively. This means that the GDB command 3202 `print $gpr5' should print the value of the OR1K general purpose 3203 register 5(1). 3204 3205 The default value for this function is `NULL', meaning undefined. 3206 It should always be defined. 3207 3208 The access should be for the specified architecture, GDBARCH. 3209 3210 3211 -- Architecture Function: struct type * register_type (struct gdbarch 3212 *GDBARCH, int REGNUM) 3213 Given a register number, this function identifies the type of data 3214 it may be holding, specified as a `struct type'. GDB allows 3215 creation of arbitrary types, but a number of built in types are 3216 provided (`builtin_type_void', `builtin_type_int32' etc), together 3217 with functions to derive types from these. 3218 3219 Typically the program counter will have a type of "pointer to 3220 function" (it points to code), the frame pointer and stack pointer 3221 will have types of "pointer to void" (they point to data on the 3222 stack) and all other integer registers will have a type of 32-bit 3223 integer or 64-bit integer. 3224 3225 This information guides the formatting when displaying register 3226 information. The default value is `NULL' meaning no information is 3227 available to guide formatting when displaying registers. 3228 3229 3230 -- Architecture Function: void print_registers_info (struct gdbarch 3231 *GDBARCH, struct ui_file *FILE, struct frame_info *FRAME, int 3232 REGNUM, int ALL) 3233 Define this function to print out one or all of the registers for 3234 the GDB `info registers' command. The default value is the 3235 function `default_print_registers_info', which uses the register 3236 type information (see `register_type' above) to determine how each 3237 register should be printed. Define a custom version of this 3238 function for fuller control over how the registers are displayed. 3239 3240 The access should be for the specified architecture, GDBARCH, with 3241 output to the the file specified by the User Interface Independent 3242 Output file handle, FILE (*note UI-Independent Output--the 3243 `ui_out' Functions: UI-Independent Output.). 3244 3245 The registers should show their values in the frame specified by 3246 FRAME. If REGNUM is -1 and ALL is zero, then all the 3247 "significant" registers should be shown (the implementer should 3248 decide which registers are "significant"). Otherwise only the 3249 value of the register specified by REGNUM should be output. If 3250 REGNUM is -1 and ALL is non-zero (true), then the value of all 3251 registers should be shown. 3252 3253 By default `default_print_registers_info' prints one register per 3254 line, and if ALL is zero omits floating-point registers. 3255 3256 3257 -- Architecture Function: void print_float_info (struct gdbarch 3258 *GDBARCH, struct ui_file *FILE, struct frame_info *FRAME, 3259 const char *ARGS) 3260 Define this function to provide output about the floating point 3261 unit and registers for the GDB `info float' command respectively. 3262 The default value is `NULL' (not defined), meaning no information 3263 will be provided. 3264 3265 The GDBARCH and FILE and FRAME arguments have the same meaning as 3266 in the `print_registers_info' function above. The string ARGS 3267 contains any supplementary arguments to the `info float' command. 3268 3269 Define this function if the target supports floating point 3270 operations. 3271 3272 3273 -- Architecture Function: void print_vector_info (struct gdbarch 3274 *GDBARCH, struct ui_file *FILE, struct frame_info *FRAME, 3275 const char *ARGS) 3276 Define this function to provide output about the vector unit and 3277 registers for the GDB `info vector' command respectively. The 3278 default value is `NULL' (not defined), meaning no information will 3279 be provided. 3280 3281 The GDBARCH, FILE and FRAME arguments have the same meaning as in 3282 the `print_registers_info' function above. The string ARGS 3283 contains any supplementary arguments to the `info vector' command. 3284 3285 Define this function if the target supports vector operations. 3286 3287 3288 -- Architecture Function: int register_reggroup_p (struct gdbarch 3289 *GDBARCH, int REGNUM, struct reggroup *GROUP) 3290 GDB groups registers into different categories (general, vector, 3291 floating point etc). This function, given a register, REGNUM, and 3292 group, GROUP, returns 1 (true) if the register is in the group and 3293 0 (false) otherwise. 3294 3295 The information should be for the specified architecture, GDBARCH 3296 3297 The default value is the function `default_register_reggroup_p' 3298 which will do a reasonable job based on the type of the register 3299 (see the function `register_type' above), with groups for general 3300 purpose registers, floating point registers, vector registers and 3301 raw (i.e not pseudo) registers. 3302 3303 3304 ---------- Footnotes ---------- 3305 3306 (1) Historically, GDB always had a concept of a frame pointer 3307 register, which could be accessed via the GDB variable, `$fp'. That 3308 concept is now deprecated, recognizing that not all architectures have 3309 a frame pointer. However if an architecture does have a frame pointer 3310 register, and defines a register or pseudo-register with the name 3311 `"fp"', then that register will be used as the value of the `$fp' 3312 variable. 3313 3314 3315 File: gdbint.info, Node: Register and Memory Data, Next: Register Caching, Prev: Register Information Functions, Up: Register Representation 3316 3317 11.6.4 Using Different Register and Memory Data Representations 3318 --------------------------------------------------------------- 3319 3320 Some architectures have different representations of data objects, 3321 depending whether the object is held in a register or memory. For 3322 example: 3323 3324 * The Alpha architecture can represent 32 bit integer values in 3325 floating-point registers. 3326 3327 * The x86 architecture supports 80-bit floating-point registers. The 3328 `long double' data type occupies 96 bits in memory but only 80 3329 bits when stored in a register. 3330 3331 3332 In general, the register representation of a data type is determined 3333 by the architecture, or GDB's interface to the architecture, while the 3334 memory representation is determined by the Application Binary Interface. 3335 3336 For almost all data types on almost all architectures, the two 3337 representations are identical, and no special handling is needed. 3338 However, they do occasionally differ. An architecture may define the 3339 following `struct gdbarch' functions to request conversions between the 3340 register and memory representations of a data type: 3341 3342 -- Architecture Function: int gdbarch_convert_register_p (struct 3343 gdbarch *GDBARCH, int REG) 3344 Return non-zero (true) if the representation of a data value 3345 stored in this register may be different to the representation of 3346 that same data value when stored in memory. The default value is 3347 `NULL' (undefined). 3348 3349 If this function is defined and returns non-zero, the `struct 3350 gdbarch' functions `gdbarch_register_to_value' and 3351 `gdbarch_value_to_register' (see below) should be used to perform 3352 any necessary conversion. 3353 3354 If defined, this function should return zero for the register's 3355 native type, when no conversion is necessary. 3356 3357 -- Architecture Function: void gdbarch_register_to_value (struct 3358 gdbarch *GDBARCH, int REG, struct type *TYPE, char *FROM, 3359 char *TO) 3360 Convert the value of register number REG to a data object of type 3361 TYPE. The buffer at FROM holds the register's value in raw 3362 format; the converted value should be placed in the buffer at TO. 3363 3364 _Note:_ `gdbarch_register_to_value' and 3365 `gdbarch_value_to_register' take their REG and TYPE arguments 3366 in different orders. 3367 3368 `gdbarch_register_to_value' should only be used with registers for 3369 which the `gdbarch_convert_register_p' function returns a non-zero 3370 value. 3371 3372 3373 -- Architecture Function: void gdbarch_value_to_register (struct 3374 gdbarch *GDBARCH, struct type *TYPE, int REG, char *FROM, 3375 char *TO) 3376 Convert a data value of type TYPE to register number REG' raw 3377 format. 3378 3379 _Note:_ `gdbarch_register_to_value' and 3380 `gdbarch_value_to_register' take their REG and TYPE arguments 3381 in different orders. 3382 3383 `gdbarch_value_to_register' should only be used with registers for 3384 which the `gdbarch_convert_register_p' function returns a non-zero 3385 value. 3386 3387 3388 3389 File: gdbint.info, Node: Register Caching, Prev: Register and Memory Data, Up: Register Representation 3390 3391 11.6.5 Register Caching 3392 ----------------------- 3393 3394 Caching of registers is used, so that the target does not need to be 3395 accessed and reanalyzed multiple times for each register in 3396 circumstances where the register value cannot have changed. 3397 3398 GDB provides `struct regcache', associated with a particular `struct 3399 gdbarch' to hold the cached values of the raw registers. A set of 3400 functions is provided to access both the raw registers (with `raw' in 3401 their name) and the full set of cooked registers (with `cooked' in 3402 their name). Functions are provided to ensure the register cache is 3403 kept synchronized with the values of the actual registers in the target. 3404 3405 Accessing registers through the `struct regcache' routines will 3406 ensure that the appropriate `struct gdbarch' functions are called when 3407 necessary to access the underlying target architecture. In general 3408 users should use the "cooked" functions, since these will map to the 3409 "raw" functions automatically as appropriate. 3410 3411 The two key functions are `regcache_cooked_read' and 3412 `regcache_cooked_write' which read or write a register from or to a 3413 byte buffer (type `gdb_byte *'). For convenience the wrapper functions 3414 `regcache_cooked_read_signed', `regcache_cooked_read_unsigned', 3415 `regcache_cooked_write_signed' and `regcache_cooked_write_unsigned' are 3416 provided, which read or write the value using the buffer and convert to 3417 or from an integral value as appropriate. 3418 3419 3420 File: gdbint.info, Node: Frame Interpretation, Next: Inferior Call Setup, Prev: Register Representation, Up: Target Architecture Definition 3421 3422 11.7 Frame Interpretation 3423 ========================= 3424 3425 * Menu: 3426 3427 * All About Stack Frames:: 3428 * Frame Handling Terminology:: 3429 * Prologue Caches:: 3430 * Functions and Variable to Analyze Frames:: 3431 * Functions to Access Frame Data:: 3432 * Analyzing Stacks---Frame Sniffers:: 3433 3434 3435 File: gdbint.info, Node: All About Stack Frames, Next: Frame Handling Terminology, Up: Frame Interpretation 3436 3437 11.7.1 All About Stack Frames 3438 ----------------------------- 3439 3440 GDB needs to understand the stack on which local (automatic) variables 3441 are stored. The area of the stack containing all the local variables 3442 for a function invocation is known as the "stack frame" for that 3443 function (or colloquially just as the "frame"). In turn the function 3444 that called the function will have its stack frame, and so on back 3445 through the chain of functions that have been called. 3446 3447 Almost all architectures have one register dedicated to point to the 3448 end of the stack (the "stack pointer"). Many have a second register 3449 which points to the start of the currently active stack frame (the 3450 "frame pointer"). The specific arrangements for an architecture are a 3451 key part of the ABI. 3452 3453 A diagram helps to explain this. Here is a simple program to compute 3454 factorials: 3455 3456 #include <stdio.h> 3457 int fact (int n) 3458 { 3459 if (0 == n) 3460 { 3461 return 1; 3462 } 3463 else 3464 { 3465 return n * fact (n - 1); 3466 } 3467 } 3468 3469 main () 3470 { 3471 int i; 3472 3473 for (i = 0; i < 10; i++) 3474 { 3475 int f = fact (i); 3476 printf ("%d! = %d\n", i, f); 3477 } 3478 } 3479 3480 Consider the state of the stack when the code reaches line 6 after 3481 the main program has called `fact (3)'. The chain of function calls 3482 will be `main ()', `fact (3)', `fact (2)', `fact (1)' and `fact (0)'. 3483 3484 In this illustration the stack is falling (as used for example by the 3485 OpenRISC 1000 ABI). The stack pointer (SP) is at the end of the stack 3486 (lowest address) and the frame pointer (FP) is at the highest address 3487 in the current stack frame. The following diagram shows how the stack 3488 looks. 3489 3490 ^ ->| | 3491 Frame | | | | 3492 Number - | | |============| int fact (int n) 3493 | | | | i = 3 | { 3494 | | | |------------| if (0 == n) { 3495 | | | | f = ? | return 1; <-------- PC 3496 #4 main() < | | |------------| } 3497 | | | | | else { 3498 | | -+->|------------| ---> return n * fact (n - 1); 3499 | -+-+--+-----o | | } 3500 = | | |============| | } 3501 | | | | n = 3 | | 3502 | | | |------------| | main () 3503 #3 fact (3) < | | | o---------+- { 3504 | -+-+->|------------| | | int i; 3505 | | | --+-----o | | | 3506 = | | |============| | | for (i = 0; i < 10; i++) { 3507 | | | | n = 2 | | -> int f = fact (i); 3508 | | | |------------| | printf ("%d! = %d\n", i , f); 3509 #2 fact (2) < | | | o------+--| } 3510 | | | ->|------------| | } 3511 | | -+--+-----o | | 3512 = | | |============| | 3513 | | | | n = 1 | | 3514 | | | |------------| | 3515 #1 fact (1) < | | | o------+--| 3516 | | | |------------| | 3517 | ---|--+-----o |<-+------- FP 3518 = | |============| | | 3519 | | | n = 0 | | | 3520 | | |------------| | | 3521 #0 fact (0) < | | o--------- | 3522 | | |------------| | 3523 | --+-----o |<--------- SP | 3524 = |============| | 3525 | | Red Zone | v 3526 | \/\/\/\/\/\/\/ Direction of 3527 #-1 < \/\/\/\/\/\/\/ stack growth 3528 | | | 3529 3530 In each stack frame, offset 0 from the stack pointer is the frame 3531 pointer of the previous frame and offset 4 (this is illustrating a 3532 32-bit architecture) from the stack pointer is the return address. 3533 Local variables are indexed from the frame pointer, with negative 3534 indexes. In the function `fact', offset -4 from the frame pointer is 3535 the argument N. In the `main' function, offset -4 from the frame 3536 pointer is the local variable I and offset -8 from the frame pointer is 3537 the local variable F(1). 3538 3539 It is very easy to get confused when examining stacks. GDB has 3540 terminology it uses rigorously throughout. The stack frame of the 3541 function currently executing, or where execution stopped is numbered 3542 zero. In this example frame #0 is the stack frame of the call to 3543 `fact (0)'. The stack frame of its calling function (`fact (1)' in 3544 this case) is numbered #1 and so on back through the chain of calls. 3545 3546 The main GDB data structure describing frames is 3547 `struct frame_info'. It is not used directly, but only via its 3548 accessor functions. `frame_info' includes information about the 3549 registers in the frame and a pointer to the code of the function with 3550 which the frame is associated. The entire stack is represented as a 3551 linked list of `frame_info' structs. 3552 3553 ---------- Footnotes ---------- 3554 3555 (1) This is a simplified example for illustrative purposes only. 3556 Good optimizing compilers would not put anything on the stack for such 3557 simple functions. Indeed they might eliminate the recursion and use of 3558 the stack entirely! 3559 3560 3561 File: gdbint.info, Node: Frame Handling Terminology, Next: Prologue Caches, Prev: All About Stack Frames, Up: Frame Interpretation 3562 3563 11.7.2 Frame Handling Terminology 3564 --------------------------------- 3565 3566 It is easy to get confused when referencing stack frames. GDB uses 3567 some precise terminology. 3568 3569 * "THIS" frame is the frame currently under consideration. 3570 3571 * The "NEXT" frame, also sometimes called the inner or newer frame 3572 is the frame of the function called by the function of THIS frame. 3573 3574 * The "PREVIOUS" frame, also sometimes called the outer or older 3575 frame is the frame of the function which called the function of 3576 THIS frame. 3577 3578 3579 So in the example in the previous section (*note All About Stack 3580 Frames: All About Stack Frames.), if THIS frame is #3 (the call to 3581 `fact (3)'), the NEXT frame is frame #2 (the call to `fact (2)') and 3582 the PREVIOUS frame is frame #4 (the call to `main ()'). 3583 3584 The "innermost" frame is the frame of the current executing 3585 function, or where the program stopped, in this example, in the middle 3586 of the call to `fact (0))'. It is always numbered frame #0. 3587 3588 The "base" of a frame is the address immediately before the start of 3589 the NEXT frame. For a stack which grows down in memory (a "falling" 3590 stack) this will be the lowest address and for a stack which grows up 3591 in memory (a "rising" stack) this will be the highest address in the 3592 frame. 3593 3594 GDB functions to analyze the stack are typically given a pointer to 3595 the NEXT frame to determine information about THIS frame. Information 3596 about THIS frame includes data on where the registers of the PREVIOUS 3597 frame are stored in this stack frame. In this example the frame 3598 pointer of the PREVIOUS frame is stored at offset 0 from the stack 3599 pointer of THIS frame. 3600 3601 The process whereby a function is given a pointer to the NEXT frame 3602 to work out information about THIS frame is referred to as "unwinding". 3603 The GDB functions involved in this typically include unwind in their 3604 name. 3605 3606 The process of analyzing a target to determine the information that 3607 should go in struct frame_info is called "sniffing". The functions 3608 that carry this out are called sniffers and typically include sniffer 3609 in their name. More than one sniffer may be required to extract all 3610 the information for a particular frame. 3611 3612 Because so many functions work using the NEXT frame, there is an 3613 issue about addressing the innermost frame--it has no NEXT frame. To 3614 solve this GDB creates a dummy frame #-1, known as the "sentinel" frame. 3615 3616 3617 File: gdbint.info, Node: Prologue Caches, Next: Functions and Variable to Analyze Frames, Prev: Frame Handling Terminology, Up: Frame Interpretation 3618 3619 11.7.3 Prologue Caches 3620 ---------------------- 3621 3622 All the frame sniffing functions typically examine the code at the 3623 start of the corresponding function, to determine the state of 3624 registers. The ABI will save old values and set new values of key 3625 registers at the start of each function in what is known as the 3626 function "prologue". 3627 3628 For any particular stack frame this data does not change, so all the 3629 standard unwinding functions, in addition to receiving a pointer to the 3630 NEXT frame as their first argument, receive a pointer to a "prologue 3631 cache" as their second argument. This can be used to store values 3632 associated with a particular frame, for reuse on subsequent calls 3633 involving the same frame. 3634 3635 It is up to the user to define the structure used (it is a `void *' 3636 pointer) and arrange allocation and deallocation of storage. However 3637 for general use, GDB provides `struct trad_frame_cache', with a set of 3638 accessor routines. This structure holds the stack and code address of 3639 THIS frame, the base address of the frame, a pointer to the struct 3640 `frame_info' for the NEXT frame and details of where the registers of 3641 the PREVIOUS frame may be found in THIS frame. 3642 3643 Typically the first time any sniffer function is called with NEXT 3644 frame, the prologue sniffer for THIS frame will be `NULL'. The sniffer 3645 will analyze the frame, allocate a prologue cache structure and 3646 populate it. Subsequent calls using the same NEXT frame will pass in 3647 this prologue cache, so the data can be returned with no additional 3648 analysis. 3649 3650 3651 File: gdbint.info, Node: Functions and Variable to Analyze Frames, Next: Functions to Access Frame Data, Prev: Prologue Caches, Up: Frame Interpretation 3652 3653 11.7.4 Functions and Variable to Analyze Frames 3654 ----------------------------------------------- 3655 3656 These struct `gdbarch' functions and variable should be defined to 3657 provide analysis of the stack frame and allow it to be adjusted as 3658 required. 3659 3660 -- Architecture Function: CORE_ADDR skip_prologue (struct gdbarch 3661 *GDBARCH, CORE_ADDR PC) 3662 The prologue of a function is the code at the beginning of the 3663 function which sets up the stack frame, saves the return address 3664 etc. The code representing the behavior of the function starts 3665 after the prologue. 3666 3667 This function skips past the prologue of a function if the program 3668 counter, PC, is within the prologue of a function. The result is 3669 the program counter immediately after the prologue. With modern 3670 optimizing compilers, this may be a far from trivial exercise. 3671 However the required information may be within the binary as 3672 DWARF2 debugging information, making the job much easier. 3673 3674 The default value is `NULL' (not defined). This function should 3675 always be provided, but can take advantage of DWARF2 debugging 3676 information, if that is available. 3677 3678 3679 -- Architecture Function: int inner_than (CORE_ADDR LHS, CORE_ADDR RHS) 3680 Given two frame or stack pointers, return non-zero (true) if the 3681 first represents the "inner" stack frame and 0 (false) otherwise. 3682 This is used to determine whether the target has a stack which 3683 grows up in memory (rising stack) or grows down in memory (falling 3684 stack). *Note All About Stack Frames: All About Stack Frames, for 3685 an explanation of "inner" frames. 3686 3687 The default value of this function is `NULL' and it should always 3688 be defined. However for almost all architectures one of the 3689 built-in functions can be used: `core_addr_lessthan' (for stacks 3690 growing down in memory) or `core_addr_greaterthan' (for stacks 3691 growing up in memory). 3692 3693 3694 -- Architecture Function: CORE_ADDR frame_align (struct gdbarch 3695 *GDBARCH, CORE_ADDR ADDRESS) 3696 The architecture may have constraints on how its frames are 3697 aligned. For example the OpenRISC 1000 ABI requires stack frames 3698 to be double-word aligned, but 32-bit versions of the architecture 3699 allocate single-word values to the stack. Thus extra padding may 3700 be needed at the end of a stack frame. 3701 3702 Given a proposed address for the stack pointer, this function 3703 returns a suitably aligned address (by expanding the stack frame). 3704 3705 The default value is `NULL' (undefined). This function should be 3706 defined for any architecture where it is possible the stack could 3707 become misaligned. The utility functions `align_down' (for falling 3708 stacks) and `align_up' (for rising stacks) will facilitate the 3709 implementation of this function. 3710 3711 3712 -- Architecture Variable: int frame_red_zone_size 3713 Some ABIs reserve space beyond the end of the stack for use by leaf 3714 functions without prologue or epilogue or by exception handlers 3715 (for example the OpenRISC 1000). 3716 3717 This is known as a "red zone" (AMD terminology). The AMD64 (nee 3718 x86-64) ABI documentation refers to the "red zone" when describing 3719 this scratch area. 3720 3721 The default value is 0. Set this field if the architecture has 3722 such a red zone. The value must be aligned as required by the ABI 3723 (see `frame_align' above for an explanation of stack frame 3724 alignment). 3725 3726 3727 3728 File: gdbint.info, Node: Functions to Access Frame Data, Next: Analyzing Stacks---Frame Sniffers, Prev: Functions and Variable to Analyze Frames, Up: Frame Interpretation 3729 3730 11.7.5 Functions to Access Frame Data 3731 ------------------------------------- 3732 3733 These functions provide access to key registers and arguments in the 3734 stack frame. 3735 3736 -- Architecture Function: CORE_ADDR unwind_pc (struct gdbarch 3737 *GDBARCH, struct frame_info *NEXT_FRAME) 3738 This function is given a pointer to the NEXT stack frame (*note 3739 All About Stack Frames: All About Stack Frames, for how frames are 3740 represented) and returns the value of the program counter in the 3741 PREVIOUS frame (i.e. the frame of the function that called THIS 3742 one). This is commonly referred to as the "return address". 3743 3744 The implementation, which must be frame agnostic (work with any 3745 frame), is typically no more than: 3746 3747 ULONGEST pc; 3748 pc = frame_unwind_register_unsigned (next_frame, ARCH_PC_REGNUM); 3749 return gdbarch_addr_bits_remove (gdbarch, pc); 3750 3751 3752 -- Architecture Function: CORE_ADDR unwind_sp (struct gdbarch 3753 *GDBARCH, struct frame_info *NEXT_FRAME) 3754 This function is given a pointer to the NEXT stack frame (*note 3755 All About Stack Frames: All About Stack Frames. for how frames are 3756 represented) and returns the value of the stack pointer in the 3757 PREVIOUS frame (i.e. the frame of the function that called THIS 3758 one). 3759 3760 The implementation, which must be frame agnostic (work with any 3761 frame), is typically no more than: 3762 3763 ULONGEST sp; 3764 sp = frame_unwind_register_unsigned (next_frame, ARCH_SP_REGNUM); 3765 return gdbarch_addr_bits_remove (gdbarch, sp); 3766 3767 3768 -- Architecture Function: int frame_num_args (struct gdbarch *GDBARCH, 3769 struct frame_info *THIS_FRAME) 3770 This function is given a pointer to THIS stack frame (*note All 3771 About Stack Frames: All About Stack Frames. for how frames are 3772 represented), and returns the number of arguments that are being 3773 passed, or -1 if not known. 3774 3775 The default value is `NULL' (undefined), in which case the number 3776 of arguments passed on any stack frame is always unknown. For many 3777 architectures this will be a suitable default. 3778 3779 3780 3781 File: gdbint.info, Node: Analyzing Stacks---Frame Sniffers, Prev: Functions to Access Frame Data, Up: Frame Interpretation 3782 3783 11.7.6 Analyzing Stacks--Frame Sniffers 3784 --------------------------------------- 3785 3786 When a program stops, GDB needs to construct the chain of struct 3787 `frame_info' representing the state of the stack using appropriate 3788 "sniffers". 3789 3790 Each architecture requires appropriate sniffers, but they do not form 3791 entries in `struct gdbarch', since more than one sniffer may be 3792 required and a sniffer may be suitable for more than one 3793 `struct gdbarch'. Instead sniffers are associated with architectures 3794 using the following functions. 3795 3796 * `frame_unwind_append_sniffer' is used to add a new sniffer to 3797 analyze THIS frame when given a pointer to the NEXT frame. 3798 3799 * `frame_base_append_sniffer' is used to add a new sniffer which can 3800 determine information about the base of a stack frame. 3801 3802 * `frame_base_set_default' is used to specify the default base 3803 sniffer. 3804 3805 3806 These functions all take a reference to `struct gdbarch', so they 3807 are associated with a specific architecture. They are usually called 3808 in the `gdbarch' initialization function, after the `gdbarch' struct 3809 has been set up. Unless a default has been set, the most recently 3810 appended sniffer will be tried first. 3811 3812 The main frame unwinding sniffer (as set by 3813 `frame_unwind_append_sniffer)' returns a structure specifying a set of 3814 sniffing functions: 3815 3816 struct frame_unwind 3817 { 3818 enum frame_type type; 3819 frame_this_id_ftype *this_id; 3820 frame_prev_register_ftype *prev_register; 3821 const struct frame_data *unwind_data; 3822 frame_sniffer_ftype *sniffer; 3823 frame_prev_pc_ftype *prev_pc; 3824 frame_dealloc_cache_ftype *dealloc_cache; 3825 }; 3826 3827 The `type' field indicates the type of frame this sniffer can 3828 handle: normal, dummy (*note Functions Creating Dummy Frames: Functions 3829 Creating Dummy Frames.), signal handler or sentinel. Signal handlers 3830 sometimes have their own simplified stack structure for efficiency, so 3831 may need their own handlers. 3832 3833 The `unwind_data' field holds additional information which may be 3834 relevant to particular types of frame. For example it may hold 3835 additional information for signal handler frames. 3836 3837 The remaining fields define functions that yield different types of 3838 information when given a pointer to the NEXT stack frame. Not all 3839 functions need be provided. If an entry is `NULL', the next sniffer 3840 will be tried instead. 3841 3842 * `this_id' determines the stack pointer and function (code entry 3843 point) for THIS stack frame. 3844 3845 * `prev_register' determines where the values of registers for the 3846 PREVIOUS stack frame are stored in THIS stack frame. 3847 3848 * `sniffer' takes a look at THIS frame's registers to determine if 3849 this is the appropriate unwinder. 3850 3851 * `prev_pc' determines the program counter for THIS frame. Only 3852 needed if the program counter is not an ordinary register (*note 3853 Functions and Variables Specifying the Register Architecture: 3854 Register Architecture Functions & Variables.). 3855 3856 * `dealloc_cache' frees any additional memory associated with the 3857 prologue cache for this frame (*note Prologue Caches: Prologue 3858 Caches.). 3859 3860 3861 In general it is only the `this_id' and `prev_register' fields that 3862 need be defined for custom sniffers. 3863 3864 The frame base sniffer is much simpler. It is a 3865 `struct frame_base', which refers to the corresponding `frame_unwind' 3866 struct and whose fields refer to functions yielding various addresses 3867 within the frame. 3868 3869 struct frame_base 3870 { 3871 const struct frame_unwind *unwind; 3872 frame_this_base_ftype *this_base; 3873 frame_this_locals_ftype *this_locals; 3874 frame_this_args_ftype *this_args; 3875 }; 3876 3877 All the functions referred to take a pointer to the NEXT frame as 3878 argument. The function referred to by `this_base' returns the base 3879 address of THIS frame, the function referred to by `this_locals' 3880 returns the base address of local variables in THIS frame and the 3881 function referred to by `this_args' returns the base address of the 3882 function arguments in this frame. 