1 \input texinfo @c -*- texinfo -*-
2 @setfilename gdbint.info
4 @dircategory Software development
6 * Gdb-Internals: (gdbint). The GNU debugger's internals.
10 Copyright @copyright{} 1990, 1991, 1992, 1993, 1994, 1996, 1998, 1999,
11 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2008, 2009
12 Free Software Foundation, Inc.
13 Contributed by Cygnus Solutions. Written by John Gilmore.
14 Second Edition by Stan Shebs.
16 Permission is granted to copy, distribute and/or modify this document
17 under the terms of the GNU Free Documentation License, Version 1.1 or
18 any later version published by the Free Software Foundation; with no
19 Invariant Sections, with no Front-Cover Texts, and with no Back-Cover
20 Texts. A copy of the license is included in the section entitled ``GNU
21 Free Documentation License''.
25 This file documents the internals of the GNU debugger @value{GDBN}.
30 @setchapternewpage off
31 @settitle @value{GDBN} Internals
37 @title @value{GDBN} Internals
38 @subtitle{A guide to the internals of the GNU debugger}
40 @author Cygnus Solutions
41 @author Second Edition:
43 @author Cygnus Solutions
46 \def\$#1${{#1}} % Kluge: collect RCS revision info without $...$
47 \xdef\manvers{\$Revision$} % For use in headers, footers too
49 \hfill Cygnus Solutions\par
51 \hfill \TeX{}info \texinfoversion\par
55 @vskip 0pt plus 1filll
62 @c Perhaps this should be the title of the document (but only for info,
63 @c not for TeX). Existing GNU manuals seem inconsistent on this point.
64 @top Scope of this Document
66 This document documents the internals of the GNU debugger, @value{GDBN}. It
67 includes description of @value{GDBN}'s key algorithms and operations, as well
68 as the mechanisms that adapt @value{GDBN} to specific hosts and targets.
81 * Target Architecture Definition::
82 * Target Descriptions::
83 * Target Vector Definition::
88 * Versions and Branches::
89 * Start of New Year Procedure::
94 * GDB Observers:: @value{GDBN} Currently available observers
95 * GNU Free Documentation License:: The license for this documentation
101 @chapter Requirements
102 @cindex requirements for @value{GDBN}
104 Before diving into the internals, you should understand the formal
105 requirements and other expectations for @value{GDBN}. Although some
106 of these may seem obvious, there have been proposals for @value{GDBN}
107 that have run counter to these requirements.
109 First of all, @value{GDBN} is a debugger. It's not designed to be a
110 front panel for embedded systems. It's not a text editor. It's not a
111 shell. It's not a programming environment.
113 @value{GDBN} is an interactive tool. Although a batch mode is
114 available, @value{GDBN}'s primary role is to interact with a human
117 @value{GDBN} should be responsive to the user. A programmer hot on
118 the trail of a nasty bug, and operating under a looming deadline, is
119 going to be very impatient of everything, including the response time
120 to debugger commands.
122 @value{GDBN} should be relatively permissive, such as for expressions.
123 While the compiler should be picky (or have the option to be made
124 picky), since source code lives for a long time usually, the
125 programmer doing debugging shouldn't be spending time figuring out to
126 mollify the debugger.
128 @value{GDBN} will be called upon to deal with really large programs.
129 Executable sizes of 50 to 100 megabytes occur regularly, and we've
130 heard reports of programs approaching 1 gigabyte in size.
132 @value{GDBN} should be able to run everywhere. No other debugger is
133 available for even half as many configurations as @value{GDBN}
137 @node Overall Structure
139 @chapter Overall Structure
141 @value{GDBN} consists of three major subsystems: user interface,
142 symbol handling (the @dfn{symbol side}), and target system handling (the
145 The user interface consists of several actual interfaces, plus
148 The symbol side consists of object file readers, debugging info
149 interpreters, symbol table management, source language expression
150 parsing, type and value printing.
152 The target side consists of execution control, stack frame analysis, and
153 physical target manipulation.
155 The target side/symbol side division is not formal, and there are a
156 number of exceptions. For instance, core file support involves symbolic
157 elements (the basic core file reader is in BFD) and target elements (it
158 supplies the contents of memory and the values of registers). Instead,
159 this division is useful for understanding how the minor subsystems
162 @section The Symbol Side
164 The symbolic side of @value{GDBN} can be thought of as ``everything
165 you can do in @value{GDBN} without having a live program running''.
166 For instance, you can look at the types of variables, and evaluate
167 many kinds of expressions.
169 @section The Target Side
171 The target side of @value{GDBN} is the ``bits and bytes manipulator''.
172 Although it may make reference to symbolic info here and there, most
173 of the target side will run with only a stripped executable
174 available---or even no executable at all, in remote debugging cases.
176 Operations such as disassembly, stack frame crawls, and register
177 display, are able to work with no symbolic info at all. In some cases,
178 such as disassembly, @value{GDBN} will use symbolic info to present addresses
179 relative to symbols rather than as raw numbers, but it will work either
182 @section Configurations
186 @dfn{Host} refers to attributes of the system where @value{GDBN} runs.
187 @dfn{Target} refers to the system where the program being debugged
188 executes. In most cases they are the same machine, in which case a
189 third type of @dfn{Native} attributes come into play.
191 Defines and include files needed to build on the host are host
192 support. Examples are tty support, system defined types, host byte
193 order, host float format. These are all calculated by @code{autoconf}
194 when the debugger is built.
196 Defines and information needed to handle the target format are target
197 dependent. Examples are the stack frame format, instruction set,
198 breakpoint instruction, registers, and how to set up and tear down the stack
201 Information that is only needed when the host and target are the same,
202 is native dependent. One example is Unix child process support; if the
203 host and target are not the same, calling @code{fork} to start the target
204 process is a bad idea. The various macros needed for finding the
205 registers in the @code{upage}, running @code{ptrace}, and such are all
206 in the native-dependent files.
208 Another example of native-dependent code is support for features that
209 are really part of the target environment, but which require
210 @code{#include} files that are only available on the host system. Core
211 file handling and @code{setjmp} handling are two common cases.
213 When you want to make @value{GDBN} work as the traditional native debugger
214 on a system, you will need to supply both target and native information.
216 @section Source Tree Structure
217 @cindex @value{GDBN} source tree structure
219 The @value{GDBN} source directory has a mostly flat structure---there
220 are only a few subdirectories. A file's name usually gives a hint as
221 to what it does; for example, @file{stabsread.c} reads stabs,
222 @file{dwarf2read.c} reads @sc{DWARF 2}, etc.
224 Files that are related to some common task have names that share
225 common substrings. For example, @file{*-thread.c} files deal with
226 debugging threads on various platforms; @file{*read.c} files deal with
227 reading various kinds of symbol and object files; @file{inf*.c} files
228 deal with direct control of the @dfn{inferior program} (@value{GDBN}
229 parlance for the program being debugged).
231 There are several dozens of files in the @file{*-tdep.c} family.
232 @samp{tdep} stands for @dfn{target-dependent code}---each of these
233 files implements debug support for a specific target architecture
234 (sparc, mips, etc). Usually, only one of these will be used in a
235 specific @value{GDBN} configuration (sometimes two, closely related).
237 Similarly, there are many @file{*-nat.c} files, each one for native
238 debugging on a specific system (e.g., @file{sparc-linux-nat.c} is for
239 native debugging of Sparc machines running the Linux kernel).
241 The few subdirectories of the source tree are:
245 Code that implements @dfn{CLI}, the @value{GDBN} Command-Line
246 Interpreter. @xref{User Interface, Command Interpreter}.
249 Code for the @value{GDBN} remote server.
252 Code for Insight, the @value{GDBN} TK-based GUI front-end.
255 The @dfn{GDB/MI}, the @value{GDBN} Machine Interface interpreter.
258 Target signal translation code.
261 Code for @dfn{TUI}, the @value{GDBN} Text-mode full-screen User
262 Interface. @xref{User Interface, TUI}.
270 @value{GDBN} uses a number of debugging-specific algorithms. They are
271 often not very complicated, but get lost in the thicket of special
272 cases and real-world issues. This chapter describes the basic
273 algorithms and mentions some of the specific target definitions that
276 @section Prologue Analysis
278 @cindex prologue analysis
279 @cindex call frame information
280 @cindex CFI (call frame information)
281 To produce a backtrace and allow the user to manipulate older frames'
282 variables and arguments, @value{GDBN} needs to find the base addresses
283 of older frames, and discover where those frames' registers have been
284 saved. Since a frame's ``callee-saves'' registers get saved by
285 younger frames if and when they're reused, a frame's registers may be
286 scattered unpredictably across younger frames. This means that
287 changing the value of a register-allocated variable in an older frame
288 may actually entail writing to a save slot in some younger frame.
290 Modern versions of GCC emit Dwarf call frame information (``CFI''),
291 which describes how to find frame base addresses and saved registers.
292 But CFI is not always available, so as a fallback @value{GDBN} uses a
293 technique called @dfn{prologue analysis} to find frame sizes and saved
294 registers. A prologue analyzer disassembles the function's machine
295 code starting from its entry point, and looks for instructions that
296 allocate frame space, save the stack pointer in a frame pointer
297 register, save registers, and so on. Obviously, this can't be done
298 accurately in general, but it's tractable to do well enough to be very
299 helpful. Prologue analysis predates the GNU toolchain's support for
300 CFI; at one time, prologue analysis was the only mechanism
301 @value{GDBN} used for stack unwinding at all, when the function
302 calling conventions didn't specify a fixed frame layout.
304 In the olden days, function prologues were generated by hand-written,
305 target-specific code in GCC, and treated as opaque and untouchable by
306 optimizers. Looking at this code, it was usually straightforward to
307 write a prologue analyzer for @value{GDBN} that would accurately
308 understand all the prologues GCC would generate. However, over time
309 GCC became more aggressive about instruction scheduling, and began to
310 understand more about the semantics of the prologue instructions
311 themselves; in response, @value{GDBN}'s analyzers became more complex
312 and fragile. Keeping the prologue analyzers working as GCC (and the
313 instruction sets themselves) evolved became a substantial task.
315 @cindex @file{prologue-value.c}
316 @cindex abstract interpretation of function prologues
317 @cindex pseudo-evaluation of function prologues
318 To try to address this problem, the code in @file{prologue-value.h}
319 and @file{prologue-value.c} provides a general framework for writing
320 prologue analyzers that are simpler and more robust than ad-hoc
321 analyzers. When we analyze a prologue using the prologue-value
322 framework, we're really doing ``abstract interpretation'' or
323 ``pseudo-evaluation'': running the function's code in simulation, but
324 using conservative approximations of the values registers and memory
325 would hold when the code actually runs. For example, if our function
326 starts with the instruction:
329 addi r1, 42 # add 42 to r1
332 we don't know exactly what value will be in @code{r1} after executing
333 this instruction, but we do know it'll be 42 greater than its original
336 If we then see an instruction like:
339 addi r1, 22 # add 22 to r1
342 we still don't know what @code{r1's} value is, but again, we can say
343 it is now 64 greater than its original value.
345 If the next instruction were:
348 mov r2, r1 # set r2 to r1's value
351 then we can say that @code{r2's} value is now the original value of
354 It's common for prologues to save registers on the stack, so we'll
355 need to track the values of stack frame slots, as well as the
356 registers. So after an instruction like this:
362 then we'd know that the stack slot four bytes above the frame pointer
363 holds the original value of @code{r1} plus 64.
367 Of course, this can only go so far before it gets unreasonable. If we
368 wanted to be able to say anything about the value of @code{r1} after
372 xor r1, r3 # exclusive-or r1 and r3, place result in r1
375 then things would get pretty complex. But remember, we're just doing
376 a conservative approximation; if exclusive-or instructions aren't
377 relevant to prologues, we can just say @code{r1}'s value is now
378 ``unknown''. We can ignore things that are too complex, if that loss of
379 information is acceptable for our application.
381 So when we say ``conservative approximation'' here, what we mean is an
382 approximation that is either accurate, or marked ``unknown'', but
385 Using this framework, a prologue analyzer is simply an interpreter for
386 machine code, but one that uses conservative approximations for the
387 contents of registers and memory instead of actual values. Starting
388 from the function's entry point, you simulate instructions up to the
389 current PC, or an instruction that you don't know how to simulate.
390 Now you can examine the state of the registers and stack slots you've
396 To see how large your stack frame is, just check the value of the
397 stack pointer register; if it's the original value of the SP
398 minus a constant, then that constant is the stack frame's size.
399 If the SP's value has been marked as ``unknown'', then that means
400 the prologue has done something too complex for us to track, and
401 we don't know the frame size.
404 To see where we've saved the previous frame's registers, we just
405 search the values we've tracked --- stack slots, usually, but
406 registers, too, if you want --- for something equal to the register's
407 original value. If the calling conventions suggest a standard place
408 to save a given register, then we can check there first, but really,
409 anything that will get us back the original value will probably work.
412 This does take some work. But prologue analyzers aren't
413 quick-and-simple pattern patching to recognize a few fixed prologue
414 forms any more; they're big, hairy functions. Along with inferior
415 function calls, prologue analysis accounts for a substantial portion
416 of the time needed to stabilize a @value{GDBN} port. So it's
417 worthwhile to look for an approach that will be easier to understand
418 and maintain. In the approach described above:
423 It's easier to see that the analyzer is correct: you just see
424 whether the analyzer properly (albeit conservatively) simulates
425 the effect of each instruction.
428 It's easier to extend the analyzer: you can add support for new
429 instructions, and know that you haven't broken anything that
430 wasn't already broken before.
433 It's orthogonal: to gather new information, you don't need to
434 complicate the code for each instruction. As long as your domain
435 of conservative values is already detailed enough to tell you
436 what you need, then all the existing instruction simulations are
437 already gathering the right data for you.
441 The file @file{prologue-value.h} contains detailed comments explaining
442 the framework and how to use it.
445 @section Breakpoint Handling
448 In general, a breakpoint is a user-designated location in the program
449 where the user wants to regain control if program execution ever reaches
452 There are two main ways to implement breakpoints; either as ``hardware''
453 breakpoints or as ``software'' breakpoints.
455 @cindex hardware breakpoints
456 @cindex program counter
457 Hardware breakpoints are sometimes available as a builtin debugging
458 features with some chips. Typically these work by having dedicated
459 register into which the breakpoint address may be stored. If the PC
460 (shorthand for @dfn{program counter})
461 ever matches a value in a breakpoint registers, the CPU raises an
462 exception and reports it to @value{GDBN}.
464 Another possibility is when an emulator is in use; many emulators
465 include circuitry that watches the address lines coming out from the
466 processor, and force it to stop if the address matches a breakpoint's
469 A third possibility is that the target already has the ability to do
470 breakpoints somehow; for instance, a ROM monitor may do its own
471 software breakpoints. So although these are not literally ``hardware
472 breakpoints'', from @value{GDBN}'s point of view they work the same;
473 @value{GDBN} need not do anything more than set the breakpoint and wait
474 for something to happen.
476 Since they depend on hardware resources, hardware breakpoints may be
477 limited in number; when the user asks for more, @value{GDBN} will
478 start trying to set software breakpoints. (On some architectures,
479 notably the 32-bit x86 platforms, @value{GDBN} cannot always know
480 whether there's enough hardware resources to insert all the hardware
481 breakpoints and watchpoints. On those platforms, @value{GDBN} prints
482 an error message only when the program being debugged is continued.)
484 @cindex software breakpoints
485 Software breakpoints require @value{GDBN} to do somewhat more work.
486 The basic theory is that @value{GDBN} will replace a program
487 instruction with a trap, illegal divide, or some other instruction
488 that will cause an exception, and then when it's encountered,
489 @value{GDBN} will take the exception and stop the program. When the
490 user says to continue, @value{GDBN} will restore the original
491 instruction, single-step, re-insert the trap, and continue on.