3883 3884 As described above, the base address of a frame is the address 3885 immediately before the start of the NEXT frame. For a falling stack, 3886 this is the lowest address in the frame and for a rising stack it is 3887 the highest address in the frame. For most architectures the same 3888 address is also the base address for local variables and arguments, in 3889 which case the same function can be used for all three entries(1). 3890 3891 ---------- Footnotes ---------- 3892 3893 (1) It is worth noting that if it cannot be determined in any other 3894 way (for example by there being a register with the name `"fp"'), then 3895 the result of the `this_base' function will be used as the value of the 3896 frame pointer variable `$fp' in GDB. This is very often not correct 3897 (for example with the OpenRISC 1000, this value is the stack pointer, 3898 `$sp'). In this case a register (raw or pseudo) with the name `"fp"' 3899 should be defined. It will be used in preference as the value of `$fp'. 3900 3901 3902 File: gdbint.info, Node: Inferior Call Setup, Next: Adding support for debugging core files, Prev: Frame Interpretation, Up: Target Architecture Definition 3903 3904 11.8 Inferior Call Setup 3905 ======================== 3906 3907 * Menu: 3908 3909 * About Dummy Frames:: 3910 * Functions Creating Dummy Frames:: 3911 3912 3913 File: gdbint.info, Node: About Dummy Frames, Next: Functions Creating Dummy Frames, Up: Inferior Call Setup 3914 3915 11.8.1 About Dummy Frames 3916 ------------------------- 3917 3918 GDB can call functions in the target code (for example by using the 3919 `call' or `print' commands). These functions may be breakpointed, and 3920 it is essential that if a function does hit a breakpoint, commands like 3921 `backtrace' work correctly. 3922 3923 This is achieved by making the stack look as though the function had 3924 been called from the point where GDB had previously stopped. This 3925 requires that GDB can set up stack frames appropriate for such function 3926 calls. 3927 3928 3929 File: gdbint.info, Node: Functions Creating Dummy Frames, Prev: About Dummy Frames, Up: Inferior Call Setup 3930 3931 11.8.2 Functions Creating Dummy Frames 3932 -------------------------------------- 3933 3934 The following functions provide the functionality to set up such 3935 "dummy" stack frames. 3936 3937 -- Architecture Function: CORE_ADDR push_dummy_call (struct gdbarch 3938 *GDBARCH, struct value *FUNCTION, struct regcache *REGCACHE, 3939 CORE_ADDR BP_ADDR, int NARGS, struct value **ARGS, CORE_ADDR 3940 SP, int STRUCT_RETURN, CORE_ADDR STRUCT_ADDR) 3941 This function sets up a dummy stack frame for the function about 3942 to be called. `push_dummy_call' is given the arguments to be 3943 passed and must copy them into registers or push them on to the 3944 stack as appropriate for the ABI. 3945 3946 FUNCTION is a pointer to the function that will be called and 3947 REGCACHE the register cache from which values should be obtained. 3948 BP_ADDR is the address to which the function should return (which 3949 is breakpointed, so GDB can regain control, hence the name). 3950 NARGS is the number of arguments to pass and ARGS an array 3951 containing the argument values. STRUCT_RETURN is non-zero (true) 3952 if the function returns a structure, and if so STRUCT_ADDR is the 3953 address in which the structure should be returned. 3954 3955 After calling this function, GDB will pass control to the target 3956 at the address of the function, which will find the stack and 3957 registers set up just as expected. 3958 3959 The default value of this function is `NULL' (undefined). If the 3960 function is not defined, then GDB will not allow the user to call 3961 functions within the target being debugged. 3962 3963 3964 -- Architecture Function: struct frame_id unwind_dummy_id (struct 3965 gdbarch *GDBARCH, struct frame_info *NEXT_FRAME) 3966 This is the inverse of `push_dummy_call' which restores the stack 3967 pointer and program counter after a call to evaluate a function 3968 using a dummy stack frame. The result is a `struct frame_id', 3969 which contains the value of the stack pointer and program counter 3970 to be used. 3971 3972 The NEXT frame pointer is provided as argument, NEXT_FRAME. THIS 3973 frame is the frame of the dummy function, which can be unwound, to 3974 yield the required stack pointer and program counter from the 3975 PREVIOUS frame. 3976 3977 The default value is `NULL' (undefined). If `push_dummy_call' is 3978 defined, then this function should also be defined. 3979 3980 3981 -- Architecture Function: CORE_ADDR push_dummy_code (struct gdbarch 3982 *GDBARCH, CORE_ADDR SP, CORE_ADDR FUNADDR, struct value 3983 **ARGS, int NARGS, struct type *VALUE_TYPE, CORE_ADDR 3984 *REAL_PC, CORE_ADDR *BP_ADDR, struct regcache *REGCACHE) 3985 If this function is not defined (its default value is `NULL'), a 3986 dummy call will use the entry point of the currently loaded code 3987 on the target as its return address. A temporary breakpoint will 3988 be set there, so the location must be writable and have room for a 3989 breakpoint. 3990 3991 It is possible that this default is not suitable. It might not be 3992 writable (in ROM possibly), or the ABI might require code to be 3993 executed on return from a call to unwind the stack before the 3994 breakpoint is encountered. 3995 3996 If either of these is the case, then push_dummy_code should be 3997 defined to push an instruction sequence onto the end of the stack 3998 to which the dummy call should return. 3999 4000 The arguments are essentially the same as those to 4001 `push_dummy_call'. However the function is provided with the type 4002 of the function result, VALUE_TYPE, BP_ADDR is used to return a 4003 value (the address at which the breakpoint instruction should be 4004 inserted) and REAL PC is used to specify the resume address when 4005 starting the call sequence. The function should return the 4006 updated innermost stack address. 4007 4008 _Note:_ This does require that code in the stack can be 4009 executed. Some Harvard architectures may not allow this. 4010 4011 4012 4013 File: gdbint.info, Node: Adding support for debugging core files, Next: Defining Other Architecture Features, Prev: Inferior Call Setup, Up: Target Architecture Definition 4014 4015 11.9 Adding support for debugging core files 4016 ============================================ 4017 4018 The prerequisite for adding core file support in GDB is to have core 4019 file support in BFD. 4020 4021 Once BFD support is available, writing the apropriate 4022 `regset_from_core_section' architecture function should be all that is 4023 needed in order to add support for core files in GDB. 4024 4025 4026 File: gdbint.info, Node: Defining Other Architecture Features, Next: Adding a New Target, Prev: Adding support for debugging core files, Up: Target Architecture Definition 4027 4028 11.10 Defining Other Architecture Features 4029 ========================================== 4030 4031 This section describes other functions and values in `gdbarch', 4032 together with some useful macros, that you can use to define the target 4033 architecture. 4034 4035 `CORE_ADDR gdbarch_addr_bits_remove (GDBARCH, ADDR)' 4036 If a raw machine instruction address includes any bits that are not 4037 really part of the address, then this function is used to zero 4038 those bits in ADDR. This is only used for addresses of 4039 instructions, and even then not in all contexts. 4040 4041 For example, the two low-order bits of the PC on the 4042 Hewlett-Packard PA 2.0 architecture contain the privilege level of 4043 the corresponding instruction. Since instructions must always be 4044 aligned on four-byte boundaries, the processor masks out these 4045 bits to generate the actual address of the instruction. 4046 `gdbarch_addr_bits_remove' would then for example look like that: 4047 arch_addr_bits_remove (CORE_ADDR addr) 4048 { 4049 return (addr &= ~0x3); 4050 } 4051 4052 `int address_class_name_to_type_flags (GDBARCH, NAME, TYPE_FLAGS_PTR)' 4053 If NAME is a valid address class qualifier name, set the `int' 4054 referenced by TYPE_FLAGS_PTR to the mask representing the qualifier 4055 and return 1. If NAME is not a valid address class qualifier name, 4056 return 0. 4057 4058 The value for TYPE_FLAGS_PTR should be one of 4059 `TYPE_FLAG_ADDRESS_CLASS_1', `TYPE_FLAG_ADDRESS_CLASS_2', or 4060 possibly some combination of these values or'd together. *Note 4061 Address Classes: Target Architecture Definition. 4062 4063 `int address_class_name_to_type_flags_p (GDBARCH)' 4064 Predicate which indicates whether 4065 `address_class_name_to_type_flags' has been defined. 4066 4067 `int gdbarch_address_class_type_flags (GDBARCH, BYTE_SIZE, DWARF2_ADDR_CLASS)' 4068 Given a pointers byte size (as described by the debug information) 4069 and the possible `DW_AT_address_class' value, return the type flags 4070 used by GDB to represent this address class. The value returned 4071 should be one of `TYPE_FLAG_ADDRESS_CLASS_1', 4072 `TYPE_FLAG_ADDRESS_CLASS_2', or possibly some combination of these 4073 values or'd together. *Note Address Classes: Target Architecture 4074 Definition. 4075 4076 `int gdbarch_address_class_type_flags_p (GDBARCH)' 4077 Predicate which indicates whether 4078 `gdbarch_address_class_type_flags_p' has been defined. 4079 4080 `const char *gdbarch_address_class_type_flags_to_name (GDBARCH, TYPE_FLAGS)' 4081 Return the name of the address class qualifier associated with the 4082 type flags given by TYPE_FLAGS. 4083 4084 `int gdbarch_address_class_type_flags_to_name_p (GDBARCH)' 4085 Predicate which indicates whether 4086 `gdbarch_address_class_type_flags_to_name' has been defined. 4087 *Note Address Classes: Target Architecture Definition. 4088 4089 `void gdbarch_address_to_pointer (GDBARCH, TYPE, BUF, ADDR)' 4090 Store in BUF a pointer of type TYPE representing the address ADDR, 4091 in the appropriate format for the current architecture. This 4092 function may safely assume that TYPE is either a pointer or a C++ 4093 reference type. *Note Pointers Are Not Always Addresses: Target 4094 Architecture Definition. 4095 4096 `int gdbarch_believe_pcc_promotion (GDBARCH)' 4097 Used to notify if the compiler promotes a `short' or `char' 4098 parameter to an `int', but still reports the parameter as its 4099 original type, rather than the promoted type. 4100 4101 `gdbarch_bits_big_endian (GDBARCH)' 4102 This is used if the numbering of bits in the targets does *not* 4103 match the endianism of the target byte order. A value of 1 means 4104 that the bits are numbered in a big-endian bit order, 0 means 4105 little-endian. 4106 4107 `set_gdbarch_bits_big_endian (GDBARCH, BITS_BIG_ENDIAN)' 4108 Calling set_gdbarch_bits_big_endian with a value of 1 indicates 4109 that the bits in the target are numbered in a big-endian bit 4110 order, 0 indicates little-endian. 4111 4112 `BREAKPOINT' 4113 This is the character array initializer for the bit pattern to put 4114 into memory where a breakpoint is set. Although it's common to 4115 use a trap instruction for a breakpoint, it's not required; for 4116 instance, the bit pattern could be an invalid instruction. The 4117 breakpoint must be no longer than the shortest instruction of the 4118 architecture. 4119 4120 `BREAKPOINT' has been deprecated in favor of 4121 `gdbarch_breakpoint_from_pc'. 4122 4123 `BIG_BREAKPOINT' 4124 `LITTLE_BREAKPOINT' 4125 Similar to BREAKPOINT, but used for bi-endian targets. 4126 4127 `BIG_BREAKPOINT' and `LITTLE_BREAKPOINT' have been deprecated in 4128 favor of `gdbarch_breakpoint_from_pc'. 4129 4130 `const gdb_byte *gdbarch_breakpoint_from_pc (GDBARCH, PCPTR, LENPTR)' 4131 Use the program counter to determine the contents and size of a 4132 breakpoint instruction. It returns a pointer to a static string 4133 of bytes that encode a breakpoint instruction, stores the length 4134 of the string to `*LENPTR', and adjusts the program counter (if 4135 necessary) to point to the actual memory location where the 4136 breakpoint should be inserted. May return `NULL' to indicate that 4137 software breakpoints are not supported. 4138 4139 Although it is common to use a trap instruction for a breakpoint, 4140 it's not required; for instance, the bit pattern could be an 4141 invalid instruction. The breakpoint must be no longer than the 4142 shortest instruction of the architecture. 4143 4144 Provided breakpoint bytes can be also used by 4145 `bp_loc_is_permanent' to detect permanent breakpoints. 4146 `gdbarch_breakpoint_from_pc' should return an unchanged memory 4147 copy if it was called for a location with permanent breakpoint as 4148 some architectures use breakpoint instructions containing 4149 arbitrary parameter value. 4150 4151 Replaces all the other BREAKPOINT macros. 4152 4153 `int gdbarch_memory_insert_breakpoint (GDBARCH, BP_TGT)' 4154 `gdbarch_memory_remove_breakpoint (GDBARCH, BP_TGT)' 4155 Insert or remove memory based breakpoints. Reasonable defaults 4156 (`default_memory_insert_breakpoint' and 4157 `default_memory_remove_breakpoint' respectively) have been 4158 provided so that it is not necessary to set these for most 4159 architectures. Architectures which may want to set 4160 `gdbarch_memory_insert_breakpoint' and 4161 `gdbarch_memory_remove_breakpoint' will likely have instructions 4162 that are oddly sized or are not stored in a conventional manner. 4163 4164 It may also be desirable (from an efficiency standpoint) to define 4165 custom breakpoint insertion and removal routines if 4166 `gdbarch_breakpoint_from_pc' needs to read the target's memory for 4167 some reason. 4168 4169 `CORE_ADDR gdbarch_adjust_breakpoint_address (GDBARCH, BPADDR)' 4170 Given an address at which a breakpoint is desired, return a 4171 breakpoint address adjusted to account for architectural 4172 constraints on breakpoint placement. This method is not needed by 4173 most targets. 4174 4175 The FR-V target (see `frv-tdep.c') requires this method. The FR-V 4176 is a VLIW architecture in which a number of RISC-like instructions 4177 are grouped (packed) together into an aggregate instruction or 4178 instruction bundle. When the processor executes one of these 4179 bundles, the component instructions are executed in parallel. 4180 4181 In the course of optimization, the compiler may group instructions 4182 from distinct source statements into the same bundle. The line 4183 number information associated with one of the latter statements 4184 will likely refer to some instruction other than the first one in 4185 the bundle. So, if the user attempts to place a breakpoint on one 4186 of these latter statements, GDB must be careful to _not_ place the 4187 break instruction on any instruction other than the first one in 4188 the bundle. (Remember though that the instructions within a 4189 bundle execute in parallel, so the _first_ instruction is the 4190 instruction at the lowest address and has nothing to do with 4191 execution order.) 4192 4193 The FR-V's `gdbarch_adjust_breakpoint_address' method will adjust a 4194 breakpoint's address by scanning backwards for the beginning of 4195 the bundle, returning the address of the bundle. 4196 4197 Since the adjustment of a breakpoint may significantly alter a 4198 user's expectation, GDB prints a warning when an adjusted 4199 breakpoint is initially set and each time that that breakpoint is 4200 hit. 4201 4202 `int gdbarch_call_dummy_location (GDBARCH)' 4203 See the file `inferior.h'. 4204 4205 This method has been replaced by `gdbarch_push_dummy_code' (*note 4206 gdbarch_push_dummy_code::). 4207 4208 `int gdbarch_cannot_fetch_register (GDBARCH, REGUM)' 4209 This function should return nonzero if REGNO cannot be fetched 4210 from an inferior process. 4211 4212 `int gdbarch_cannot_store_register (GDBARCH, REGNUM)' 4213 This function should return nonzero if REGNO should not be written 4214 to the target. This is often the case for program counters, 4215 status words, and other special registers. This function returns 4216 0 as default so that GDB will assume that all registers may be 4217 written. 4218 4219 `int gdbarch_convert_register_p (GDBARCH, REGNUM, struct type *TYPE)' 4220 Return non-zero if register REGNUM represents data values of type 4221 TYPE in a non-standard form. *Note Using Different Register and 4222 Memory Data Representations: Target Architecture Definition. 4223 4224 `int gdbarch_fp0_regnum (GDBARCH)' 4225 This function returns the number of the first floating point 4226 register, if the machine has such registers. Otherwise, it 4227 returns -1. 4228 4229 `CORE_ADDR gdbarch_decr_pc_after_break (GDBARCH)' 4230 This function shall return the amount by which to decrement the PC 4231 after the program encounters a breakpoint. This is often the 4232 number of bytes in `BREAKPOINT', though not always. For most 4233 targets this value will be 0. 4234 4235 `DISABLE_UNSETTABLE_BREAK (ADDR)' 4236 If defined, this should evaluate to 1 if ADDR is in a shared 4237 library in which breakpoints cannot be set and so should be 4238 disabled. 4239 4240 `int gdbarch_dwarf2_reg_to_regnum (GDBARCH, DWARF2_REGNR)' 4241 Convert DWARF2 register number DWARF2_REGNR into GDB regnum. If 4242 not defined, no conversion will be performed. 4243 4244 `int gdbarch_ecoff_reg_to_regnum (GDBARCH, ECOFF_REGNR)' 4245 Convert ECOFF register number ECOFF_REGNR into GDB regnum. If 4246 not defined, no conversion will be performed. 4247 4248 `GCC_COMPILED_FLAG_SYMBOL' 4249 `GCC2_COMPILED_FLAG_SYMBOL' 4250 If defined, these are the names of the symbols that GDB will look 4251 for to detect that GCC compiled the file. The default symbols are 4252 `gcc_compiled.' and `gcc2_compiled.', respectively. (Currently 4253 only defined for the Delta 68.) 4254 4255 `gdbarch_get_longjmp_target' 4256 This function determines the target PC address that `longjmp' will 4257 jump to, assuming that we have just stopped at a `longjmp' 4258 breakpoint. It takes a `CORE_ADDR *' as argument, and stores the 4259 target PC value through this pointer. It examines the current 4260 state of the machine as needed, typically by using a 4261 manually-determined offset into the `jmp_buf'. (While we might 4262 like to get the offset from the target's `jmpbuf.h', that header 4263 file cannot be assumed to be available when building a 4264 cross-debugger.) 4265 4266 `DEPRECATED_IBM6000_TARGET' 4267 Shows that we are configured for an IBM RS/6000 system. This 4268 conditional should be eliminated (FIXME) and replaced by 4269 feature-specific macros. It was introduced in haste and we are 4270 repenting at leisure. 4271 4272 `I386_USE_GENERIC_WATCHPOINTS' 4273 An x86-based target can define this to use the generic x86 4274 watchpoint support; see *note I386_USE_GENERIC_WATCHPOINTS: 4275 Algorithms. 4276 4277 `gdbarch_in_function_epilogue_p (GDBARCH, ADDR)' 4278 Returns non-zero if the given ADDR is in the epilogue of a 4279 function. The epilogue of a function is defined as the part of a 4280 function where the stack frame of the function already has been 4281 destroyed up to the final `return from function call' instruction. 4282 4283 `int gdbarch_in_solib_return_trampoline (GDBARCH, PC, NAME)' 4284 Define this function to return nonzero if the program is stopped 4285 in the trampoline that returns from a shared library. 4286 4287 `target_so_ops.in_dynsym_resolve_code (PC)' 4288 Define this to return nonzero if the program is stopped in the 4289 dynamic linker. 4290 4291 `SKIP_SOLIB_RESOLVER (PC)' 4292 Define this to evaluate to the (nonzero) address at which execution 4293 should continue to get past the dynamic linker's symbol resolution 4294 function. A zero value indicates that it is not important or 4295 necessary to set a breakpoint to get through the dynamic linker 4296 and that single stepping will suffice. 4297 4298 `CORE_ADDR gdbarch_integer_to_address (GDBARCH, TYPE, BUF)' 4299 Define this when the architecture needs to handle non-pointer to 4300 address conversions specially. Converts that value to an address 4301 according to the current architectures conventions. 4302 4303 _Pragmatics: When the user copies a well defined expression from 4304 their source code and passes it, as a parameter, to GDB's `print' 4305 command, they should get the same value as would have been 4306 computed by the target program. Any deviation from this rule can 4307 cause major confusion and annoyance, and needs to be justified 4308 carefully. In other words, GDB doesn't really have the freedom to 4309 do these conversions in clever and useful ways. It has, however, 4310 been pointed out that users aren't complaining about how GDB casts 4311 integers to pointers; they are complaining that they can't take an 4312 address from a disassembly listing and give it to `x/i'. Adding 4313 an architecture method like `gdbarch_integer_to_address' certainly 4314 makes it possible for GDB to "get it right" in all circumstances._ 4315 4316 *Note Pointers Are Not Always Addresses: Target Architecture 4317 Definition. 4318 4319 `CORE_ADDR gdbarch_pointer_to_address (GDBARCH, TYPE, BUF)' 4320 Assume that BUF holds a pointer of type TYPE, in the appropriate 4321 format for the current architecture. Return the byte address the 4322 pointer refers to. *Note Pointers Are Not Always Addresses: 4323 Target Architecture Definition. 4324 4325 `void gdbarch_register_to_value(GDBARCH, FRAME, REGNUM, TYPE, FUR)' 4326 Convert the raw contents of register REGNUM into a value of type 4327 TYPE. *Note Using Different Register and Memory Data 4328 Representations: Target Architecture Definition. 4329 4330 `REGISTER_CONVERT_TO_VIRTUAL(REG, TYPE, FROM, TO)' 4331 Convert the value of register REG from its raw form to its virtual 4332 form. *Note Raw and Virtual Register Representations: Target 4333 Architecture Definition. 4334 4335 `REGISTER_CONVERT_TO_RAW(TYPE, REG, FROM, TO)' 4336 Convert the value of register REG from its virtual form to its raw 4337 form. *Note Raw and Virtual Register Representations: Target 4338 Architecture Definition. 4339 4340 `const struct regset *regset_from_core_section (struct gdbarch * GDBARCH, const char * SECT_NAME, size_t SECT_SIZE)' 4341 Return the appropriate register set for a core file section with 4342 name SECT_NAME and size SECT_SIZE. 4343 4344 `SOFTWARE_SINGLE_STEP_P()' 4345 Define this as 1 if the target does not have a hardware single-step 4346 mechanism. The macro `SOFTWARE_SINGLE_STEP' must also be defined. 4347 4348 `SOFTWARE_SINGLE_STEP(SIGNAL, INSERT_BREAKPOINTS_P)' 4349 A function that inserts or removes (depending on 4350 INSERT_BREAKPOINTS_P) breakpoints at each possible destinations of 4351 the next instruction. See `sparc-tdep.c' and `rs6000-tdep.c' for 4352 examples. 4353 4354 `set_gdbarch_sofun_address_maybe_missing (GDBARCH, SET)' 4355 Somebody clever observed that, the more actual addresses you have 4356 in the debug information, the more time the linker has to spend 4357 relocating them. So whenever there's some other way the debugger 4358 could find the address it needs, you should omit it from the debug 4359 info, to make linking faster. 4360 4361 Calling `set_gdbarch_sofun_address_maybe_missing' with a non-zero 4362 argument SET indicates that a particular set of hacks of this sort 4363 are in use, affecting `N_SO' and `N_FUN' entries in stabs-format 4364 debugging information. `N_SO' stabs mark the beginning and ending 4365 addresses of compilation units in the text segment. `N_FUN' stabs 4366 mark the starts and ends of functions. 4367 4368 In this case, GDB assumes two things: 4369 4370 * `N_FUN' stabs have an address of zero. Instead of using those 4371 addresses, you should find the address where the function 4372 starts by taking the function name from the stab, and then 4373 looking that up in the minsyms (the linker/assembler symbol 4374 table). In other words, the stab has the name, and the 4375 linker/assembler symbol table is the only place that carries 4376 the address. 4377 4378 * `N_SO' stabs have an address of zero, too. You just look at 4379 the `N_FUN' stabs that appear before and after the `N_SO' 4380 stab, and guess the starting and ending addresses of the 4381 compilation unit from them. 4382 4383 `int gdbarch_stabs_argument_has_addr (GDBARCH, TYPE)' 4384 Define this function to return nonzero if a function argument of 4385 type TYPE is passed by reference instead of value. 4386 4387 `CORE_ADDR gdbarch_push_dummy_call (GDBARCH, FUNCTION, REGCACHE, BP_ADDR, NARGS, ARGS, SP, STRUCT_RETURN, STRUCT_ADDR)' 4388 Define this to push the dummy frame's call to the inferior 4389 function onto the stack. In addition to pushing NARGS, the code 4390 should push STRUCT_ADDR (when STRUCT_RETURN is non-zero), and the 4391 return address (BP_ADDR). 4392 4393 FUNCTION is a pointer to a `struct value'; on architectures that 4394 use function descriptors, this contains the function descriptor 4395 value. 4396 4397 Returns the updated top-of-stack pointer. 4398 4399 `CORE_ADDR gdbarch_push_dummy_code (GDBARCH, SP, FUNADDR, USING_GCC, ARGS, NARGS, VALUE_TYPE, REAL_PC, BP_ADDR, REGCACHE)' 4400 Given a stack based call dummy, push the instruction sequence 4401 (including space for a breakpoint) to which the called function 4402 should return. 4403 4404 Set BP_ADDR to the address at which the breakpoint instruction 4405 should be inserted, REAL_PC to the resume address when starting 4406 the call sequence, and return the updated inner-most stack address. 4407 4408 By default, the stack is grown sufficient to hold a frame-aligned 4409 (*note frame_align::) breakpoint, BP_ADDR is set to the address 4410 reserved for that breakpoint, and REAL_PC set to FUNADDR. 4411 4412 This method replaces `gdbarch_call_dummy_location (GDBARCH)'. 4413 4414 `int gdbarch_sdb_reg_to_regnum (GDBARCH, SDB_REGNR)' 4415 Use this function to convert sdb register SDB_REGNR into GDB 4416 regnum. If not defined, no conversion will be done. 4417 4418 `enum return_value_convention gdbarch_return_value (struct gdbarch *GDBARCH, struct type *VALTYPE, struct regcache *REGCACHE, void *READBUF, const void *WRITEBUF)' 4419 Given a function with a return-value of type RETTYPE, return which 4420 return-value convention that function would use. 4421 4422 GDB currently recognizes two function return-value conventions: 4423 `RETURN_VALUE_REGISTER_CONVENTION' where the return value is found 4424 in registers; and `RETURN_VALUE_STRUCT_CONVENTION' where the return 4425 value is found in memory and the address of that memory location is 4426 passed in as the function's first parameter. 4427 4428 If the register convention is being used, and WRITEBUF is 4429 non-`NULL', also copy the return-value in WRITEBUF into REGCACHE. 4430 4431 If the register convention is being used, and READBUF is 4432 non-`NULL', also copy the return value from REGCACHE into READBUF 4433 (REGCACHE contains a copy of the registers from the just returned 4434 function). 4435 4436 _Maintainer note: This method replaces separate predicate, extract, 4437 store methods. By having only one method, the logic needed to 4438 determine the return-value convention need only be implemented in 4439 one place. If GDB were written in an OO language, this method 4440 would instead return an object that knew how to perform the 4441 register return-value extract and store._ 4442 4443 _Maintainer note: This method does not take a GCC_P parameter, and 4444 such a parameter should not be added. If an architecture that 4445 requires per-compiler or per-function information be identified, 4446 then the replacement of RETTYPE with `struct value' FUNCTION 4447 should be pursued._ 4448 4449 _Maintainer note: The REGCACHE parameter limits this methods to 4450 the inner most frame. While replacing REGCACHE with a `struct 4451 frame_info' FRAME parameter would remove that limitation there has 4452 yet to be a demonstrated need for such a change._ 4453 4454 `void gdbarch_skip_permanent_breakpoint (GDBARCH, REGCACHE)' 4455 Advance the inferior's PC past a permanent breakpoint. GDB 4456 normally steps over a breakpoint by removing it, stepping one 4457 instruction, and re-inserting the breakpoint. However, permanent 4458 breakpoints are hardwired into the inferior, and can't be removed, 4459 so this strategy doesn't work. Calling 4460 `gdbarch_skip_permanent_breakpoint' adjusts the processor's state 4461 so that execution will resume just after the breakpoint. This 4462 function does the right thing even when the breakpoint is in the 4463 delay slot of a branch or jump. 4464 4465 `CORE_ADDR gdbarch_skip_trampoline_code (GDBARCH, FRAME, PC)' 4466 If the target machine has trampoline code that sits between 4467 callers and the functions being called, then define this function 4468 to return a new PC that is at the start of the real function. 4469 4470 `int gdbarch_deprecated_fp_regnum (GDBARCH)' 4471 If the frame pointer is in a register, use this function to return 4472 the number of that register. 4473 4474 `int gdbarch_stab_reg_to_regnum (GDBARCH, STAB_REGNR)' 4475 Use this function to convert stab register STAB_REGNR into GDB 4476 regnum. If not defined, no conversion will be done. 4477 4478 `SYMBOL_RELOADING_DEFAULT' 4479 The default value of the "symbol-reloading" variable. (Never 4480 defined in current sources.) 4481 4482 `TARGET_CHAR_BIT' 4483 Number of bits in a char; defaults to 8. 4484 4485 `int gdbarch_char_signed (GDBARCH)' 4486 Non-zero if `char' is normally signed on this architecture; zero if 4487 it should be unsigned. 4488 4489 The ISO C standard requires the compiler to treat `char' as 4490 equivalent to either `signed char' or `unsigned char'; any 4491 character in the standard execution set is supposed to be positive. 4492 Most compilers treat `char' as signed, but `char' is unsigned on 4493 the IBM S/390, RS6000, and PowerPC targets. 4494 4495 `int gdbarch_double_bit (GDBARCH)' 4496 Number of bits in a double float; defaults to 4497 `8 * TARGET_CHAR_BIT'. 4498 4499 `int gdbarch_float_bit (GDBARCH)' 4500 Number of bits in a float; defaults to `4 * TARGET_CHAR_BIT'. 4501 4502 `int gdbarch_int_bit (GDBARCH)' 4503 Number of bits in an integer; defaults to `4 * TARGET_CHAR_BIT'. 4504 4505 `int gdbarch_long_bit (GDBARCH)' 4506 Number of bits in a long integer; defaults to 4507 `4 * TARGET_CHAR_BIT'. 4508 4509 `int gdbarch_long_double_bit (GDBARCH)' 4510 Number of bits in a long double float; defaults to 4511 `2 * gdbarch_double_bit (GDBARCH)'. 4512 4513 `int gdbarch_long_long_bit (GDBARCH)' 4514 Number of bits in a long long integer; defaults to 4515 `2 * gdbarch_long_bit (GDBARCH)'. 4516 4517 `int gdbarch_ptr_bit (GDBARCH)' 4518 Number of bits in a pointer; defaults to 4519 `gdbarch_int_bit (GDBARCH)'. 4520 4521 `int gdbarch_short_bit (GDBARCH)' 4522 Number of bits in a short integer; defaults to 4523 `2 * TARGET_CHAR_BIT'. 4524 4525 `void gdbarch_virtual_frame_pointer (GDBARCH, PC, FRAME_REGNUM, FRAME_OFFSET)' 4526 Returns a `(REGISTER, OFFSET)' pair representing the virtual frame 4527 pointer in use at the code address PC. If virtual frame pointers 4528 are not used, a default definition simply returns 4529 `gdbarch_deprecated_fp_regnum' (or `gdbarch_sp_regnum', if no 4530 frame pointer is defined), with an offset of zero. 4531 4532 `TARGET_HAS_HARDWARE_WATCHPOINTS' 4533 If non-zero, the target has support for hardware-assisted 4534 watchpoints. *Note watchpoints: Algorithms, for more details and 4535 other related macros. 4536 4537 `int gdbarch_print_insn (GDBARCH, VMA, INFO)' 4538 This is the function used by GDB to print an assembly instruction. 4539 It prints the instruction at address VMA in debugged memory and 4540 returns the length of the instruction, in bytes. This usually 4541 points to a function in the `opcodes' library (*note Opcodes: 4542 Support Libraries.). INFO is a structure (of type 4543 `disassemble_info') defined in the header file 4544 `include/dis-asm.h', and used to pass information to the 4545 instruction decoding routine. 4546 4547 `frame_id gdbarch_dummy_id (GDBARCH, FRAME)' 4548 Given FRAME return a `struct frame_id' that uniquely identifies an 4549 inferior function call's dummy frame. The value returned must 4550 match the dummy frame stack value previously saved by 4551 `call_function_by_hand'. 4552 4553 `void gdbarch_value_to_register (GDBARCH, FRAME, TYPE, BUF)' 4554 Convert a value of type TYPE into the raw contents of a register. 4555 *Note Using Different Register and Memory Data Representations: 4556 Target Architecture Definition. 4557 4558 4559 Motorola M68K target conditionals. 4560 4561 `BPT_VECTOR' 4562 Define this to be the 4-bit location of the breakpoint trap 4563 vector. If not defined, it will default to `0xf'. 4564 4565 `REMOTE_BPT_VECTOR' 4566 Defaults to `1'. 4567 4568 4569 4570 File: gdbint.info, Node: Adding a New Target, Prev: Defining Other Architecture Features, Up: Target Architecture Definition 4571 4572 11.11 Adding a New Target 4573 ========================= 4574 4575 The following files add a target to GDB: 4576 4577 `gdb/TTT-tdep.c' 4578 Contains any miscellaneous code required for this target machine. 4579 On some machines it doesn't exist at all. 4580 4581 `gdb/ARCH-tdep.c' 4582 `gdb/ARCH-tdep.h' 4583 This is required to describe the basic layout of the target 4584 machine's processor chip (registers, stack, etc.). It can be 4585 shared among many targets that use the same processor architecture. 4586 4587 4588 (Target header files such as `gdb/config/ARCH/tm-TTT.h', 4589 `gdb/config/ARCH/tm-ARCH.h', and `config/tm-OS.h' are no longer used.) 4590 4591 A GDB description for a new architecture, arch is created by 4592 defining a global function `_initialize_ARCH_tdep', by convention in 4593 the source file `ARCH-tdep.c'. For example, in the case of the 4594 OpenRISC 1000, this function is called `_initialize_or1k_tdep' and is 4595 found in the file `or1k-tdep.c'. 4596 4597 The object file resulting from compiling this source file, which will 4598 contain the implementation of the `_initialize_ARCH_tdep' function is 4599 specified in the GDB `configure.tgt' file, which includes a large case 4600 statement pattern matching against the `--target' option of the 4601 `configure' script. 4602 4603 _Note:_ If the architecture requires multiple source files, the 4604 corresponding binaries should be included in `configure.tgt'. 4605 However if there are header files, the dependencies on these will 4606 not be picked up from the entries in `configure.tgt'. The 4607 `Makefile.in' file will need extending to show these dependencies. 4608 4609 A new struct gdbarch, defining the new architecture, is created 4610 within the `_initialize_ARCH_tdep' function by calling 4611 `gdbarch_register': 4612 4613 void gdbarch_register (enum bfd_architecture architecture, 4614 gdbarch_init_ftype *init_func, 4615 gdbarch_dump_tdep_ftype *tdep_dump_func); 4616 4617 This function has been described fully in an earlier section. *Note 4618 How an Architecture is Represented: How an Architecture is Represented. 4619 4620 The new `struct gdbarch' should contain implementations of the 4621 necessary functions (described in the previous sections) to describe 4622 the basic layout of the target machine's processor chip (registers, 4623 stack, etc.). It can be shared among many targets that use the same 4624 processor architecture. 4625 4626 4627 File: gdbint.info, Node: Target Descriptions, Next: Target Vector Definition, Prev: Target Architecture Definition, Up: Top 4628 4629 12 Target Descriptions 4630 ********************** 4631 4632 The target architecture definition (*note Target Architecture 4633 Definition::) contains GDB's hard-coded knowledge about an 4634 architecture. For some platforms, it is handy to have more flexible 4635 knowledge about a specific instance of the architecture--for instance, 4636 a processor or development board. "Target descriptions" provide a 4637 mechanism for the user to tell GDB more about what their target 4638 supports, or for the target to tell GDB directly. 4639 4640 For details on writing, automatically supplying, and manually 4641 selecting target descriptions, see *note Target Descriptions: 4642 (gdb)Target Descriptions. This section will cover some related topics 4643 about the GDB internals. 4644 4645 * Menu: 4646 4647 * Target Descriptions Implementation:: 4648 * Adding Target Described Register Support:: 4649 4650 4651 File: gdbint.info, Node: Target Descriptions Implementation, Next: Adding Target Described Register Support, Up: Target Descriptions 4652 4653 12.1 Target Descriptions Implementation 4654 ======================================= 4655 4656 Before GDB connects to a new target, or runs a new program on an 4657 existing target, it discards any existing target description and 4658 reverts to a default gdbarch. Then, after connecting, it looks for a 4659 new target description by calling `target_find_description'. 4660 4661 A description may come from a user specified file (XML), the remote 4662 `qXfer:features:read' packet (also XML), or from any custom 4663 `to_read_description' routine in the target vector. For instance, the 4664 remote target supports guessing whether a MIPS target is 32-bit or 4665 64-bit based on the size of the `g' packet. 4666 4667 If any target description is found, GDB creates a new gdbarch 4668 incorporating the description by calling `gdbarch_update_p'. Any 4669 `<architecture>' element is handled first, to determine which 4670 architecture's gdbarch initialization routine is called to create the 4671 new architecture. Then the initialization routine is called, and has a 4672 chance to adjust the constructed architecture based on the contents of 4673 the target description. For instance, it can recognize any properties 4674 set by a `to_read_description' routine. Also see *note Adding Target 4675 Described Register Support::. 4676 4677 4678 File: gdbint.info, Node: Adding Target Described Register Support, Prev: Target Descriptions Implementation, Up: Target Descriptions 4679 4680 12.2 Adding Target Described Register Support 4681 ============================================= 4682 4683 Target descriptions can report additional registers specific to an 4684 instance of the target. But it takes a little work in the architecture 4685 specific routines to support this. 4686 4687 A target description must either have no registers or a complete 4688 set--this avoids complexity in trying to merge standard registers with 4689 the target defined registers. It is the architecture's responsibility 4690 to validate that a description with registers has everything it needs. 4691 To keep architecture code simple, the same mechanism is used to assign 4692 fixed internal register numbers to standard registers. 4693 4694 If `tdesc_has_registers' returns 1, the description contains 4695 registers. The architecture's `gdbarch_init' routine should: 4696 4697 * Call `tdesc_data_alloc' to allocate storage, early, before 4698 searching for a matching gdbarch or allocating a new one. 4699 4700 * Use `tdesc_find_feature' to locate standard features by name. 4701 4702 * Use `tdesc_numbered_register' and `tdesc_numbered_register_choices' 4703 to locate the expected registers in the standard features. 4704 4705 * Return `NULL' if a required feature is missing, or if any standard 4706 feature is missing expected registers. This will produce a 4707 warning that the description was incomplete. 4708 4709 * Free the allocated data before returning, unless 4710 `tdesc_use_registers' is called. 4711 4712 * Call `set_gdbarch_num_regs' as usual, with a number higher than any 4713 fixed number passed to `tdesc_numbered_register'. 4714 4715 * Call `tdesc_use_registers' after creating a new gdbarch, before 4716 returning it. 4717 4718 4719 After `tdesc_use_registers' has been called, the architecture's 4720 `register_name', `register_type', and `register_reggroup_p' routines 4721 will not be called; that information will be taken from the target 4722 description. `num_regs' may be increased to account for any additional 4723 registers in the description. 4724 4725 Pseudo-registers require some extra care: 4726 4727 * Using `tdesc_numbered_register' allows the architecture to give 4728 constant register numbers to standard architectural registers, e.g. 4729 as an `enum' in `ARCH-tdep.h'. But because pseudo-registers are 4730 always numbered above `num_regs', which may be increased by the 4731 description, constant numbers can not be used for pseudos. They 4732 must be numbered relative to `num_regs' instead. 4733 4734 * The description will not describe pseudo-registers, so the 4735 architecture must call `set_tdesc_pseudo_register_name', 4736 `set_tdesc_pseudo_register_type', and 4737 `set_tdesc_pseudo_register_reggroup_p' to supply routines 4738 describing pseudo registers. These routines will be passed 4739 internal register numbers, so the same routines used for the 4740 gdbarch equivalents are usually suitable. 4741 4742 4743 4744 File: gdbint.info, Node: Target Vector Definition, Next: Native Debugging, Prev: Target Descriptions, Up: Top 4745 4746 13 Target Vector Definition 4747 *************************** 4748 4749 The target vector defines the interface between GDB's abstract handling 4750 of target systems, and the nitty-gritty code that actually exercises 4751 control over a process or a serial port. GDB includes some 30-40 4752 different target vectors; however, each configuration of GDB includes 4753 only a few of them. 4754 4755 * Menu: 4756 4757 * Managing Execution State:: 4758 * Existing Targets:: 4759 4760 4761 File: gdbint.info, Node: Managing Execution State, Next: Existing Targets, Up: Target Vector Definition 4762 4763 13.1 Managing Execution State 4764 ============================= 4765 4766 A target vector can be completely inactive (not pushed on the target 4767 stack), active but not running (pushed, but not connected to a fully 4768 manifested inferior), or completely active (pushed, with an accessible 4769 inferior). Most targets are only completely inactive or completely 4770 active, but some support persistent connections to a target even when 4771 the target has exited or not yet started. 4772 4773 For example, connecting to the simulator using `target sim' does not 4774 create a running program. Neither registers nor memory are accessible 4775 until `run'. Similarly, after `kill', the program can not continue 4776 executing. But in both cases GDB remains connected to the simulator, 4777 and target-specific commands are directed to the simulator. 4778 4779 A target which only supports complete activation should push itself 4780 onto the stack in its `to_open' routine (by calling `push_target'), and 4781 unpush itself from the stack in its `to_mourn_inferior' routine (by 4782 calling `unpush_target'). 4783 4784 A target which supports both partial and complete activation should 4785 still call `push_target' in `to_open', but not call `unpush_target' in 4786 `to_mourn_inferior'. Instead, it should call either 4787 `target_mark_running' or `target_mark_exited' in its `to_open', 4788 depending on whether the target is fully active after connection. It 4789 should also call `target_mark_running' any time the inferior becomes 4790 fully active (e.g. in `to_create_inferior' and `to_attach'), and 4791 `target_mark_exited' when the inferior becomes inactive (in 4792 `to_mourn_inferior'). The target should also make sure to call 4793 `target_mourn_inferior' from its `to_kill', to return the target to 4794 inactive state. 4795 4796 4797 File: gdbint.info, Node: Existing Targets, Prev: Managing Execution State, Up: Target Vector Definition 4798 4799 13.2 Existing Targets 4800 ===================== 4801 4802 13.2.1 File Targets 4803 ------------------- 4804 4805 Both executables and core files have target vectors. 4806 4807 13.2.2 Standard Protocol and Remote Stubs 4808 ----------------------------------------- 4809 4810 GDB's file `remote.c' talks a serial protocol to code that runs in the 4811 target system. GDB provides several sample "stubs" that can be 4812 integrated into target programs or operating systems for this purpose; 4813 they are named `CPU-stub.c'. Many operating systems, embedded targets, 4814 emulators, and simulators already have a GDB stub built into them, and 4815 maintenance of the remote protocol must be careful to preserve 4816 compatibility. 4817 4818 The GDB user's manual describes how to put such a stub into your 4819 target code. What follows is a discussion of integrating the SPARC 4820 stub into a complicated operating system (rather than a simple 4821 program), by Stu Grossman, the author of this stub. 4822 4823 The trap handling code in the stub assumes the following upon entry 4824 to `trap_low': 4825 4826 1. %l1 and %l2 contain pc and npc respectively at the time of the 4827 trap; 4828 4829 2. traps are disabled; 4830 4831 3. you are in the correct trap window. 4832 4833 As long as your trap handler can guarantee those conditions, then 4834 there is no reason why you shouldn't be able to "share" traps with the 4835 stub. The stub has no requirement that it be jumped to directly from 4836 the hardware trap vector. That is why it calls `exceptionHandler()', 4837 which is provided by the external environment. For instance, this could 4838 set up the hardware traps to actually execute code which calls the stub 4839 first, and then transfers to its own trap handler. 4840 4841 For the most point, there probably won't be much of an issue with 4842 "sharing" traps, as the traps we use are usually not used by the kernel, 4843 and often indicate unrecoverable error conditions. Anyway, this is all 4844 controlled by a table, and is trivial to modify. The most important 4845 trap for us is for `ta 1'. Without that, we can't single step or do 4846 breakpoints. Everything else is unnecessary for the proper operation 4847 of the debugger/stub. 4848 4849 From reading the stub, it's probably not obvious how breakpoints 4850 work. They are simply done by deposit/examine operations from GDB. 4851 4852 13.2.3 ROM Monitor Interface 4853 ---------------------------- 4854 4855 13.2.4 Custom Protocols 4856 ----------------------- 4857 4858 13.2.5 Transport Layer 4859 ---------------------- 4860 4861 13.2.6 Builtin Simulator 4862 ------------------------ 4863 4864 4865 File: gdbint.info, Node: Native Debugging, Next: Support Libraries, Prev: Target Vector Definition, Up: Top 4866 4867 14 Native Debugging 4868 ******************* 4869 4870 Several files control GDB's configuration for native support: 4871 4872 `gdb/config/ARCH/XYZ.mh' 4873 Specifies Makefile fragments needed by a _native_ configuration on 4874 machine XYZ. In particular, this lists the required 4875 native-dependent object files, by defining `NATDEPFILES=...'. 4876 Also specifies the header file which describes native support on 4877 XYZ, by defining `NAT_FILE= nm-XYZ.h'. You can also define 4878 `NAT_CFLAGS', `NAT_ADD_FILES', `NAT_CLIBS', `NAT_CDEPS', 4879 `NAT_GENERATED_FILES', etc.; see `Makefile.in'. 4880 4881 _Maintainer's note: The `.mh' suffix is because this file 4882 originally contained `Makefile' fragments for hosting GDB on 4883 machine XYZ. While the file is no longer used for this purpose, 4884 the `.mh' suffix remains. Perhaps someone will eventually rename 4885 these fragments so that they have a `.mn' suffix._ 4886 4887 `gdb/config/ARCH/nm-XYZ.h' 4888 (`nm.h' is a link to this file, created by `configure'). Contains 4889 C macro definitions describing the native system environment, such 4890 as child process control and core file support. 4891 4892 `gdb/XYZ-nat.c' 4893 Contains any miscellaneous C code required for this native support 4894 of this machine. On some machines it doesn't exist at all. 4895 4896 There are some "generic" versions of routines that can be used by 4897 various systems. These can be customized in various ways by macros 4898 defined in your `nm-XYZ.h' file. If these routines work for the XYZ 4899 host, you can just include the generic file's name (with `.o', not 4900 `.c') in `NATDEPFILES'. 4901 4902 Otherwise, if your machine needs custom support routines, you will 4903 need to write routines that perform the same functions as the generic 4904 file. Put them into `XYZ-nat.c', and put `XYZ-nat.o' into 4905 `NATDEPFILES'. 4906 4907 `inftarg.c' 4908 This contains the _target_ops vector_ that supports Unix child 4909 processes on systems which use ptrace and wait to control the 4910 child. 4911 4912 `procfs.c' 4913 This contains the _target_ops vector_ that supports Unix child 4914 processes on systems which use /proc to control the child. 4915 4916 `fork-child.c' 4917 This does the low-level grunge that uses Unix system calls to do a 4918 "fork and exec" to start up a child process. 4919 4920 `infptrace.c' 4921 This is the low level interface to inferior processes for systems 4922 using the Unix `ptrace' call in a vanilla way. 4923 4924 14.1 ptrace 4925 =========== 4926 4927 14.2 /proc 4928 ========== 4929 4930 14.3 win32 4931 ========== 4932 4933 14.4 shared libraries 4934 ===================== 4935 4936 14.5 Native Conditionals 4937 ======================== 4938 4939 When GDB is configured and compiled, various macros are defined or left 4940 undefined, to control compilation when the host and target systems are 4941 the same. These macros should be defined (or left undefined) in 4942 `nm-SYSTEM.h'. 4943 4944 `I386_USE_GENERIC_WATCHPOINTS' 4945 An x86-based machine can define this to use the generic x86 4946 watchpoint support; see *note I386_USE_GENERIC_WATCHPOINTS: 4947 Algorithms. 4948 4949 `SOLIB_ADD (FILENAME, FROM_TTY, TARG, READSYMS)' 4950 Define this to expand into an expression that will cause the 4951 symbols in FILENAME to be added to GDB's symbol table. If 4952 READSYMS is zero symbols are not read but any necessary low level 4953 processing for FILENAME is still done. 4954 4955 `SOLIB_CREATE_INFERIOR_HOOK' 4956 Define this to expand into any shared-library-relocation code that 4957 you want to be run just after the child process has been forked. 4958 4959 `START_INFERIOR_TRAPS_EXPECTED' 4960 When starting an inferior, GDB normally expects to trap twice; 4961 once when the shell execs, and once when the program itself execs. 4962 If the actual number of traps is something other than 2, then 4963 define this macro to expand into the number expected. 4964 4965 4966 4967 File: gdbint.info, Node: Support Libraries, Next: Coding, Prev: Native Debugging, Up: Top 4968 4969 15 Support Libraries 4970 ******************** 4971 4972 15.1 BFD 4973 ======== 4974 4975 BFD provides support for GDB in several ways: 4976 4977 _identifying executable and core files_ 4978 BFD will identify a variety of file types, including a.out, coff, 4979 and several variants thereof, as well as several kinds of core 4980 files. 4981 4982 _access to sections of files_ 4983 BFD parses the file headers to determine the names, virtual 4984 addresses, sizes, and file locations of all the various named 4985 sections in files (such as the text section or the data section). 4986 GDB simply calls BFD to read or write section X at byte offset Y 4987 for length Z. 4988 4989 _specialized core file support_ 4990 BFD provides routines to determine the failing command name stored 4991 in a core file, the signal with which the program failed, and 4992 whether a core file matches (i.e. could be a core dump of) a 4993 particular executable file. 4994 4995 _locating the symbol information_ 4996 GDB uses an internal interface of BFD to determine where to find 4997 the symbol information in an executable file or symbol-file. GDB 4998 itself handles the reading of symbols, since BFD does not 4999 "understand" debug symbols, but GDB uses BFD's cached information 5000 to find the symbols, string table, etc. 5001 5002 15.2 opcodes 5003 ============ 5004 5005 The opcodes library provides GDB's disassembler. (It's a separate 5006 library because it's also used in binutils, for `objdump'). 5007 5008 15.3 readline 5009 ============= 5010 5011 The `readline' library provides a set of functions for use by 5012 applications that allow users to edit command lines as they are typed 5013 in. 5014 5015 15.4 libiberty 5016 ============== 5017 5018 The `libiberty' library provides a set of functions and features that 5019 integrate and improve on functionality found in modern operating 5020 systems. Broadly speaking, such features can be divided into three 5021 groups: supplemental functions (functions that may be missing in some 5022 environments and operating systems), replacement functions (providing a 5023 uniform and easier to use interface for commonly used standard 5024 functions), and extensions (which provide additional functionality 5025 beyond standard functions). 5026 5027 GDB uses various features provided by the `libiberty' library, for 5028 instance the C++ demangler, the IEEE floating format support functions, 5029 the input options parser `getopt', the `obstack' extension, and other 5030 functions. 5031 5032 15.4.1 `obstacks' in GDB 5033 ------------------------ 5034 5035 The obstack mechanism provides a convenient way to allocate and free 5036 chunks of memory. Each obstack is a pool of memory that is managed 5037 like a stack. Objects (of any nature, size and alignment) are 5038 allocated and freed in a LIFO fashion on an obstack (see `libiberty''s 5039 documentation for a more detailed explanation of `obstacks'). 5040 5041 The most noticeable use of the `obstacks' in GDB is in object files. 5042 There is an obstack associated with each internal representation of an 5043 object file. Lots of things get allocated on these `obstacks': 5044 dictionary entries, blocks, blockvectors, symbols, minimal symbols, 5045 types, vectors of fundamental types, class fields of types, object 5046 files section lists, object files section offset lists, line tables, 5047 symbol tables, partial symbol tables, string tables, symbol table 5048 private data, macros tables, debug information sections and entries, 5049 import and export lists (som), unwind information (hppa), dwarf2 5050 location expressions data. Plus various strings such as directory 5051 names strings, debug format strings, names of types. 5052 5053 An essential and convenient property of all data on `obstacks' is 5054 that memory for it gets allocated (with `obstack_alloc') at various 5055 times during a debugging session, but it is released all at once using 5056 the `obstack_free' function. The `obstack_free' function takes a 5057 pointer to where in the stack it must start the deletion from (much 5058 like the cleanup chains have a pointer to where to start the cleanups). 5059 Because of the stack like structure of the `obstacks', this allows to 5060 free only a top portion of the obstack. There are a few instances in 5061 GDB where such thing happens. Calls to `obstack_free' are done after 5062 some local data is allocated to the obstack. Only the local data is 5063 deleted from the obstack. Of course this assumes that nothing between 5064 the `obstack_alloc' and the `obstack_free' allocates anything else on 5065 the same obstack. For this reason it is best and safest to use 5066 temporary `obstacks'. 5067 5068 Releasing the whole obstack is also not safe per se. It is safe only 5069 under the condition that we know the `obstacks' memory is no longer 5070 needed. In GDB we get rid of the `obstacks' only when we get rid of 5071 the whole objfile(s), for instance upon reading a new symbol file. 5072 5073 15.5 gnu-regex 5074 ============== 5075 5076 Regex conditionals. 5077 5078 `C_ALLOCA' 5079 5080 `NFAILURES' 5081 5082 `RE_NREGS' 5083 5084 `SIGN_EXTEND_CHAR' 5085 5086 `SWITCH_ENUM_BUG' 5087 5088 `SYNTAX_TABLE' 5089 5090 `Sword' 5091 5092 `sparc' 5093 5094 15.6 Array Containers 5095 ===================== 5096 5097 Often it is necessary to manipulate a dynamic array of a set of 5098 objects. C forces some bookkeeping on this, which can get cumbersome 5099 and repetitive. The `vec.h' file contains macros for defining and 5100 using a typesafe vector type. The functions defined will be inlined 5101 when compiling, and so the abstraction cost should be zero. Domain 5102 checks are added to detect programming errors. 5103 5104 An example use would be an array of symbols or section information. 5105 The array can be grown as symbols are read in (or preallocated), and 5106 the accessor macros provided keep care of all the necessary 5107 bookkeeping. Because the arrays are type safe, there is no danger of 5108 accidentally mixing up the contents. Think of these as C++ templates, 5109 but implemented in C. 5110 5111 Because of the different behavior of structure objects, scalar 5112 objects and of pointers, there are three flavors of vector, one for 5113 each of these variants. Both the structure object and pointer variants 5114 pass pointers to objects around -- in the former case the pointers are 5115 stored into the vector and in the latter case the pointers are 5116 dereferenced and the objects copied into the vector. The scalar object 5117 variant is suitable for `int'-like objects, and the vector elements are 5118 returned by value. 5119 5120 There are both `index' and `iterate' accessors. The iterator 5121 returns a boolean iteration condition and updates the iteration 5122 variable passed by reference. Because the iterator will be inlined, 5123 the address-of can be optimized away. 5124 5125 The vectors are implemented using the trailing array idiom, thus they 5126 are not resizeable without changing the address of the vector object 5127 itself. This means you cannot have variables or fields of vector type 5128 -- always use a pointer to a vector. The one exception is the final 5129 field of a structure, which could be a vector type. You will have to 5130 use the `embedded_size' & `embedded_init' calls to create such objects, 5131 and they will probably not be resizeable (so don't use the "safe" 5132 allocation variants). The trailing array idiom is used (rather than a 5133 pointer to an array of data), because, if we allow `NULL' to also 5134 represent an empty vector, empty vectors occupy minimal space in the 5135 structure containing them. 5136 5137 Each operation that increases the number of active elements is 5138 available in "quick" and "safe" variants. The former presumes that 5139 there is sufficient allocated space for the operation to succeed (it 5140 dies if there is not). The latter will reallocate the vector, if 5141 needed. Reallocation causes an exponential increase in vector size. 5142 If you know you will be adding N elements, it would be more efficient 5143 to use the reserve operation before adding the elements with the 5144 "quick" operation. This will ensure there are at least as many 5145 elements as you ask for, it will exponentially increase if there are 5146 too few spare slots. If you want reserve a specific number of slots, 5147 but do not want the exponential increase (for instance, you know this 5148 is the last allocation), use a negative number for reservation. You 5149 can also create a vector of a specific size from the get go. 5150 5151 You should prefer the push and pop operations, as they append and 5152 remove from the end of the vector. If you need to remove several items 5153 in one go, use the truncate operation. The insert and remove 5154 operations allow you to change elements in the middle of the vector. 5155 There are two remove operations, one which preserves the element 5156 ordering `ordered_remove', and one which does not `unordered_remove'. 5157 The latter function copies the end element into the removed slot, 5158 rather than invoke a memmove operation. The `lower_bound' function 5159 will determine where to place an item in the array using insert that 5160 will maintain sorted order. 5161 5162 If you need to directly manipulate a vector, then the `address' 5163 accessor will return the address of the start of the vector. Also the 5164 `space' predicate will tell you whether there is spare capacity in the 5165 vector. You will not normally need to use these two functions. 5166 5167 Vector types are defined using a `DEF_VEC_{O,P,I}(TYPENAME)' macro. 