493 Since it literally overwrites the program being tested, the program area
494 must be writable, so this technique won't work on programs in ROM. It
495 can also distort the behavior of programs that examine themselves,
496 although such a situation would be highly unusual.
498 Also, the software breakpoint instruction should be the smallest size of
499 instruction, so it doesn't overwrite an instruction that might be a jump
500 target, and cause disaster when the program jumps into the middle of the
501 breakpoint instruction. (Strictly speaking, the breakpoint must be no
502 larger than the smallest interval between instructions that may be jump
503 targets; perhaps there is an architecture where only even-numbered
504 instructions may jumped to.) Note that it's possible for an instruction
505 set not to have any instructions usable for a software breakpoint,
506 although in practice only the ARC has failed to define such an
509 Basic breakpoint object handling is in @file{breakpoint.c}. However,
510 much of the interesting breakpoint action is in @file{infrun.c}.
513 @cindex insert or remove software breakpoint
514 @findex target_remove_breakpoint
515 @findex target_insert_breakpoint
516 @item target_remove_breakpoint (@var{bp_tgt})
517 @itemx target_insert_breakpoint (@var{bp_tgt})
518 Insert or remove a software breakpoint at address
519 @code{@var{bp_tgt}->placed_address}. Returns zero for success,
520 non-zero for failure. On input, @var{bp_tgt} contains the address of the
521 breakpoint, and is otherwise initialized to zero. The fields of the
522 @code{struct bp_target_info} pointed to by @var{bp_tgt} are updated
523 to contain other information about the breakpoint on output. The field
524 @code{placed_address} may be updated if the breakpoint was placed at a
525 related address; the field @code{shadow_contents} contains the real
526 contents of the bytes where the breakpoint has been inserted,
527 if reading memory would return the breakpoint instead of the
528 underlying memory; the field @code{shadow_len} is the length of
529 memory cached in @code{shadow_contents}, if any; and the field
530 @code{placed_size} is optionally set and used by the target, if
531 it could differ from @code{shadow_len}.
533 For example, the remote target @samp{Z0} packet does not require
534 shadowing memory, so @code{shadow_len} is left at zero. However,
535 the length reported by @code{gdbarch_breakpoint_from_pc} is cached in
536 @code{placed_size}, so that a matching @samp{z0} packet can be
537 used to remove the breakpoint.
539 @cindex insert or remove hardware breakpoint
540 @findex target_remove_hw_breakpoint
541 @findex target_insert_hw_breakpoint
542 @item target_remove_hw_breakpoint (@var{bp_tgt})
543 @itemx target_insert_hw_breakpoint (@var{bp_tgt})
544 Insert or remove a hardware-assisted breakpoint at address
545 @code{@var{bp_tgt}->placed_address}. Returns zero for success,
546 non-zero for failure. See @code{target_insert_breakpoint} for
547 a description of the @code{struct bp_target_info} pointed to by
548 @var{bp_tgt}; the @code{shadow_contents} and
549 @code{shadow_len} members are not used for hardware breakpoints,
550 but @code{placed_size} may be.
553 @section Single Stepping
555 @section Signal Handling
557 @section Thread Handling
559 @section Inferior Function Calls
561 @section Longjmp Support
563 @cindex @code{longjmp} debugging
564 @value{GDBN} has support for figuring out that the target is doing a
565 @code{longjmp} and for stopping at the target of the jump, if we are
566 stepping. This is done with a few specialized internal breakpoints,
567 which are visible in the output of the @samp{maint info breakpoint}
570 @findex gdbarch_get_longjmp_target
571 To make this work, you need to define a function called
572 @code{gdbarch_get_longjmp_target}, which will examine the
573 @code{jmp_buf} structure and extract the @code{longjmp} target address.
574 Since @code{jmp_buf} is target specific and typically defined in a
575 target header not available to @value{GDBN}, you will need to
576 determine the offset of the PC manually and return that; many targets
577 define a @code{jb_pc_offset} field in the tdep structure to save the
578 value once calculated.
583 Watchpoints are a special kind of breakpoints (@pxref{Algorithms,
584 breakpoints}) which break when data is accessed rather than when some
585 instruction is executed. When you have data which changes without
586 your knowing what code does that, watchpoints are the silver bullet to
587 hunt down and kill such bugs.
589 @cindex hardware watchpoints
590 @cindex software watchpoints
591 Watchpoints can be either hardware-assisted or not; the latter type is
592 known as ``software watchpoints.'' @value{GDBN} always uses
593 hardware-assisted watchpoints if they are available, and falls back on
594 software watchpoints otherwise. Typical situations where @value{GDBN}
595 will use software watchpoints are:
599 The watched memory region is too large for the underlying hardware
600 watchpoint support. For example, each x86 debug register can watch up
601 to 4 bytes of memory, so trying to watch data structures whose size is
602 more than 16 bytes will cause @value{GDBN} to use software
606 The value of the expression to be watched depends on data held in
607 registers (as opposed to memory).
610 Too many different watchpoints requested. (On some architectures,
611 this situation is impossible to detect until the debugged program is
612 resumed.) Note that x86 debug registers are used both for hardware
613 breakpoints and for watchpoints, so setting too many hardware
614 breakpoints might cause watchpoint insertion to fail.
617 No hardware-assisted watchpoints provided by the target
621 Software watchpoints are very slow, since @value{GDBN} needs to
622 single-step the program being debugged and test the value of the
623 watched expression(s) after each instruction. The rest of this
624 section is mostly irrelevant for software watchpoints.
626 When the inferior stops, @value{GDBN} tries to establish, among other
627 possible reasons, whether it stopped due to a watchpoint being hit.
628 It first uses @code{STOPPED_BY_WATCHPOINT} to see if any watchpoint
629 was hit. If not, all watchpoint checking is skipped.
631 Then @value{GDBN} calls @code{target_stopped_data_address} exactly
632 once. This method returns the address of the watchpoint which
633 triggered, if the target can determine it. If the triggered address
634 is available, @value{GDBN} compares the address returned by this
635 method with each watched memory address in each active watchpoint.
636 For data-read and data-access watchpoints, @value{GDBN} announces
637 every watchpoint that watches the triggered address as being hit.
638 For this reason, data-read and data-access watchpoints
639 @emph{require} that the triggered address be available; if not, read
640 and access watchpoints will never be considered hit. For data-write
641 watchpoints, if the triggered address is available, @value{GDBN}
642 considers only those watchpoints which match that address;
643 otherwise, @value{GDBN} considers all data-write watchpoints. For
644 each data-write watchpoint that @value{GDBN} considers, it evaluates
645 the expression whose value is being watched, and tests whether the
646 watched value has changed. Watchpoints whose watched values have
647 changed are announced as hit.
649 @c FIXME move these to the main lists of target/native defns
651 @value{GDBN} uses several macros and primitives to support hardware
655 @findex TARGET_HAS_HARDWARE_WATCHPOINTS
656 @item TARGET_HAS_HARDWARE_WATCHPOINTS
657 If defined, the target supports hardware watchpoints.
658 (Currently only used for several native configs.)
660 @findex TARGET_CAN_USE_HARDWARE_WATCHPOINT
661 @item TARGET_CAN_USE_HARDWARE_WATCHPOINT (@var{type}, @var{count}, @var{other})
662 Return the number of hardware watchpoints of type @var{type} that are
663 possible to be set. The value is positive if @var{count} watchpoints
664 of this type can be set, zero if setting watchpoints of this type is
665 not supported, and negative if @var{count} is more than the maximum
666 number of watchpoints of type @var{type} that can be set. @var{other}
667 is non-zero if other types of watchpoints are currently enabled (there
668 are architectures which cannot set watchpoints of different types at
671 @findex TARGET_REGION_OK_FOR_HW_WATCHPOINT
672 @item TARGET_REGION_OK_FOR_HW_WATCHPOINT (@var{addr}, @var{len})
673 Return non-zero if hardware watchpoints can be used to watch a region
674 whose address is @var{addr} and whose length in bytes is @var{len}.
676 @cindex insert or remove hardware watchpoint
677 @findex target_insert_watchpoint
678 @findex target_remove_watchpoint
679 @item target_insert_watchpoint (@var{addr}, @var{len}, @var{type})
680 @itemx target_remove_watchpoint (@var{addr}, @var{len}, @var{type})
681 Insert or remove a hardware watchpoint starting at @var{addr}, for
682 @var{len} bytes. @var{type} is the watchpoint type, one of the
683 possible values of the enumerated data type @code{target_hw_bp_type},
684 defined by @file{breakpoint.h} as follows:
687 enum target_hw_bp_type
689 hw_write = 0, /* Common (write) HW watchpoint */
690 hw_read = 1, /* Read HW watchpoint */
691 hw_access = 2, /* Access (read or write) HW watchpoint */
692 hw_execute = 3 /* Execute HW breakpoint */
697 These two macros should return 0 for success, non-zero for failure.
699 @findex target_stopped_data_address
700 @item target_stopped_data_address (@var{addr_p})
701 If the inferior has some watchpoint that triggered, place the address
702 associated with the watchpoint at the location pointed to by
703 @var{addr_p} and return non-zero. Otherwise, return zero. This
704 is required for data-read and data-access watchpoints. It is
705 not required for data-write watchpoints, but @value{GDBN} uses
706 it to improve handling of those also.
708 @value{GDBN} will only call this method once per watchpoint stop,
709 immediately after calling @code{STOPPED_BY_WATCHPOINT}. If the
710 target's watchpoint indication is sticky, i.e., stays set after
711 resuming, this method should clear it. For instance, the x86 debug
712 control register has sticky triggered flags.
714 @findex target_watchpoint_addr_within_range
715 @item target_watchpoint_addr_within_range (@var{target}, @var{addr}, @var{start}, @var{length})
716 Check whether @var{addr} (as returned by @code{target_stopped_data_address})
717 lies within the hardware-defined watchpoint region described by
718 @var{start} and @var{length}. This only needs to be provided if the
719 granularity of a watchpoint is greater than one byte, i.e., if the
720 watchpoint can also trigger on nearby addresses outside of the watched
723 @findex HAVE_STEPPABLE_WATCHPOINT
724 @item HAVE_STEPPABLE_WATCHPOINT
725 If defined to a non-zero value, it is not necessary to disable a
726 watchpoint to step over it. Like @code{gdbarch_have_nonsteppable_watchpoint},
727 this is usually set when watchpoints trigger at the instruction
728 which will perform an interesting read or write. It should be
729 set if there is a temporary disable bit which allows the processor
730 to step over the interesting instruction without raising the
731 watchpoint exception again.
733 @findex gdbarch_have_nonsteppable_watchpoint
734 @item int gdbarch_have_nonsteppable_watchpoint (@var{gdbarch})
735 If it returns a non-zero value, @value{GDBN} should disable a
736 watchpoint to step the inferior over it. This is usually set when
737 watchpoints trigger at the instruction which will perform an
738 interesting read or write.
740 @findex HAVE_CONTINUABLE_WATCHPOINT
741 @item HAVE_CONTINUABLE_WATCHPOINT
742 If defined to a non-zero value, it is possible to continue the
743 inferior after a watchpoint has been hit. This is usually set
744 when watchpoints trigger at the instruction following an interesting
747 @findex CANNOT_STEP_HW_WATCHPOINTS
748 @item CANNOT_STEP_HW_WATCHPOINTS
749 If this is defined to a non-zero value, @value{GDBN} will remove all
750 watchpoints before stepping the inferior.
752 @findex STOPPED_BY_WATCHPOINT
753 @item STOPPED_BY_WATCHPOINT (@var{wait_status})
754 Return non-zero if stopped by a watchpoint. @var{wait_status} is of
755 the type @code{struct target_waitstatus}, defined by @file{target.h}.
756 Normally, this macro is defined to invoke the function pointed to by
757 the @code{to_stopped_by_watchpoint} member of the structure (of the
758 type @code{target_ops}, defined on @file{target.h}) that describes the
759 target-specific operations; @code{to_stopped_by_watchpoint} ignores
760 the @var{wait_status} argument.
762 @value{GDBN} does not require the non-zero value returned by
763 @code{STOPPED_BY_WATCHPOINT} to be 100% correct, so if a target cannot
764 determine for sure whether the inferior stopped due to a watchpoint,
765 it could return non-zero ``just in case''.
768 @subsection Watchpoints and Threads
769 @cindex watchpoints, with threads
771 @value{GDBN} only supports process-wide watchpoints, which trigger
772 in all threads. @value{GDBN} uses the thread ID to make watchpoints
773 act as if they were thread-specific, but it cannot set hardware
774 watchpoints that only trigger in a specific thread. Therefore, even
775 if the target supports threads, per-thread debug registers, and
776 watchpoints which only affect a single thread, it should set the
777 per-thread debug registers for all threads to the same value. On
778 @sc{gnu}/Linux native targets, this is accomplished by using
779 @code{ALL_LWPS} in @code{target_insert_watchpoint} and
780 @code{target_remove_watchpoint} and by using
781 @code{linux_set_new_thread} to register a handler for newly created
784 @value{GDBN}'s @sc{gnu}/Linux support only reports a single event
785 at a time, although multiple events can trigger simultaneously for
786 multi-threaded programs. When multiple events occur, @file{linux-nat.c}
787 queues subsequent events and returns them the next time the program
788 is resumed. This means that @code{STOPPED_BY_WATCHPOINT} and
789 @code{target_stopped_data_address} only need to consult the current
790 thread's state---the thread indicated by @code{inferior_ptid}. If
791 two threads have hit watchpoints simultaneously, those routines
792 will be called a second time for the second thread.
794 @subsection x86 Watchpoints
795 @cindex x86 debug registers
796 @cindex watchpoints, on x86
798 The 32-bit Intel x86 (a.k.a.@: ia32) processors feature special debug
799 registers designed to facilitate debugging. @value{GDBN} provides a
800 generic library of functions that x86-based ports can use to implement
801 support for watchpoints and hardware-assisted breakpoints. This
802 subsection documents the x86 watchpoint facilities in @value{GDBN}.
804 (At present, the library functions read and write debug registers directly, and are
805 thus only available for native configurations.)
807 To use the generic x86 watchpoint support, a port should do the
811 @findex I386_USE_GENERIC_WATCHPOINTS
813 Define the macro @code{I386_USE_GENERIC_WATCHPOINTS} somewhere in the
814 target-dependent headers.
817 Include the @file{config/i386/nm-i386.h} header file @emph{after}
818 defining @code{I386_USE_GENERIC_WATCHPOINTS}.
821 Add @file{i386-nat.o} to the value of the Make variable
822 @code{NATDEPFILES} (@pxref{Native Debugging, NATDEPFILES}).
825 Provide implementations for the @code{I386_DR_LOW_*} macros described
826 below. Typically, each macro should call a target-specific function
827 which does the real work.
830 The x86 watchpoint support works by maintaining mirror images of the
831 debug registers. Values are copied between the mirror images and the
832 real debug registers via a set of macros which each target needs to
836 @findex I386_DR_LOW_SET_CONTROL
837 @item I386_DR_LOW_SET_CONTROL (@var{val})
838 Set the Debug Control (DR7) register to the value @var{val}.
840 @findex I386_DR_LOW_SET_ADDR
841 @item I386_DR_LOW_SET_ADDR (@var{idx}, @var{addr})
842 Put the address @var{addr} into the debug register number @var{idx}.
844 @findex I386_DR_LOW_RESET_ADDR
845 @item I386_DR_LOW_RESET_ADDR (@var{idx})
846 Reset (i.e.@: zero out) the address stored in the debug register
849 @findex I386_DR_LOW_GET_STATUS
850 @item I386_DR_LOW_GET_STATUS
851 Return the value of the Debug Status (DR6) register. This value is
852 used immediately after it is returned by
853 @code{I386_DR_LOW_GET_STATUS}, so as to support per-thread status
857 For each one of the 4 debug registers (whose indices are from 0 to 3)
858 that store addresses, a reference count is maintained by @value{GDBN},
859 to allow sharing of debug registers by several watchpoints. This
860 allows users to define several watchpoints that watch the same
861 expression, but with different conditions and/or commands, without
862 wasting debug registers which are in short supply. @value{GDBN}
863 maintains the reference counts internally, targets don't have to do
864 anything to use this feature.