5168 Variables of vector type are declared using a `VEC(TYPENAME)' macro. 5169 The characters `O', `P' and `I' indicate whether TYPENAME is an object 5170 (`O'), pointer (`P') or integral (`I') type. Be careful to pick the 5171 correct one, as you'll get an awkward and inefficient API if you use 5172 the wrong one. There is a check, which results in a compile-time 5173 warning, for the `P' and `I' versions, but there is no check for the 5174 `O' versions, as that is not possible in plain C. 5175 5176 An example of their use would be, 5177 5178 DEF_VEC_P(tree); // non-managed tree vector. 5179 5180 struct my_struct { 5181 VEC(tree) *v; // A (pointer to) a vector of tree pointers. 5182 }; 5183 5184 struct my_struct *s; 5185 5186 if (VEC_length(tree, s->v)) { we have some contents } 5187 VEC_safe_push(tree, s->v, decl); // append some decl onto the end 5188 for (ix = 0; VEC_iterate(tree, s->v, ix, elt); ix++) 5189 { do something with elt } 5190 5191 The `vec.h' file provides details on how to invoke the various 5192 accessors provided. They are enumerated here: 5193 5194 `VEC_length' 5195 Return the number of items in the array, 5196 5197 `VEC_empty' 5198 Return true if the array has no elements. 5199 5200 `VEC_last' 5201 `VEC_index' 5202 Return the last or arbitrary item in the array. 5203 5204 `VEC_iterate' 5205 Access an array element and indicate whether the array has been 5206 traversed. 5207 5208 `VEC_alloc' 5209 `VEC_free' 5210 Create and destroy an array. 5211 5212 `VEC_embedded_size' 5213 `VEC_embedded_init' 5214 Helpers for embedding an array as the final element of another 5215 struct. 5216 5217 `VEC_copy' 5218 Duplicate an array. 5219 5220 `VEC_space' 5221 Return the amount of free space in an array. 5222 5223 `VEC_reserve' 5224 Ensure a certain amount of free space. 5225 5226 `VEC_quick_push' 5227 `VEC_safe_push' 5228 Append to an array, either assuming the space is available, or 5229 making sure that it is. 5230 5231 `VEC_pop' 5232 Remove the last item from an array. 5233 5234 `VEC_truncate' 5235 Remove several items from the end of an array. 5236 5237 `VEC_safe_grow' 5238 Add several items to the end of an array. 5239 5240 `VEC_replace' 5241 Overwrite an item in the array. 5242 5243 `VEC_quick_insert' 5244 `VEC_safe_insert' 5245 Insert an item into the middle of the array. Either the space must 5246 already exist, or the space is created. 5247 5248 `VEC_ordered_remove' 5249 `VEC_unordered_remove' 5250 Remove an item from the array, preserving order or not. 5251 5252 `VEC_block_remove' 5253 Remove a set of items from the array. 5254 5255 `VEC_address' 5256 Provide the address of the first element. 5257 5258 `VEC_lower_bound' 5259 Binary search the array. 5260 5261 5262 15.7 include 5263 ============ 5264 5265 5266 File: gdbint.info, Node: Coding, Next: Porting GDB, Prev: Support Libraries, Up: Top 5267 5268 16 Coding 5269 ********* 5270 5271 This chapter covers topics that are lower-level than the major 5272 algorithms of GDB. 5273 5274 16.1 Cleanups 5275 ============= 5276 5277 Cleanups are a structured way to deal with things that need to be done 5278 later. 5279 5280 When your code does something (e.g., `xmalloc' some memory, or 5281 `open' a file) that needs to be undone later (e.g., `xfree' the memory 5282 or `close' the file), it can make a cleanup. The cleanup will be done 5283 at some future point: when the command is finished and control returns 5284 to the top level; when an error occurs and the stack is unwound; or 5285 when your code decides it's time to explicitly perform cleanups. 5286 Alternatively you can elect to discard the cleanups you created. 5287 5288 Syntax: 5289 5290 `struct cleanup *OLD_CHAIN;' 5291 Declare a variable which will hold a cleanup chain handle. 5292 5293 `OLD_CHAIN = make_cleanup (FUNCTION, ARG);' 5294 Make a cleanup which will cause FUNCTION to be called with ARG (a 5295 `char *') later. The result, OLD_CHAIN, is a handle that can 5296 later be passed to `do_cleanups' or `discard_cleanups'. Unless 5297 you are going to call `do_cleanups' or `discard_cleanups', you can 5298 ignore the result from `make_cleanup'. 5299 5300 `do_cleanups (OLD_CHAIN);' 5301 Do all cleanups added to the chain since the corresponding 5302 `make_cleanup' call was made. 5303 5304 `discard_cleanups (OLD_CHAIN);' 5305 Same as `do_cleanups' except that it just removes the cleanups from 5306 the chain and does not call the specified functions. 5307 5308 Cleanups are implemented as a chain. The handle returned by 5309 `make_cleanups' includes the cleanup passed to the call and any later 5310 cleanups appended to the chain (but not yet discarded or performed). 5311 E.g.: 5312 5313 make_cleanup (a, 0); 5314 { 5315 struct cleanup *old = make_cleanup (b, 0); 5316 make_cleanup (c, 0) 5317 ... 5318 do_cleanups (old); 5319 } 5320 5321 will call `c()' and `b()' but will not call `a()'. The cleanup that 5322 calls `a()' will remain in the cleanup chain, and will be done later 5323 unless otherwise discarded. 5324 5325 Your function should explicitly do or discard the cleanups it 5326 creates. Failing to do this leads to non-deterministic behavior since 5327 the caller will arbitrarily do or discard your functions cleanups. 5328 This need leads to two common cleanup styles. 5329 5330 The first style is try/finally. Before it exits, your code-block 5331 calls `do_cleanups' with the old cleanup chain and thus ensures that 5332 your code-block's cleanups are always performed. For instance, the 5333 following code-segment avoids a memory leak problem (even when `error' 5334 is called and a forced stack unwind occurs) by ensuring that the 5335 `xfree' will always be called: 5336 5337 struct cleanup *old = make_cleanup (null_cleanup, 0); 5338 data = xmalloc (sizeof blah); 5339 make_cleanup (xfree, data); 5340 ... blah blah ... 5341 do_cleanups (old); 5342 5343 The second style is try/except. Before it exits, your code-block 5344 calls `discard_cleanups' with the old cleanup chain and thus ensures 5345 that any created cleanups are not performed. For instance, the 5346 following code segment, ensures that the file will be closed but only 5347 if there is an error: 5348 5349 FILE *file = fopen ("afile", "r"); 5350 struct cleanup *old = make_cleanup (close_file, file); 5351 ... blah blah ... 5352 discard_cleanups (old); 5353 return file; 5354 5355 Some functions, e.g., `fputs_filtered()' or `error()', specify that 5356 they "should not be called when cleanups are not in place". This means 5357 that any actions you need to reverse in the case of an error or 5358 interruption must be on the cleanup chain before you call these 5359 functions, since they might never return to your code (they `longjmp' 5360 instead). 5361 5362 16.2 Per-architecture module data 5363 ================================= 5364 5365 The multi-arch framework includes a mechanism for adding module 5366 specific per-architecture data-pointers to the `struct gdbarch' 5367 architecture object. 5368 5369 A module registers one or more per-architecture data-pointers using: 5370 5371 -- Architecture Function: struct gdbarch_data * 5372 gdbarch_data_register_pre_init (gdbarch_data_pre_init_ftype *PRE_INIT) 5373 PRE_INIT is used to, on-demand, allocate an initial value for a 5374 per-architecture data-pointer using the architecture's obstack 5375 (passed in as a parameter). Since PRE_INIT can be called during 5376 architecture creation, it is not parameterized with the 5377 architecture. and must not call modules that use per-architecture 5378 data. 5379 5380 -- Architecture Function: struct gdbarch_data * 5381 gdbarch_data_register_post_init (gdbarch_data_post_init_ftype 5382 *POST_INIT) 5383 POST_INIT is used to obtain an initial value for a 5384 per-architecture data-pointer _after_. Since POST_INIT is always 5385 called after architecture creation, it both receives the fully 5386 initialized architecture and is free to call modules that use 5387 per-architecture data (care needs to be taken to ensure that those 5388 other modules do not try to call back to this module as that will 5389 create in cycles in the initialization call graph). 5390 5391 These functions return a `struct gdbarch_data' that is used to 5392 identify the per-architecture data-pointer added for that module. 5393 5394 The per-architecture data-pointer is accessed using the function: 5395 5396 -- Architecture Function: void * gdbarch_data (struct gdbarch 5397 *GDBARCH, struct gdbarch_data *DATA_HANDLE) 5398 Given the architecture ARCH and module data handle DATA_HANDLE 5399 (returned by `gdbarch_data_register_pre_init' or 5400 `gdbarch_data_register_post_init'), this function returns the 5401 current value of the per-architecture data-pointer. If the data 5402 pointer is `NULL', it is first initialized by calling the 5403 corresponding PRE_INIT or POST_INIT method. 5404 5405 The examples below assume the following definitions: 5406 5407 struct nozel { int total; }; 5408 static struct gdbarch_data *nozel_handle; 5409 5410 A module can extend the architecture vector, adding additional 5411 per-architecture data, using the PRE_INIT method. The module's 5412 per-architecture data is then initialized during architecture creation. 5413 5414 In the below, the module's per-architecture _nozel_ is added. An 5415 architecture can specify its nozel by calling `set_gdbarch_nozel' from 5416 `gdbarch_init'. 5417 5418 static void * 5419 nozel_pre_init (struct obstack *obstack) 5420 { 5421 struct nozel *data = OBSTACK_ZALLOC (obstack, struct nozel); 5422 return data; 5423 } 5424 5425 extern void 5426 set_gdbarch_nozel (struct gdbarch *gdbarch, int total) 5427 { 5428 struct nozel *data = gdbarch_data (gdbarch, nozel_handle); 5429 data->total = nozel; 5430 } 5431 5432 A module can on-demand create architecture dependent data structures 5433 using `post_init'. 5434 5435 In the below, the nozel's total is computed on-demand by 5436 `nozel_post_init' using information obtained from the architecture. 5437 5438 static void * 5439 nozel_post_init (struct gdbarch *gdbarch) 5440 { 5441 struct nozel *data = GDBARCH_OBSTACK_ZALLOC (gdbarch, struct nozel); 5442 nozel->total = gdbarch... (gdbarch); 5443 return data; 5444 } 5445 5446 extern int 5447 nozel_total (struct gdbarch *gdbarch) 5448 { 5449 struct nozel *data = gdbarch_data (gdbarch, nozel_handle); 5450 return data->total; 5451 } 5452 5453 16.3 Wrapping Output Lines 5454 ========================== 5455 5456 Output that goes through `printf_filtered' or `fputs_filtered' or 5457 `fputs_demangled' needs only to have calls to `wrap_here' added in 5458 places that would be good breaking points. The utility routines will 5459 take care of actually wrapping if the line width is exceeded. 5460 5461 The argument to `wrap_here' is an indentation string which is 5462 printed _only_ if the line breaks there. This argument is saved away 5463 and used later. It must remain valid until the next call to 5464 `wrap_here' or until a newline has been printed through the 5465 `*_filtered' functions. Don't pass in a local variable and then return! 5466 5467 It is usually best to call `wrap_here' after printing a comma or 5468 space. If you call it before printing a space, make sure that your 5469 indentation properly accounts for the leading space that will print if 5470 the line wraps there. 5471 5472 Any function or set of functions that produce filtered output must 5473 finish by printing a newline, to flush the wrap buffer, before switching 5474 to unfiltered (`printf') output. Symbol reading routines that print 5475 warnings are a good example. 5476 5477 16.4 GDB Coding Standards 5478 ========================= 5479 5480 GDB follows the GNU coding standards, as described in 5481 `etc/standards.texi'. This file is also available for anonymous FTP 5482 from GNU archive sites. GDB takes a strict interpretation of the 5483 standard; in general, when the GNU standard recommends a practice but 5484 does not require it, GDB requires it. 5485 5486 GDB follows an additional set of coding standards specific to GDB, 5487 as described in the following sections. 5488 5489 16.4.1 ISO C 5490 ------------ 5491 5492 GDB assumes an ISO/IEC 9899:1990 (a.k.a. ISO C90) compliant compiler. 5493 5494 GDB does not assume an ISO C or POSIX compliant C library. 5495 5496 16.4.2 Memory Management 5497 ------------------------ 5498 5499 GDB does not use the functions `malloc', `realloc', `calloc', `free' 5500 and `asprintf'. 5501 5502 GDB uses the functions `xmalloc', `xrealloc' and `xcalloc' when 5503 allocating memory. Unlike `malloc' et.al. these functions do not 5504 return when the memory pool is empty. Instead, they unwind the stack 5505 using cleanups. These functions return `NULL' when requested to 5506 allocate a chunk of memory of size zero. 5507 5508 _Pragmatics: By using these functions, the need to check every 5509 memory allocation is removed. These functions provide portable 5510 behavior._ 5511 5512 GDB does not use the function `free'. 5513 5514 GDB uses the function `xfree' to return memory to the memory pool. 5515 Consistent with ISO-C, this function ignores a request to free a `NULL' 5516 pointer. 5517 5518 _Pragmatics: On some systems `free' fails when passed a `NULL' 5519 pointer._ 5520 5521 GDB can use the non-portable function `alloca' for the allocation of 5522 small temporary values (such as strings). 5523 5524 _Pragmatics: This function is very non-portable. Some systems 5525 restrict the memory being allocated to no more than a few kilobytes._ 5526 5527 GDB uses the string function `xstrdup' and the print function 5528 `xstrprintf'. 5529 5530 _Pragmatics: `asprintf' and `strdup' can fail. Print functions such 5531 as `sprintf' are very prone to buffer overflow errors._ 5532 5533 16.4.3 Compiler Warnings 5534 ------------------------ 5535 5536 With few exceptions, developers should avoid the configuration option 5537 `--disable-werror' when building GDB. The exceptions are listed in the 5538 file `gdb/MAINTAINERS'. The default, when building with GCC, is 5539 `--enable-werror'. 5540 5541 This option causes GDB (when built using GCC) to be compiled with a 5542 carefully selected list of compiler warning flags. Any warnings from 5543 those flags are treated as errors. 5544 5545 The current list of warning flags includes: 5546 5547 `-Wall' 5548 Recommended GCC warnings. 5549 5550 `-Wdeclaration-after-statement' 5551 GCC 3.x (and later) and C99 allow declarations mixed with code, 5552 but GCC 2.x and C89 do not. 5553 5554 `-Wpointer-arith' 5555 5556 `-Wformat-nonliteral' 5557 Non-literal format strings, with a few exceptions, are bugs - they 5558 might contain unintended user-supplied format specifiers. Since 5559 GDB uses the `format printf' attribute on all `printf' like 5560 functions this checks not just `printf' calls but also calls to 5561 functions such as `fprintf_unfiltered'. 5562 5563 `-Wno-pointer-sign' 5564 In version 4.0, GCC began warning about pointer argument passing or 5565 assignment even when the source and destination differed only in 5566 signedness. However, most GDB code doesn't distinguish carefully 5567 between `char' and `unsigned char'. In early 2006 the GDB 5568 developers decided correcting these warnings wasn't worth the time 5569 it would take. 5570 5571 `-Wno-unused-parameter' 5572 Due to the way that GDB is implemented many functions have unused 5573 parameters. Consequently this warning is avoided. The macro 5574 `ATTRIBUTE_UNUSED' is not used as it leads to false negatives -- 5575 it is not an error to have `ATTRIBUTE_UNUSED' on a parameter that 5576 is being used. 5577 5578 `-Wno-unused' 5579 `-Wno-switch' 5580 `-Wno-char-subscripts' 5581 These are warnings which might be useful for GDB, but are 5582 currently too noisy to enable with `-Werror'. 5583 5584 5585 16.4.4 Formatting 5586 ----------------- 5587 5588 The standard GNU recommendations for formatting must be followed 5589 strictly. 5590 5591 A function declaration should not have its name in column zero. A 5592 function definition should have its name in column zero. 5593 5594 /* Declaration */ 5595 static void foo (void); 5596 /* Definition */ 5597 void 5598 foo (void) 5599 { 5600 } 5601 5602 _Pragmatics: This simplifies scripting. Function definitions can be 5603 found using `^function-name'._ 5604 5605 There must be a space between a function or macro name and the 5606 opening parenthesis of its argument list (except for macro definitions, 5607 as required by C). There must not be a space after an open 5608 paren/bracket or before a close paren/bracket. 5609 5610 While additional whitespace is generally helpful for reading, do not 5611 use more than one blank line to separate blocks, and avoid adding 5612 whitespace after the end of a program line (as of 1/99, some 600 lines 5613 had whitespace after the semicolon). Excess whitespace causes 5614 difficulties for `diff' and `patch' utilities. 5615 5616 Pointers are declared using the traditional K&R C style: 5617 5618 void *foo; 5619 5620 and not: 5621 5622 void * foo; 5623 void* foo; 5624 5625 16.4.5 Comments 5626 --------------- 5627 5628 The standard GNU requirements on comments must be followed strictly. 5629 5630 Block comments must appear in the following form, with no `/*'- or 5631 `*/'-only lines, and no leading `*': 5632 5633 /* Wait for control to return from inferior to debugger. If inferior 5634 gets a signal, we may decide to start it up again instead of 5635 returning. That is why there is a loop in this function. When 5636 this function actually returns it means the inferior should be left 5637 stopped and GDB should read more commands. */ 5638 5639 (Note that this format is encouraged by Emacs; tabbing for a 5640 multi-line comment works correctly, and `M-q' fills the block 5641 consistently.) 5642 5643 Put a blank line between the block comments preceding function or 5644 variable definitions, and the definition itself. 5645 5646 In general, put function-body comments on lines by themselves, rather 5647 than trying to fit them into the 20 characters left at the end of a 5648 line, since either the comment or the code will inevitably get longer 5649 than will fit, and then somebody will have to move it anyhow. 5650 5651 16.4.6 C Usage 5652 -------------- 5653 5654 Code must not depend on the sizes of C data types, the format of the 5655 host's floating point numbers, the alignment of anything, or the order 5656 of evaluation of expressions. 5657 5658 Use functions freely. There are only a handful of compute-bound 5659 areas in GDB that might be affected by the overhead of a function call, 5660 mainly in symbol reading. Most of GDB's performance is limited by the 5661 target interface (whether serial line or system call). 5662 5663 However, use functions with moderation. A thousand one-line 5664 functions are just as hard to understand as a single thousand-line 5665 function. 5666 5667 _Macros are bad, M'kay._ (But if you have to use a macro, make sure 5668 that the macro arguments are protected with parentheses.) 5669 5670 Declarations like `struct foo *' should be used in preference to 5671 declarations like `typedef struct foo { ... } *foo_ptr'. 5672 5673 16.4.7 Function Prototypes 5674 -------------------------- 5675 5676 Prototypes must be used when both _declaring_ and _defining_ a 5677 function. Prototypes for GDB functions must include both the argument 5678 type and name, with the name matching that used in the actual function 5679 definition. 5680 5681 All external functions should have a declaration in a header file 5682 that callers include, except for `_initialize_*' functions, which must 5683 be external so that `init.c' construction works, but shouldn't be 5684 visible to random source files. 5685 5686 Where a source file needs a forward declaration of a static function, 5687 that declaration must appear in a block near the top of the source file. 5688 5689 16.4.8 Internal Error Recovery 5690 ------------------------------ 5691 5692 During its execution, GDB can encounter two types of errors. User 5693 errors and internal errors. User errors include not only a user 5694 entering an incorrect command but also problems arising from corrupt 5695 object files and system errors when interacting with the target. 5696 Internal errors include situations where GDB has detected, at run time, 5697 a corrupt or erroneous situation. 5698 5699 When reporting an internal error, GDB uses `internal_error' and 5700 `gdb_assert'. 5701 5702 GDB must not call `abort' or `assert'. 5703 5704 _Pragmatics: There is no `internal_warning' function. Either the 5705 code detected a user error, recovered from it and issued a `warning' or 5706 the code failed to correctly recover from the user error and issued an 5707 `internal_error'._ 5708 5709 16.4.9 File Names 5710 ----------------- 5711 5712 Any file used when building the core of GDB must be in lower case. Any 5713 file used when building the core of GDB must be 8.3 unique. These 5714 requirements apply to both source and generated files. 5715 5716 _Pragmatics: The core of GDB must be buildable on many platforms 5717 including DJGPP and MacOS/HFS. Every time an unfriendly file is 5718 introduced to the build process both `Makefile.in' and `configure.in' 5719 need to be modified accordingly. Compare the convoluted conversion 5720 process needed to transform `COPYING' into `copying.c' with the 5721 conversion needed to transform `version.in' into `version.c'._ 5722 5723 Any file non 8.3 compliant file (that is not used when building the 5724 core of GDB) must be added to `gdb/config/djgpp/fnchange.lst'. 5725 5726 _Pragmatics: This is clearly a compromise._ 5727 5728 When GDB has a local version of a system header file (ex `string.h') 5729 the file name based on the POSIX header prefixed with `gdb_' 5730 (`gdb_string.h'). These headers should be relatively independent: they 5731 should use only macros defined by `configure', the compiler, or the 5732 host; they should include only system headers; they should refer only 5733 to system types. They may be shared between multiple programs, e.g. 5734 GDB and GDBSERVER. 5735 5736 For other files `-' is used as the separator. 5737 5738 16.4.10 Include Files 5739 --------------------- 5740 5741 A `.c' file should include `defs.h' first. 5742 5743 A `.c' file should directly include the `.h' file of every 5744 declaration and/or definition it directly refers to. It cannot rely on 5745 indirect inclusion. 5746 5747 A `.h' file should directly include the `.h' file of every 5748 declaration and/or definition it directly refers to. It cannot rely on 5749 indirect inclusion. Exception: The file `defs.h' does not need to be 5750 directly included. 5751 5752 An external declaration should only appear in one include file. 5753 5754 An external declaration should never appear in a `.c' file. 5755 Exception: a declaration for the `_initialize' function that pacifies 5756 `-Wmissing-declaration'. 5757 5758 A `typedef' definition should only appear in one include file. 5759 5760 An opaque `struct' declaration can appear in multiple `.h' files. 5761 Where possible, a `.h' file should use an opaque `struct' declaration 5762 instead of an include. 5763 5764 All `.h' files should be wrapped in: 5765 5766 #ifndef INCLUDE_FILE_NAME_H 5767 #define INCLUDE_FILE_NAME_H 5768 header body 5769 #endif 5770 5771 16.4.11 Clean Design and Portable Implementation 5772 ------------------------------------------------ 5773 5774 In addition to getting the syntax right, there's the little question of 5775 semantics. Some things are done in certain ways in GDB because long 5776 experience has shown that the more obvious ways caused various kinds of 5777 trouble. 5778 5779 You can't assume the byte order of anything that comes from a target 5780 (including VALUEs, object files, and instructions). Such things must 5781 be byte-swapped using `SWAP_TARGET_AND_HOST' in GDB, or one of the swap 5782 routines defined in `bfd.h', such as `bfd_get_32'. 5783 5784 You can't assume that you know what interface is being used to talk 5785 to the target system. All references to the target must go through the 5786 current `target_ops' vector. 5787 5788 You can't assume that the host and target machines are the same 5789 machine (except in the "native" support modules). In particular, you 5790 can't assume that the target machine's header files will be available 5791 on the host machine. Target code must bring along its own header files 5792 - written from scratch or explicitly donated by their owner, to avoid 5793 copyright problems. 5794 5795 Insertion of new `#ifdef''s will be frowned upon. It's much better 5796 to write the code portably than to conditionalize it for various 5797 systems. 5798 5799 New `#ifdef''s which test for specific compilers or manufacturers or 5800 operating systems are unacceptable. All `#ifdef''s should test for 5801 features. The information about which configurations contain which 5802 features should be segregated into the configuration files. Experience 5803 has proven far too often that a feature unique to one particular system 5804 often creeps into other systems; and that a conditional based on some 5805 predefined macro for your current system will become worthless over 5806 time, as new versions of your system come out that behave differently 5807 with regard to this feature. 5808 5809 Adding code that handles specific architectures, operating systems, 5810 target interfaces, or hosts, is not acceptable in generic code. 5811 5812 One particularly notorious area where system dependencies tend to 5813 creep in is handling of file names. The mainline GDB code assumes 5814 Posix semantics of file names: absolute file names begin with a forward 5815 slash `/', slashes are used to separate leading directories, 5816 case-sensitive file names. These assumptions are not necessarily true 5817 on non-Posix systems such as MS-Windows. To avoid system-dependent 5818 code where you need to take apart or construct a file name, use the 5819 following portable macros: 5820 5821 `HAVE_DOS_BASED_FILE_SYSTEM' 5822 This preprocessing symbol is defined to a non-zero value on hosts 5823 whose filesystems belong to the MS-DOS/MS-Windows family. Use this 5824 symbol to write conditional code which should only be compiled for 5825 such hosts. 5826 5827 `IS_DIR_SEPARATOR (C)' 5828 Evaluates to a non-zero value if C is a directory separator 5829 character. On Unix and GNU/Linux systems, only a slash `/' is 5830 such a character, but on Windows, both `/' and `\' will pass. 5831 5832 `IS_ABSOLUTE_PATH (FILE)' 5833 Evaluates to a non-zero value if FILE is an absolute file name. 5834 For Unix and GNU/Linux hosts, a name which begins with a slash `/' 5835 is absolute. On DOS and Windows, `d:/foo' and `x:\bar' are also 5836 absolute file names. 5837 5838 `FILENAME_CMP (F1, F2)' 5839 Calls a function which compares file names F1 and F2 as 5840 appropriate for the underlying host filesystem. For Posix systems, 5841 this simply calls `strcmp'; on case-insensitive filesystems it 5842 will call `strcasecmp' instead. 5843 5844 `DIRNAME_SEPARATOR' 5845 Evaluates to a character which separates directories in 5846 `PATH'-style lists, typically held in environment variables. This 5847 character is `:' on Unix, `;' on DOS and Windows. 5848 5849 `SLASH_STRING' 5850 This evaluates to a constant string you should use to produce an 5851 absolute filename from leading directories and the file's basename. 5852 `SLASH_STRING' is `"/"' on most systems, but might be `"\\"' for 5853 some Windows-based ports. 5854 5855 In addition to using these macros, be sure to use portable library 5856 functions whenever possible. For example, to extract a directory or a 5857 basename part from a file name, use the `dirname' and `basename' 5858 library functions (available in `libiberty' for platforms which don't 5859 provide them), instead of searching for a slash with `strrchr'. 5860 5861 Another way to generalize GDB along a particular interface is with an 5862 attribute struct. For example, GDB has been generalized to handle 5863 multiple kinds of remote interfaces--not by `#ifdef's everywhere, but 5864 by defining the `target_ops' structure and having a current target (as 5865 well as a stack of targets below it, for memory references). Whenever 5866 something needs to be done that depends on which remote interface we are 5867 using, a flag in the current target_ops structure is tested (e.g., 5868 `target_has_stack'), or a function is called through a pointer in the 5869 current target_ops structure. In this way, when a new remote interface 5870 is added, only one module needs to be touched--the one that actually 5871 implements the new remote interface. Other examples of 5872 attribute-structs are BFD access to multiple kinds of object file 5873 formats, or GDB's access to multiple source languages. 5874 5875 Please avoid duplicating code. For example, in GDB 3.x all the code 5876 interfacing between `ptrace' and the rest of GDB was duplicated in 5877 `*-dep.c', and so changing something was very painful. In GDB 4.x, 5878 these have all been consolidated into `infptrace.c'. `infptrace.c' can 5879 deal with variations between systems the same way any system-independent 5880 file would (hooks, `#if defined', etc.), and machines which are 5881 radically different don't need to use `infptrace.c' at all. 5882 5883 All debugging code must be controllable using the `set debug MODULE' 5884 command. Do not use `printf' to print trace messages. Use 5885 `fprintf_unfiltered(gdb_stdlog, ...'. Do not use `#ifdef DEBUG'. 5886 5887 5888 File: gdbint.info, Node: Porting GDB, Next: Versions and Branches, Prev: Coding, Up: Top 5889 5890 17 Porting GDB 5891 ************** 5892 5893 Most of the work in making GDB compile on a new machine is in 5894 specifying the configuration of the machine. Porting a new 5895 architecture to GDB can be broken into a number of steps. 5896 5897 * Ensure a BFD exists for executables of the target architecture in 5898 the `bfd' directory. If one does not exist, create one by 5899 modifying an existing similar one. 5900 5901 * Implement a disassembler for the target architecture in the 5902 `opcodes' directory. 