866 The x86 debug registers can each watch a region that is 1, 2, or 4
867 bytes long. The ia32 architecture requires that each watched region
868 be appropriately aligned: 2-byte region on 2-byte boundary, 4-byte
869 region on 4-byte boundary. However, the x86 watchpoint support in
870 @value{GDBN} can watch unaligned regions and regions larger than 4
871 bytes (up to 16 bytes) by allocating several debug registers to watch
872 a single region. This allocation of several registers per a watched
873 region is also done automatically without target code intervention.
875 The generic x86 watchpoint support provides the following API for the
876 @value{GDBN}'s application code:
879 @findex i386_region_ok_for_watchpoint
880 @item i386_region_ok_for_watchpoint (@var{addr}, @var{len})
881 The macro @code{TARGET_REGION_OK_FOR_HW_WATCHPOINT} is set to call
882 this function. It counts the number of debug registers required to
883 watch a given region, and returns a non-zero value if that number is
884 less than 4, the number of debug registers available to x86
887 @findex i386_stopped_data_address
888 @item i386_stopped_data_address (@var{addr_p})
890 @code{target_stopped_data_address} is set to call this function.
892 function examines the breakpoint condition bits in the DR6 Debug
893 Status register, as returned by the @code{I386_DR_LOW_GET_STATUS}
894 macro, and returns the address associated with the first bit that is
897 @findex i386_stopped_by_watchpoint
898 @item i386_stopped_by_watchpoint (void)
899 The macro @code{STOPPED_BY_WATCHPOINT}
900 is set to call this function. The
901 argument passed to @code{STOPPED_BY_WATCHPOINT} is ignored. This
902 function examines the breakpoint condition bits in the DR6 Debug
903 Status register, as returned by the @code{I386_DR_LOW_GET_STATUS}
904 macro, and returns true if any bit is set. Otherwise, false is
907 @findex i386_insert_watchpoint
908 @findex i386_remove_watchpoint
909 @item i386_insert_watchpoint (@var{addr}, @var{len}, @var{type})
910 @itemx i386_remove_watchpoint (@var{addr}, @var{len}, @var{type})
911 Insert or remove a watchpoint. The macros
912 @code{target_insert_watchpoint} and @code{target_remove_watchpoint}
913 are set to call these functions. @code{i386_insert_watchpoint} first
914 looks for a debug register which is already set to watch the same
915 region for the same access types; if found, it just increments the
916 reference count of that debug register, thus implementing debug
917 register sharing between watchpoints. If no such register is found,
918 the function looks for a vacant debug register, sets its mirrored
919 value to @var{addr}, sets the mirrored value of DR7 Debug Control
920 register as appropriate for the @var{len} and @var{type} parameters,
921 and then passes the new values of the debug register and DR7 to the
922 inferior by calling @code{I386_DR_LOW_SET_ADDR} and
923 @code{I386_DR_LOW_SET_CONTROL}. If more than one debug register is
924 required to cover the given region, the above process is repeated for
927 @code{i386_remove_watchpoint} does the opposite: it resets the address
928 in the mirrored value of the debug register and its read/write and
929 length bits in the mirrored value of DR7, then passes these new
930 values to the inferior via @code{I386_DR_LOW_RESET_ADDR} and
931 @code{I386_DR_LOW_SET_CONTROL}. If a register is shared by several
932 watchpoints, each time a @code{i386_remove_watchpoint} is called, it
933 decrements the reference count, and only calls
934 @code{I386_DR_LOW_RESET_ADDR} and @code{I386_DR_LOW_SET_CONTROL} when
935 the count goes to zero.
937 @findex i386_insert_hw_breakpoint
938 @findex i386_remove_hw_breakpoint
939 @item i386_insert_hw_breakpoint (@var{bp_tgt})
940 @itemx i386_remove_hw_breakpoint (@var{bp_tgt})
941 These functions insert and remove hardware-assisted breakpoints. The
942 macros @code{target_insert_hw_breakpoint} and
943 @code{target_remove_hw_breakpoint} are set to call these functions.
944 The argument is a @code{struct bp_target_info *}, as described in
945 the documentation for @code{target_insert_breakpoint}.
946 These functions work like @code{i386_insert_watchpoint} and
947 @code{i386_remove_watchpoint}, respectively, except that they set up
948 the debug registers to watch instruction execution, and each
949 hardware-assisted breakpoint always requires exactly one debug
952 @findex i386_stopped_by_hwbp
953 @item i386_stopped_by_hwbp (void)
954 This function returns non-zero if the inferior has some watchpoint or
955 hardware breakpoint that triggered. It works like
956 @code{i386_stopped_data_address}, except that it doesn't record the
957 address whose watchpoint triggered.
959 @findex i386_cleanup_dregs
960 @item i386_cleanup_dregs (void)
961 This function clears all the reference counts, addresses, and control
962 bits in the mirror images of the debug registers. It doesn't affect
963 the actual debug registers in the inferior process.
970 x86 processors support setting watchpoints on I/O reads or writes.
971 However, since no target supports this (as of March 2001), and since
972 @code{enum target_hw_bp_type} doesn't even have an enumeration for I/O
973 watchpoints, this feature is not yet available to @value{GDBN} running
977 x86 processors can enable watchpoints locally, for the current task
978 only, or globally, for all the tasks. For each debug register,
979 there's a bit in the DR7 Debug Control register that determines
980 whether the associated address is watched locally or globally. The
981 current implementation of x86 watchpoint support in @value{GDBN}
982 always sets watchpoints to be locally enabled, since global
983 watchpoints might interfere with the underlying OS and are probably
984 unavailable in many platforms.
990 In the abstract, a checkpoint is a point in the execution history of
991 the program, which the user may wish to return to at some later time.
993 Internally, a checkpoint is a saved copy of the program state, including
994 whatever information is required in order to restore the program to that
995 state at a later time. This can be expected to include the state of
996 registers and memory, and may include external state such as the state
997 of open files and devices.
999 There are a number of ways in which checkpoints may be implemented
1000 in gdb, e.g.@: as corefiles, as forked processes, and as some opaque
1001 method implemented on the target side.
1003 A corefile can be used to save an image of target memory and register
1004 state, which can in principle be restored later --- but corefiles do
1005 not typically include information about external entities such as
1006 open files. Currently this method is not implemented in gdb.
1008 A forked process can save the state of user memory and registers,
1009 as well as some subset of external (kernel) state. This method
1010 is used to implement checkpoints on Linux, and in principle might
1011 be used on other systems.
1013 Some targets, e.g.@: simulators, might have their own built-in
1014 method for saving checkpoints, and gdb might be able to take
1015 advantage of that capability without necessarily knowing any
1016 details of how it is done.
1019 @section Observing changes in @value{GDBN} internals
1020 @cindex observer pattern interface
1021 @cindex notifications about changes in internals
1023 In order to function properly, several modules need to be notified when
1024 some changes occur in the @value{GDBN} internals. Traditionally, these
1025 modules have relied on several paradigms, the most common ones being
1026 hooks and gdb-events. Unfortunately, none of these paradigms was
1027 versatile enough to become the standard notification mechanism in
1028 @value{GDBN}. The fact that they only supported one ``client'' was also
1029 a strong limitation.
1031 A new paradigm, based on the Observer pattern of the @cite{Design
1032 Patterns} book, has therefore been implemented. The goal was to provide
1033 a new interface overcoming the issues with the notification mechanisms
1034 previously available. This new interface needed to be strongly typed,
1035 easy to extend, and versatile enough to be used as the standard
1036 interface when adding new notifications.
1038 See @ref{GDB Observers} for a brief description of the observers
1039 currently implemented in GDB. The rationale for the current
1040 implementation is also briefly discussed.
1042 @node User Interface
1044 @chapter User Interface
1046 @value{GDBN} has several user interfaces, of which the traditional
1047 command-line interface is perhaps the most familiar.
1049 @section Command Interpreter
1051 @cindex command interpreter
1053 The command interpreter in @value{GDBN} is fairly simple. It is designed to
1054 allow for the set of commands to be augmented dynamically, and also
1055 has a recursive subcommand capability, where the first argument to
1056 a command may itself direct a lookup on a different command list.
1058 For instance, the @samp{set} command just starts a lookup on the
1059 @code{setlist} command list, while @samp{set thread} recurses
1060 to the @code{set_thread_cmd_list}.
1064 To add commands in general, use @code{add_cmd}. @code{add_com} adds to
1065 the main command list, and should be used for those commands. The usual
1066 place to add commands is in the @code{_initialize_@var{xyz}} routines at
1067 the ends of most source files.
1069 @findex add_setshow_cmd
1070 @findex add_setshow_cmd_full
1071 To add paired @samp{set} and @samp{show} commands, use
1072 @code{add_setshow_cmd} or @code{add_setshow_cmd_full}. The former is
1073 a slightly simpler interface which is useful when you don't need to
1074 further modify the new command structures, while the latter returns
1075 the new command structures for manipulation.
1077 @cindex deprecating commands
1078 @findex deprecate_cmd
1079 Before removing commands from the command set it is a good idea to
1080 deprecate them for some time. Use @code{deprecate_cmd} on commands or
1081 aliases to set the deprecated flag. @code{deprecate_cmd} takes a
1082 @code{struct cmd_list_element} as it's first argument. You can use the
1083 return value from @code{add_com} or @code{add_cmd} to deprecate the
1084 command immediately after it is created.
1086 The first time a command is used the user will be warned and offered a
1087 replacement (if one exists). Note that the replacement string passed to
1088 @code{deprecate_cmd} should be the full name of the command, i.e., the
1089 entire string the user should type at the command line.
1091 @section UI-Independent Output---the @code{ui_out} Functions
1092 @c This section is based on the documentation written by Fernando
1093 @c Nasser <fnasser@redhat.com>.
1095 @cindex @code{ui_out} functions
1096 The @code{ui_out} functions present an abstraction level for the
1097 @value{GDBN} output code. They hide the specifics of different user
1098 interfaces supported by @value{GDBN}, and thus free the programmer
1099 from the need to write several versions of the same code, one each for
1100 every UI, to produce output.
1102 @subsection Overview and Terminology
1104 In general, execution of each @value{GDBN} command produces some sort
1105 of output, and can even generate an input request.
1107 Output can be generated for the following purposes:
1111 to display a @emph{result} of an operation;
1114 to convey @emph{info} or produce side-effects of a requested
1118 to provide a @emph{notification} of an asynchronous event (including
1119 progress indication of a prolonged asynchronous operation);
1122 to display @emph{error messages} (including warnings);
1125 to show @emph{debug data};
1128 to @emph{query} or prompt a user for input (a special case).
1132 This section mainly concentrates on how to build result output,
1133 although some of it also applies to other kinds of output.
1135 Generation of output that displays the results of an operation
1136 involves one or more of the following:
1140 output of the actual data
1143 formatting the output as appropriate for console output, to make it
1144 easily readable by humans
1147 machine oriented formatting--a more terse formatting to allow for easy
1148 parsing by programs which read @value{GDBN}'s output
1151 annotation, whose purpose is to help legacy GUIs to identify interesting
1155 The @code{ui_out} routines take care of the first three aspects.
1156 Annotations are provided by separate annotation routines. Note that use
1157 of annotations for an interface between a GUI and @value{GDBN} is
1160 Output can be in the form of a single item, which we call a @dfn{field};
1161 a @dfn{list} consisting of identical fields; a @dfn{tuple} consisting of
1162 non-identical fields; or a @dfn{table}, which is a tuple consisting of a
1163 header and a body. In a BNF-like form:
1166 @item <table> @expansion{}
1167 @code{<header> <body>}
1168 @item <header> @expansion{}
1169 @code{@{ <column> @}}
1170 @item <column> @expansion{}
1171 @code{<width> <alignment> <title>}
1172 @item <body> @expansion{}
1177 @subsection General Conventions
1179 Most @code{ui_out} routines are of type @code{void}, the exceptions are
1180 @code{ui_out_stream_new} (which returns a pointer to the newly created
1181 object) and the @code{make_cleanup} routines.
1183 The first parameter is always the @code{ui_out} vector object, a pointer
1184 to a @code{struct ui_out}.
1186 The @var{format} parameter is like in @code{printf} family of functions.
1187 When it is present, there must also be a variable list of arguments
1188 sufficient used to satisfy the @code{%} specifiers in the supplied
1191 When a character string argument is not used in a @code{ui_out} function
1192 call, a @code{NULL} pointer has to be supplied instead.
1195 @subsection Table, Tuple and List Functions
1197 @cindex list output functions
1198 @cindex table output functions
1199 @cindex tuple output functions
1200 This section introduces @code{ui_out} routines for building lists,
1201 tuples and tables. The routines to output the actual data items
1202 (fields) are presented in the next section.
1204 To recap: A @dfn{tuple} is a sequence of @dfn{fields}, each field
1205 containing information about an object; a @dfn{list} is a sequence of
1206 fields where each field describes an identical object.
1208 Use the @dfn{table} functions when your output consists of a list of
1209 rows (tuples) and the console output should include a heading. Use this
1210 even when you are listing just one object but you still want the header.
1212 @cindex nesting level in @code{ui_out} functions
1213 Tables can not be nested. Tuples and lists can be nested up to a
1214 maximum of five levels.
1216 The overall structure of the table output code is something like this:
1231 Here is the description of table-, tuple- and list-related @code{ui_out}
1234 @deftypefun void ui_out_table_begin (struct ui_out *@var{uiout}, int @var{nbrofcols}, int @var{nr_rows}, const char *@var{tblid})
1235 The function @code{ui_out_table_begin} marks the beginning of the output
1236 of a table. It should always be called before any other @code{ui_out}
1237 function for a given table. @var{nbrofcols} is the number of columns in
1238 the table. @var{nr_rows} is the number of rows in the table.
1239 @var{tblid} is an optional string identifying the table. The string
1240 pointed to by @var{tblid} is copied by the implementation of
1241 @code{ui_out_table_begin}, so the application can free the string if it
1242 was @code{malloc}ed.
1244 The companion function @code{ui_out_table_end}, described below, marks
1245 the end of the table's output.
1248 @deftypefun void ui_out_table_header (struct ui_out *@var{uiout}, int @var{width}, enum ui_align @var{alignment}, const char *@var{colhdr})
1249 @code{ui_out_table_header} provides the header information for a single
1250 table column. You call this function several times, one each for every
1251 column of the table, after @code{ui_out_table_begin}, but before
1252 @code{ui_out_table_body}.
1254 The value of @var{width} gives the column width in characters. The
1255 value of @var{alignment} is one of @code{left}, @code{center}, and
1256 @code{right}, and it specifies how to align the header: left-justify,
1257 center, or right-justify it. @var{colhdr} points to a string that
1258 specifies the column header; the implementation copies that string, so
1259 column header strings in @code{malloc}ed storage can be freed after the
1263 @deftypefun void ui_out_table_body (struct ui_out *@var{uiout})
1264 This function delimits the table header from the table body.
1267 @deftypefun void ui_out_table_end (struct ui_out *@var{uiout})
1268 This function signals the end of a table's output. It should be called
1269 after the table body has been produced by the list and field output
1272 There should be exactly one call to @code{ui_out_table_end} for each
1273 call to @code{ui_out_table_begin}, otherwise the @code{ui_out} functions
1274 will signal an internal error.
1277 The output of the tuples that represent the table rows must follow the
1278 call to @code{ui_out_table_body} and precede the call to
1279 @code{ui_out_table_end}. You build a tuple by calling
1280 @code{ui_out_tuple_begin} and @code{ui_out_tuple_end}, with suitable
1281 calls to functions which actually output fields between them.
1283 @deftypefun void ui_out_tuple_begin (struct ui_out *@var{uiout}, const char *@var{id})
1284 This function marks the beginning of a tuple output. @var{id} points
1285 to an optional string that identifies the tuple; it is copied by the
1286 implementation, and so strings in @code{malloc}ed storage can be freed
1290 @deftypefun void ui_out_tuple_end (struct ui_out *@var{uiout})
1291 This function signals an end of a tuple output. There should be exactly
1292 one call to @code{ui_out_tuple_end} for each call to
1293 @code{ui_out_tuple_begin}, otherwise an internal @value{GDBN} error will
1297 @deftypefun struct cleanup *make_cleanup_ui_out_tuple_begin_end (struct ui_out *@var{uiout}, const char *@var{id})
1298 This function first opens the tuple and then establishes a cleanup
1299 (@pxref{Coding, Cleanups}) to close the tuple. It provides a convenient
1300 and correct implementation of the non-portable@footnote{The function
1301 cast is not portable ISO C.} code sequence:
1303 struct cleanup *old_cleanup;
1304 ui_out_tuple_begin (uiout, "...");
1305 old_cleanup = make_cleanup ((void(*)(void *)) ui_out_tuple_end,
1310 @deftypefun void ui_out_list_begin (struct ui_out *@var{uiout}, const char *@var{id})
1311 This function marks the beginning of a list output. @var{id} points to
1312 an optional string that identifies the list; it is copied by the
1313 implementation, and so strings in @code{malloc}ed storage can be freed
1317 @deftypefun void ui_out_list_end (struct ui_out *@var{uiout})
1318 This function signals an end of a list output. There should be exactly
1319 one call to @code{ui_out_list_end} for each call to
1320 @code{ui_out_list_begin}, otherwise an internal @value{GDBN} error will
1324 @deftypefun struct cleanup *make_cleanup_ui_out_list_begin_end (struct ui_out *@var{uiout}, const char *@var{id})
1325 Similar to @code{make_cleanup_ui_out_tuple_begin_end}, this function
1326 opens a list and then establishes cleanup (@pxref{Coding, Cleanups})
1327 that will close the list.
1330 @subsection Item Output Functions
1332 @cindex item output functions
1333 @cindex field output functions
1335 The functions described below produce output for the actual data
1336 items, or fields, which contain information about the object.
1338 Choose the appropriate function accordingly to your particular needs.
1340 @deftypefun void ui_out_field_fmt (struct ui_out *@var{uiout}, char *@var{fldname}, char *@var{format}, ...)
1341 This is the most general output function. It produces the
1342 representation of the data in the variable-length argument list
1343 according to formatting specifications in @var{format}, a
1344 @code{printf}-like format string. The optional argument @var{fldname}
1345 supplies the name of the field. The data items themselves are
1346 supplied as additional arguments after @var{format}.
1348 This generic function should be used only when it is not possible to
1349 use one of the specialized versions (see below).
1352 @deftypefun void ui_out_field_int (struct ui_out *@var{uiout}, const char *@var{fldname}, int @var{value})
1353 This function outputs a value of an @code{int} variable. It uses the
1354 @code{"%d"} output conversion specification. @var{fldname} specifies
1355 the name of the field.
1358 @deftypefun void ui_out_field_fmt_int (struct ui_out *@var{uiout}, int @var{width}, enum ui_align @var{alignment}, const char *@var{fldname}, int @var{value})
1359 This function outputs a value of an @code{int} variable. It differs from
1360 @code{ui_out_field_int} in that the caller specifies the desired @var{width} and @var{alignment} of the output.
1361 @var{fldname} specifies
1362 the name of the field.
1365 @deftypefun void ui_out_field_core_addr (struct ui_out *@var{uiout}, const char *@var{fldname}, CORE_ADDR @var{address})
1366 This function outputs an address.
1369 @deftypefun void ui_out_field_string (struct ui_out *@var{uiout}, const char *@var{fldname}, const char *@var{string})
1370 This function outputs a string using the @code{"%s"} conversion
1374 Sometimes, there's a need to compose your output piece by piece using
1375 functions that operate on a stream, such as @code{value_print} or
1376 @code{fprintf_symbol_filtered}. These functions accept an argument of
1377 the type @code{struct ui_file *}, a pointer to a @code{ui_file} object
1378 used to store the data stream used for the output. When you use one
1379 of these functions, you need a way to pass their results stored in a
1380 @code{ui_file} object to the @code{ui_out} functions. To this end,
1381 you first create a @code{ui_stream} object by calling
1382 @code{ui_out_stream_new}, pass the @code{stream} member of that
1383 @code{ui_stream} object to @code{value_print} and similar functions,
1384 and finally call @code{ui_out_field_stream} to output the field you
1385 constructed. When the @code{ui_stream} object is no longer needed,
1386 you should destroy it and free its memory by calling
1387 @code{ui_out_stream_delete}.
1389 @deftypefun struct ui_stream *ui_out_stream_new (struct ui_out *@var{uiout})
1390 This function creates a new @code{ui_stream} object which uses the
1391 same output methods as the @code{ui_out} object whose pointer is
1392 passed in @var{uiout}. It returns a pointer to the newly created
1393 @code{ui_stream} object.
1396 @deftypefun void ui_out_stream_delete (struct ui_stream *@var{streambuf})
1397 This functions destroys a @code{ui_stream} object specified by
1401 @deftypefun void ui_out_field_stream (struct ui_out *@var{uiout}, const char *@var{fieldname}, struct ui_stream *@var{streambuf})
1402 This function consumes all the data accumulated in
1403 @code{streambuf->stream} and outputs it like
1404 @code{ui_out_field_string} does. After a call to
1405 @code{ui_out_field_stream}, the accumulated data no longer exists, but
1406 the stream is still valid and may be used for producing more fields.
1409 @strong{Important:} If there is any chance that your code could bail
1410 out before completing output generation and reaching the point where
1411 @code{ui_out_stream_delete} is called, it is necessary to set up a
1412 cleanup, to avoid leaking memory and other resources. Here's a
1413 skeleton code to do that:
1416 struct ui_stream *mybuf = ui_out_stream_new (uiout);
1417 struct cleanup *old = make_cleanup (ui_out_stream_delete, mybuf);
1422 If the function already has the old cleanup chain set (for other kinds
1423 of cleanups), you just have to add your cleanup to it:
1426 mybuf = ui_out_stream_new (uiout);
1427 make_cleanup (ui_out_stream_delete, mybuf);
1430 Note that with cleanups in place, you should not call
1431 @code{ui_out_stream_delete} directly, or you would attempt to free the
1434 @subsection Utility Output Functions
1436 @deftypefun void ui_out_field_skip (struct ui_out *@var{uiout}, const char *@var{fldname})
1437 This function skips a field in a table. Use it if you have to leave
1438 an empty field without disrupting the table alignment. The argument
1439 @var{fldname} specifies a name for the (missing) filed.
1442 @deftypefun void ui_out_text (struct ui_out *@var{uiout}, const char *@var{string})
1443 This function outputs the text in @var{string} in a way that makes it
1444 easy to be read by humans. For example, the console implementation of
1445 this method filters the text through a built-in pager, to prevent it
1446 from scrolling off the visible portion of the screen.
1448 Use this function for printing relatively long chunks of text around
1449 the actual field data: the text it produces is not aligned according
1450 to the table's format. Use @code{ui_out_field_string} to output a
1451 string field, and use @code{ui_out_message}, described below, to
1452 output short messages.
1455 @deftypefun void ui_out_spaces (struct ui_out *@var{uiout}, int @var{nspaces})
1456 This function outputs @var{nspaces} spaces. It is handy to align the
1457 text produced by @code{ui_out_text} with the rest of the table or
1461 @deftypefun void ui_out_message (struct ui_out *@var{uiout}, int @var{verbosity}, const char *@var{format}, ...)
1462 This function produces a formatted message, provided that the current
1463 verbosity level is at least as large as given by @var{verbosity}. The
1464 current verbosity level is specified by the user with the @samp{set
1465 verbositylevel} command.@footnote{As of this writing (April 2001),
1466 setting verbosity level is not yet implemented, and is always returned
1467 as zero. So calling @code{ui_out_message} with a @var{verbosity}
1468 argument more than zero will cause the message to never be printed.}
1471 @deftypefun void ui_out_wrap_hint (struct ui_out *@var{uiout}, char *@var{indent})
1472 This function gives the console output filter (a paging filter) a hint
1473 of where to break lines which are too long. Ignored for all other
1474 output consumers. @var{indent}, if non-@code{NULL}, is the string to
1475 be printed to indent the wrapped text on the next line; it must remain
1476 accessible until the next call to @code{ui_out_wrap_hint}, or until an
1477 explicit newline is produced by one of the other functions. If
1478 @var{indent} is @code{NULL}, the wrapped text will not be indented.
1481 @deftypefun void ui_out_flush (struct ui_out *@var{uiout})
1482 This function flushes whatever output has been accumulated so far, if
1483 the UI buffers output.
1487 @subsection Examples of Use of @code{ui_out} functions
1489 @cindex using @code{ui_out} functions
1490 @cindex @code{ui_out} functions, usage examples
1491 This section gives some practical examples of using the @code{ui_out}
1492 functions to generalize the old console-oriented code in
1493 @value{GDBN}. The examples all come from functions defined on the
1494 @file{breakpoints.c} file.
1496 This example, from the @code{breakpoint_1} function, shows how to
1499 The original code was:
1502 if (!found_a_breakpoint++)
1504 annotate_breakpoints_headers ();
1507 printf_filtered ("Num ");
1509 printf_filtered ("Type ");
1511 printf_filtered ("Disp ");
1513 printf_filtered ("Enb ");
1517 printf_filtered ("Address ");
1520 printf_filtered ("What\n");
1522 annotate_breakpoints_table ();
1526 Here's the new version:
1529 nr_printable_breakpoints = @dots{};
1532 ui_out_table_begin (ui, 6, nr_printable_breakpoints, "BreakpointTable");
1534 ui_out_table_begin (ui, 5, nr_printable_breakpoints, "BreakpointTable");
1536 if (nr_printable_breakpoints > 0)
1537 annotate_breakpoints_headers ();
1538 if (nr_printable_breakpoints > 0)
1540 ui_out_table_header (uiout, 3, ui_left, "number", "Num"); /* 1 */
1541 if (nr_printable_breakpoints > 0)
1543 ui_out_table_header (uiout, 14, ui_left, "type", "Type"); /* 2 */
1544 if (nr_printable_breakpoints > 0)
1546 ui_out_table_header (uiout, 4, ui_left, "disp", "Disp"); /* 3 */
1547 if (nr_printable_breakpoints > 0)
1549 ui_out_table_header (uiout, 3, ui_left, "enabled", "Enb"); /* 4 */
1552 if (nr_printable_breakpoints > 0)
1554 if (gdbarch_addr_bit (current_gdbarch) <= 32)
1555 ui_out_table_header (uiout, 10, ui_left, "addr", "Address");/* 5 */
1557 ui_out_table_header (uiout, 18, ui_left, "addr", "Address");/* 5 */
1559 if (nr_printable_breakpoints > 0)
1561 ui_out_table_header (uiout, 40, ui_noalign, "what", "What"); /* 6 */
1562 ui_out_table_body (uiout);
1563 if (nr_printable_breakpoints > 0)
1564 annotate_breakpoints_table ();
1567 This example, from the @code{print_one_breakpoint} function, shows how
1568 to produce the actual data for the table whose structure was defined
1569 in the above example. The original code was:
1574 printf_filtered ("%-3d ", b->number);
1576 if ((int)b->type > (sizeof(bptypes)/sizeof(bptypes[0]))
1577 || ((int) b->type != bptypes[(int) b->type].type))
1578 internal_error ("bptypes table does not describe type #%d.",
1580 printf_filtered ("%-14s ", bptypes[(int)b->type].description);
1582 printf_filtered ("%-4s ", bpdisps[(int)b->disposition]);
1584 printf_filtered ("%-3c ", bpenables[(int)b->enable]);
1588 This is the new version:
1592 ui_out_tuple_begin (uiout, "bkpt");
1594 ui_out_field_int (uiout, "number", b->number);
1596 if (((int) b->type > (sizeof (bptypes) / sizeof (bptypes[0])))
1597 || ((int) b->type != bptypes[(int) b->type].type))
1598 internal_error ("bptypes table does not describe type #%d.",
1600 ui_out_field_string (uiout, "type", bptypes[(int)b->type].description);
1602 ui_out_field_string (uiout, "disp", bpdisps[(int)b->disposition]);
1604 ui_out_field_fmt (uiout, "enabled", "%c", bpenables[(int)b->enable]);
1608 This example, also from @code{print_one_breakpoint}, shows how to
1609 produce a complicated output field using the @code{print_expression}
1610 functions which requires a stream to be passed. It also shows how to
1611 automate stream destruction with cleanups. The original code was:
1615 print_expression (b->exp, gdb_stdout);
1621 struct ui_stream *stb = ui_out_stream_new (uiout);
1622 struct cleanup *old_chain = make_cleanup_ui_out_stream_delete (stb);
1625 print_expression (b->exp, stb->stream);
1626 ui_out_field_stream (uiout, "what", local_stream);
1629 This example, also from @code{print_one_breakpoint}, shows how to use
1630 @code{ui_out_text} and @code{ui_out_field_string}. The original code
1635 if (b->dll_pathname == NULL)
1636 printf_filtered ("<any library> ");
1638 printf_filtered ("library \"%s\" ", b->dll_pathname);
1645 if (b->dll_pathname == NULL)
1647 ui_out_field_string (uiout, "what", "<any library>");
1648 ui_out_spaces (uiout, 1);
1652 ui_out_text (uiout, "library \"");
1653 ui_out_field_string (uiout, "what", b->dll_pathname);
1654 ui_out_text (uiout, "\" ");
1658 The following example from @code{print_one_breakpoint} shows how to
1659 use @code{ui_out_field_int} and @code{ui_out_spaces}. The original
1664 if (b->forked_inferior_pid != 0)
1665 printf_filtered ("process %d ", b->forked_inferior_pid);
1672 if (b->forked_inferior_pid != 0)
1674 ui_out_text (uiout, "process ");
1675 ui_out_field_int (uiout, "what", b->forked_inferior_pid);
1676 ui_out_spaces (uiout, 1);
1680 Here's an example of using @code{ui_out_field_string}. The original
1685 if (b->exec_pathname != NULL)
1686 printf_filtered ("program \"%s\" ", b->exec_pathname);
1693 if (b->exec_pathname != NULL)
1695 ui_out_text (uiout, "program \"");
1696 ui_out_field_string (uiout, "what", b->exec_pathname);
1697 ui_out_text (uiout, "\" ");
1701 Finally, here's an example of printing an address. The original code:
1705 printf_filtered ("%s ",
1706 hex_string_custom ((unsigned long) b->address, 8));
1713 ui_out_field_core_addr (uiout, "Address", b->address);
1717 @section Console Printing
1726 @cindex @code{libgdb}
1727 @code{libgdb} 1.0 was an abortive project of years ago. The theory was
1728 to provide an API to @value{GDBN}'s functionality.
1731 @cindex @code{libgdb}
1732 @code{libgdb} 2.0 is an ongoing effort to update @value{GDBN} so that is
1733 better able to support graphical and other environments.