5903 5904 * Define the target architecture in the `gdb' directory (*note 5905 Adding a New Target: Adding a New Target.). Add the pattern for 5906 the new target to `configure.tgt' with the names of the files that 5907 contain the code. By convention the target architecture 5908 definition for an architecture ARCH is placed in `ARCH-tdep.c'. 5909 5910 Within `ARCH-tdep.c' define the function `_initialize_ARCH_tdep' 5911 which calls `gdbarch_register' to create the new `struct gdbarch' 5912 for the architecture. 5913 5914 * If a new remote target is needed, consider adding a new remote 5915 target by defining a function `_initialize_remote_ARCH'. However 5916 if at all possible use the GDB _Remote Serial Protocol_ for this 5917 and implement the server side protocol independently with the 5918 target. 5919 5920 * If desired implement a simulator in the `sim' directory. This 5921 should create the library `libsim.a' implementing the interface in 5922 `remote-sim.h' (found in the `include' directory). 5923 5924 * Build and test. If desired, lobby the GDB steering group to have 5925 the new port included in the main distribution! 5926 5927 * Add a description of the new architecture to the main GDB user 5928 guide (*note Configuration Specific Information: 5929 (gdb)Configuration Specific Information.). 5930 5931 5932 5933 File: gdbint.info, Node: Versions and Branches, Next: Start of New Year Procedure, Prev: Porting GDB, Up: Top 5934 5935 18 Versions and Branches 5936 ************************ 5937 5938 18.1 Versions 5939 ============= 5940 5941 GDB's version is determined by the file `gdb/version.in' and takes one 5942 of the following forms: 5943 5944 MAJOR.MINOR 5945 MAJOR.MINOR.PATCHLEVEL 5946 an official release (e.g., 6.2 or 6.2.1) 5947 5948 MAJOR.MINOR.PATCHLEVEL.YYYYMMDD 5949 a snapshot taken at YYYY-MM-DD-gmt (e.g., 6.1.50.20020302, 5950 6.1.90.20020304, or 6.1.0.20020308) 5951 5952 MAJOR.MINOR.PATCHLEVEL.YYYYMMDD-cvs 5953 a CVS check out drawn on YYYY-MM-DD (e.g., 6.1.50.20020302-cvs, 5954 6.1.90.20020304-cvs, or 6.1.0.20020308-cvs) 5955 5956 MAJOR.MINOR.PATCHLEVEL.YYYYMMDD (VENDOR) 5957 a vendor specific release of GDB, that while based on 5958 MAJOR.MINOR.PATCHLEVEL.YYYYMMDD, may include additional changes 5959 5960 GDB's mainline uses the MAJOR and MINOR version numbers from the 5961 most recent release branch, with a PATCHLEVEL of 50. At the time each 5962 new release branch is created, the mainline's MAJOR and MINOR version 5963 numbers are updated. 5964 5965 GDB's release branch is similar. When the branch is cut, the 5966 PATCHLEVEL is changed from 50 to 90. As draft releases are drawn from 5967 the branch, the PATCHLEVEL is incremented. Once the first release 5968 (MAJOR.MINOR) has been made, the PATCHLEVEL is set to 0 and updates 5969 have an incremented PATCHLEVEL. 5970 5971 For snapshots, and CVS check outs, it is also possible to identify 5972 the CVS origin: 5973 5974 MAJOR.MINOR.50.YYYYMMDD 5975 drawn from the HEAD of mainline CVS (e.g., 6.1.50.20020302) 5976 5977 MAJOR.MINOR.90.YYYYMMDD 5978 MAJOR.MINOR.91.YYYYMMDD ... 5979 drawn from a release branch prior to the release (e.g., 5980 6.1.90.20020304) 5981 5982 MAJOR.MINOR.0.YYYYMMDD 5983 MAJOR.MINOR.1.YYYYMMDD ... 5984 drawn from a release branch after the release (e.g., 5985 6.2.0.20020308) 5986 5987 If the previous GDB version is 6.1 and the current version is 6.2, 5988 then, substituting 6 for MAJOR and 1 or 2 for MINOR, here's an 5989 illustration of a typical sequence: 5990 5991 <HEAD> 5992 | 5993 6.1.50.20020302-cvs 5994 | 5995 +--------------------------. 5996 | <gdb_6_2-branch> 5997 | | 5998 6.2.50.20020303-cvs 6.1.90 (draft #1) 5999 | | 6000 6.2.50.20020304-cvs 6.1.90.20020304-cvs 6001 | | 6002 6.2.50.20020305-cvs 6.1.91 (draft #2) 6003 | | 6004 6.2.50.20020306-cvs 6.1.91.20020306-cvs 6005 | | 6006 6.2.50.20020307-cvs 6.2 (release) 6007 | | 6008 6.2.50.20020308-cvs 6.2.0.20020308-cvs 6009 | | 6010 6.2.50.20020309-cvs 6.2.1 (update) 6011 | | 6012 6.2.50.20020310-cvs <branch closed> 6013 | 6014 6.2.50.20020311-cvs 6015 | 6016 +--------------------------. 6017 | <gdb_6_3-branch> 6018 | | 6019 6.3.50.20020312-cvs 6.2.90 (draft #1) 6020 | | 6021 6022 18.2 Release Branches 6023 ===================== 6024 6025 GDB draws a release series (6.2, 6.2.1, ...) from a single release 6026 branch, and identifies that branch using the CVS branch tags: 6027 6028 gdb_MAJOR_MINOR-YYYYMMDD-branchpoint 6029 gdb_MAJOR_MINOR-branch 6030 gdb_MAJOR_MINOR-YYYYMMDD-release 6031 6032 _Pragmatics: To help identify the date at which a branch or release 6033 is made, both the branchpoint and release tags include the date that 6034 they are cut (YYYYMMDD) in the tag. The branch tag, denoting the head 6035 of the branch, does not need this._ 6036 6037 18.3 Vendor Branches 6038 ==================== 6039 6040 To avoid version conflicts, vendors are expected to modify the file 6041 `gdb/version.in' to include a vendor unique alphabetic identifier (an 6042 official GDB release never uses alphabetic characters in its version 6043 identifier). E.g., `6.2widgit2', or `6.2 (Widgit Inc Patch 2)'. 6044 6045 18.4 Experimental Branches 6046 ========================== 6047 6048 18.4.1 Guidelines 6049 ----------------- 6050 6051 GDB permits the creation of branches, cut from the CVS repository, for 6052 experimental development. Branches make it possible for developers to 6053 share preliminary work, and maintainers to examine significant new 6054 developments. 6055 6056 The following are a set of guidelines for creating such branches: 6057 6058 _a branch has an owner_ 6059 The owner can set further policy for a branch, but may not change 6060 the ground rules. In particular, they can set a policy for 6061 commits (be it adding more reviewers or deciding who can commit). 6062 6063 _all commits are posted_ 6064 All changes committed to a branch shall also be posted to the GDB 6065 patches mailing list <gdb-patches (a] sourceware.org>. While 6066 commentary on such changes are encouraged, people should remember 6067 that the changes only apply to a branch. 6068 6069 _all commits are covered by an assignment_ 6070 This ensures that all changes belong to the Free Software 6071 Foundation, and avoids the possibility that the branch may become 6072 contaminated. 6073 6074 _a branch is focused_ 6075 A focused branch has a single objective or goal, and does not 6076 contain unnecessary or irrelevant changes. Cleanups, where 6077 identified, being be pushed into the mainline as soon as possible. 6078 6079 _a branch tracks mainline_ 6080 This keeps the level of divergence under control. It also keeps 6081 the pressure on developers to push cleanups and other stuff into 6082 the mainline. 6083 6084 _a branch shall contain the entire GDB module_ 6085 The GDB module `gdb' should be specified when creating a branch 6086 (branches of individual files should be avoided). *Note Tags::. 6087 6088 _a branch shall be branded using `version.in'_ 6089 The file `gdb/version.in' shall be modified so that it identifies 6090 the branch OWNER and branch NAME, e.g., 6091 `6.2.50.20030303_owner_name' or `6.2 (Owner Name)'. 6092 6093 6094 18.4.2 Tags 6095 ----------- 6096 6097 To simplify the identification of GDB branches, the following branch 6098 tagging convention is strongly recommended: 6099 6100 `OWNER_NAME-YYYYMMDD-branchpoint' 6101 `OWNER_NAME-YYYYMMDD-branch' 6102 The branch point and corresponding branch tag. YYYYMMDD is the 6103 date that the branch was created. A branch is created using the 6104 sequence: 6105 cvs rtag OWNER_NAME-YYYYMMDD-branchpoint gdb 6106 cvs rtag -b -r OWNER_NAME-YYYYMMDD-branchpoint \ 6107 OWNER_NAME-YYYYMMDD-branch gdb 6108 6109 `OWNER_NAME-YYYYMMDD-mergepoint' 6110 The tagged point, on the mainline, that was used when merging the 6111 branch on YYYYMMDD. To merge in all changes since the branch was 6112 cut, use a command sequence like: 6113 cvs rtag OWNER_NAME-YYYYMMDD-mergepoint gdb 6114 cvs update \ 6115 -jOWNER_NAME-YYYYMMDD-branchpoint 6116 -jOWNER_NAME-YYYYMMDD-mergepoint 6117 Similar sequences can be used to just merge in changes since the 6118 last merge. 6119 6120 6121 For further information on CVS, see Concurrent Versions System 6122 (http://www.gnu.org/software/cvs/). 6123 6124 6125 File: gdbint.info, Node: Start of New Year Procedure, Next: Releasing GDB, Prev: Versions and Branches, Up: Top 6126 6127 19 Start of New Year Procedure 6128 ****************************** 6129 6130 At the start of each new year, the following actions should be 6131 performed: 6132 6133 * Rotate the ChangeLog file 6134 6135 The current `ChangeLog' file should be renamed into 6136 `ChangeLog-YYYY' where YYYY is the year that has just passed. A 6137 new `ChangeLog' file should be created, and its contents should 6138 contain a reference to the previous ChangeLog. The following 6139 should also be preserved at the end of the new ChangeLog, in order 6140 to provide the appropriate settings when editing this file with 6141 Emacs: 6142 Local Variables: 6143 mode: change-log 6144 left-margin: 8 6145 fill-column: 74 6146 version-control: never 6147 coding: utf-8 6148 End: 6149 6150 * Add an entry for the newly created ChangeLog file 6151 (`ChangeLog-YYYY') in `gdb/config/djgpp/fnchange.lst'. 6152 6153 * Update the copyright year in the startup message 6154 6155 Update the copyright year in: 6156 * file `top.c', function `print_gdb_version' 6157 6158 * file `gdbserver/server.c', function `gdbserver_version' 6159 6160 * file `gdbserver/gdbreplay.c', function `gdbreplay_version' 6161 6162 * Run the `copyright.sh' script to add the new year in the copyright 6163 notices of most source files. This script requires Emacs 22 or 6164 later to be installed. 6165 6166 * The new year also needs to be added manually in all other files 6167 that are not already taken care of by the `copyright.sh' script: 6168 * `*.s' 6169 6170 * `*.f' 6171 6172 * `*.f90' 6173 6174 * `*.igen' 6175 6176 * `*.ac' 6177 6178 * `*.texi' 6179 6180 * `*.texinfo' 6181 6182 * `*.tex' 6183 6184 * `*.defs' 6185 6186 * `*.1' 6187 6188 6189 6190 File: gdbint.info, Node: Releasing GDB, Next: Testsuite, Prev: Start of New Year Procedure, Up: Top 6191 6192 20 Releasing GDB 6193 **************** 6194 6195 20.1 Branch Commit Policy 6196 ========================= 6197 6198 The branch commit policy is pretty slack. GDB releases 5.0, 5.1 and 6199 5.2 all used the below: 6200 6201 * The `gdb/MAINTAINERS' file still holds. 6202 6203 * Don't fix something on the branch unless/until it is also fixed in 6204 the trunk. If this isn't possible, mentioning it in the 6205 `gdb/PROBLEMS' file is better than committing a hack. 6206 6207 * When considering a patch for the branch, suggested criteria 6208 include: Does it fix a build? Does it fix the sequence `break 6209 main; run' when debugging a static binary? 6210 6211 * The further a change is from the core of GDB, the less likely the 6212 change will worry anyone (e.g., target specific code). 6213 6214 * Only post a proposal to change the core of GDB after you've sent 6215 individual bribes to all the people listed in the `MAINTAINERS' 6216 file ;-) 6217 6218 _Pragmatics: Provided updates are restricted to non-core 6219 functionality there is little chance that a broken change will be fatal. 6220 This means that changes such as adding a new architectures or (within 6221 reason) support for a new host are considered acceptable._ 6222 6223 20.2 Obsoleting code 6224 ==================== 6225 6226 Before anything else, poke the other developers (and around the source 6227 code) to see if there is anything that can be removed from GDB (an old 6228 target, an unused file). 6229 6230 Obsolete code is identified by adding an `OBSOLETE' prefix to every 6231 line. Doing this means that it is easy to identify something that has 6232 been obsoleted when greping through the sources. 6233 6234 The process is done in stages -- this is mainly to ensure that the 6235 wider GDB community has a reasonable opportunity to respond. Remember, 6236 everything on the Internet takes a week. 6237 6238 1. Post the proposal on the GDB mailing list <gdb (a] sourceware.org> 6239 Creating a bug report to track the task's state, is also highly 6240 recommended. 6241 6242 2. Wait a week or so. 6243 6244 3. Post the proposal on the GDB Announcement mailing list 6245 <gdb-announce (a] sourceware.org>. 6246 6247 4. Wait a week or so. 6248 6249 5. Go through and edit all relevant files and lines so that they are 6250 prefixed with the word `OBSOLETE'. 6251 6252 6. Wait until the next GDB version, containing this obsolete code, 6253 has been released. 6254 6255 7. Remove the obsolete code. 6256 6257 _Maintainer note: While removing old code is regrettable it is 6258 hopefully better for GDB's long term development. Firstly it helps the 6259 developers by removing code that is either no longer relevant or simply 6260 wrong. Secondly since it removes any history associated with the file 6261 (effectively clearing the slate) the developer has a much freer hand 6262 when it comes to fixing broken files._ 6263 6264 20.3 Before the Branch 6265 ====================== 6266 6267 The most important objective at this stage is to find and fix simple 6268 changes that become a pain to track once the branch is created. For 6269 instance, configuration problems that stop GDB from even building. If 6270 you can't get the problem fixed, document it in the `gdb/PROBLEMS' file. 6271 6272 Prompt for `gdb/NEWS' 6273 --------------------- 6274 6275 People always forget. Send a post reminding them but also if you know 6276 something interesting happened add it yourself. The `schedule' script 6277 will mention this in its e-mail. 6278 6279 Review `gdb/README' 6280 ------------------- 6281 6282 Grab one of the nightly snapshots and then walk through the 6283 `gdb/README' looking for anything that can be improved. The `schedule' 6284 script will mention this in its e-mail. 6285 6286 Refresh any imported files. 6287 --------------------------- 6288 6289 A number of files are taken from external repositories. They include: 6290 6291 * `texinfo/texinfo.tex' 6292 6293 * `config.guess' et. al. (see the top-level `MAINTAINERS' file) 6294 6295 * `etc/standards.texi', `etc/make-stds.texi' 6296 6297 Check the ARI 6298 ------------- 6299 6300 A.R.I. is an `awk' script (Awk Regression Index ;-) that checks for a 6301 number of errors and coding conventions. The checks include things 6302 like using `malloc' instead of `xmalloc' and file naming problems. 6303 There shouldn't be any regressions. 6304 6305 20.3.1 Review the bug data base 6306 ------------------------------- 6307 6308 Close anything obviously fixed. 6309 6310 20.3.2 Check all cross targets build 6311 ------------------------------------ 6312 6313 The targets are listed in `gdb/MAINTAINERS'. 6314 6315 20.4 Cut the Branch 6316 =================== 6317 6318 Create the branch 6319 ----------------- 6320 6321 $ u=5.1 6322 $ v=5.2 6323 $ V=`echo $v | sed 's/\./_/g'` 6324 $ D=`date -u +%Y-%m-%d` 6325 $ echo $u $V $D 6326 5.1 5_2 2002-03-03 6327 $ echo cvs -f -d :ext:sourceware.org:/cvs/src rtag \ 6328 -D $D-gmt gdb_$V-$D-branchpoint insight 6329 cvs -f -d :ext:sourceware.org:/cvs/src rtag 6330 -D 2002-03-03-gmt gdb_5_2-2002-03-03-branchpoint insight 6331 $ ^echo ^^ 6332 ... 6333 $ echo cvs -f -d :ext:sourceware.org:/cvs/src rtag \ 6334 -b -r gdb_$V-$D-branchpoint gdb_$V-branch insight 6335 cvs -f -d :ext:sourceware.org:/cvs/src rtag \ 6336 -b -r gdb_5_2-2002-03-03-branchpoint gdb_5_2-branch insight 6337 $ ^echo ^^ 6338 ... 6339 $ 6340 6341 * By using `-D YYYY-MM-DD-gmt', the branch is forced to an exact 6342 date/time. 6343 6344 * The trunk is first tagged so that the branch point can easily be 6345 found. 6346 6347 * Insight, which includes GDB, is tagged at the same time. 6348 6349 * `version.in' gets bumped to avoid version number conflicts. 6350 6351 * The reading of `.cvsrc' is disabled using `-f'. 6352 6353 Update `version.in' 6354 ------------------- 6355 6356 $ u=5.1 6357 $ v=5.2 6358 $ V=`echo $v | sed 's/\./_/g'` 6359 $ echo $u $v$V 6360 5.1 5_2 6361 $ cd /tmp 6362 $ echo cvs -f -d :ext:sourceware.org:/cvs/src co \ 6363 -r gdb_$V-branch src/gdb/version.in 6364 cvs -f -d :ext:sourceware.org:/cvs/src co 6365 -r gdb_5_2-branch src/gdb/version.in 6366 $ ^echo ^^ 6367 U src/gdb/version.in 6368 $ cd src/gdb 6369 $ echo $u.90-0000-00-00-cvs > version.in 6370 $ cat version.in 6371 5.1.90-0000-00-00-cvs 6372 $ cvs -f commit version.in 6373 6374 * `0000-00-00' is used as a date to pump prime the version.in update 6375 mechanism. 6376 6377 * `.90' and the previous branch version are used as fairly arbitrary 6378 initial branch version number. 6379 6380 Update the web and news pages 6381 ----------------------------- 6382 6383 Something? 6384 6385 Tweak cron to track the new branch 6386 ---------------------------------- 6387 6388 The file `gdbadmin/cron/crontab' contains gdbadmin's cron table. This 6389 file needs to be updated so that: 6390 6391 * A daily timestamp is added to the file `version.in'. 6392 6393 * The new branch is included in the snapshot process. 6394 6395 See the file `gdbadmin/cron/README' for how to install the updated cron 6396 table. 6397 6398 The file `gdbadmin/ss/README' should also be reviewed to reflect any 6399 changes. That file is copied to both the branch/ and current/ snapshot 6400 directories. 6401 6402 Update the NEWS and README files 6403 -------------------------------- 6404 6405 The `NEWS' file needs to be updated so that on the branch it refers to 6406 _changes in the current release_ while on the trunk it also refers to 6407 _changes since the current release_. 6408 6409 The `README' file needs to be updated so that it refers to the 6410 current release. 6411 6412 Post the branch info 6413 -------------------- 6414 6415 Send an announcement to the mailing lists: 6416 6417 * GDB Announcement mailing list <gdb-announce (a] sourceware.org> 6418 6419 * GDB Discussion mailing list <gdb (a] sourceware.org> and GDB Testers 6420 mailing list <gdb-testers (a] sourceware.org> 6421 6422 _Pragmatics: The branch creation is sent to the announce list to 6423 ensure that people people not subscribed to the higher volume discussion 6424 list are alerted._ 6425 6426 The announcement should include: 6427 6428 * The branch tag. 6429 6430 * How to check out the branch using CVS. 6431 6432 * The date/number of weeks until the release. 6433 6434 * The branch commit policy still holds. 6435 6436 20.5 Stabilize the branch 6437 ========================= 6438 6439 Something goes here. 6440 6441 20.6 Create a Release 6442 ===================== 6443 6444 The process of creating and then making available a release is broken 6445 down into a number of stages. The first part addresses the technical 6446 process of creating a releasable tar ball. The later stages address the 6447 process of releasing that tar ball. 6448 6449 When making a release candidate just the first section is needed. 6450 6451 20.6.1 Create a release candidate 6452 --------------------------------- 6453 6454 The objective at this stage is to create a set of tar balls that can be 6455 made available as a formal release (or as a less formal release 6456 candidate). 6457 6458 Freeze the branch 6459 ................. 6460 6461 Send out an e-mail notifying everyone that the branch is frozen to 6462 <gdb-patches (a] sourceware.org>. 6463 6464 Establish a few defaults. 6465 ......................... 6466 6467 $ b=gdb_5_2-branch 6468 $ v=5.2 6469 $ t=/sourceware/snapshot-tmp/gdbadmin-tmp 6470 $ echo $t/$b/$v 6471 /sourceware/snapshot-tmp/gdbadmin-tmp/gdb_5_2-branch/5.2 6472 $ mkdir -p $t/$b/$v 6473 $ cd $t/$b/$v 6474 $ pwd 6475 /sourceware/snapshot-tmp/gdbadmin-tmp/gdb_5_2-branch/5.2 6476 $ which autoconf 6477 /home/gdbadmin/bin/autoconf 6478 $ 6479 6480 Notes: 6481 6482 * Check the `autoconf' version carefully. You want to be using the 6483 version documented in the toplevel `README-maintainer-mode' file. 6484 It is very unlikely that the version of `autoconf' installed in 6485 system directories (e.g., `/usr/bin/autoconf') is correct. 6486 6487 Check out the relevant modules: 6488 ............................... 6489 6490 $ for m in gdb insight 6491 do 6492 ( mkdir -p $m && cd $m && cvs -q -f -d /cvs/src co -P -r $b $m ) 6493 done 6494 $ 6495 6496 Note: 6497 6498 * The reading of `.cvsrc' is disabled (`-f') so that there isn't any 6499 confusion between what is written here and what your local `cvs' 6500 really does. 6501 6502 Update relevant files. 6503 ...................... 6504 6505 `gdb/NEWS' 6506 Major releases get their comments added as part of the mainline. 6507 Minor releases should probably mention any significant bugs that 6508 were fixed. 6509 6510 Don't forget to include the `ChangeLog' entry. 6511 6512 $ emacs gdb/src/gdb/NEWS 6513 ... 6514 c-x 4 a 6515 ... 6516 c-x c-s c-x c-c 6517 $ cp gdb/src/gdb/NEWS insight/src/gdb/NEWS 6518 $ cp gdb/src/gdb/ChangeLog insight/src/gdb/ChangeLog 6519 6520 `gdb/README' 6521 You'll need to update: 6522 6523 * The version. 6524 6525 * The update date. 6526 6527 * Who did it. 6528 6529 $ emacs gdb/src/gdb/README 6530 ... 6531 c-x 4 a 6532 ... 6533 c-x c-s c-x c-c 6534 $ cp gdb/src/gdb/README insight/src/gdb/README 6535 $ cp gdb/src/gdb/ChangeLog insight/src/gdb/ChangeLog 6536 6537 _Maintainer note: Hopefully the `README' file was reviewed before 6538 the initial branch was cut so just a simple substitute is needed 6539 to get it updated._ 6540 6541 _Maintainer note: Other projects generate `README' and `INSTALL' 6542 from the core documentation. This might be worth pursuing._ 6543 6544 `gdb/version.in' 6545 $ echo $v > gdb/src/gdb/version.in 6546 $ cat gdb/src/gdb/version.in 6547 5.2 6548 $ emacs gdb/src/gdb/version.in 6549 ... 6550 c-x 4 a 6551 ... Bump to version ... 6552 c-x c-s c-x c-c 6553 $ cp gdb/src/gdb/version.in insight/src/gdb/version.in 6554 $ cp gdb/src/gdb/ChangeLog insight/src/gdb/ChangeLog 6555 6556 6557 Do the dirty work 6558 ................. 6559 6560 This is identical to the process used to create the daily snapshot. 6561 6562 $ for m in gdb insight 6563 do 6564 ( cd $m/src && gmake -f src-release $m.tar ) 6565 done 6566 6567 If the top level source directory does not have `src-release' (GDB 6568 version 5.3.1 or earlier), try these commands instead: 6569 6570 $ for m in gdb insight 6571 do 6572 ( cd $m/src && gmake -f Makefile.in $m.tar ) 6573 done 6574 6575 Check the source files 6576 ...................... 6577 6578 You're looking for files that have mysteriously disappeared. 6579 `distclean' has the habit of deleting files it shouldn't. Watch out 6580 for the `version.in' update `cronjob'. 6581 6582 $ ( cd gdb/src && cvs -f -q -n update ) 6583 M djunpack.bat 6584 ? gdb-5.1.91.tar 6585 ? proto-toplev 6586 ... lots of generated files ... 6587 M gdb/ChangeLog 6588 M gdb/NEWS 6589 M gdb/README 6590 M gdb/version.in 6591 ... lots of generated files ... 6592 $ 6593 6594 _Don't worry about the `gdb.info-??' or `gdb/p-exp.tab.c'. They were 6595 generated (and yes `gdb.info-1' was also generated only something 6596 strange with CVS means that they didn't get suppressed). Fixing it 6597 would be nice though._ 6598 6599 Create compressed versions of the release 6600 ......................................... 6601 6602 $ cp */src/*.tar . 6603 $ cp */src/*.bz2 . 6604 $ ls -F 6605 gdb/ gdb-5.2.tar insight/ insight-5.2.tar 6606 $ for m in gdb insight 6607 do 6608 bzip2 -v -9 -c $m-$v.tar > $m-$v.tar.bz2 6609 gzip -v -9 -c $m-$v.tar > $m-$v.tar.gz 6610 done 6611 $ 6612 6613 Note: 6614 6615 * A pipe such as `bunzip2 < xxx.bz2 | gzip -9 > xxx.gz' is not since, 6616 in that mode, `gzip' does not know the name of the file and, hence, 6617 can not include it in the compressed file. This is also why the 6618 release process runs `tar' and `bzip2' as separate passes. 6619 6620 20.6.2 Sanity check the tar ball 6621 -------------------------------- 6622 6623 Pick a popular machine (Solaris/PPC?) and try the build on that. 6624 6625 $ bunzip2 < gdb-5.2.tar.bz2 | tar xpf - 6626 $ cd gdb-5.2 6627 $ ./configure 6628 $ make 6629 ... 6630 $ ./gdb/gdb ./gdb/gdb 6631 GNU gdb 5.2 6632 ... 6633 (gdb) b main 6634 Breakpoint 1 at 0x80732bc: file main.c, line 734. 6635 (gdb) run 6636 Starting program: /tmp/gdb-5.2/gdb/gdb 6637 6638 Breakpoint 1, main (argc=1, argv=0xbffff8b4) at main.c:734 6639 734 catch_errors (captured_main, &args, "", RETURN_MASK_ALL); 6640 (gdb) print args 6641 $1 = {argc = 136426532, argv = 0x821b7f0} 6642 (gdb) 6643 6644 20.6.3 Make a release candidate available 6645 ----------------------------------------- 6646 6647 If this is a release candidate then the only remaining steps are: 6648 6649 1. Commit `version.in' and `ChangeLog' 6650 6651 2. Tweak `version.in' (and `ChangeLog' to read L.M.N-0000-00-00-cvs 6652 so that the version update process can restart. 6653 6654 3. Make the release candidate available in 6655 `ftp://sourceware.org/pub/gdb/snapshots/branch' 6656 6657 4. Notify the relevant mailing lists ( <gdb (a] sourceware.org> and 6658 <gdb-testers (a] sourceware.org> that the candidate is available. 6659 6660 20.6.4 Make a formal release available 6661 -------------------------------------- 6662 6663 (And you thought all that was required was to post an e-mail.) 6664 6665 Install on sware 6666 ................ 6667 6668 Copy the new files to both the release and the old release directory: 6669 6670 $ cp *.bz2 *.gz ~ftp/pub/gdb/old-releases/ 6671 $ cp *.bz2 *.gz ~ftp/pub/gdb/releases 6672 6673 Clean up the releases directory so that only the most recent releases 6674 are available (e.g. keep 5.2 and 5.2.1 but remove 5.1): 6675 6676 $ cd ~ftp/pub/gdb/releases 6677 $ rm ... 6678 6679 Update the file `README' and `.message' in the releases directory: 6680 6681 $ vi README 6682 ... 6683 $ rm -f .message 6684 $ ln README .message 6685 6686 Update the web pages. 6687 ..................... 6688 6689 `htdocs/download/ANNOUNCEMENT' 6690 This file, which is posted as the official announcement, includes: 6691 * General announcement. 6692 6693 * News. If making an M.N.1 release, retain the news from 6694 earlier M.N release. 6695 6696 * Errata. 6697 6698 `htdocs/index.html' 6699 `htdocs/news/index.html' 6700 `htdocs/download/index.html' 6701 These files include: 6702 * Announcement of the most recent release. 6703 6704 * News entry (remember to update both the top level and the 6705 news directory). 6706 These pages also need to be regenerate using `index.sh'. 6707 6708 `download/onlinedocs/' 6709 You need to find the magic command that is used to generate the 6710 online docs from the `.tar.bz2'. The best way is to look in the 6711 output from one of the nightly `cron' jobs and then just edit 6712 accordingly. Something like: 6713 6714 $ ~/ss/update-web-docs \ 6715 ~ftp/pub/gdb/releases/gdb-5.2.tar.bz2 \ 6716 $PWD/www \ 6717 /www/sourceware/htdocs/gdb/download/onlinedocs \ 6718 gdb 6719 6720 `download/ari/' 6721 Just like the online documentation. Something like: 6722 6723 $ /bin/sh ~/ss/update-web-ari \ 6724 ~ftp/pub/gdb/releases/gdb-5.2.tar.bz2 \ 6725 $PWD/www \ 6726 /www/sourceware/htdocs/gdb/download/ari \ 6727 gdb 6728 6729 6730 Shadow the pages onto gnu 6731 ......................... 6732 6733 Something goes here. 6734 6735 Install the GDB tar ball on GNU 6736 ............................... 6737 6738 At the time of writing, the GNU machine was `gnudist.gnu.org' in 6739 `~ftp/gnu/gdb'. 6740 6741 Make the `ANNOUNCEMENT' 6742 ....................... 6743 6744 Post the `ANNOUNCEMENT' file you created above to: 6745 6746 * GDB Announcement mailing list <gdb-announce (a] sourceware.org> 6747 6748 * General GNU Announcement list <info-gnu (a] gnu.org> (but delay it a 6749 day or so to let things get out) 6750 6751 * GDB Bug Report mailing list <bug-gdb (a] gnu.org> 6752 6753 20.6.5 Cleanup 6754 -------------- 6755 6756 The release is out but you're still not finished. 6757 6758 Commit outstanding changes 6759 .......................... 6760 6761 In particular you'll need to commit any changes to: 6762 6763 * `gdb/ChangeLog' 6764 6765 * `gdb/version.in' 6766 6767 * `gdb/NEWS' 6768 6769 * `gdb/README' 6770 6771 Tag the release 6772 ............... 6773 6774 Something like: 6775 6776 $ d=`date -u +%Y-%m-%d` 6777 $ echo $d 6778 2002-01-24 6779 $ ( cd insight/src/gdb && cvs -f -q update ) 6780 $ ( cd insight/src && cvs -f -q tag gdb_5_2-$d-release ) 6781 6782 Insight is used since that contains more of the release than GDB. 6783 6784 Mention the release on the trunk 6785 ................................ 6786 6787 Just put something in the `ChangeLog' so that the trunk also indicates 6788 when the release was made. 6789 6790 Restart `gdb/version.in' 6791 ........................ 6792 6793 If `gdb/version.in' does not contain an ISO date such as `2002-01-24' 6794 then the daily `cronjob' won't update it. Having committed all the 6795 release changes it can be set to `5.2.0_0000-00-00-cvs' which will 6796 restart things (yes the `_' is important - it affects the snapshot 6797 process). 6798 6799 Don't forget the `ChangeLog'. 6800 6801 Merge into trunk 6802 ................ 6803 6804 The files committed to the branch may also need changes merged into the 6805 trunk. 6806 6807 Revise the release schedule 6808 ........................... 6809 6810 Post a revised release schedule to GDB Discussion List 6811 <gdb (a] sourceware.org> with an updated announcement. The schedule can be 6812 generated by running: 6813 6814 $ ~/ss/schedule `date +%s` schedule 6815 6816 The first parameter is approximate date/time in seconds (from the epoch) 6817 of the most recent release. 6818 6819 Also update the schedule `cronjob'. 6820 6821 20.7 Post release 6822 ================= 6823 6824 Remove any `OBSOLETE' code. 6825 6826 6827 File: gdbint.info, Node: Testsuite, Next: Hints, Prev: Releasing GDB, Up: Top 6828 6829 21 Testsuite 6830 ************ 6831 6832 The testsuite is an important component of the GDB package. While it 6833 is always worthwhile to encourage user testing, in practice this is 6834 rarely sufficient; users typically use only a small subset of the 6835 available commands, and it has proven all too common for a change to 6836 cause a significant regression that went unnoticed for some time. 6837 6838 The GDB testsuite uses the DejaGNU testing framework. The tests 6839 themselves are calls to various `Tcl' procs; the framework runs all the 6840 procs and summarizes the passes and fails. 