1735 Since @code{libgdb} development is on-going, its architecture is still
1736 evolving. The following components have so far been identified:
1740 Observer - @file{gdb-events.h}.
1742 Builder - @file{ui-out.h}
1744 Event Loop - @file{event-loop.h}
1746 Library - @file{gdb.h}
1749 The model that ties these components together is described below.
1751 @section The @code{libgdb} Model
1753 A client of @code{libgdb} interacts with the library in two ways.
1757 As an observer (using @file{gdb-events}) receiving notifications from
1758 @code{libgdb} of any internal state changes (break point changes, run
1761 As a client querying @code{libgdb} (using the @file{ui-out} builder) to
1762 obtain various status values from @value{GDBN}.
1765 Since @code{libgdb} could have multiple clients (e.g., a GUI supporting
1766 the existing @value{GDBN} CLI), those clients must co-operate when
1767 controlling @code{libgdb}. In particular, a client must ensure that
1768 @code{libgdb} is idle (i.e. no other client is using @code{libgdb})
1769 before responding to a @file{gdb-event} by making a query.
1771 @section CLI support
1773 At present @value{GDBN}'s CLI is very much entangled in with the core of
1774 @code{libgdb}. Consequently, a client wishing to include the CLI in
1775 their interface needs to carefully co-ordinate its own and the CLI's
1778 It is suggested that the client set @code{libgdb} up to be bi-modal
1779 (alternate between CLI and client query modes). The notes below sketch
1784 The client registers itself as an observer of @code{libgdb}.
1786 The client create and install @code{cli-out} builder using its own
1787 versions of the @code{ui-file} @code{gdb_stderr}, @code{gdb_stdtarg} and
1788 @code{gdb_stdout} streams.
1790 The client creates a separate custom @code{ui-out} builder that is only
1791 used while making direct queries to @code{libgdb}.
1794 When the client receives input intended for the CLI, it simply passes it
1795 along. Since the @code{cli-out} builder is installed by default, all
1796 the CLI output in response to that command is routed (pronounced rooted)
1797 through to the client controlled @code{gdb_stdout} et.@: al.@: streams.
1798 At the same time, the client is kept abreast of internal changes by
1799 virtue of being a @code{libgdb} observer.
1801 The only restriction on the client is that it must wait until
1802 @code{libgdb} becomes idle before initiating any queries (using the
1803 client's custom builder).
1805 @section @code{libgdb} components
1807 @subheading Observer - @file{gdb-events.h}
1808 @file{gdb-events} provides the client with a very raw mechanism that can
1809 be used to implement an observer. At present it only allows for one
1810 observer and that observer must, internally, handle the need to delay
1811 the processing of any event notifications until after @code{libgdb} has
1812 finished the current command.
1814 @subheading Builder - @file{ui-out.h}
1815 @file{ui-out} provides the infrastructure necessary for a client to
1816 create a builder. That builder is then passed down to @code{libgdb}
1817 when doing any queries.
1819 @subheading Event Loop - @file{event-loop.h}
1820 @c There could be an entire section on the event-loop
1821 @file{event-loop}, currently non-re-entrant, provides a simple event
1822 loop. A client would need to either plug its self into this loop or,
1823 implement a new event-loop that GDB would use.
1825 The event-loop will eventually be made re-entrant. This is so that
1826 @value{GDBN} can better handle the problem of some commands blocking
1827 instead of returning.
1829 @subheading Library - @file{gdb.h}
1830 @file{libgdb} is the most obvious component of this system. It provides
1831 the query interface. Each function is parameterized by a @code{ui-out}
1832 builder. The result of the query is constructed using that builder
1833 before the query function returns.
1840 @cindex @code{value} structure
1841 @value{GDBN} uses @code{struct value}, or @dfn{values}, as an internal
1842 abstraction for the representation of a variety of inferior objects
1843 and @value{GDBN} convenience objects.
1845 Values have an associated @code{struct type}, that describes a virtual
1846 view of the raw data or object stored in or accessed through the
1849 A value is in addition discriminated by its lvalue-ness, given its
1850 @code{enum lval_type} enumeration type:
1852 @cindex @code{lval_type} enumeration, for values.
1854 @item @code{not_lval}
1855 This value is not an lval. It can't be assigned to.
1857 @item @code{lval_memory}
1858 This value represents an object in memory.
1860 @item @code{lval_register}
1861 This value represents an object that lives in a register.
1863 @item @code{lval_internalvar}
1864 Represents the value of an internal variable.
1866 @item @code{lval_internalvar_component}
1867 Represents part of a @value{GDBN} internal variable. E.g., a
1870 @cindex computed values
1871 @item @code{lval_computed}
1872 These are ``computed'' values. They allow creating specialized value
1873 objects for specific purposes, all abstracted away from the core value
1874 support code. The creator of such a value writes specialized
1875 functions to handle the reading and writing to/from the value's
1876 backend data, and optionally, a ``copy operator'' and a
1879 Pointers to these functions are stored in a @code{struct lval_funcs}
1880 instance (declared in @file{value.h}), and passed to the
1881 @code{allocate_computed_value} function, as in the example below.
1885 nil_value_read (struct value *v)
1887 /* This callback reads data from some backend, and stores it in V.
1888 In this case, we always read null data. You'll want to fill in
1889 something more interesting. */
1891 memset (value_contents_all_raw (v),
1893 TYPE_LENGTH (value_type (v)));
1897 nil_value_write (struct value *v, struct value *fromval)
1899 /* Takes the data from FROMVAL and stores it in the backend of V. */
1901 to_oblivion (value_contents_all_raw (fromval),
1903 TYPE_LENGTH (value_type (fromval)));
1906 static struct lval_funcs nil_value_funcs =
1913 make_nil_value (void)
1918 type = make_nils_type ();
1919 v = allocate_computed_value (type, &nil_value_funcs, NULL);
1925 See the implementation of the @code{$_siginfo} convenience variable in
1926 @file{infrun.c} as a real example use of lval_computed.
1931 @chapter Stack Frames
1934 @cindex call stack frame
1935 A frame is a construct that @value{GDBN} uses to keep track of calling
1936 and called functions.
1938 @cindex unwind frame
1939 @value{GDBN}'s frame model, a fresh design, was implemented with the
1940 need to support @sc{dwarf}'s Call Frame Information in mind. In fact,
1941 the term ``unwind'' is taken directly from that specification.
1942 Developers wishing to learn more about unwinders, are encouraged to
1943 read the @sc{dwarf} specification, available from
1944 @url{http://www.dwarfstd.org}.
1946 @findex frame_register_unwind
1947 @findex get_frame_register
1948 @value{GDBN}'s model is that you find a frame's registers by
1949 ``unwinding'' them from the next younger frame. That is,
1950 @samp{get_frame_register} which returns the value of a register in
1951 frame #1 (the next-to-youngest frame), is implemented by calling frame
1952 #0's @code{frame_register_unwind} (the youngest frame). But then the
1953 obvious question is: how do you access the registers of the youngest
1956 @cindex sentinel frame
1957 @findex get_frame_type
1958 @vindex SENTINEL_FRAME
1959 To answer this question, GDB has the @dfn{sentinel} frame, the
1960 ``-1st'' frame. Unwinding registers from the sentinel frame gives you
1961 the current values of the youngest real frame's registers. If @var{f}
1962 is a sentinel frame, then @code{get_frame_type (@var{f}) @equiv{}
1965 @section Selecting an Unwinder
1967 @findex frame_unwind_prepend_unwinder
1968 @findex frame_unwind_append_unwinder
1969 The architecture registers a list of frame unwinders (@code{struct
1970 frame_unwind}), using the functions
1971 @code{frame_unwind_prepend_unwinder} and
1972 @code{frame_unwind_append_unwinder}. Each unwinder includes a
1973 sniffer. Whenever @value{GDBN} needs to unwind a frame (to fetch the
1974 previous frame's registers or the current frame's ID), it calls
1975 registered sniffers in order to find one which recognizes the frame.
1976 The first time a sniffer returns non-zero, the corresponding unwinder
1977 is assigned to the frame.
1979 @section Unwinding the Frame ID
1982 Every frame has an associated ID, of type @code{struct frame_id}.
1983 The ID includes the stack base and function start address for
1984 the frame. The ID persists through the entire life of the frame,
1985 including while other called frames are running; it is used to
1986 locate an appropriate @code{struct frame_info} from the cache.
1988 Every time the inferior stops, and at various other times, the frame
1989 cache is flushed. Because of this, parts of @value{GDBN} which need
1990 to keep track of individual frames cannot use pointers to @code{struct
1991 frame_info}. A frame ID provides a stable reference to a frame, even
1992 when the unwinder must be run again to generate a new @code{struct
1993 frame_info} for the same frame.
1995 The frame's unwinder's @code{this_id} method is called to find the ID.
1996 Note that this is different from register unwinding, where the next
1997 frame's @code{prev_register} is called to unwind this frame's
2000 Both stack base and function address are required to identify the
2001 frame, because a recursive function has the same function address for
2002 two consecutive frames and a leaf function may have the same stack
2003 address as its caller. On some platforms, a third address is part of
2004 the ID to further disambiguate frames---for instance, on IA-64
2005 the separate register stack address is included in the ID.
2007 An invalid frame ID (@code{null_frame_id}) returned from the
2008 @code{this_id} method means to stop unwinding after this frame.
2010 @section Unwinding Registers
2012 Each unwinder includes a @code{prev_register} method. This method
2013 takes a frame, an associated cache pointer, and a register number.
2014 It returns a @code{struct value *} describing the requested register,
2015 as saved by this frame. This is the value of the register that is
2016 current in this frame's caller.
2018 The returned value must have the same type as the register. It may
2019 have any lvalue type. In most circumstances one of these routines
2020 will generate the appropriate value:
2023 @item frame_unwind_got_optimized
2024 @findex frame_unwind_got_optimized
2025 This register was not saved.
2027 @item frame_unwind_got_register
2028 @findex frame_unwind_got_register
2029 This register was copied into another register in this frame. This
2030 is also used for unchanged registers; they are ``copied'' into the
2033 @item frame_unwind_got_memory
2034 @findex frame_unwind_got_memory
2035 This register was saved in memory.
2037 @item frame_unwind_got_constant
2038 @findex frame_unwind_got_constant
2039 This register was not saved, but the unwinder can compute the previous
2040 value some other way.
2042 @item frame_unwind_got_address
2043 @findex frame_unwind_got_address
2044 Same as @code{frame_unwind_got_constant}, except that the value is a target
2045 address. This is frequently used for the stack pointer, which is not
2046 explicitly saved but has a known offset from this frame's stack
2047 pointer. For architectures with a flat unified address space, this is
2048 generally the same as @code{frame_unwind_got_constant}.
2051 @node Symbol Handling
2053 @chapter Symbol Handling
2055 Symbols are a key part of @value{GDBN}'s operation. Symbols include
2056 variables, functions, and types.
2058 Symbol information for a large program can be truly massive, and
2059 reading of symbol information is one of the major performance
2060 bottlenecks in @value{GDBN}; it can take many minutes to process it
2061 all. Studies have shown that nearly all the time spent is
2062 computational, rather than file reading.
2064 One of the ways for @value{GDBN} to provide a good user experience is
2065 to start up quickly, taking no more than a few seconds. It is simply
2066 not possible to process all of a program's debugging info in that
2067 time, and so we attempt to handle symbols incrementally. For instance,
2068 we create @dfn{partial symbol tables} consisting of only selected
2069 symbols, and only expand them to full symbol tables when necessary.
2071 @section Symbol Reading
2073 @cindex symbol reading
2074 @cindex reading of symbols
2075 @cindex symbol files
2076 @value{GDBN} reads symbols from @dfn{symbol files}. The usual symbol
2077 file is the file containing the program which @value{GDBN} is
2078 debugging. @value{GDBN} can be directed to use a different file for
2079 symbols (with the @samp{symbol-file} command), and it can also read
2080 more symbols via the @samp{add-file} and @samp{load} commands. In
2081 addition, it may bring in more symbols while loading shared
2084 @findex find_sym_fns
2085 Symbol files are initially opened by code in @file{symfile.c} using
2086 the BFD library (@pxref{Support Libraries}). BFD identifies the type
2087 of the file by examining its header. @code{find_sym_fns} then uses
2088 this identification to locate a set of symbol-reading functions.
2090 @findex add_symtab_fns
2091 @cindex @code{sym_fns} structure
2092 @cindex adding a symbol-reading module
2093 Symbol-reading modules identify themselves to @value{GDBN} by calling
2094 @code{add_symtab_fns} during their module initialization. The argument
2095 to @code{add_symtab_fns} is a @code{struct sym_fns} which contains the
2096 name (or name prefix) of the symbol format, the length of the prefix,
2097 and pointers to four functions. These functions are called at various
2098 times to process symbol files whose identification matches the specified
2101 The functions supplied by each module are:
2104 @item @var{xyz}_symfile_init(struct sym_fns *sf)
2106 @cindex secondary symbol file
2107 Called from @code{symbol_file_add} when we are about to read a new
2108 symbol file. This function should clean up any internal state (possibly
2109 resulting from half-read previous files, for example) and prepare to
2110 read a new symbol file. Note that the symbol file which we are reading
2111 might be a new ``main'' symbol file, or might be a secondary symbol file
2112 whose symbols are being added to the existing symbol table.
2114 The argument to @code{@var{xyz}_symfile_init} is a newly allocated
2115 @code{struct sym_fns} whose @code{bfd} field contains the BFD for the
2116 new symbol file being read. Its @code{private} field has been zeroed,
2117 and can be modified as desired. Typically, a struct of private
2118 information will be @code{malloc}'d, and a pointer to it will be placed
2119 in the @code{private} field.
2121 There is no result from @code{@var{xyz}_symfile_init}, but it can call
2122 @code{error} if it detects an unavoidable problem.
2124 @item @var{xyz}_new_init()
2126 Called from @code{symbol_file_add} when discarding existing symbols.
2127 This function needs only handle the symbol-reading module's internal
2128 state; the symbol table data structures visible to the rest of
2129 @value{GDBN} will be discarded by @code{symbol_file_add}. It has no
2130 arguments and no result. It may be called after
2131 @code{@var{xyz}_symfile_init}, if a new symbol table is being read, or
2132 may be called alone if all symbols are simply being discarded.
2134 @item @var{xyz}_symfile_read(struct sym_fns *sf, CORE_ADDR addr, int mainline)
2136 Called from @code{symbol_file_add} to actually read the symbols from a
2137 symbol-file into a set of psymtabs or symtabs.
2139 @code{sf} points to the @code{struct sym_fns} originally passed to
2140 @code{@var{xyz}_sym_init} for possible initialization. @code{addr} is
2141 the offset between the file's specified start address and its true
2142 address in memory. @code{mainline} is 1 if this is the main symbol
2143 table being read, and 0 if a secondary symbol file (e.g., shared library
2144 or dynamically loaded file) is being read.@refill
2147 In addition, if a symbol-reading module creates psymtabs when
2148 @var{xyz}_symfile_read is called, these psymtabs will contain a pointer
2149 to a function @code{@var{xyz}_psymtab_to_symtab}, which can be called
2150 from any point in the @value{GDBN} symbol-handling code.
2153 @item @var{xyz}_psymtab_to_symtab (struct partial_symtab *pst)
2155 Called from @code{psymtab_to_symtab} (or the @code{PSYMTAB_TO_SYMTAB} macro) if
2156 the psymtab has not already been read in and had its @code{pst->symtab}
2157 pointer set. The argument is the psymtab to be fleshed-out into a
2158 symtab. Upon return, @code{pst->readin} should have been set to 1, and
2159 @code{pst->symtab} should contain a pointer to the new corresponding symtab, or
2160 zero if there were no symbols in that part of the symbol file.
2163 @section Partial Symbol Tables
2165 @value{GDBN} has three types of symbol tables:
2168 @cindex full symbol table
2171 Full symbol tables (@dfn{symtabs}). These contain the main
2172 information about symbols and addresses.
2176 Partial symbol tables (@dfn{psymtabs}). These contain enough
2177 information to know when to read the corresponding part of the full
2180 @cindex minimal symbol table
2183 Minimal symbol tables (@dfn{msymtabs}). These contain information
2184 gleaned from non-debugging symbols.
2187 @cindex partial symbol table
2188 This section describes partial symbol tables.
2190 A psymtab is constructed by doing a very quick pass over an executable
2191 file's debugging information. Small amounts of information are
2192 extracted---enough to identify which parts of the symbol table will
2193 need to be re-read and fully digested later, when the user needs the
2194 information. The speed of this pass causes @value{GDBN} to start up very
2195 quickly. Later, as the detailed rereading occurs, it occurs in small
2196 pieces, at various times, and the delay therefrom is mostly invisible to
2198 @c (@xref{Symbol Reading}.)
2200 The symbols that show up in a file's psymtab should be, roughly, those
2201 visible to the debugger's user when the program is not running code from
2202 that file. These include external symbols and types, static symbols and
2203 types, and @code{enum} values declared at file scope.
2205 The psymtab also contains the range of instruction addresses that the
2206 full symbol table would represent.
2208 @cindex finding a symbol
2209 @cindex symbol lookup
2210 The idea is that there are only two ways for the user (or much of the
2211 code in the debugger) to reference a symbol:
2214 @findex find_pc_function
2215 @findex find_pc_line
2217 By its address (e.g., execution stops at some address which is inside a
2218 function in this file). The address will be noticed to be in the
2219 range of this psymtab, and the full symtab will be read in.
2220 @code{find_pc_function}, @code{find_pc_line}, and other
2221 @code{find_pc_@dots{}} functions handle this.
2223 @cindex lookup_symbol
2226 (e.g., the user asks to print a variable, or set a breakpoint on a
2227 function). Global names and file-scope names will be found in the
2228 psymtab, which will cause the symtab to be pulled in. Local names will
2229 have to be qualified by a global name, or a file-scope name, in which
2230 case we will have already read in the symtab as we evaluated the
2231 qualifier. Or, a local symbol can be referenced when we are ``in'' a
2232 local scope, in which case the first case applies. @code{lookup_symbol}
2233 does most of the work here.
2236 The only reason that psymtabs exist is to cause a symtab to be read in
2237 at the right moment. Any symbol that can be elided from a psymtab,
2238 while still causing that to happen, should not appear in it. Since
2239 psymtabs don't have the idea of scope, you can't put local symbols in
2240 them anyway. Psymtabs don't have the idea of the type of a symbol,
2241 either, so types need not appear, unless they will be referenced by
2244 It is a bug for @value{GDBN} to behave one way when only a psymtab has
2245 been read, and another way if the corresponding symtab has been read
2246 in. Such bugs are typically caused by a psymtab that does not contain
2247 all the visible symbols, or which has the wrong instruction address
2250 The psymtab for a particular section of a symbol file (objfile) could be
2251 thrown away after the symtab has been read in. The symtab should always
2252 be searched before the psymtab, so the psymtab will never be used (in a
2253 bug-free environment). Currently, psymtabs are allocated on an obstack,
2254 and all the psymbols themselves are allocated in a pair of large arrays
2255 on an obstack, so there is little to be gained by trying to free them
2256 unless you want to do a lot more work.
2260 @unnumberedsubsec Fundamental Types (e.g., @code{FT_VOID}, @code{FT_BOOLEAN}).
2262 @cindex fundamental types
2263 These are the fundamental types that @value{GDBN} uses internally. Fundamental
2264 types from the various debugging formats (stabs, ELF, etc) are mapped
2265 into one of these. They are basically a union of all fundamental types
2266 that @value{GDBN} knows about for all the languages that @value{GDBN}
2269 @unnumberedsubsec Type Codes (e.g., @code{TYPE_CODE_PTR}, @code{TYPE_CODE_ARRAY}).
2272 Each time @value{GDBN} builds an internal type, it marks it with one
2273 of these types. The type may be a fundamental type, such as
2274 @code{TYPE_CODE_INT}, or a derived type, such as @code{TYPE_CODE_PTR}
2275 which is a pointer to another type. Typically, several @code{FT_*}
2276 types map to one @code{TYPE_CODE_*} type, and are distinguished by
2277 other members of the type struct, such as whether the type is signed
2278 or unsigned, and how many bits it uses.
2280 @unnumberedsubsec Builtin Types (e.g., @code{builtin_type_void}, @code{builtin_type_char}).
2282 These are instances of type structs that roughly correspond to
2283 fundamental types and are created as global types for @value{GDBN} to
2284 use for various ugly historical reasons. We eventually want to
2285 eliminate these. Note for example that @code{builtin_type_int}
2286 initialized in @file{gdbtypes.c} is basically the same as a
2287 @code{TYPE_CODE_INT} type that is initialized in @file{c-lang.c} for
2288 an @code{FT_INTEGER} fundamental type. The difference is that the
2289 @code{builtin_type} is not associated with any particular objfile, and
2290 only one instance exists, while @file{c-lang.c} builds as many
2291 @code{TYPE_CODE_INT} types as needed, with each one associated with
2292 some particular objfile.
2294 @section Object File Formats
2295 @cindex object file formats
2299 @cindex @code{a.out} format
2300 The @code{a.out} format is the original file format for Unix. It
2301 consists of three sections: @code{text}, @code{data}, and @code{bss},
2302 which are for program code, initialized data, and uninitialized data,
2305 The @code{a.out} format is so simple that it doesn't have any reserved
2306 place for debugging information. (Hey, the original Unix hackers used
2307 @samp{adb}, which is a machine-language debugger!) The only debugging
2308 format for @code{a.out} is stabs, which is encoded as a set of normal
2309 symbols with distinctive attributes.
2311 The basic @code{a.out} reader is in @file{dbxread.c}.
2316 The COFF format was introduced with System V Release 3 (SVR3) Unix.
2317 COFF files may have multiple sections, each prefixed by a header. The
2318 number of sections is limited.
2320 The COFF specification includes support for debugging. Although this
2321 was a step forward, the debugging information was woefully limited.
2322 For instance, it was not possible to represent code that came from an
2323 included file. GNU's COFF-using configs often use stabs-type info,
2324 encapsulated in special sections.
2326 The COFF reader is in @file{coffread.c}.
2330 @cindex ECOFF format
2331 ECOFF is an extended COFF originally introduced for Mips and Alpha
2334 The basic ECOFF reader is in @file{mipsread.c}.
2338 @cindex XCOFF format
2339 The IBM RS/6000 running AIX uses an object file format called XCOFF.
2340 The COFF sections, symbols, and line numbers are used, but debugging
2341 symbols are @code{dbx}-style stabs whose strings are located in the
2342 @code{.debug} section (rather than the string table). For more
2343 information, see @ref{Top,,,stabs,The Stabs Debugging Format}.
2345 The shared library scheme has a clean interface for figuring out what
2346 shared libraries are in use, but the catch is that everything which
2347 refers to addresses (symbol tables and breakpoints at least) needs to be
2348 relocated for both shared libraries and the main executable. At least
2349 using the standard mechanism this can only be done once the program has
2350 been run (or the core file has been read).
2354 @cindex PE-COFF format
2355 Windows 95 and NT use the PE (@dfn{Portable Executable}) format for their
2356 executables. PE is basically COFF with additional headers.
2358 While BFD includes special PE support, @value{GDBN} needs only the basic
2364 The ELF format came with System V Release 4 (SVR4) Unix. ELF is
2365 similar to COFF in being organized into a number of sections, but it
2366 removes many of COFF's limitations. Debugging info may be either stabs
2367 encapsulated in ELF sections, or more commonly these days, DWARF.
2369 The basic ELF reader is in @file{elfread.c}.
2374 SOM is HP's object file and debug format (not to be confused with IBM's
2375 SOM, which is a cross-language ABI).
2377 The SOM reader is in @file{somread.c}.
2379 @section Debugging File Formats
2381 This section describes characteristics of debugging information that
2382 are independent of the object file format.
2386 @cindex stabs debugging info
2387 @code{stabs} started out as special symbols within the @code{a.out}
2388 format. Since then, it has been encapsulated into other file
2389 formats, such as COFF and ELF.
2391 While @file{dbxread.c} does some of the basic stab processing,
2392 including for encapsulated versions, @file{stabsread.c} does
2397 @cindex COFF debugging info
2398 The basic COFF definition includes debugging information. The level
2399 of support is minimal and non-extensible, and is not often used.
2401 @subsection Mips debug (Third Eye)
2403 @cindex ECOFF debugging info
2404 ECOFF includes a definition of a special debug format.
2406 The file @file{mdebugread.c} implements reading for this format.
2408 @c mention DWARF 1 as a formerly-supported format
2412 @cindex DWARF 2 debugging info
2413 DWARF 2 is an improved but incompatible version of DWARF 1.
2415 The DWARF 2 reader is in @file{dwarf2read.c}.
2417 @subsection Compressed DWARF 2
2419 @cindex Compressed DWARF 2 debugging info
2420 Compressed DWARF 2 is not technically a separate debugging format, but
2421 merely DWARF 2 debug information that has been compressed. In this
2422 format, every object-file section holding DWARF 2 debugging
2423 information is compressed and prepended with a header. (The section
2424 is also typically renamed, so a section called @code{.debug_info} in a
2425 DWARF 2 binary would be called @code{.zdebug_info} in a compressed
2426 DWARF 2 binary.) The header is 12 bytes long:
2430 4 bytes: the literal string ``ZLIB''
2432 8 bytes: the uncompressed size of the section, in big-endian byte
2436 The same reader is used for both compressed an normal DWARF 2 info.
2437 Section decompression is done in @code{zlib_decompress_section} in
2438 @file{dwarf2read.c}.
2442 @cindex DWARF 3 debugging info
2443 DWARF 3 is an improved version of DWARF 2.
2447 @cindex SOM debugging info
2448 Like COFF, the SOM definition includes debugging information.
2450 @section Adding a New Symbol Reader to @value{GDBN}
2452 @cindex adding debugging info reader
2453 If you are using an existing object file format (@code{a.out}, COFF, ELF, etc),
2454 there is probably little to be done.
2456 If you need to add a new object file format, you must first add it to
2457 BFD. This is beyond the scope of this document.
2459 You must then arrange for the BFD code to provide access to the
2460 debugging symbols. Generally @value{GDBN} will have to call swapping
2461 routines from BFD and a few other BFD internal routines to locate the
2462 debugging information. As much as possible, @value{GDBN} should not
2463 depend on the BFD internal data structures.
2465 For some targets (e.g., COFF), there is a special transfer vector used
2466 to call swapping routines, since the external data structures on various
2467 platforms have different sizes and layouts. Specialized routines that
2468 will only ever be implemented by one object file format may be called
2469 directly. This interface should be described in a file
2470 @file{bfd/lib@var{xyz}.h}, which is included by @value{GDBN}.
2472 @section Memory Management for Symbol Files
2474 Most memory associated with a loaded symbol file is stored on
2475 its @code{objfile_obstack}. This includes symbols, types,
2476 namespace data, and other information produced by the symbol readers.
2478 Because this data lives on the objfile's obstack, it is automatically
2479 released when the objfile is unloaded or reloaded. Therefore one
2480 objfile must not reference symbol or type data from another objfile;
2481 they could be unloaded at different times.
2483 User convenience variables, et cetera, have associated types. Normally
2484 these types live in the associated objfile. However, when the objfile
2485 is unloaded, those types are deep copied to global memory, so that
2486 the values of the user variables and history items are not lost.
2489 @node Language Support
2491 @chapter Language Support
2493 @cindex language support
2494 @value{GDBN}'s language support is mainly driven by the symbol reader,
2495 although it is possible for the user to set the source language
2498 @value{GDBN} chooses the source language by looking at the extension
2499 of the file recorded in the debug info; @file{.c} means C, @file{.f}
2500 means Fortran, etc. It may also use a special-purpose language
2501 identifier if the debug format supports it, like with DWARF.
2503 @section Adding a Source Language to @value{GDBN}
2505 @cindex adding source language
2506 To add other languages to @value{GDBN}'s expression parser, follow the
2510 @item Create the expression parser.
2512 @cindex expression parser
2513 This should reside in a file @file{@var{lang}-exp.y}. Routines for
2514 building parsed expressions into a @code{union exp_element} list are in
2517 @cindex language parser
2518 Since we can't depend upon everyone having Bison, and YACC produces
2519 parsers that define a bunch of global names, the following lines
2520 @strong{must} be included at the top of the YACC parser, to prevent the
2521 various parsers from defining the same global names:
2524 #define yyparse @var{lang}_parse
2525 #define yylex @var{lang}_lex
2526 #define yyerror @var{lang}_error
2527 #define yylval @var{lang}_lval
2528 #define yychar @var{lang}_char
2529 #define yydebug @var{lang}_debug
2530 #define yypact @var{lang}_pact
2531 #define yyr1 @var{lang}_r1
2532 #define yyr2 @var{lang}_r2
2533 #define yydef @var{lang}_def
2534 #define yychk @var{lang}_chk
2535 #define yypgo @var{lang}_pgo
2536 #define yyact @var{lang}_act
2537 #define yyexca @var{lang}_exca
2538 #define yyerrflag @var{lang}_errflag
2539 #define yynerrs @var{lang}_nerrs
2542 At the bottom of your parser, define a @code{struct language_defn} and
2543 initialize it with the right values for your language. Define an
2544 @code{initialize_@var{lang}} routine and have it call
2545 @samp{add_language(@var{lang}_language_defn)} to tell the rest of @value{GDBN}
2546 that your language exists. You'll need some other supporting variables
2547 and functions, which will be used via pointers from your
2548 @code{@var{lang}_language_defn}. See the declaration of @code{struct
2549 language_defn} in @file{language.h}, and the other @file{*-exp.y} files,
2550 for more information.
2552 @item Add any evaluation routines, if necessary
2554 @cindex expression evaluation routines
2555 @findex evaluate_subexp
2556 @findex prefixify_subexp
2557 @findex length_of_subexp
2558 If you need new opcodes (that represent the operations of the language),
2559 add them to the enumerated type in @file{expression.h}. Add support
2560 code for these operations in the @code{evaluate_subexp} function
2561 defined in the file @file{eval.c}. Add cases
2562 for new opcodes in two functions from @file{parse.c}:
2563 @code{prefixify_subexp} and @code{length_of_subexp}. These compute
2564 the number of @code{exp_element}s that a given operation takes up.
2566 @item Update some existing code
2568 Add an enumerated identifier for your language to the enumerated type
2569 @code{enum language} in @file{defs.h}.
2571 Update the routines in @file{language.c} so your language is included.
2572 These routines include type predicates and such, which (in some cases)
2573 are language dependent. If your language does not appear in the switch
2574 statement, an error is reported.
2576 @vindex current_language
2577 Also included in @file{language.c} is the code that updates the variable
2578 @code{current_language}, and the routines that translate the
2579 @code{language_@var{lang}} enumerated identifier into a printable
2582 @findex _initialize_language
2583 Update the function @code{_initialize_language} to include your
2584 language. This function picks the default language upon startup, so is
2585 dependent upon which languages that @value{GDBN} is built for.
2587 @findex allocate_symtab
2588 Update @code{allocate_symtab} in @file{symfile.c} and/or symbol-reading
2589 code so that the language of each symtab (source file) is set properly.
2590 This is used to determine the language to use at each stack frame level.
2591 Currently, the language is set based upon the extension of the source
2592 file. If the language can be better inferred from the symbol
2593 information, please set the language of the symtab in the symbol-reading
2596 @findex print_subexp
2597 @findex op_print_tab
2598 Add helper code to @code{print_subexp} (in @file{expprint.c}) to handle any new
2599 expression opcodes you have added to @file{expression.h}. Also, add the
2600 printed representations of your operators to @code{op_print_tab}.