6841 6842 21.1 Using the Testsuite 6843 ======================== 6844 6845 To run the testsuite, simply go to the GDB object directory (or to the 6846 testsuite's objdir) and type `make check'. This just sets up some 6847 environment variables and invokes DejaGNU's `runtest' script. While 6848 the testsuite is running, you'll get mentions of which test file is in 6849 use, and a mention of any unexpected passes or fails. When the 6850 testsuite is finished, you'll get a summary that looks like this: 6851 6852 === gdb Summary === 6853 6854 # of expected passes 6016 6855 # of unexpected failures 58 6856 # of unexpected successes 5 6857 # of expected failures 183 6858 # of unresolved testcases 3 6859 # of untested testcases 5 6860 6861 To run a specific test script, type: 6862 make check RUNTESTFLAGS='TESTS' 6863 where TESTS is a list of test script file names, separated by spaces. 6864 6865 If you use GNU make, you can use its `-j' option to run the 6866 testsuite in parallel. This can greatly reduce the amount of time it 6867 takes for the testsuite to run. In this case, if you set 6868 `RUNTESTFLAGS' then, by default, the tests will be run serially even 6869 under `-j'. You can override this and force a parallel run by setting 6870 the `make' variable `FORCE_PARALLEL' to any non-empty value. Note that 6871 the parallel `make check' assumes that you want to run the entire 6872 testsuite, so it is not compatible with some dejagnu options, like 6873 `--directory'. 6874 6875 The ideal test run consists of expected passes only; however, reality 6876 conspires to keep us from this ideal. Unexpected failures indicate 6877 real problems, whether in GDB or in the testsuite. Expected failures 6878 are still failures, but ones which have been decided are too hard to 6879 deal with at the time; for instance, a test case might work everywhere 6880 except on AIX, and there is no prospect of the AIX case being fixed in 6881 the near future. Expected failures should not be added lightly, since 6882 you may be masking serious bugs in GDB. Unexpected successes are 6883 expected fails that are passing for some reason, while unresolved and 6884 untested cases often indicate some minor catastrophe, such as the 6885 compiler being unable to deal with a test program. 6886 6887 When making any significant change to GDB, you should run the 6888 testsuite before and after the change, to confirm that there are no 6889 regressions. Note that truly complete testing would require that you 6890 run the testsuite with all supported configurations and a variety of 6891 compilers; however this is more than really necessary. In many cases 6892 testing with a single configuration is sufficient. Other useful 6893 options are to test one big-endian (Sparc) and one little-endian (x86) 6894 host, a cross config with a builtin simulator (powerpc-eabi, mips-elf), 6895 or a 64-bit host (Alpha). 6896 6897 If you add new functionality to GDB, please consider adding tests 6898 for it as well; this way future GDB hackers can detect and fix their 6899 changes that break the functionality you added. Similarly, if you fix 6900 a bug that was not previously reported as a test failure, please add a 6901 test case for it. Some cases are extremely difficult to test, such as 6902 code that handles host OS failures or bugs in particular versions of 6903 compilers, and it's OK not to try to write tests for all of those. 6904 6905 DejaGNU supports separate build, host, and target machines. However, 6906 some GDB test scripts do not work if the build machine and the host 6907 machine are not the same. In such an environment, these scripts will 6908 give a result of "UNRESOLVED", like this: 6909 6910 UNRESOLVED: gdb.base/example.exp: This test script does not work on a remote host. 6911 6912 21.2 Testsuite Parameters 6913 ========================= 6914 6915 Several variables exist to modify the behavior of the testsuite. 6916 6917 * `TRANSCRIPT' 6918 6919 Sometimes it is convenient to get a transcript of the commands 6920 which the testsuite sends to GDB. For example, if GDB crashes 6921 during testing, a transcript can be used to more easily 6922 reconstruct the failure when running GDB under GDB. 6923 6924 You can instruct the GDB testsuite to write transcripts by setting 6925 the DejaGNU variable `TRANSCRIPT' (to any value) before invoking 6926 `runtest' or `make check'. The transcripts will be written into 6927 DejaGNU's output directory. One transcript will be made for each 6928 invocation of GDB; they will be named `transcript.N', where N is 6929 an integer. The first line of the transcript file will show how 6930 GDB was invoked; each subsequent line is a command sent as input 6931 to GDB. 6932 6933 make check RUNTESTFLAGS=TRANSCRIPT=y 6934 6935 Note that the transcript is not always complete. In particular, 6936 tests of completion can yield partial command lines. 6937 6938 * `GDB' 6939 6940 Sometimes one wishes to test a different GDB than the one in the 6941 build directory. For example, one may wish to run the testsuite on 6942 `/usr/bin/gdb'. 6943 6944 make check RUNTESTFLAGS=GDB=/usr/bin/gdb 6945 6946 * `GDBSERVER' 6947 6948 When testing a different GDB, it is often useful to also test a 6949 different gdbserver. 6950 6951 make check RUNTESTFLAGS="GDB=/usr/bin/gdb GDBSERVER=/usr/bin/gdbserver" 6952 6953 * `INTERNAL_GDBFLAGS' 6954 6955 When running the testsuite normally one doesn't want whatever is in 6956 `~/.gdbinit' to interfere with the tests, therefore the test 6957 harness passes `-nx' to GDB. One also doesn't want any windowed 6958 version of GDB, e.g., `gdbtui', to run. This is achieved via 6959 `INTERNAL_GDBFLAGS'. 6960 6961 set INTERNAL_GDBFLAGS "-nw -nx" 6962 6963 This is all well and good, except when testing an installed GDB 6964 that has been configured with `--with-system-gdbinit'. Here one 6965 does not want `~/.gdbinit' loaded but one may want the system 6966 `.gdbinit' file loaded. This can be achieved by pointing `$HOME' 6967 at a directory without a `.gdbinit' and by overriding 6968 `INTERNAL_GDBFLAGS' and removing `-nx'. 6969 6970 cd testsuite 6971 HOME=`pwd` runtest \ 6972 GDB=/usr/bin/gdb \ 6973 GDBSERVER=/usr/bin/gdbserver \ 6974 INTERNAL_GDBFLAGS=-nw 6975 6976 6977 There are two ways to run the testsuite and pass additional 6978 parameters to DejaGnu. The first is with `make check' and specifying 6979 the makefile variable `RUNTESTFLAGS'. 6980 6981 make check RUNTESTFLAGS=TRANSCRIPT=y 6982 6983 The second is to cd to the `testsuite' directory and invoke the 6984 DejaGnu `runtest' command directly. 6985 6986 cd testsuite 6987 make site.exp 6988 runtest TRANSCRIPT=y 6989 6990 21.3 Testsuite Configuration 6991 ============================ 6992 6993 It is possible to adjust the behavior of the testsuite by defining the 6994 global variables listed below, either in a `site.exp' file, or in a 6995 board file. 6996 6997 * `gdb_test_timeout' 6998 6999 Defining this variable changes the default timeout duration used 7000 during communication with GDB. More specifically, the global 7001 variable used during testing is `timeout', but this variable gets 7002 reset to `gdb_test_timeout' at the beginning of each testcase, 7003 making sure that any local change to `timeout' in a testcase does 7004 not affect subsequent testcases. 7005 7006 This global variable comes in handy when the debugger is slower 7007 than normal due to the testing environment, triggering unexpected 7008 `TIMEOUT' test failures. Examples include when testing on a 7009 remote machine, or against a system where communications are slow. 7010 7011 If not specifically defined, this variable gets automatically 7012 defined to the same value as `timeout' during the testsuite 7013 initialization. The default value of the timeout is defined in 7014 the file `gdb/testsuite/config/unix.exp' that is part of the GDB 7015 test suite(1). 7016 7017 7018 21.4 Testsuite Organization 7019 =========================== 7020 7021 The testsuite is entirely contained in `gdb/testsuite'. While the 7022 testsuite includes some makefiles and configury, these are very minimal, 7023 and used for little besides cleaning up, since the tests themselves 7024 handle the compilation of the programs that GDB will run. The file 7025 `testsuite/lib/gdb.exp' contains common utility procs useful for all 7026 GDB tests, while the directory `testsuite/config' contains 7027 configuration-specific files, typically used for special-purpose 7028 definitions of procs like `gdb_load' and `gdb_start'. 7029 7030 The tests themselves are to be found in `testsuite/gdb.*' and 7031 subdirectories of those. The names of the test files must always end 7032 with `.exp'. DejaGNU collects the test files by wildcarding in the 7033 test directories, so both subdirectories and individual files get 7034 chosen and run in alphabetical order. 7035 7036 The following table lists the main types of subdirectories and what 7037 they are for. Since DejaGNU finds test files no matter where they are 7038 located, and since each test file sets up its own compilation and 7039 execution environment, this organization is simply for convenience and 7040 intelligibility. 7041 7042 `gdb.base' 7043 This is the base testsuite. The tests in it should apply to all 7044 configurations of GDB (but generic native-only tests may live 7045 here). The test programs should be in the subset of C that is 7046 valid K&R, ANSI/ISO, and C++ (`#ifdef's are allowed if necessary, 7047 for instance for prototypes). 7048 7049 `gdb.LANG' 7050 Language-specific tests for any language LANG besides C. Examples 7051 are `gdb.cp' and `gdb.java'. 7052 7053 `gdb.PLATFORM' 7054 Non-portable tests. The tests are specific to a specific 7055 configuration (host or target), such as HP-UX or eCos. Example is 7056 `gdb.hp', for HP-UX. 7057 7058 `gdb.COMPILER' 7059 Tests specific to a particular compiler. As of this writing (June 7060 1999), there aren't currently any groups of tests in this category 7061 that couldn't just as sensibly be made platform-specific, but one 7062 could imagine a `gdb.gcc', for tests of GDB's handling of GCC 7063 extensions. 7064 7065 `gdb.SUBSYSTEM' 7066 Tests that exercise a specific GDB subsystem in more depth. For 7067 instance, `gdb.disasm' exercises various disassemblers, while 7068 `gdb.stabs' tests pathways through the stabs symbol reader. 7069 7070 21.5 Writing Tests 7071 ================== 7072 7073 In many areas, the GDB tests are already quite comprehensive; you 7074 should be able to copy existing tests to handle new cases. 7075 7076 You should try to use `gdb_test' whenever possible, since it 7077 includes cases to handle all the unexpected errors that might happen. 7078 However, it doesn't cost anything to add new test procedures; for 7079 instance, `gdb.base/exprs.exp' defines a `test_expr' that calls 7080 `gdb_test' multiple times. 7081 7082 Only use `send_gdb' and `gdb_expect' when absolutely necessary. 7083 Even if GDB has several valid responses to a command, you can use 7084 `gdb_test_multiple'. Like `gdb_test', `gdb_test_multiple' recognizes 7085 internal errors and unexpected prompts. 7086 7087 Do not write tests which expect a literal tab character from GDB. 7088 On some operating systems (e.g. OpenBSD) the TTY layer expands tabs to 7089 spaces, so by the time GDB's output reaches expect the tab is gone. 7090 7091 The source language programs do _not_ need to be in a consistent 7092 style. Since GDB is used to debug programs written in many different 7093 styles, it's worth having a mix of styles in the testsuite; for 7094 instance, some GDB bugs involving the display of source lines would 7095 never manifest themselves if the programs used GNU coding style 7096 uniformly. 7097 7098 ---------- Footnotes ---------- 7099 7100 (1) If you are using a board file, it could override the test-suite 7101 default; search the board file for "timeout". 7102 7103 7104 File: gdbint.info, Node: Hints, Next: GDB Observers, Prev: Testsuite, Up: Top 7105 7106 22 Hints 7107 ******** 7108 7109 Check the `README' file, it often has useful information that does not 7110 appear anywhere else in the directory. 7111 7112 * Menu: 7113 7114 * Getting Started:: Getting started working on GDB 7115 * Debugging GDB:: Debugging GDB with itself 7116 7117 7118 File: gdbint.info, Node: Getting Started, Up: Hints 7119 7120 22.1 Getting Started 7121 ==================== 7122 7123 GDB is a large and complicated program, and if you first starting to 7124 work on it, it can be hard to know where to start. Fortunately, if you 7125 know how to go about it, there are ways to figure out what is going on. 7126 7127 This manual, the GDB Internals manual, has information which applies 7128 generally to many parts of GDB. 7129 7130 Information about particular functions or data structures are 7131 located in comments with those functions or data structures. If you 7132 run across a function or a global variable which does not have a 7133 comment correctly explaining what is does, this can be thought of as a 7134 bug in GDB; feel free to submit a bug report, with a suggested comment 7135 if you can figure out what the comment should say. If you find a 7136 comment which is actually wrong, be especially sure to report that. 7137 7138 Comments explaining the function of macros defined in host, target, 7139 or native dependent files can be in several places. Sometimes they are 7140 repeated every place the macro is defined. Sometimes they are where the 7141 macro is used. Sometimes there is a header file which supplies a 7142 default definition of the macro, and the comment is there. This manual 7143 also documents all the available macros. 7144 7145 Start with the header files. Once you have some idea of how GDB's 7146 internal symbol tables are stored (see `symtab.h', `gdbtypes.h'), you 7147 will find it much easier to understand the code which uses and creates 7148 those symbol tables. 7149 7150 You may wish to process the information you are getting somehow, to 7151 enhance your understanding of it. Summarize it, translate it to another 7152 language, add some (perhaps trivial or non-useful) feature to GDB, use 7153 the code to predict what a test case would do and write the test case 7154 and verify your prediction, etc. If you are reading code and your eyes 7155 are starting to glaze over, this is a sign you need to use a more active 7156 approach. 7157 7158 Once you have a part of GDB to start with, you can find more 7159 specifically the part you are looking for by stepping through each 7160 function with the `next' command. Do not use `step' or you will 7161 quickly get distracted; when the function you are stepping through 7162 calls another function try only to get a big-picture understanding 7163 (perhaps using the comment at the beginning of the function being 7164 called) of what it does. This way you can identify which of the 7165 functions being called by the function you are stepping through is the 7166 one which you are interested in. You may need to examine the data 7167 structures generated at each stage, with reference to the comments in 7168 the header files explaining what the data structures are supposed to 7169 look like. 7170 7171 Of course, this same technique can be used if you are just reading 7172 the code, rather than actually stepping through it. The same general 7173 principle applies--when the code you are looking at calls something 7174 else, just try to understand generally what the code being called does, 7175 rather than worrying about all its details. 7176 7177 A good place to start when tracking down some particular area is with 7178 a command which invokes that feature. Suppose you want to know how 7179 single-stepping works. As a GDB user, you know that the `step' command 7180 invokes single-stepping. The command is invoked via command tables 7181 (see `command.h'); by convention the function which actually performs 7182 the command is formed by taking the name of the command and adding 7183 `_command', or in the case of an `info' subcommand, `_info'. For 7184 example, the `step' command invokes the `step_command' function and the 7185 `info display' command invokes `display_info'. When this convention is 7186 not followed, you might have to use `grep' or `M-x tags-search' in 7187 emacs, or run GDB on itself and set a breakpoint in `execute_command'. 7188 7189 If all of the above fail, it may be appropriate to ask for 7190 information on `bug-gdb'. But _never_ post a generic question like "I 7191 was wondering if anyone could give me some tips about understanding 7192 GDB"--if we had some magic secret we would put it in this manual. 7193 Suggestions for improving the manual are always welcome, of course. 7194 7195 7196 File: gdbint.info, Node: Debugging GDB, Up: Hints 7197 7198 22.2 Debugging GDB with itself 7199 ============================== 7200 7201 If GDB is limping on your machine, this is the preferred way to get it 7202 fully functional. Be warned that in some ancient Unix systems, like 7203 Ultrix 4.2, a program can't be running in one process while it is being 7204 debugged in another. Rather than typing the command `./gdb ./gdb', 7205 which works on Suns and such, you can copy `gdb' to `gdb2' and then 7206 type `./gdb ./gdb2'. 7207 7208 When you run GDB in the GDB source directory, it will read a 7209 `.gdbinit' file that sets up some simple things to make debugging gdb 7210 easier. The `info' command, when executed without a subcommand in a 7211 GDB being debugged by gdb, will pop you back up to the top level gdb. 7212 See `.gdbinit' for details. 7213 7214 If you use emacs, you will probably want to do a `make TAGS' after 7215 you configure your distribution; this will put the machine dependent 7216 routines for your local machine where they will be accessed first by 7217 `M-.' 7218 7219 Also, make sure that you've either compiled GDB with your local cc, 7220 or have run `fixincludes' if you are compiling with gcc. 7221 7222 22.3 Submitting Patches 7223 ======================= 7224 7225 Thanks for thinking of offering your changes back to the community of 7226 GDB users. In general we like to get well designed enhancements. 7227 Thanks also for checking in advance about the best way to transfer the 7228 changes. 7229 7230 The GDB maintainers will only install "cleanly designed" patches. 7231 This manual summarizes what we believe to be clean design for GDB. 7232 7233 If the maintainers don't have time to put the patch in when it 7234 arrives, or if there is any question about a patch, it goes into a 7235 large queue with everyone else's patches and bug reports. 7236 7237 The legal issue is that to incorporate substantial changes requires a 7238 copyright assignment from you and/or your employer, granting ownership 7239 of the changes to the Free Software Foundation. You can get the 7240 standard documents for doing this by sending mail to `gnu (a] gnu.org' and 7241 asking for it. We recommend that people write in "All programs owned 7242 by the Free Software Foundation" as "NAME OF PROGRAM", so that changes 7243 in many programs (not just GDB, but GAS, Emacs, GCC, etc) can be 7244 contributed with only one piece of legalese pushed through the 7245 bureaucracy and filed with the FSF. We can't start merging changes 7246 until this paperwork is received by the FSF (their rules, which we 7247 follow since we maintain it for them). 7248 7249 Technically, the easiest way to receive changes is to receive each 7250 feature as a small context diff or unidiff, suitable for `patch'. Each 7251 message sent to me should include the changes to C code and header 7252 files for a single feature, plus `ChangeLog' entries for each directory 7253 where files were modified, and diffs for any changes needed to the 7254 manuals (`gdb/doc/gdb.texinfo' or `gdb/doc/gdbint.texinfo'). If there 7255 are a lot of changes for a single feature, they can be split down into 7256 multiple messages. 7257 7258 In this way, if we read and like the feature, we can add it to the 7259 sources with a single patch command, do some testing, and check it in. 7260 If you leave out the `ChangeLog', we have to write one. If you leave 7261 out the doc, we have to puzzle out what needs documenting. Etc., etc. 7262 7263 The reason to send each change in a separate message is that we will 7264 not install some of the changes. They'll be returned to you with 7265 questions or comments. If we're doing our job correctly, the message 7266 back to you will say what you have to fix in order to make the change 7267 acceptable. The reason to have separate messages for separate features 7268 is so that the acceptable changes can be installed while one or more 7269 changes are being reworked. If multiple features are sent in a single 7270 message, we tend to not put in the effort to sort out the acceptable 7271 changes from the unacceptable, so none of the features get installed 7272 until all are acceptable. 7273 7274 If this sounds painful or authoritarian, well, it is. But we get a 7275 lot of bug reports and a lot of patches, and many of them don't get 7276 installed because we don't have the time to finish the job that the bug 7277 reporter or the contributor could have done. Patches that arrive 7278 complete, working, and well designed, tend to get installed on the day 7279 they arrive. The others go into a queue and get installed as time 7280 permits, which, since the maintainers have many demands to meet, may not 7281 be for quite some time. 7282 7283 Please send patches directly to the GDB maintainers 7284 <gdb-patches (a] sourceware.org>. 7285 7286 22.4 Build Script 7287 ================= 7288 7289 The script `gdb_buildall.sh' builds GDB with flag 7290 `--enable-targets=all' set. This builds GDB with all supported targets 7291 activated. This helps testing GDB when doing changes that affect more 7292 than one architecture and is much faster than using `gdb_mbuild.sh'. 7293 7294 After building GDB the script checks which architectures are 7295 supported and then switches the current architecture to each of those 7296 to get information about the architecture. The test results are stored 7297 in log files in the directory the script was called from. 7298 7299 7300 File: gdbint.info, Node: GDB Observers, Next: GNU Free Documentation License, Prev: Hints, Up: Top 7301 7302 Appendix A GDB Currently available observers 7303 ******************************************** 7304 7305 A.1 Implementation rationale 7306 ============================ 7307 7308 An "observer" is an entity which is interested in being notified when 7309 GDB reaches certain states, or certain events occur in GDB. The entity 7310 being observed is called the "subject". To receive notifications, the 7311 observer attaches a callback to the subject. One subject can have 7312 several observers. 7313 7314 `observer.c' implements an internal generic low-level event 7315 notification mechanism. This generic event notification mechanism is 7316 then re-used to implement the exported high-level notification 7317 management routines for all possible notifications. 7318 7319 The current implementation of the generic observer provides support 7320 for contextual data. This contextual data is given to the subject when 7321 attaching the callback. In return, the subject will provide this 7322 contextual data back to the observer as a parameter of the callback. 7323 7324 Note that the current support for the contextual data is only 7325 partial, as it lacks a mechanism that would deallocate this data when 7326 the callback is detached. This is not a problem so far, as this 7327 contextual data is only used internally to hold a function pointer. 7328 Later on, if a certain observer needs to provide support for user-level 7329 contextual data, then the generic notification mechanism will need to be 7330 enhanced to allow the observer to provide a routine to deallocate the 7331 data when attaching the callback. 7332 7333 The observer implementation is also currently not reentrant. In 7334 particular, it is therefore not possible to call the attach or detach 7335 routines during a notification. 7336 7337 A.2 Debugging 7338 ============= 7339 7340 Observer notifications can be traced using the command `set debug 7341 observer 1' (*note Optional messages about internal happenings: 7342 (gdb)Debugging Output.). 7343 7344 A.3 `normal_stop' Notifications 7345 =============================== 7346 7347 GDB notifies all `normal_stop' observers when the inferior execution 7348 has just stopped, the associated messages and annotations have been 7349 printed, and the control is about to be returned to the user. 7350 7351 Note that the `normal_stop' notification is not emitted when the 7352 execution stops due to a breakpoint, and this breakpoint has a 7353 condition that is not met. If the breakpoint has any associated 7354 commands list, the commands are executed after the notification is 7355 emitted. 7356 7357 The following interfaces are available to manage observers: 7358 7359 -- Function: extern struct observer *observer_attach_EVENT 7360 (observer_EVENT_ftype *F) 7361 Using the function F, create an observer that is notified when 7362 ever EVENT occurs, return the observer. 7363 7364 -- Function: extern void observer_detach_EVENT (struct observer 7365 *OBSERVER); 7366 Remove OBSERVER from the list of observers to be notified when 7367 EVENT occurs. 7368 7369 -- Function: extern void observer_notify_EVENT (void); 7370 Send a notification to all EVENT observers. 7371 7372 The following observable events are defined: 7373 7374 -- Function: void normal_stop (struct bpstats *BS, int PRINT_FRAME) 7375 The inferior has stopped for real. The BS argument describes the 7376 breakpoints were are stopped at, if any. Second argument 7377 PRINT_FRAME non-zero means display the location where the inferior 7378 has stopped. 7379 7380 -- Function: void target_changed (struct target_ops *TARGET) 7381 The target's register contents have changed. 7382 7383 -- Function: void executable_changed (void) 7384 The executable being debugged by GDB has changed: The user decided 7385 to debug a different program, or the program he was debugging has 7386 been modified since being loaded by the debugger (by being 7387 recompiled, for instance). 7388 7389 -- Function: void inferior_created (struct target_ops *OBJFILE, int 7390 FROM_TTY) 7391 GDB has just connected to an inferior. For `run', GDB calls this 7392 observer while the inferior is still stopped at the entry-point 7393 instruction. For `attach' and `core', GDB calls this observer 7394 immediately after connecting to the inferior, and before any 7395 information on the inferior has been printed. 7396 7397 -- Function: void solib_loaded (struct so_list *SOLIB) 7398 The shared library specified by SOLIB has been loaded. Note that 7399 when GDB calls this observer, the library's symbols probably 7400 haven't been loaded yet. 7401 7402 -- Function: void solib_unloaded (struct so_list *SOLIB) 7403 The shared library specified by SOLIB has been unloaded. Note 7404 that when GDB calls this observer, the library's symbols have not 7405 been unloaded yet, and thus are still available. 7406 7407 -- Function: void new_objfile (struct objfile *OBJFILE) 7408 The symbol file specified by OBJFILE has been loaded. Called with 7409 OBJFILE equal to `NULL' to indicate previously loaded symbol table 7410 data has now been invalidated. 7411 7412 -- Function: void new_thread (struct thread_info *T) 7413 The thread specified by T has been created. 7414 7415 -- Function: void thread_exit (struct thread_info *T, int SILENT) 7416 The thread specified by T has exited. The SILENT argument 7417 indicates that GDB is removing the thread from its tables without 7418 wanting to notify the user about it. 7419 7420 -- Function: void thread_stop_requested (ptid_t PTID) 7421 An explicit stop request was issued to PTID. If PTID equals 7422 MINUS_ONE_PTID, the request applied to all threads. If 7423 `ptid_is_pid(ptid)' returns true, the request applied to all 7424 threads of the process pointed at by PTID. Otherwise, the request 7425 applied to the single thread pointed at by PTID. 7426 7427 -- Function: void target_resumed (ptid_t PTID) 7428 The target was resumed. The PTID parameter specifies which thread 7429 was resume, and may be RESUME_ALL if all threads are resumed. 7430 7431 -- Function: void about_to_proceed (void) 7432 The target is about to be proceeded. 7433 7434 -- Function: void breakpoint_created (int BPNUM) 7435 A new breakpoint has been created. The argument BPNUM is the 7436 number of the newly-created breakpoint. 7437 7438 -- Function: void breakpoint_deleted (int BPNUM) 7439 A breakpoint has been destroyed. The argument BPNUM is the number 7440 of the newly-destroyed breakpoint. 7441 7442 -- Function: void breakpoint_modified (int BPNUM) 7443 A breakpoint has been modified in some way. The argument BPNUM is 7444 the number of the modified breakpoint. 7445 7446 -- Function: void tracepoint_created (int TPNUM) 7447 A new tracepoint has been created. The argument TPNUM is the 7448 number of the newly-created tracepoint. 7449 7450 -- Function: void tracepoint_deleted (int TPNUM) 7451 A tracepoint has been destroyed. The argument TPNUM is the number 7452 of the newly-destroyed tracepoint. 7453 7454 -- Function: void tracepoint_modified (int TPNUM) 7455 A tracepoint has been modified in some way. The argument TPNUM is 7456 the number of the modified tracepoint. 7457 7458 -- Function: void architecture_changed (struct gdbarch *NEWARCH) 7459 The current architecture has changed. The argument NEWARCH is a 7460 pointer to the new architecture. 7461 7462 -- Function: void thread_ptid_changed (ptid_t OLD_PTID, ptid_t 7463 NEW_PTID) 7464 The thread's ptid has changed. The OLD_PTID parameter specifies 7465 the old value, and NEW_PTID specifies the new value. 