2602 @item Add a place of call
2605 Add a call to @code{@var{lang}_parse()} and @code{@var{lang}_error} in
2606 @code{parse_exp_1} (defined in @file{parse.c}).
2608 @item Edit @file{Makefile.in}
2610 Add dependencies in @file{Makefile.in}. Make sure you update the macro
2611 variables such as @code{HFILES} and @code{OBJS}, otherwise your code may
2612 not get linked in, or, worse yet, it may not get @code{tar}red into the
2617 @node Host Definition
2619 @chapter Host Definition
2621 With the advent of Autoconf, it's rarely necessary to have host
2622 definition machinery anymore. The following information is provided,
2623 mainly, as an historical reference.
2625 @section Adding a New Host
2627 @cindex adding a new host
2628 @cindex host, adding
2629 @value{GDBN}'s host configuration support normally happens via Autoconf.
2630 New host-specific definitions should not be needed. Older hosts
2631 @value{GDBN} still use the host-specific definitions and files listed
2632 below, but these mostly exist for historical reasons, and will
2633 eventually disappear.
2636 @item gdb/config/@var{arch}/@var{xyz}.mh
2637 This file is a Makefile fragment that once contained both host and
2638 native configuration information (@pxref{Native Debugging}) for the
2639 machine @var{xyz}. The host configuration information is now handled
2642 Host configuration information included definitions for @code{CC},
2643 @code{SYSV_DEFINE}, @code{XM_CFLAGS}, @code{XM_ADD_FILES},
2644 @code{XM_CLIBS}, @code{XM_CDEPS}, etc.; see @file{Makefile.in}.
2646 New host-only configurations do not need this file.
2650 (Files named @file{gdb/config/@var{arch}/xm-@var{xyz}.h} were once
2651 used to define host-specific macros, but were no longer needed and
2652 have all been removed.)
2654 @subheading Generic Host Support Files
2656 @cindex generic host support
2657 There are some ``generic'' versions of routines that can be used by
2661 @cindex remote debugging support
2662 @cindex serial line support
2664 This contains serial line support for Unix systems. It is included by
2665 default on all Unix-like hosts.
2668 This contains serial pipe support for Unix systems. It is included by
2669 default on all Unix-like hosts.
2672 This contains serial line support for 32-bit programs running under
2673 Windows using MinGW.
2676 This contains serial line support for 32-bit programs running under DOS,
2677 using the DJGPP (a.k.a.@: GO32) execution environment.
2679 @cindex TCP remote support
2681 This contains generic TCP support using sockets. It is included by
2682 default on all Unix-like hosts and with MinGW.
2685 @section Host Conditionals
2687 When @value{GDBN} is configured and compiled, various macros are
2688 defined or left undefined, to control compilation based on the
2689 attributes of the host system. While formerly they could be set in
2690 host-specific header files, at present they can be changed only by
2691 setting @code{CFLAGS} when building, or by editing the source code.
2693 These macros and their meanings (or if the meaning is not documented
2694 here, then one of the source files where they are used is indicated)
2698 @item @value{GDBN}INIT_FILENAME
2699 The default name of @value{GDBN}'s initialization file (normally
2702 @item SIGWINCH_HANDLER
2703 If your host defines @code{SIGWINCH}, you can define this to be the name
2704 of a function to be called if @code{SIGWINCH} is received.
2706 @item SIGWINCH_HANDLER_BODY
2707 Define this to expand into code that will define the function named by
2708 the expansion of @code{SIGWINCH_HANDLER}.
2710 @item CRLF_SOURCE_FILES
2711 @cindex DOS text files
2712 Define this if host files use @code{\r\n} rather than @code{\n} as a
2713 line terminator. This will cause source file listings to omit @code{\r}
2714 characters when printing and it will allow @code{\r\n} line endings of files
2715 which are ``sourced'' by gdb. It must be possible to open files in binary
2716 mode using @code{O_BINARY} or, for fopen, @code{"rb"}.
2718 @item DEFAULT_PROMPT
2720 The default value of the prompt string (normally @code{"(gdb) "}).
2723 @cindex terminal device
2724 The name of the generic TTY device, defaults to @code{"/dev/tty"}.
2727 Substitute for isatty, if not available.
2730 Define this if binary files are opened the same way as text files.
2732 @item CC_HAS_LONG_LONG
2733 @cindex @code{long long} data type
2734 Define this if the host C compiler supports @code{long long}. This is set
2735 by the @code{configure} script.
2737 @item PRINTF_HAS_LONG_LONG
2738 Define this if the host can handle printing of long long integers via
2739 the printf format conversion specifier @code{ll}. This is set by the
2740 @code{configure} script.
2742 @item LSEEK_NOT_LINEAR
2743 Define this if @code{lseek (n)} does not necessarily move to byte number
2744 @code{n} in the file. This is only used when reading source files. It
2745 is normally faster to define @code{CRLF_SOURCE_FILES} when possible.
2748 If defined, this should be one or more tokens, such as @code{volatile},
2749 that can be used in both the declaration and definition of functions to
2750 indicate that they never return. The default is already set correctly
2751 if compiling with GCC. This will almost never need to be defined.
2754 If defined, this should be one or more tokens, such as
2755 @code{__attribute__ ((noreturn))}, that can be used in the declarations
2756 of functions to indicate that they never return. The default is already
2757 set correctly if compiling with GCC. This will almost never need to be
2761 Define this to help placate @code{lint} in some situations.
2764 Define this to override the defaults of @code{__volatile__} or
2769 @node Target Architecture Definition
2771 @chapter Target Architecture Definition
2773 @cindex target architecture definition
2774 @value{GDBN}'s target architecture defines what sort of
2775 machine-language programs @value{GDBN} can work with, and how it works
2778 The target architecture object is implemented as the C structure
2779 @code{struct gdbarch *}. The structure, and its methods, are generated
2780 using the Bourne shell script @file{gdbarch.sh}.
2783 * OS ABI Variant Handling::
2784 * Initialize New Architecture::
2785 * Registers and Memory::
2786 * Pointers and Addresses::
2788 * Raw and Virtual Registers::
2789 * Register and Memory Data::
2790 * Frame Interpretation::
2791 * Inferior Call Setup::
2792 * Compiler Characteristics::
2793 * Target Conditionals::
2794 * Adding a New Target::
2797 @node OS ABI Variant Handling
2798 @section Operating System ABI Variant Handling
2799 @cindex OS ABI variants
2801 @value{GDBN} provides a mechanism for handling variations in OS
2802 ABIs. An OS ABI variant may have influence over any number of
2803 variables in the target architecture definition. There are two major
2804 components in the OS ABI mechanism: sniffers and handlers.
2806 A @dfn{sniffer} examines a file matching a BFD architecture/flavour pair
2807 (the architecture may be wildcarded) in an attempt to determine the
2808 OS ABI of that file. Sniffers with a wildcarded architecture are considered
2809 to be @dfn{generic}, while sniffers for a specific architecture are
2810 considered to be @dfn{specific}. A match from a specific sniffer
2811 overrides a match from a generic sniffer. Multiple sniffers for an
2812 architecture/flavour may exist, in order to differentiate between two
2813 different operating systems which use the same basic file format. The
2814 OS ABI framework provides a generic sniffer for ELF-format files which
2815 examines the @code{EI_OSABI} field of the ELF header, as well as note
2816 sections known to be used by several operating systems.
2818 @cindex fine-tuning @code{gdbarch} structure
2819 A @dfn{handler} is used to fine-tune the @code{gdbarch} structure for the
2820 selected OS ABI. There may be only one handler for a given OS ABI
2821 for each BFD architecture.
2823 The following OS ABI variants are defined in @file{defs.h}:
2827 @findex GDB_OSABI_UNINITIALIZED
2828 @item GDB_OSABI_UNINITIALIZED
2829 Used for struct gdbarch_info if ABI is still uninitialized.
2831 @findex GDB_OSABI_UNKNOWN
2832 @item GDB_OSABI_UNKNOWN
2833 The ABI of the inferior is unknown. The default @code{gdbarch}
2834 settings for the architecture will be used.
2836 @findex GDB_OSABI_SVR4
2837 @item GDB_OSABI_SVR4
2838 UNIX System V Release 4.
2840 @findex GDB_OSABI_HURD
2841 @item GDB_OSABI_HURD
2842 GNU using the Hurd kernel.
2844 @findex GDB_OSABI_SOLARIS
2845 @item GDB_OSABI_SOLARIS
2848 @findex GDB_OSABI_OSF1
2849 @item GDB_OSABI_OSF1
2850 OSF/1, including Digital UNIX and Compaq Tru64 UNIX.
2852 @findex GDB_OSABI_LINUX
2853 @item GDB_OSABI_LINUX
2854 GNU using the Linux kernel.
2856 @findex GDB_OSABI_FREEBSD_AOUT
2857 @item GDB_OSABI_FREEBSD_AOUT
2858 FreeBSD using the @code{a.out} executable format.
2860 @findex GDB_OSABI_FREEBSD_ELF
2861 @item GDB_OSABI_FREEBSD_ELF
2862 FreeBSD using the ELF executable format.
2864 @findex GDB_OSABI_NETBSD_AOUT
2865 @item GDB_OSABI_NETBSD_AOUT
2866 NetBSD using the @code{a.out} executable format.
2868 @findex GDB_OSABI_NETBSD_ELF
2869 @item GDB_OSABI_NETBSD_ELF
2870 NetBSD using the ELF executable format.
2872 @findex GDB_OSABI_OPENBSD_ELF
2873 @item GDB_OSABI_OPENBSD_ELF
2874 OpenBSD using the ELF executable format.
2876 @findex GDB_OSABI_WINCE
2877 @item GDB_OSABI_WINCE
2880 @findex GDB_OSABI_GO32
2881 @item GDB_OSABI_GO32
2884 @findex GDB_OSABI_IRIX
2885 @item GDB_OSABI_IRIX
2888 @findex GDB_OSABI_INTERIX
2889 @item GDB_OSABI_INTERIX
2890 Interix (Posix layer for MS-Windows systems).
2892 @findex GDB_OSABI_HPUX_ELF
2893 @item GDB_OSABI_HPUX_ELF
2894 HP/UX using the ELF executable format.
2896 @findex GDB_OSABI_HPUX_SOM
2897 @item GDB_OSABI_HPUX_SOM
2898 HP/UX using the SOM executable format.
2900 @findex GDB_OSABI_QNXNTO
2901 @item GDB_OSABI_QNXNTO
2904 @findex GDB_OSABI_CYGWIN
2905 @item GDB_OSABI_CYGWIN
2908 @findex GDB_OSABI_AIX
2914 Here are the functions that make up the OS ABI framework:
2916 @deftypefun const char *gdbarch_osabi_name (enum gdb_osabi @var{osabi})
2917 Return the name of the OS ABI corresponding to @var{osabi}.
2920 @deftypefun void gdbarch_register_osabi (enum bfd_architecture @var{arch}, unsigned long @var{machine}, enum gdb_osabi @var{osabi}, void (*@var{init_osabi})(struct gdbarch_info @var{info}, struct gdbarch *@var{gdbarch}))
2921 Register the OS ABI handler specified by @var{init_osabi} for the
2922 architecture, machine type and OS ABI specified by @var{arch},
2923 @var{machine} and @var{osabi}. In most cases, a value of zero for the
2924 machine type, which implies the architecture's default machine type,
2928 @deftypefun void gdbarch_register_osabi_sniffer (enum bfd_architecture @var{arch}, enum bfd_flavour @var{flavour}, enum gdb_osabi (*@var{sniffer})(bfd *@var{abfd}))
2929 Register the OS ABI file sniffer specified by @var{sniffer} for the
2930 BFD architecture/flavour pair specified by @var{arch} and @var{flavour}.
2931 If @var{arch} is @code{bfd_arch_unknown}, the sniffer is considered to
2932 be generic, and is allowed to examine @var{flavour}-flavoured files for
2936 @deftypefun enum gdb_osabi gdbarch_lookup_osabi (bfd *@var{abfd})
2937 Examine the file described by @var{abfd} to determine its OS ABI.
2938 The value @code{GDB_OSABI_UNKNOWN} is returned if the OS ABI cannot
2942 @deftypefun void gdbarch_init_osabi (struct gdbarch info @var{info}, struct gdbarch *@var{gdbarch}, enum gdb_osabi @var{osabi})
2943 Invoke the OS ABI handler corresponding to @var{osabi} to fine-tune the
2944 @code{gdbarch} structure specified by @var{gdbarch}. If a handler
2945 corresponding to @var{osabi} has not been registered for @var{gdbarch}'s
2946 architecture, a warning will be issued and the debugging session will continue
2947 with the defaults already established for @var{gdbarch}.
2950 @deftypefun void generic_elf_osabi_sniff_abi_tag_sections (bfd *@var{abfd}, asection *@var{sect}, void *@var{obj})
2951 Helper routine for ELF file sniffers. Examine the file described by
2952 @var{abfd} and look at ABI tag note sections to determine the OS ABI
2953 from the note. This function should be called via
2954 @code{bfd_map_over_sections}.
2957 @node Initialize New Architecture
2958 @section Initializing a New Architecture
2960 Each @code{gdbarch} is associated with a single @sc{bfd} architecture,
2961 via a @code{bfd_arch_@var{arch}} constant. The @code{gdbarch} is
2962 registered by a call to @code{register_gdbarch_init}, usually from
2963 the file's @code{_initialize_@var{filename}} routine, which will
2964 be automatically called during @value{GDBN} startup. The arguments
2965 are a @sc{bfd} architecture constant and an initialization function.
2967 The initialization function has this type:
2970 static struct gdbarch *
2971 @var{arch}_gdbarch_init (struct gdbarch_info @var{info},
2972 struct gdbarch_list *@var{arches})
2975 The @var{info} argument contains parameters used to select the correct
2976 architecture, and @var{arches} is a list of architectures which
2977 have already been created with the same @code{bfd_arch_@var{arch}}
2980 The initialization function should first make sure that @var{info}
2981 is acceptable, and return @code{NULL} if it is not. Then, it should
2982 search through @var{arches} for an exact match to @var{info}, and
2983 return one if found. Lastly, if no exact match was found, it should
2984 create a new architecture based on @var{info} and return it.
2986 Only information in @var{info} should be used to choose the new
2987 architecture. Historically, @var{info} could be sparse, and
2988 defaults would be collected from the first element on @var{arches}.
2989 However, @value{GDBN} now fills in @var{info} more thoroughly,
2990 so new @code{gdbarch} initialization functions should not take
2991 defaults from @var{arches}.
2993 @node Registers and Memory
2994 @section Registers and Memory
2996 @value{GDBN}'s model of the target machine is rather simple.
2997 @value{GDBN} assumes the machine includes a bank of registers and a
2998 block of memory. Each register may have a different size.
3000 @value{GDBN} does not have a magical way to match up with the
3001 compiler's idea of which registers are which; however, it is critical
3002 that they do match up accurately. The only way to make this work is
3003 to get accurate information about the order that the compiler uses,
3004 and to reflect that in the @code{gdbarch_register_name} and related functions.
3006 @value{GDBN} can handle big-endian, little-endian, and bi-endian architectures.
3008 @node Pointers and Addresses
3009 @section Pointers Are Not Always Addresses
3010 @cindex pointer representation
3011 @cindex address representation
3012 @cindex word-addressed machines
3013 @cindex separate data and code address spaces
3014 @cindex spaces, separate data and code address
3015 @cindex address spaces, separate data and code
3016 @cindex code pointers, word-addressed
3017 @cindex converting between pointers and addresses
3018 @cindex D10V addresses
3020 On almost all 32-bit architectures, the representation of a pointer is
3021 indistinguishable from the representation of some fixed-length number
3022 whose value is the byte address of the object pointed to. On such
3023 machines, the words ``pointer'' and ``address'' can be used interchangeably.