7466 7467 -- Function: void inferior_appeared (int PID) 7468 GDB has attached to a new inferior identified by PID. 7469 7470 -- Function: void inferior_exit (int PID) 7471 Either GDB detached from the inferior, or the inferior exited. 7472 The argument PID identifies the inferior. 7473 7474 -- Function: void memory_changed (CORE_ADDR ADDR, int LEN, const 7475 bfd_byte *DATA) 7476 Bytes from DATA to DATA + LEN have been written to the current 7477 inferior at ADDR. 7478 7479 -- Function: void test_notification (int SOMEARG) 7480 This observer is used for internal testing. Do not use. See 7481 testsuite/gdb.gdb/observer.exp. 7482 7483 7484 File: gdbint.info, Node: GNU Free Documentation License, Next: Index, Prev: GDB Observers, Up: Top 7485 7486 Appendix B GNU Free Documentation License 7487 ***************************************** 7488 7489 Version 1.2, November 2002 7490 7491 Copyright (C) 2000,2001,2002 Free Software Foundation, Inc. 7492 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. 7493 7494 Everyone is permitted to copy and distribute verbatim copies 7495 of this license document, but changing it is not allowed. 7496 7497 0. PREAMBLE 7498 7499 The purpose of this License is to make a manual, textbook, or other 7500 functional and useful document "free" in the sense of freedom: to 7501 assure everyone the effective freedom to copy and redistribute it, 7502 with or without modifying it, either commercially or 7503 noncommercially. Secondarily, this License preserves for the 7504 author and publisher a way to get credit for their work, while not 7505 being considered responsible for modifications made by others. 7506 7507 This License is a kind of "copyleft", which means that derivative 7508 works of the document must themselves be free in the same sense. 7509 It complements the GNU General Public License, which is a copyleft 7510 license designed for free software. 7511 7512 We have designed this License in order to use it for manuals for 7513 free software, because free software needs free documentation: a 7514 free program should come with manuals providing the same freedoms 7515 that the software does. But this License is not limited to 7516 software manuals; it can be used for any textual work, regardless 7517 of subject matter or whether it is published as a printed book. 7518 We recommend this License principally for works whose purpose is 7519 instruction or reference. 7520 7521 1. APPLICABILITY AND DEFINITIONS 7522 7523 This License applies to any manual or other work, in any medium, 7524 that contains a notice placed by the copyright holder saying it 7525 can be distributed under the terms of this License. Such a notice 7526 grants a world-wide, royalty-free license, unlimited in duration, 7527 to use that work under the conditions stated herein. The 7528 "Document", below, refers to any such manual or work. Any member 7529 of the public is a licensee, and is addressed as "you". You 7530 accept the license if you copy, modify or distribute the work in a 7531 way requiring permission under copyright law. 7532 7533 A "Modified Version" of the Document means any work containing the 7534 Document or a portion of it, either copied verbatim, or with 7535 modifications and/or translated into another language. 7536 7537 A "Secondary Section" is a named appendix or a front-matter section 7538 of the Document that deals exclusively with the relationship of the 7539 publishers or authors of the Document to the Document's overall 7540 subject (or to related matters) and contains nothing that could 7541 fall directly within that overall subject. (Thus, if the Document 7542 is in part a textbook of mathematics, a Secondary Section may not 7543 explain any mathematics.) The relationship could be a matter of 7544 historical connection with the subject or with related matters, or 7545 of legal, commercial, philosophical, ethical or political position 7546 regarding them. 7547 7548 The "Invariant Sections" are certain Secondary Sections whose 7549 titles are designated, as being those of Invariant Sections, in 7550 the notice that says that the Document is released under this 7551 License. If a section does not fit the above definition of 7552 Secondary then it is not allowed to be designated as Invariant. 7553 The Document may contain zero Invariant Sections. If the Document 7554 does not identify any Invariant Sections then there are none. 7555 7556 The "Cover Texts" are certain short passages of text that are 7557 listed, as Front-Cover Texts or Back-Cover Texts, in the notice 7558 that says that the Document is released under this License. A 7559 Front-Cover Text may be at most 5 words, and a Back-Cover Text may 7560 be at most 25 words. 7561 7562 A "Transparent" copy of the Document means a machine-readable copy, 7563 represented in a format whose specification is available to the 7564 general public, that is suitable for revising the document 7565 straightforwardly with generic text editors or (for images 7566 composed of pixels) generic paint programs or (for drawings) some 7567 widely available drawing editor, and that is suitable for input to 7568 text formatters or for automatic translation to a variety of 7569 formats suitable for input to text formatters. A copy made in an 7570 otherwise Transparent file format whose markup, or absence of 7571 markup, has been arranged to thwart or discourage subsequent 7572 modification by readers is not Transparent. An image format is 7573 not Transparent if used for any substantial amount of text. A 7574 copy that is not "Transparent" is called "Opaque". 7575 7576 Examples of suitable formats for Transparent copies include plain 7577 ASCII without markup, Texinfo input format, LaTeX input format, 7578 SGML or XML using a publicly available DTD, and 7579 standard-conforming simple HTML, PostScript or PDF designed for 7580 human modification. Examples of transparent image formats include 7581 PNG, XCF and JPG. Opaque formats include proprietary formats that 7582 can be read and edited only by proprietary word processors, SGML or 7583 XML for which the DTD and/or processing tools are not generally 7584 available, and the machine-generated HTML, PostScript or PDF 7585 produced by some word processors for output purposes only. 7586 7587 The "Title Page" means, for a printed book, the title page itself, 7588 plus such following pages as are needed to hold, legibly, the 7589 material this License requires to appear in the title page. For 7590 works in formats which do not have any title page as such, "Title 7591 Page" means the text near the most prominent appearance of the 7592 work's title, preceding the beginning of the body of the text. 7593 7594 A section "Entitled XYZ" means a named subunit of the Document 7595 whose title either is precisely XYZ or contains XYZ in parentheses 7596 following text that translates XYZ in another language. (Here XYZ 7597 stands for a specific section name mentioned below, such as 7598 "Acknowledgements", "Dedications", "Endorsements", or "History".) 7599 To "Preserve the Title" of such a section when you modify the 7600 Document means that it remains a section "Entitled XYZ" according 7601 to this definition. 7602 7603 The Document may include Warranty Disclaimers next to the notice 7604 which states that this License applies to the Document. These 7605 Warranty Disclaimers are considered to be included by reference in 7606 this License, but only as regards disclaiming warranties: any other 7607 implication that these Warranty Disclaimers may have is void and 7608 has no effect on the meaning of this License. 7609 7610 2. VERBATIM COPYING 7611 7612 You may copy and distribute the Document in any medium, either 7613 commercially or noncommercially, provided that this License, the 7614 copyright notices, and the license notice saying this License 7615 applies to the Document are reproduced in all copies, and that you 7616 add no other conditions whatsoever to those of this License. You 7617 may not use technical measures to obstruct or control the reading 7618 or further copying of the copies you make or distribute. However, 7619 you may accept compensation in exchange for copies. If you 7620 distribute a large enough number of copies you must also follow 7621 the conditions in section 3. 7622 7623 You may also lend copies, under the same conditions stated above, 7624 and you may publicly display copies. 7625 7626 3. COPYING IN QUANTITY 7627 7628 If you publish printed copies (or copies in media that commonly 7629 have printed covers) of the Document, numbering more than 100, and 7630 the Document's license notice requires Cover Texts, you must 7631 enclose the copies in covers that carry, clearly and legibly, all 7632 these Cover Texts: Front-Cover Texts on the front cover, and 7633 Back-Cover Texts on the back cover. Both covers must also clearly 7634 and legibly identify you as the publisher of these copies. The 7635 front cover must present the full title with all words of the 7636 title equally prominent and visible. You may add other material 7637 on the covers in addition. Copying with changes limited to the 7638 covers, as long as they preserve the title of the Document and 7639 satisfy these conditions, can be treated as verbatim copying in 7640 other respects. 7641 7642 If the required texts for either cover are too voluminous to fit 7643 legibly, you should put the first ones listed (as many as fit 7644 reasonably) on the actual cover, and continue the rest onto 7645 adjacent pages. 7646 7647 If you publish or distribute Opaque copies of the Document 7648 numbering more than 100, you must either include a 7649 machine-readable Transparent copy along with each Opaque copy, or 7650 state in or with each Opaque copy a computer-network location from 7651 which the general network-using public has access to download 7652 using public-standard network protocols a complete Transparent 7653 copy of the Document, free of added material. If you use the 7654 latter option, you must take reasonably prudent steps, when you 7655 begin distribution of Opaque copies in quantity, to ensure that 7656 this Transparent copy will remain thus accessible at the stated 7657 location until at least one year after the last time you 7658 distribute an Opaque copy (directly or through your agents or 7659 retailers) of that edition to the public. 7660 7661 It is requested, but not required, that you contact the authors of 7662 the Document well before redistributing any large number of 7663 copies, to give them a chance to provide you with an updated 7664 version of the Document. 7665 7666 4. MODIFICATIONS 7667 7668 You may copy and distribute a Modified Version of the Document 7669 under the conditions of sections 2 and 3 above, provided that you 7670 release the Modified Version under precisely this License, with 7671 the Modified Version filling the role of the Document, thus 7672 licensing distribution and modification of the Modified Version to 7673 whoever possesses a copy of it. In addition, you must do these 7674 things in the Modified Version: 7675 7676 A. Use in the Title Page (and on the covers, if any) a title 7677 distinct from that of the Document, and from those of 7678 previous versions (which should, if there were any, be listed 7679 in the History section of the Document). You may use the 7680 same title as a previous version if the original publisher of 7681 that version gives permission. 7682 7683 B. List on the Title Page, as authors, one or more persons or 7684 entities responsible for authorship of the modifications in 7685 the Modified Version, together with at least five of the 7686 principal authors of the Document (all of its principal 7687 authors, if it has fewer than five), unless they release you 7688 from this requirement. 7689 7690 C. State on the Title page the name of the publisher of the 7691 Modified Version, as the publisher. 7692 7693 D. Preserve all the copyright notices of the Document. 7694 7695 E. Add an appropriate copyright notice for your modifications 7696 adjacent to the other copyright notices. 7697 7698 F. Include, immediately after the copyright notices, a license 7699 notice giving the public permission to use the Modified 7700 Version under the terms of this License, in the form shown in 7701 the Addendum below. 7702 7703 G. Preserve in that license notice the full lists of Invariant 7704 Sections and required Cover Texts given in the Document's 7705 license notice. 7706 7707 H. Include an unaltered copy of this License. 7708 7709 I. Preserve the section Entitled "History", Preserve its Title, 7710 and add to it an item stating at least the title, year, new 7711 authors, and publisher of the Modified Version as given on 7712 the Title Page. If there is no section Entitled "History" in 7713 the Document, create one stating the title, year, authors, 7714 and publisher of the Document as given on its Title Page, 7715 then add an item describing the Modified Version as stated in 7716 the previous sentence. 7717 7718 J. Preserve the network location, if any, given in the Document 7719 for public access to a Transparent copy of the Document, and 7720 likewise the network locations given in the Document for 7721 previous versions it was based on. These may be placed in 7722 the "History" section. You may omit a network location for a 7723 work that was published at least four years before the 7724 Document itself, or if the original publisher of the version 7725 it refers to gives permission. 7726 7727 K. For any section Entitled "Acknowledgements" or "Dedications", 7728 Preserve the Title of the section, and preserve in the 7729 section all the substance and tone of each of the contributor 7730 acknowledgements and/or dedications given therein. 7731 7732 L. Preserve all the Invariant Sections of the Document, 7733 unaltered in their text and in their titles. Section numbers 7734 or the equivalent are not considered part of the section 7735 titles. 7736 7737 M. Delete any section Entitled "Endorsements". Such a section 7738 may not be included in the Modified Version. 7739 7740 N. Do not retitle any existing section to be Entitled 7741 "Endorsements" or to conflict in title with any Invariant 7742 Section. 7743 7744 O. Preserve any Warranty Disclaimers. 7745 7746 If the Modified Version includes new front-matter sections or 7747 appendices that qualify as Secondary Sections and contain no 7748 material copied from the Document, you may at your option 7749 designate some or all of these sections as invariant. To do this, 7750 add their titles to the list of Invariant Sections in the Modified 7751 Version's license notice. These titles must be distinct from any 7752 other section titles. 7753 7754 You may add a section Entitled "Endorsements", provided it contains 7755 nothing but endorsements of your Modified Version by various 7756 parties--for example, statements of peer review or that the text 7757 has been approved by an organization as the authoritative 7758 definition of a standard. 7759 7760 You may add a passage of up to five words as a Front-Cover Text, 7761 and a passage of up to 25 words as a Back-Cover Text, to the end 7762 of the list of Cover Texts in the Modified Version. Only one 7763 passage of Front-Cover Text and one of Back-Cover Text may be 7764 added by (or through arrangements made by) any one entity. If the 7765 Document already includes a cover text for the same cover, 7766 previously added by you or by arrangement made by the same entity 7767 you are acting on behalf of, you may not add another; but you may 7768 replace the old one, on explicit permission from the previous 7769 publisher that added the old one. 7770 7771 The author(s) and publisher(s) of the Document do not by this 7772 License give permission to use their names for publicity for or to 7773 assert or imply endorsement of any Modified Version. 7774 7775 5. COMBINING DOCUMENTS 7776 7777 You may combine the Document with other documents released under 7778 this License, under the terms defined in section 4 above for 7779 modified versions, provided that you include in the combination 7780 all of the Invariant Sections of all of the original documents, 7781 unmodified, and list them all as Invariant Sections of your 7782 combined work in its license notice, and that you preserve all 7783 their Warranty Disclaimers. 7784 7785 The combined work need only contain one copy of this License, and 7786 multiple identical Invariant Sections may be replaced with a single 7787 copy. If there are multiple Invariant Sections with the same name 7788 but different contents, make the title of each such section unique 7789 by adding at the end of it, in parentheses, the name of the 7790 original author or publisher of that section if known, or else a 7791 unique number. Make the same adjustment to the section titles in 7792 the list of Invariant Sections in the license notice of the 7793 combined work. 7794 7795 In the combination, you must combine any sections Entitled 7796 "History" in the various original documents, forming one section 7797 Entitled "History"; likewise combine any sections Entitled 7798 "Acknowledgements", and any sections Entitled "Dedications". You 7799 must delete all sections Entitled "Endorsements." 7800 7801 6. COLLECTIONS OF DOCUMENTS 7802 7803 You may make a collection consisting of the Document and other 7804 documents released under this License, and replace the individual 7805 copies of this License in the various documents with a single copy 7806 that is included in the collection, provided that you follow the 7807 rules of this License for verbatim copying of each of the 7808 documents in all other respects. 7809 7810 You may extract a single document from such a collection, and 7811 distribute it individually under this License, provided you insert 7812 a copy of this License into the extracted document, and follow 7813 this License in all other respects regarding verbatim copying of 7814 that document. 7815 7816 7. AGGREGATION WITH INDEPENDENT WORKS 7817 7818 A compilation of the Document or its derivatives with other 7819 separate and independent documents or works, in or on a volume of 7820 a storage or distribution medium, is called an "aggregate" if the 7821 copyright resulting from the compilation is not used to limit the 7822 legal rights of the compilation's users beyond what the individual 7823 works permit. When the Document is included in an aggregate, this 7824 License does not apply to the other works in the aggregate which 7825 are not themselves derivative works of the Document. 7826 7827 If the Cover Text requirement of section 3 is applicable to these 7828 copies of the Document, then if the Document is less than one half 7829 of the entire aggregate, the Document's Cover Texts may be placed 7830 on covers that bracket the Document within the aggregate, or the 7831 electronic equivalent of covers if the Document is in electronic 7832 form. Otherwise they must appear on printed covers that bracket 7833 the whole aggregate. 7834 7835 8. TRANSLATION 7836 7837 Translation is considered a kind of modification, so you may 7838 distribute translations of the Document under the terms of section 7839 4. Replacing Invariant Sections with translations requires special 7840 permission from their copyright holders, but you may include 7841 translations of some or all Invariant Sections in addition to the 7842 original versions of these Invariant Sections. You may include a 7843 translation of this License, and all the license notices in the 7844 Document, and any Warranty Disclaimers, provided that you also 7845 include the original English version of this License and the 7846 original versions of those notices and disclaimers. In case of a 7847 disagreement between the translation and the original version of 7848 this License or a notice or disclaimer, the original version will 7849 prevail. 7850 7851 If a section in the Document is Entitled "Acknowledgements", 7852 "Dedications", or "History", the requirement (section 4) to 7853 Preserve its Title (section 1) will typically require changing the 7854 actual title. 7855 7856 9. TERMINATION 7857 7858 You may not copy, modify, sublicense, or distribute the Document 7859 except as expressly provided for under this License. Any other 7860 attempt to copy, modify, sublicense or distribute the Document is 7861 void, and will automatically terminate your rights under this 7862 License. However, parties who have received copies, or rights, 7863 from you under this License will not have their licenses 7864 terminated so long as such parties remain in full compliance. 7865 7866 10. FUTURE REVISIONS OF THIS LICENSE 7867 7868 The Free Software Foundation may publish new, revised versions of 7869 the GNU Free Documentation License from time to time. Such new 7870 versions will be similar in spirit to the present version, but may 7871 differ in detail to address new problems or concerns. See 7872 `http://www.gnu.org/copyleft/'. 7873 7874 Each version of the License is given a distinguishing version 7875 number. If the Document specifies that a particular numbered 7876 version of this License "or any later version" applies to it, you 7877 have the option of following the terms and conditions either of 7878 that specified version or of any later version that has been 7879 published (not as a draft) by the Free Software Foundation. If 7880 the Document does not specify a version number of this License, 7881 you may choose any version ever published (not as a draft) by the 7882 Free Software Foundation. 7883 7884 B.1 ADDENDUM: How to use this License for your documents 7885 ======================================================== 7886 7887 To use this License in a document you have written, include a copy of 7888 the License in the document and put the following copyright and license 7889 notices just after the title page: 7890 7891 Copyright (C) YEAR YOUR NAME. 7892 Permission is granted to copy, distribute and/or modify this document 7893 under the terms of the GNU Free Documentation License, Version 1.2 7894 or any later version published by the Free Software Foundation; 7895 with no Invariant Sections, no Front-Cover Texts, and no Back-Cover 7896 Texts. A copy of the license is included in the section entitled ``GNU 7897 Free Documentation License''. 7898 7899 If you have Invariant Sections, Front-Cover Texts and Back-Cover 7900 Texts, replace the "with...Texts." line with this: 7901 7902 with the Invariant Sections being LIST THEIR TITLES, with 7903 the Front-Cover Texts being LIST, and with the Back-Cover Texts 7904 being LIST. 7905 7906 If you have Invariant Sections without Cover Texts, or some other 7907 combination of the three, merge those two alternatives to suit the 7908 situation. 7909 7910 If your document contains nontrivial examples of program code, we 7911 recommend releasing these examples in parallel under your choice of 7912 free software license, such as the GNU General Public License, to 7913 permit their use in free software. 7914 7915 7916 File: gdbint.info, Node: Index, Prev: GNU Free Documentation License, Up: Top 7917 7918 Index 7919 ***** 7920 7921 [index] 7922 * Menu: 7923 7924 * $fp: Register Information Functions. 7925 (line 126) 7926 * $pc: Register Architecture Functions & Variables. 7927 (line 58) 7928 * $ps: Register Architecture Functions & Variables. 7929 (line 69) 7930 * $sp: Register Architecture Functions & Variables. 7931 (line 49) 7932 * _initialize_ARCH_tdep <1>: Adding a New Target. (line 22) 7933 * _initialize_ARCH_tdep: How an Architecture is Represented. 7934 (line 13) 7935 * _initialize_language: Language Support. (line 79) 7936 * a.out format: Symbol Handling. (line 213) 7937 * about_to_proceed: GDB Observers. (line 133) 7938 * abstract interpretation of function prologues: Algorithms. (line 48) 7939 * add_cmd: User Interface. (line 21) 7940 * add_com: User Interface. (line 21) 7941 * add_setshow_cmd: User Interface. (line 26) 7942 * add_setshow_cmd_full: User Interface. (line 26) 7943 * add_symtab_fns: Symbol Handling. (line 37) 7944 * adding a new host: Host Definition. (line 13) 7945 * adding a symbol-reading module: Symbol Handling. (line 37) 7946 * adding a target: Adding a New Target. (line 6) 7947 * adding debugging info reader: Symbol Handling. (line 360) 7948 * adding source language: Language Support. (line 17) 7949 * address classes: Address Classes. (line 6) 7950 * address representation: Pointers and Addresses. 7951 (line 6) 7952 * address spaces, separate data and code: Pointers and Addresses. 7953 (line 6) 7954 * address_class_name_to_type_flags: Defining Other Architecture Features. 7955 (line 28) 7956 * address_class_name_to_type_flags_p: Defining Other Architecture Features. 7957 (line 39) 7958 * algorithms: Algorithms. (line 6) 7959 * align_down: Functions and Variable to Analyze Frames. 7960 (line 46) 7961 * align_up: Functions and Variable to Analyze Frames. 7962 (line 46) 7963 * allocate_symtab: Language Support. (line 83) 7964 * ARCH-tdep.c: How an Architecture is Represented. 7965 (line 13) 7966 * architecture representation: How an Architecture is Represented. 7967 (line 6) 7968 * architecture_changed: GDB Observers. (line 160) 7969 * Array Containers: Support Libraries. (line 131) 7970 * assumptions about targets: Coding. (line 514) 7971 * ATTR_NORETURN: Host Definition. (line 126) 7972 * base of a frame: Frame Handling Terminology. 7973 (line 28) 7974 * BFD library: Support Libraries. (line 9) 7975 * bfd_arch_info: Looking Up an Existing Architecture. 7976 (line 41) 7977 * BIG_BREAKPOINT: Defining Other Architecture Features. 7978 (line 100) 7979 * BPT_VECTOR: Defining Other Architecture Features. 7980 (line 536) 7981 * BREAKPOINT: Defining Other Architecture Features. 7982 (line 88) 7983 * breakpoint address adjusted: Defining Other Architecture Features. 7984 (line 145) 7985 * breakpoint_created: GDB Observers. (line 136) 7986 * breakpoint_deleted: GDB Observers. (line 140) 7987 * breakpoint_modified: GDB Observers. (line 144) 7988 * breakpoints: Algorithms. (line 151) 7989 * bug-gdb mailing list: Getting Started. (line 72) 7990 * build script: Debugging GDB. (line 94) 7991 * C data types: Coding. (line 389) 7992 * call frame information: Algorithms. (line 14) 7993 * call stack frame: Stack Frames. (line 6) 7994 * calls to the inferior: Inferior Call Setup. (line 6) 7995 * CANNOT_STEP_HW_WATCHPOINTS: Algorithms. (line 408) 7996 * CC_HAS_LONG_LONG: Host Definition. (line 105) 7997 * CFI (call frame information): Algorithms. (line 14) 7998 * checkpoints: Algorithms. (line 604) 7999 * cleanups: Coding. (line 12) 8000 * CLI: User Interface. (line 12) 8001 * code pointers, word-addressed: Pointers and Addresses. 8002 (line 6) 8003 * coding standards: Coding. (line 215) 8004 * COFF debugging info: Symbol Handling. (line 310) 8005 * COFF format: Symbol Handling. (line 228) 8006 * command implementation: Getting Started. (line 60) 8007 * command interpreter: User Interface. (line 12) 8008 * comment formatting: Coding. (line 363) 8009 * compiler warnings: Coding. (line 271) 8010 * Compressed DWARF 2 debugging info: Symbol Handling. (line 330) 8011 * computed values: Values. (line 35) 8012 * configure.tgt: How an Architecture is Represented. 8013 (line 19) 8014 * converting between pointers and addresses: Pointers and Addresses. 8015 (line 6) 8016 * converting integers to addresses: Defining Other Architecture Features. 8017 (line 274) 8018 * cooked register representation: Raw and Cooked Registers. 8019 (line 6) 8020 * core files: Adding support for debugging core files. 8021 (line 6) 8022 * core_addr_greaterthan: Functions and Variable to Analyze Frames. 8023 (line 30) 8024 * core_addr_lessthan: Functions and Variable to Analyze Frames. 8025 (line 30) 8026 * CRLF_SOURCE_FILES: Host Definition. (line 86) 8027 * current_language: Language Support. (line 75) 8028 * D10V addresses: Pointers and Addresses. 8029 (line 6) 8030 * data output: User Interface. (line 254) 8031 * data-pointer, per-architecture/per-module: Coding. (line 100) 8032 * debugging GDB: Debugging GDB. (line 6) 8033 * DEFAULT_PROMPT: Host Definition. (line 93) 8034 * deprecate_cmd: User Interface. (line 32) 8035 * DEPRECATED_IBM6000_TARGET: Defining Other Architecture Features. 8036 (line 242) 8037 * deprecating commands: User Interface. (line 32) 8038 * design: Coding. (line 509) 8039 * DEV_TTY: Host Definition. (line 96) 8040 * DIRNAME_SEPARATOR: Coding. (line 579) 8041 * DISABLE_UNSETTABLE_BREAK: Defining Other Architecture Features. 8042 (line 211) 8043 * discard_cleanups: Coding. (line 39) 8044 * do_cleanups: Coding. (line 35) 8045 * DOS text files: Host Definition. (line 87) 8046 * dummy frames: About Dummy Frames. (line 6) 8047 * DW_AT_address_class: Address Classes. (line 6) 8048 * DW_AT_byte_size: Address Classes. (line 6) 8049 * DWARF 2 debugging info: Symbol Handling. (line 323) 8050 * DWARF 3 debugging info: Symbol Handling. (line 350) 8051 * ECOFF debugging info: Symbol Handling. (line 316) 8052 * ECOFF format: Symbol Handling. (line 243) 8053 * ELF format: Symbol Handling. (line 276) 8054 * evaluate_subexp: Language Support. (line 58) 8055 * executable_changed: GDB Observers. (line 85) 8056 * execution state: Managing Execution State. 8057 (line 6) 8058 * experimental branches: Versions and Branches. 8059 (line 116) 8060 * expression evaluation routines: Language Support. (line 58) 8061 * expression parser: Language Support. (line 21) 8062 * extract_typed_address: Pointers and Addresses. 8063 (line 52) 8064 * FDL, GNU Free Documentation License: GNU Free Documentation License. 8065 (line 6) 8066 * field output functions: User Interface. (line 254) 8067 * file names, portability: Coding. (line 547) 8068 * FILENAME_CMP: Coding. (line 573) 8069 * find_pc_function: Symbol Handling. (line 136) 8070 * find_pc_line: Symbol Handling. (line 136) 8071 * find_sym_fns: Symbol Handling. (line 32) 8072 * finding a symbol: Symbol Handling. (line 133) 8073 * fine-tuning gdbarch structure: OS ABI Variant Handling. 8074 (line 23) 8075 * first floating point register: Register Architecture Functions & Variables. 8076 (line 78) 8077 * FOPEN_RB: Host Definition. (line 102) 8078 * fp0_regnum: Register Architecture Functions & Variables. 8079 (line 78) 8080 * frame: Stack Frames. (line 6) 8081 * frame ID: Stack Frames. (line 41) 8082 * frame pointer: Register Information Functions. 8083 (line 126) 8084 * frame, definition of base of a frame: Frame Handling Terminology. 8085 (line 28) 8086 * frame, definition of innermost frame: Frame Handling Terminology. 8087 (line 24) 8088 * frame, definition of NEXT frame: Frame Handling Terminology. 8089 (line 11) 8090 * frame, definition of PREVIOUS frame: Frame Handling Terminology. 8091 (line 14) 8092 * frame, definition of sentinel frame: Frame Handling Terminology. 8093 (line 52) 8094 * frame, definition of sniffing: Frame Handling Terminology. 8095 (line 46) 8096 * frame, definition of THIS frame: Frame Handling Terminology. 8097 (line 9) 8098 * frame, definition of unwinding: Frame Handling Terminology. 8099 (line 41) 8100 * frame_align: Functions and Variable to Analyze Frames. 8101 (line 46) 8102 * frame_base: Analyzing Stacks---Frame Sniffers. 8103 (line 89) 8104 * frame_base_append_sniffer: Analyzing Stacks---Frame Sniffers. 8105 (line 19) 8106 * frame_base_set_default: Analyzing Stacks---Frame Sniffers. 8107 (line 22) 8108 * frame_num_args: Functions to Access Frame Data. 8109 (line 43) 8110 * frame_red_zone_size: Functions and Variable to Analyze Frames. 8111 (line 63) 8112 * frame_register_unwind: Stack Frames. (line 15) 8113 * frame_unwind: Analyzing Stacks---Frame Sniffers. 8114 (line 36) 8115 * frame_unwind_append_sniffer: Analyzing Stacks---Frame Sniffers. 8116 (line 16) 8117 * frame_unwind_append_unwinder: Stack Frames. (line 30) 8118 * frame_unwind_got_address: Stack Frames. (line 105) 8119 * frame_unwind_got_constant: Stack Frames. (line 101) 8120 * frame_unwind_got_memory: Stack Frames. (line 98) 8121 * frame_unwind_got_optimized: Stack Frames. (line 90) 8122 * frame_unwind_got_register: Stack Frames. (line 93) 8123 * frame_unwind_prepend_unwinder: Stack Frames. (line 30) 8124 * full symbol table: Symbol Handling. (line 104) 8125 * function prologue: Prologue Caches. (line 6) 8126 * function prototypes: Coding. (line 411) 8127 * function usage: Coding. (line 393) 8128 * fundamental types: Symbol Handling. (line 178) 8129 * GCC2_COMPILED_FLAG_SYMBOL: Defining Other Architecture Features. 8130 (line 225) 8131 * GCC_COMPILED_FLAG_SYMBOL: Defining Other Architecture Features. 8132 (line 225) 8133 * GDB source tree structure: Overall Structure. (line 83) 8134 * gdb_byte: Register Caching. (line 23) 8135 * GDB_OSABI_AIX: OS ABI Variant Handling. 8136 (line 90) 8137 * GDB_OSABI_CYGWIN: OS ABI Variant Handling. 8138 (line 87) 8139 * GDB_OSABI_FREEBSD_AOUT: OS ABI Variant Handling. 8140 (line 51) 8141 * GDB_OSABI_FREEBSD_ELF: OS ABI Variant Handling. 8142 (line 54) 8143 * GDB_OSABI_GO32: OS ABI Variant Handling. 8144 (line 69) 8145 * GDB_OSABI_HPUX_ELF: OS ABI Variant Handling. 8146 (line 78) 8147 * GDB_OSABI_HPUX_SOM: OS ABI Variant Handling. 8148 (line 81) 8149 * GDB_OSABI_HURD: OS ABI Variant Handling. 8150 (line 39) 8151 * GDB_OSABI_INTERIX: OS ABI Variant Handling. 8152 (line 75) 8153 * GDB_OSABI_IRIX: OS ABI Variant Handling. 8154 (line 72) 8155 * GDB_OSABI_LINUX: OS ABI Variant Handling. 8156 (line 48) 8157 * GDB_OSABI_NETBSD_AOUT: OS ABI Variant Handling. 8158 (line 57) 8159 * GDB_OSABI_NETBSD_ELF: OS ABI Variant Handling. 8160 (line 60) 8161 * GDB_OSABI_OPENBSD_ELF: OS ABI Variant Handling. 8162 (line 63) 8163 * GDB_OSABI_OSF1: OS ABI Variant Handling. 8164 (line 45) 8165 * GDB_OSABI_QNXNTO: OS ABI Variant Handling. 8166 (line 84) 8167 * GDB_OSABI_SOLARIS: OS ABI Variant Handling. 8168 (line 42) 8169 * GDB_OSABI_SVR4: OS ABI Variant Handling. 8170 (line 36) 8171 * GDB_OSABI_UNINITIALIZED: OS ABI Variant Handling. 8172 (line 29) 8173 * GDB_OSABI_UNKNOWN: OS ABI Variant Handling. 8174 (line 32) 8175 * GDB_OSABI_WINCE: OS ABI Variant Handling. 8176 (line 66) 8177 * gdbarch: How an Architecture is Represented. 8178 (line 19) 8179 * gdbarch accessor functions: Creating a New Architecture. 8180 (line 14) 8181 * gdbarch lookup: Looking Up an Existing Architecture. 8182 (line 6) 8183 * gdbarch register architecture functions: Register Architecture Functions & Variables. 8184 (line 6) 8185 * gdbarch register information functions: Register Information Functions. 8186 (line 6) 8187 * gdbarch_addr_bits_remove: Defining Other Architecture Features. 8188 (line 11) 8189 * gdbarch_address_class_name_to_type_flags: Address Classes. (line 30) 8190 * gdbarch_address_class_type_flags <1>: Defining Other Architecture Features. 8191 (line 43) 8192 * gdbarch_address_class_type_flags: Address Classes. (line 18) 8193 * gdbarch_address_class_type_flags_p: Defining Other Architecture Features. 8194 (line 52) 8195 * gdbarch_address_class_type_flags_to_name <1>: Defining Other Architecture Features. 8196 (line 56) 8197 * gdbarch_address_class_type_flags_to_name: Address Classes. (line 25) 8198 * gdbarch_address_class_type_flags_to_name_p: Defining Other Architecture Features. 8199 (line 60) 8200 * gdbarch_address_to_pointer <1>: Defining Other Architecture Features. 8201 (line 65) 8202 * gdbarch_address_to_pointer: Pointers and Addresses. 8203 (line 114) 8204 * gdbarch_adjust_breakpoint_address: Defining Other Architecture Features. 8205 (line 145) 8206 * gdbarch_alloc: Creating a New Architecture. 8207 (line 6) 8208 * gdbarch_believe_pcc_promotion: Defining Other Architecture Features. 8209 (line 72) 8210 * gdbarch_bits_big_endian: Defining Other Architecture Features. 8211 (line 77) 8212 * gdbarch_breakpoint_from_pc: Defining Other Architecture Features. 8213 (line 106) 8214 * gdbarch_call_dummy_location: Defining Other Architecture Features. 8215 (line 178) 8216 * gdbarch_cannot_fetch_register: Defining Other Architecture Features. 8217 (line 184) 8218 * gdbarch_cannot_store_register: Defining Other Architecture Features. 8219 (line 188) 8220 * gdbarch_char_signed: Defining Other Architecture Features. 8221 (line 461) 8222 * gdbarch_convert_register_p <1>: Defining Other Architecture Features. 8223 (line 195) 8224 * gdbarch_convert_register_p: Register and Memory Data. 8225 (line 30) 8226 * gdbarch_data: Coding. (line 133) 8227 * gdbarch_data_register_post_init: Coding. (line 118) 8228 * gdbarch_data_register_pre_init: Coding. (line 108) 8229 * gdbarch_decr_pc_after_break: Defining Other Architecture Features. 8230 (line 205) 8231 * gdbarch_deprecated_fp_regnum: Defining Other Architecture Features. 8232 (line 446) 8233 * gdbarch_double_bit: Defining Other Architecture Features. 8234 (line 471) 8235 * gdbarch_dummy_id: Defining Other Architecture Features. 8236 (line 523) 8237 * gdbarch_dwarf2_reg_to_regnum: Defining Other Architecture Features. 8238 (line 216) 8239 * gdbarch_ecoff_reg_to_regnum: Defining Other Architecture Features. 8240 (line 220) 8241 * gdbarch_float_bit: Defining Other Architecture Features. 8242 (line 475) 8243 * gdbarch_fp0_regnum: Defining Other Architecture Features. 8244 (line 200) 8245 * gdbarch_get_longjmp_target <1>: Defining Other Architecture Features. 8246 (line 231) 8247 * gdbarch_get_longjmp_target: Algorithms. (line 263) 8248 * gdbarch_have_nonsteppable_watchpoint: Algorithms. (line 396) 8249 * gdbarch_in_function_epilogue_p: Defining Other Architecture Features. 8250 (line 253) 8251 * gdbarch_in_solib_return_trampoline: Defining Other Architecture Features. 8252 (line 259) 8253 * gdbarch_info: Looking Up an Existing Architecture. 8254 (line 22) 8255 * gdbarch_init_osabi: OS ABI Variant Handling. 8256 (line 125) 8257 * gdbarch_int_bit: Defining Other Architecture Features. 8258 (line 478) 8259 * gdbarch_integer_to_address: Defining Other Architecture Features. 8260 (line 274) 8261 * gdbarch_list_lookup_by_info: Looking Up an Existing Architecture. 8262 (line 22) 8263 * gdbarch_long_bit: Defining Other Architecture Features. 8264 (line 481) 8265 * gdbarch_long_double_bit: Defining Other Architecture Features. 8266 (line 485) 8267 * gdbarch_long_long_bit: Defining Other Architecture Features. 8268 (line 489) 8269 * gdbarch_lookup_osabi: OS ABI Variant Handling. 8270 (line 119) 8271 * gdbarch_memory_insert_breakpoint: Defining Other Architecture Features. 8272 (line 130) 8273 * gdbarch_memory_remove_breakpoint: Defining Other Architecture Features. 8274 (line 130) 8275 * gdbarch_osabi_name: OS ABI Variant Handling. 8276 (line 97) 8277 * gdbarch_pointer_to_address <1>: Defining Other Architecture Features. 8278 (line 295) 8279 * gdbarch_pointer_to_address: Pointers and Addresses. 8280 (line 105) 8281 * gdbarch_print_insn: Defining Other Architecture Features. 8282 (line 513) 8283 * gdbarch_ptr_bit: Defining Other Architecture Features. 8284 (line 493) 8285 * gdbarch_push_dummy_call: Defining Other Architecture Features. 8286 (line 363) 8287 * gdbarch_push_dummy_code: Defining Other Architecture Features. 8288 (line 375) 8289 * gdbarch_register <1>: Adding a New Target. (line 40) 8290 * gdbarch_register: How an Architecture is Represented. 8291 (line 19) 8292 * gdbarch_register_osabi: OS ABI Variant Handling. 8293 (line 103) 8294 * gdbarch_register_osabi_sniffer: OS ABI Variant Handling. 8295 (line 112) 8296 * gdbarch_register_to_value <1>: Defining Other Architecture Features. 8297 (line 301) 8298 * gdbarch_register_to_value: Register and Memory Data. 8299 (line 46) 8300 * gdbarch_return_value: Defining Other Architecture Features. 8301 (line 394) 8302 * gdbarch_sdb_reg_to_regnum: Defining Other Architecture Features. 8303 (line 390) 8304 * gdbarch_short_bit: Defining Other Architecture Features. 8305 (line 497) 8306 * gdbarch_skip_permanent_breakpoint: Defining Other Architecture Features. 8307 (line 430) 8308 * gdbarch_skip_trampoline_code: Defining Other Architecture Features. 8309 (line 441) 8310 * gdbarch_stab_reg_to_regnum: Defining Other Architecture Features. 8311 (line 450) 8312 * gdbarch_stabs_argument_has_addr: Defining Other Architecture Features. 8313 (line 359) 8314 * gdbarch_tdep definition: Creating a New Architecture. 8315 (line 34) 8316 * gdbarch_tdep when allocating new gdbarch: Creating a New Architecture. 8317 (line 6) 8318 * gdbarch_value_to_register <1>: Defining Other Architecture Features. 8319 (line 529) 8320 * gdbarch_value_to_register: Register and Memory Data. 8321 (line 62) 8322 * gdbarch_virtual_frame_pointer: Defining Other Architecture Features. 8323 (line 501) 8324 * GDBINIT_FILENAME: Host Definition. (line 74) 8325 * generic host support: Host Definition. (line 38) 8326 * generic_elf_osabi_sniff_abi_tag_sections: OS ABI Variant Handling. 8327 (line 133) 8328 * get_frame_register: Stack Frames. (line 15) 8329 * get_frame_type: Stack Frames. (line 22) 8330 * hardware breakpoints: Algorithms. (line 158) 8331 * hardware watchpoints: Algorithms. (line 280) 8332 * HAVE_CONTINUABLE_WATCHPOINT: Algorithms. (line 402) 8333 * HAVE_DOS_BASED_FILE_SYSTEM: Coding. (line 556) 8334 * HAVE_STEPPABLE_WATCHPOINT: Algorithms. (line 386) 8335 * host: Overall Structure. (line 50) 8336 * host, adding: Host Definition. (line 13) 8337 * i386_cleanup_dregs: Algorithms. (line 580) 8338 * I386_DR_LOW_GET_STATUS: Algorithms. (line 493) 8339 * I386_DR_LOW_RESET_ADDR: Algorithms. (line 489) 8340 * I386_DR_LOW_SET_ADDR: Algorithms. (line 486) 8341 * I386_DR_LOW_SET_CONTROL: Algorithms. (line 483) 8342 * i386_insert_hw_breakpoint: Algorithms. (line 568) 8343 * i386_insert_watchpoint: Algorithms. (line 540) 8344 * i386_region_ok_for_watchpoint: Algorithms. (line 518) 8345 * i386_remove_hw_breakpoint: Algorithms. (line 568) 8346 * i386_remove_watchpoint: Algorithms. (line 540) 8347 * i386_stopped_by_watchpoint: Algorithms. (line 532) 8348 * i386_stopped_data_address: Algorithms. (line 525) 8349 * I386_USE_GENERIC_WATCHPOINTS: Algorithms. (line 465) 8350 * in_dynsym_resolve_code: Defining Other Architecture Features. 8351 (line 263) 8352 * inferior_appeared: GDB Observers. (line 169) 8353 * inferior_created: GDB Observers. (line 92) 8354 * inferior_exit: GDB Observers. (line 172) 8355 * inner_than: Functions and Variable to Analyze Frames. 8356 (line 30) 8357 * innermost frame: Frame Handling Terminology. 8358 (line 24) 8359 * insert or remove hardware breakpoint: Algorithms. (line 234) 8360 * insert or remove hardware watchpoint: Algorithms. (line 347) 8361 * insert or remove software breakpoint: Algorithms. (line 211) 8362 * IS_ABSOLUTE_PATH: Coding. (line 567) 8363 * IS_DIR_SEPARATOR: Coding. (line 562) 8364 * ISATTY: Host Definition. (line 99) 8365 * item output functions: User Interface. (line 254) 8366 * language parser: Language Support. (line 25) 8367 * language support: Language Support. (line 6) 8368 * legal papers for code contributions: Debugging GDB. (line 42) 8369 * length_of_subexp: Language Support. (line 58) 8370 * libgdb: libgdb. (line 9) 8371 * libiberty library: Support Libraries. (line 52) 8372 * line wrap in output: Coding. (line 191) 8373 * lint: Host Definition. (line 133) 8374 * list output functions: User Interface. (line 131) 8375 * LITTLE_BREAKPOINT: Defining Other Architecture Features. 8376 (line 100) 8377 * long long data type: Host Definition. (line 106) 8378 * longjmp debugging: Algorithms. (line 258) 8379 * lookup_symbol: Symbol Handling. (line 142) 8380 * LSEEK_NOT_LINEAR: Host Definition. (line 114) 8381 * lval_type enumeration, for values.: Values. (line 19) 8382 * make_cleanup: Coding. (line 28) 8383 * make_cleanup_ui_out_list_begin_end: User Interface. (line 247) 8384 * make_cleanup_ui_out_tuple_begin_end: User Interface. (line 223) 8385 * making a new release of gdb: Releasing GDB. (line 6) 8386 * memory representation: Register and Memory Data. 8387 (line 6) 8388 * memory_changed: GDB Observers. (line 177) 8389 * minimal symbol table: Symbol Handling. (line 111) 8390 * minsymtabs: Symbol Handling. (line 111) 8391 * multi-arch data: Coding. (line 100) 8392 * NATDEPFILES: Native Debugging. (line 8) 8393 * native conditionals: Native Debugging. (line 75) 8394 * native debugging: Native Debugging. (line 6) 8395 * nesting level in ui_out functions: User Interface. (line 143) 8396 * new year procedure: Start of New Year Procedure. 8397 (line 6) 8398 * new_objfile: GDB Observers. (line 109) 8399 * new_thread: GDB Observers. (line 114) 8400 * NEXT frame: Frame Handling Terminology. 8401 (line 11) 8402 * NORETURN: Host Definition. (line 119) 8403 * normal_stop: GDB Observers. (line 76) 8404 * normal_stop observer: GDB Observers. (line 48) 8405 * notification about inferior execution stop: GDB Observers. (line 48) 8406 * notifications about changes in internals: Algorithms. (line 634) 8407 * object file formats: Symbol Handling. (line 210) 8408 * observer pattern interface: Algorithms. (line 634) 8409 * observers implementation rationale: GDB Observers. (line 9) 8410 * obstacks: Support Libraries. (line 69) 8411 * op_print_tab: Language Support. (line 91) 8412 * opcodes library: Support Libraries. (line 39) 8413 * OS ABI variants: OS ABI Variant Handling. 8414 (line 6) 8415 * parse_exp_1: Language Support. (line 97) 8416 * partial symbol table: Symbol Handling. (line 114) 8417 * pc_regnum: Register Architecture Functions & Variables. 8418 (line 58) 8419 * PE-COFF format: Symbol Handling. (line 267) 8420 * per-architecture module data: Coding. (line 100) 8421 * pointer representation: Pointers and Addresses. 8422 (line 6) 8423 * portability: Coding. (line 530) 8424 * portable file name handling: Coding. (line 547) 8425 * porting to new machines: Porting GDB. (line 6) 8426 * prefixify_subexp: Language Support. (line 58) 8427 * PREVIOUS frame: Frame Handling Terminology. 8428 (line 14) 8429 * print_float_info: Register Information Functions. 8430 (line 80) 8431 * print_registers_info: Register Information Functions. 8432 (line 53) 8433 * print_subexp: Language Support. (line 91) 8434 * print_vector_info: Register Information Functions. 8435 (line 96) 8436 * PRINTF_HAS_LONG_LONG: Host Definition. (line 109) 8437 * processor status register: Register Architecture Functions & Variables. 8438 (line 69) 8439 * program counter <1>: Register Architecture Functions & Variables. 8440 (line 58) 8441 * program counter: Algorithms. (line 158) 8442 * prologue analysis: Algorithms. (line 14) 8443 * prologue cache: Prologue Caches. (line 12) 8444 * prologue of a function: Prologue Caches. (line 6) 8445 * prologue-value.c: Algorithms. (line 48) 8446 * prompt: Host Definition. (line 94) 8447 * ps_regnum: Register Architecture Functions & Variables. 8448 (line 69) 8449 * pseudo-evaluation of function prologues: Algorithms. (line 48) 8450 * pseudo_register_read: Register Architecture Functions & Variables. 8451 (line 29) 8452 * pseudo_register_write: Register Architecture Functions & Variables. 8453 (line 33) 8454 * psymtabs: Symbol Handling. (line 107) 8455 * push_dummy_call: Functions Creating Dummy Frames. 8456 (line 13) 8457 * push_dummy_code: Functions Creating Dummy Frames. 8458 (line 57) 8459 * raw register representation: Raw and Cooked Registers. 8460 (line 6) 8461 * read_pc: Register Architecture Functions & Variables. 8462 (line 10) 8463 * reading of symbols: Symbol Handling. (line 25) 8464 * readline library: Support Libraries. (line 45) 8465 * regcache_cooked_read: Register Caching. (line 23) 8466 * regcache_cooked_read_signed: Register Caching. (line 23) 8467 * regcache_cooked_read_unsigned: Register Caching. (line 23) 8468 * regcache_cooked_write: Register Caching. (line 23) 8469 * regcache_cooked_write_signed: Register Caching. (line 23) 8470 * regcache_cooked_write_unsigned: Register Caching. (line 23) 8471 * register caching: Register Caching. (line 6) 8472 * register data formats, converting: Register and Memory Data. 8473 (line 6) 8474 * register representation: Register and Memory Data. 8475 (line 6) 8476 * REGISTER_CONVERT_TO_RAW: Defining Other Architecture Features. 8477 (line 311) 8478 * REGISTER_CONVERT_TO_VIRTUAL: Defining Other Architecture Features. 8479 (line 306) 8480 * register_name: Register Information Functions. 8481 (line 10) 8482 * register_reggroup_p: Register Information Functions. 8483 (line 110) 8484 * register_type: Register Information Functions. 8485 (line 33) 8486 * regset_from_core_section: Defining Other Architecture Features. 8487 (line 316) 8488 * regular expressions library: Support Libraries. (line 110) 8489 * Release Branches: Versions and Branches. 8490 (line 93) 8491 * remote debugging support: Host Definition. (line 41) 8492 * REMOTE_BPT_VECTOR: Defining Other Architecture Features. 8493 (line 540) 8494 * representation of architecture: How an Architecture is Represented. 8495 (line 6) 8496 * representations, raw and cooked registers: Raw and Cooked Registers. 8497 (line 6) 8498 * representations, register and memory: Register and Memory Data. 8499 (line 6) 8500 * requirements for GDB: Requirements. (line 6) 8501 * restart: Algorithms. (line 604) 8502 * running the test suite: Testsuite. (line 19) 8503 * secondary symbol file: Symbol Handling. (line 47) 8504 * sentinel frame <1>: Frame Handling Terminology. 8505 (line 52) 8506 * sentinel frame: Stack Frames. (line 22) 8507 * SENTINEL_FRAME: Stack Frames. (line 22) 8508 * separate data and code address spaces: Pointers and Addresses. 8509 (line 6) 8510 * serial line support: Host Definition. (line 41) 8511 * set_gdbarch functions: Creating a New Architecture. 8512 (line 14) 8513 * set_gdbarch_bits_big_endian: Defining Other Architecture Features. 8514 (line 83) 8515 * set_gdbarch_sofun_address_maybe_missing: Defining Other Architecture Features. 8516 (line 330) 8517 * SIGWINCH_HANDLER: Host Definition. (line 78) 8518 * SIGWINCH_HANDLER_BODY: Host Definition. (line 82) 8519 * skip_prologue: Functions and Variable to Analyze Frames. 8520 (line 12) 8521 * SKIP_SOLIB_RESOLVER: Defining Other Architecture Features. 8522 (line 267) 8523 * SLASH_STRING: Coding. (line 584) 8524 * sniffing: Frame Handling Terminology. 8525 (line 46) 8526 * software breakpoints: Algorithms. (line 184) 8527 * software watchpoints: Algorithms. (line 280) 8528 * SOFTWARE_SINGLE_STEP: Defining Other Architecture Features. 8529 (line 324) 8530 * SOFTWARE_SINGLE_STEP_P: Defining Other Architecture Features. 8531 (line 320) 8532 * SOLIB_ADD: Native Debugging. (line 86) 8533 * SOLIB_CREATE_INFERIOR_HOOK: Native Debugging. (line 92) 8534 * solib_loaded: GDB Observers. (line 99) 8535 * solib_unloaded: GDB Observers. (line 104) 8536 * SOM debugging info: Symbol Handling. (line 355) 8537 * SOM format: Symbol Handling. (line 286) 8538 * source code formatting: Coding. (line 323) 8539 * sp_regnum: Register Architecture Functions & Variables. 8540 (line 49) 8541 * spaces, separate data and code address: Pointers and Addresses. 8542 (line 6) 8543 * stabs debugging info: Symbol Handling. (line 300) 8544 * stack frame, definition of base of a frame: Frame Handling Terminology. 8545 (line 28) 8546 * stack frame, definition of innermost frame: Frame Handling Terminology. 8547 (line 24) 8548 * stack frame, definition of NEXT frame: Frame Handling Terminology. 8549 (line 11) 8550 * stack frame, definition of PREVIOUS frame: Frame Handling Terminology. 8551 (line 14) 8552 * stack frame, definition of sentinel frame: Frame Handling Terminology. 8553 (line 52) 8554 * stack frame, definition of sniffing: Frame Handling Terminology. 8555 (line 46) 8556 * stack frame, definition of THIS frame: Frame Handling Terminology. 8557 (line 9) 8558 * stack frame, definition of unwinding: Frame Handling Terminology. 8559 (line 41) 8560 * stack pointer: Register Architecture Functions & Variables. 8561 (line 49) 8562 * START_INFERIOR_TRAPS_EXPECTED: Native Debugging. (line 96) 8563 * status register: Register Architecture Functions & Variables. 8564 (line 69) 8565 * STOPPED_BY_WATCHPOINT: Algorithms. (line 412) 8566 * store_typed_address: Pointers and Addresses. 8567 (line 70) 8568 * struct: GDB Observers. (line 62) 8569 * struct gdbarch creation: Creating a New Architecture. 8570 (line 6) 8571 * struct regcache: Register Caching. (line 10) 8572 * struct value, converting register contents to: Register and Memory Data. 8573 (line 6) 8574 * submitting patches: Debugging GDB. (line 30) 8575 * sym_fns structure: Symbol Handling. (line 37) 8576 * symbol files: Symbol Handling. (line 25) 8577 * symbol lookup: Symbol Handling. (line 133) 8578 * symbol reading: Symbol Handling. (line 25) 8579 * SYMBOL_RELOADING_DEFAULT: Defining Other Architecture Features. 8580 (line 454) 8581 * symtabs: Symbol Handling. (line 104) 8582 * system dependencies: Coding. (line 534) 8583 * table output functions: User Interface. (line 131) 8584 * target: Overall Structure. (line 50) 8585 * target architecture definition: Target Architecture Definition. 8586 (line 6) 8587 * target dependent files: Adding a New Target. (line 8) 8588 * target descriptions: Target Descriptions. (line 6) 8589 * target descriptions, adding register support: Adding Target Described Register Support. 8590 (line 6) 8591 * target descriptions, implementation: Target Descriptions Implementation. 8592 (line 6) 8593 * target vector: Target Vector Definition. 8594 (line 6) 8595 * TARGET_CAN_USE_HARDWARE_WATCHPOINT: Algorithms. (line 333) 8596 * target_changed: GDB Observers. (line 82) 8597 * TARGET_CHAR_BIT: Defining Other Architecture Features. 8598 (line 458) 8599 * target_insert_breakpoint: Algorithms. (line 211) 8600 * target_insert_hw_breakpoint: Algorithms. (line 234) 8601 * target_insert_watchpoint: Algorithms. (line 347) 8602 * TARGET_REGION_OK_FOR_HW_WATCHPOINT: Algorithms. (line 343) 8603 * target_remove_breakpoint: Algorithms. (line 211) 8604 * target_remove_hw_breakpoint: Algorithms. (line 234) 8605 * target_remove_watchpoint: Algorithms. (line 347) 8606 * target_resumed: GDB Observers. (line 129) 8607 * target_stopped_data_address: Algorithms. (line 364) 8608 * target_watchpoint_addr_within_range: Algorithms. (line 378) 8609 * targets: Existing Targets. (line 6) 8610 * TCP remote support: Host Definition. (line 57) 8611 * terminal device: Host Definition. (line 97) 8612 * test suite: Testsuite. (line 6) 8613 * test suite organization: Testsuite. (line 195) 8614 * test_notification: GDB Observers. (line 181) 8615 * Testsuite Configuration: Testsuite. (line 167) 8616 * THIS frame: Frame Handling Terminology. 8617 (line 9) 8618 * thread_exit: GDB Observers. (line 117) 8619 * thread_ptid_changed: GDB Observers. (line 165) 8620 * thread_stop_requested: GDB Observers. (line 122) 8621 * tracepoint_created: GDB Observers. (line 148) 8622 * tracepoint_deleted: GDB Observers. (line 152) 8623 * tracepoint_modified: GDB Observers. (line 156) 8624 * tuple output functions: User Interface. (line 131) 8625 * type codes: Symbol Handling. (line 186) 8626 * types: Coding. (line 405) 8627 * ui_out functions: User Interface. (line 47) 8628 * ui_out functions, usage examples: User Interface. (line 398) 8629 * ui_out_field_core_addr: User Interface. (line 287) 8630 * ui_out_field_fmt: User Interface. (line 261) 8631 * ui_out_field_fmt_int: User Interface. (line 280) 8632 * ui_out_field_int: User Interface. (line 273) 8633 * ui_out_field_skip: User Interface. (line 352) 8634 * ui_out_field_stream: User Interface. (line 320) 8635 * ui_out_field_string: User Interface. (line 291) 8636 * ui_out_flush: User Interface. (line 392) 8637 * ui_out_list_begin: User Interface. (line 234) 8638 * ui_out_list_end: User Interface. (line 240) 8639 * ui_out_message: User Interface. (line 376) 8640 * ui_out_spaces: User Interface. (line 371) 8641 * ui_out_stream_delete: User Interface. (line 315) 8642 * ui_out_stream_new: User Interface. (line 309) 8643 * ui_out_table_begin: User Interface. (line 165) 8644 * ui_out_table_body: User Interface. (line 191) 8645 * ui_out_table_end: User Interface. (line 194) 8646 * ui_out_table_header: User Interface. (line 178) 8647 * ui_out_text: User Interface. (line 358) 8648 * ui_out_tuple_begin: User Interface. (line 210) 8649 * ui_out_tuple_end: User Interface. (line 216) 8650 * ui_out_wrap_hint: User Interface. (line 382) 8651 * unwind frame: Stack Frames. (line 9) 8652 * unwind_dummy_id: Functions Creating Dummy Frames. 8653 (line 38) 8654 * unwind_pc: Functions to Access Frame Data. 8655 (line 11) 8656 * unwind_sp: Functions to Access Frame Data. 8657 (line 27) 8658 * unwinding: Frame Handling Terminology. 8659 (line 41) 8660 * using ui_out functions: User Interface. (line 398) 8661 * value structure: Values. (line 9) 8662 * value_as_address: Pointers and Addresses. 8663 (line 84) 8664 * value_from_pointer: Pointers and Addresses. 8665 (line 93) 8666 * values: Values. (line 9) 8667 * VEC: Support Libraries. (line 131) 8668 * vendor branches: Versions and Branches. 8669 (line 108) 8670 * void: GDB Observers. (line 67) 8671 * volatile: Host Definition. (line 136) 8672 * watchpoints: Algorithms. (line 274) 8673 * watchpoints, on x86: Algorithms. (line 453) 8674 * watchpoints, with threads: Algorithms. (line 429) 8675 * word-addressed machines: Pointers and Addresses. 8676 (line 6) 8677 * wrap_here: Coding. (line 191) 8678 * write_pc: Register Architecture Functions & Variables. 8679 (line 13) 8680 * writing tests: Testsuite. (line 247) 8681 * x86 debug registers: Algorithms. (line 453) 8682 * XCOFF format: Symbol Handling. (line 251) 8683 8684 8685 8686 Tag Table: 8687 Node: Top1589 8688 Node: Summary2470 8689 Node: Requirements2620 8690 Node: Contributors4099 8691 Node: Overall Structure5692 8692 Node: Algorithms10715 8693 Node: User Interface42299 8694 Ref: UI-Independent Output44154 8695 Ref: User Interface-Footnote-166125 8696 Ref: User Interface-Footnote-266174 8697 Node: libgdb66409 8698 Node: Values70360 8699 Node: Stack Frames73204 8700 Node: Symbol Handling78186 8701 Node: Language Support94720 8702 Node: Host Definition99446 8703 Node: Target Architecture Definition104437 8704 Node: OS ABI Variant Handling105257 8705 Node: Initialize New Architecture110102 8706 Node: How an Architecture is Represented110453 8707 Node: Looking Up an Existing Architecture112410 8708 Node: Creating a New Architecture115329 8709 Node: Registers and Memory117367 8710 Node: Pointers and Addresses118159 8711 Ref: Pointers and Addresses-Footnote-1124160 8712 Node: Address Classes124403 8713 Node: Register Representation127648 8714 Node: Raw and Cooked Registers128022 8715 Node: Register Architecture Functions & Variables129206 8716 Node: Register Information Functions132815 8717 Ref: Register Information Functions-Footnote-1138721 8718 Node: Register and Memory Data139140 8719 Node: Register Caching142289 8720 Node: Frame Interpretation143825 8721 Node: All About Stack Frames144231 8722 Ref: All About Stack Frames-Footnote-1149523 8723 Node: Frame Handling Terminology149755 8724 Node: Prologue Caches152282 8725 Node: Functions and Variable to Analyze Frames153963 8726 Ref: frame_align156061 8727 Node: Functions to Access Frame Data157575 8728 Node: Analyzing Stacks---Frame Sniffers159866 8729 Ref: Analyzing Stacks---Frame Sniffers-Footnote-1164516 8730 Node: Inferior Call Setup165013 8731 Node: About Dummy Frames165296 8732 Node: Functions Creating Dummy Frames165922 8733 Node: Adding support for debugging core files169979 8734 Node: Defining Other Architecture Features170523 8735 Ref: gdbarch_breakpoint_from_pc175370 8736 Ref: gdbarch_stabs_argument_has_addr187764 8737 Ref: gdbarch_push_dummy_call188011 8738 Ref: gdbarch_push_dummy_code188571 8739 Ref: gdbarch_return_value189553 8740 Ref: gdbarch_dummy_id195319 8741 Node: Adding a New Target196007 8742 Node: Target Descriptions198474 8743 Node: Target Descriptions Implementation199413 8744 Node: Adding Target Described Register Support200787 8745 Node: Target Vector Definition203733 8746 Node: Managing Execution State204265 8747 Node: Existing Targets206078 8748 Node: Native Debugging208593 8749 Node: Support Libraries212421 8750 Node: Coding223936 8751 Node: Porting GDB248945 8752 Node: Versions and Branches250814 8753 Ref: Tags256770 8754 Ref: experimental branch tags257101 8755 Node: Start of New Year Procedure257833 8756 Node: Releasing GDB259639 8757 Node: Testsuite277871 8758 Ref: Testsuite-Footnote-1289736 8759 Node: Hints289854 8760 Node: Getting Started290176 8761 Node: Debugging GDB294319 8762 Node: GDB Observers299384 8763 Node: GNU Free Documentation License307286 8764 Node: Index329730 8765 8766 End Tag Table 8767