3024 However, architectures with smaller word sizes are often cramped for
3025 address space, so they may choose a pointer representation that breaks this
3026 identity, and allows a larger code address space.
3028 @c D10V is gone from sources - more current example?
3030 For example, the Renesas D10V is a 16-bit VLIW processor whose
3031 instructions are 32 bits long@footnote{Some D10V instructions are
3032 actually pairs of 16-bit sub-instructions. However, since you can't
3033 jump into the middle of such a pair, code addresses can only refer to
3034 full 32 bit instructions, which is what matters in this explanation.}.
3035 If the D10V used ordinary byte addresses to refer to code locations,
3036 then the processor would only be able to address 64kb of instructions.
3037 However, since instructions must be aligned on four-byte boundaries, the
3038 low two bits of any valid instruction's byte address are always
3039 zero---byte addresses waste two bits. So instead of byte addresses,
3040 the D10V uses word addresses---byte addresses shifted right two bits---to
3041 refer to code. Thus, the D10V can use 16-bit words to address 256kb of
3044 However, this means that code pointers and data pointers have different
3045 forms on the D10V. The 16-bit word @code{0xC020} refers to byte address
3046 @code{0xC020} when used as a data address, but refers to byte address
3047 @code{0x30080} when used as a code address.
3049 (The D10V also uses separate code and data address spaces, which also
3050 affects the correspondence between pointers and addresses, but we're
3051 going to ignore that here; this example is already too long.)
3053 To cope with architectures like this---the D10V is not the only
3054 one!---@value{GDBN} tries to distinguish between @dfn{addresses}, which are
3055 byte numbers, and @dfn{pointers}, which are the target's representation
3056 of an address of a particular type of data. In the example above,
3057 @code{0xC020} is the pointer, which refers to one of the addresses
3058 @code{0xC020} or @code{0x30080}, depending on the type imposed upon it.
3059 @value{GDBN} provides functions for turning a pointer into an address
3060 and vice versa, in the appropriate way for the current architecture.
3062 Unfortunately, since addresses and pointers are identical on almost all
3063 processors, this distinction tends to bit-rot pretty quickly. Thus,
3064 each time you port @value{GDBN} to an architecture which does
3065 distinguish between pointers and addresses, you'll probably need to
3066 clean up some architecture-independent code.
3068 Here are functions which convert between pointers and addresses:
3070 @deftypefun CORE_ADDR extract_typed_address (void *@var{buf}, struct type *@var{type})
3071 Treat the bytes at @var{buf} as a pointer or reference of type
3072 @var{type}, and return the address it represents, in a manner
3073 appropriate for the current architecture. This yields an address
3074 @value{GDBN} can use to read target memory, disassemble, etc. Note that
3075 @var{buf} refers to a buffer in @value{GDBN}'s memory, not the
3078 For example, if the current architecture is the Intel x86, this function
3079 extracts a little-endian integer of the appropriate length from
3080 @var{buf} and returns it. However, if the current architecture is the
3081 D10V, this function will return a 16-bit integer extracted from
3082 @var{buf}, multiplied by four if @var{type} is a pointer to a function.
3084 If @var{type} is not a pointer or reference type, then this function
3085 will signal an internal error.
3088 @deftypefun CORE_ADDR store_typed_address (void *@var{buf}, struct type *@var{type}, CORE_ADDR @var{addr})
3089 Store the address @var{addr} in @var{buf}, in the proper format for a
3090 pointer of type @var{type} in the current architecture. Note that
3091 @var{buf} refers to a buffer in @value{GDBN}'s memory, not the
3094 For example, if the current architecture is the Intel x86, this function
3095 stores @var{addr} unmodified as a little-endian integer of the
3096 appropriate length in @var{buf}. However, if the current architecture
3097 is the D10V, this function divides @var{addr} by four if @var{type} is
3098 a pointer to a function, and then stores it in @var{buf}.
3100 If @var{type} is not a pointer or reference type, then this function
3101 will signal an internal error.
3104 @deftypefun CORE_ADDR value_as_address (struct value *@var{val})
3105 Assuming that @var{val} is a pointer, return the address it represents,
3106 as appropriate for the current architecture.
3108 This function actually works on integral values, as well as pointers.
3109 For pointers, it performs architecture-specific conversions as
3110 described above for @code{extract_typed_address}.
3113 @deftypefun CORE_ADDR value_from_pointer (struct type *@var{type}, CORE_ADDR @var{addr})
3114 Create and return a value representing a pointer of type @var{type} to
3115 the address @var{addr}, as appropriate for the current architecture.
3116 This function performs architecture-specific conversions as described
3117 above for @code{store_typed_address}.
3120 Here are two functions which architectures can define to indicate the
3121 relationship between pointers and addresses. These have default
3122 definitions, appropriate for architectures on which all pointers are
3123 simple unsigned byte addresses.
3125 @deftypefun CORE_ADDR gdbarch_pointer_to_address (struct gdbarch *@var{current_gdbarch}, struct type *@var{type}, char *@var{buf})
3126 Assume that @var{buf} holds a pointer of type @var{type}, in the
3127 appropriate format for the current architecture. Return the byte
3128 address the pointer refers to.
3130 This function may safely assume that @var{type} is either a pointer or a
3131 C@t{++} reference type.
3134 @deftypefun void gdbarch_address_to_pointer (struct gdbarch *@var{current_gdbarch}, struct type *@var{type}, char *@var{buf}, CORE_ADDR @var{addr})
3135 Store in @var{buf} a pointer of type @var{type} representing the address
3136 @var{addr}, in the appropriate format for the current architecture.
3138 This function may safely assume that @var{type} is either a pointer or a
3139 C@t{++} reference type.
3142 @node Address Classes
3143 @section Address Classes
3144 @cindex address classes
3145 @cindex DW_AT_byte_size
3146 @cindex DW_AT_address_class
3148 Sometimes information about different kinds of addresses is available
3149 via the debug information. For example, some programming environments
3150 define addresses of several different sizes. If the debug information
3151 distinguishes these kinds of address classes through either the size
3152 info (e.g, @code{DW_AT_byte_size} in @w{DWARF 2}) or through an explicit
3153 address class attribute (e.g, @code{DW_AT_address_class} in @w{DWARF 2}), the
3154 following macros should be defined in order to disambiguate these
3155 types within @value{GDBN} as well as provide the added information to
3156 a @value{GDBN} user when printing type expressions.
3158 @deftypefun int gdbarch_address_class_type_flags (struct gdbarch *@var{current_gdbarch}, int @var{byte_size}, int @var{dwarf2_addr_class})
3159 Returns the type flags needed to construct a pointer type whose size
3160 is @var{byte_size} and whose address class is @var{dwarf2_addr_class}.
3161 This function is normally called from within a symbol reader. See
3162 @file{dwarf2read.c}.
3165 @deftypefun char *gdbarch_address_class_type_flags_to_name (struct gdbarch *@var{current_gdbarch}, int @var{type_flags})
3166 Given the type flags representing an address class qualifier, return
3169 @deftypefun int gdbarch_address_class_name_to_type_flags (struct gdbarch *@var{current_gdbarch}, int @var{name}, int *@var{type_flags_ptr})
3170 Given an address qualifier name, set the @code{int} referenced by @var{type_flags_ptr} to the type flags
3171 for that address class qualifier.
3174 Since the need for address classes is rather rare, none of
3175 the address class functions are defined by default. Predicate
3176 functions are provided to detect when they are defined.
3178 Consider a hypothetical architecture in which addresses are normally
3179 32-bits wide, but 16-bit addresses are also supported. Furthermore,
3180 suppose that the @w{DWARF 2} information for this architecture simply
3181 uses a @code{DW_AT_byte_size} value of 2 to indicate the use of one
3182 of these "short" pointers. The following functions could be defined
3183 to implement the address class functions:
3186 somearch_address_class_type_flags (int byte_size,
3187 int dwarf2_addr_class)
3190 return TYPE_FLAG_ADDRESS_CLASS_1;
3196 somearch_address_class_type_flags_to_name (int type_flags)
3198 if (type_flags & TYPE_FLAG_ADDRESS_CLASS_1)
3205 somearch_address_class_name_to_type_flags (char *name,
3206 int *type_flags_ptr)
3208 if (strcmp (name, "short") == 0)
3210 *type_flags_ptr = TYPE_FLAG_ADDRESS_CLASS_1;
3218 The qualifier @code{@@short} is used in @value{GDBN}'s type expressions
3219 to indicate the presence of one of these "short" pointers. E.g, if
3220 the debug information indicates that @code{short_ptr_var} is one of these
3221 short pointers, @value{GDBN} might show the following behavior:
3224 (gdb) ptype short_ptr_var
3225 type = int * @@short
3229 @node Raw and Virtual Registers
3230 @section Raw and Virtual Register Representations
3231 @cindex raw register representation
3232 @cindex virtual register representation
3233 @cindex representations, raw and virtual registers
3235 @emph{Maintainer note: This section is pretty much obsolete. The
3236 functionality described here has largely been replaced by
3237 pseudo-registers and the mechanisms described in @ref{Register and
3238 Memory Data, , Using Different Register and Memory Data
3239 Representations}. See also @uref{http://www.gnu.org/software/gdb/bugs/,
3240 Bug Tracking Database} and
3241 @uref{http://sources.redhat.com/gdb/current/ari/, ARI Index} for more
3242 up-to-date information.}
3244 Some architectures use one representation for a value when it lives in a
3245 register, but use a different representation when it lives in memory.
3246 In @value{GDBN}'s terminology, the @dfn{raw} representation is the one used in
3247 the target registers, and the @dfn{virtual} representation is the one
3248 used in memory, and within @value{GDBN} @code{struct value} objects.
3250 @emph{Maintainer note: Notice that the same mechanism is being used to
3251 both convert a register to a @code{struct value} and alternative
3254 For almost all data types on almost all architectures, the virtual and
3255 raw representations are identical, and no special handling is needed.
3256 However, they do occasionally differ. For example:
3260 The x86 architecture supports an 80-bit @code{long double} type. However, when
3261 we store those values in memory, they occupy twelve bytes: the
3262 floating-point number occupies the first ten, and the final two bytes
3263 are unused. This keeps the values aligned on four-byte boundaries,
3264 allowing more efficient access. Thus, the x86 80-bit floating-point
3265 type is the raw representation, and the twelve-byte loosely-packed
3266 arrangement is the virtual representation.
3269 Some 64-bit MIPS targets present 32-bit registers to @value{GDBN} as 64-bit
3270 registers, with garbage in their upper bits. @value{GDBN} ignores the top 32
3271 bits. Thus, the 64-bit form, with garbage in the upper 32 bits, is the
3272 raw representation, and the trimmed 32-bit representation is the
3273 virtual representation.
3276 In general, the raw representation is determined by the architecture, or
3277 @value{GDBN}'s interface to the architecture, while the virtual representation
3278 can be chosen for @value{GDBN}'s convenience. @value{GDBN}'s register file,
3279 @code{registers}, holds the register contents in raw format, and the
3280 @value{GDBN} remote protocol transmits register values in raw format.
3282 Your architecture may define the following macros to request
3283 conversions between the raw and virtual format:
3285 @deftypefn {Target Macro} int REGISTER_CONVERTIBLE (int @var{reg})
3286 Return non-zero if register number @var{reg}'s value needs different raw
3287 and virtual formats.
3289 You should not use @code{REGISTER_CONVERT_TO_VIRTUAL} for a register
3290 unless this macro returns a non-zero value for that register.
3293 @deftypefn {Target Macro} void REGISTER_CONVERT_TO_VIRTUAL (int @var{reg}, struct type *@var{type}, char *@var{from}, char *@var{to})
3294 Convert the value of register number @var{reg} to @var{type}, which
3295 should always be @code{gdbarch_register_type (@var{reg})}. The buffer
3296 at @var{from} holds the register's value in raw format; the macro should
3297 convert the value to virtual format, and place it at @var{to}.
3299 Note that @code{REGISTER_CONVERT_TO_VIRTUAL} and
3300 @code{REGISTER_CONVERT_TO_RAW} take their @var{reg} and @var{type}
3301 arguments in different orders.
3303 You should only use @code{REGISTER_CONVERT_TO_VIRTUAL} with registers
3304 for which the @code{REGISTER_CONVERTIBLE} macro returns a non-zero
3308 @deftypefn {Target Macro} void REGISTER_CONVERT_TO_RAW (struct type *@var{type}, int @var{reg}, char *@var{from}, char *@var{to})
3309 Convert the value of register number @var{reg} to @var{type}, which
3310 should always be @code{gdbarch_register_type (@var{reg})}. The buffer
3311 at @var{from} holds the register's value in raw format; the macro should
3312 convert the value to virtual format, and place it at @var{to}.
3314 Note that REGISTER_CONVERT_TO_VIRTUAL and REGISTER_CONVERT_TO_RAW take
3315 their @var{reg} and @var{type} arguments in different orders.
3319 @node Register and Memory Data
3320 @section Using Different Register and Memory Data Representations
3321 @cindex register representation
3322 @cindex memory representation
3323 @cindex representations, register and memory
3324 @cindex register data formats, converting
3325 @cindex @code{struct value}, converting register contents to
3327 @emph{Maintainer's note: The way GDB manipulates registers is undergoing
3328 significant change. Many of the macros and functions referred to in this
3329 section are likely to be subject to further revision. See
3330 @uref{http://sources.redhat.com/gdb/current/ari/, A.R. Index} and
3331 @uref{http://www.gnu.org/software/gdb/bugs, Bug Tracking Database} for
3332 further information. cagney/2002-05-06.}
3334 Some architectures can represent a data object in a register using a
3335 form that is different to the objects more normal memory representation.
3341 The Alpha architecture can represent 32 bit integer values in
3342 floating-point registers.
3345 The x86 architecture supports 80-bit floating-point registers. The
3346 @code{long double} data type occupies 96 bits in memory but only 80 bits
3347 when stored in a register.
3351 In general, the register representation of a data type is determined by
3352 the architecture, or @value{GDBN}'s interface to the architecture, while
3353 the memory representation is determined by the Application Binary
3356 For almost all data types on almost all architectures, the two
3357 representations are identical, and no special handling is needed.
3358 However, they do occasionally differ. Your architecture may define the
3359 following macros to request conversions between the register and memory
3360 representations of a data type:
3362 @deftypefun int gdbarch_convert_register_p (struct gdbarch *@var{gdbarch}, int @var{reg})
3363 Return non-zero if the representation of a data value stored in this
3364 register may be different to the representation of that same data value
3365 when stored in memory.
3367 When non-zero, the macros @code{gdbarch_register_to_value} and
3368 @code{value_to_register} are used to perform any necessary conversion.
3370 This function should return zero for the register's native type, when
3371 no conversion is necessary.
3374 @deftypefun void gdbarch_register_to_value (struct gdbarch *@var{gdbarch}, int @var{reg}, struct type *@var{type}, char *@var{from}, char *@var{to})
3375 Convert the value of register number @var{reg} to a data object of type
3376 @var{type}. The buffer at @var{from} holds the register's value in raw
3377 format; the converted value should be placed in the buffer at @var{to}.
3379 Note that @code{gdbarch_register_to_value} and @code{gdbarch_value_to_register}
3380 take their @var{reg} and @var{type} arguments in different orders.
3382 You should only use @code{gdbarch_register_to_value} with registers for which
3383 the @code{gdbarch_convert_register_p} function returns a non-zero value.
3386 @deftypefun void gdbarch_value_to_register (struct gdbarch *@var{gdbarch}, struct type *@var{type}, int @var{reg}, char *@var{from}, char *@var{to})
3387 Convert a data value of type @var{type} to register number @var{reg}'
3390 Note that @code{gdbarch_register_to_value} and @code{gdbarch_value_to_register}
3391 take their @var{reg} and @var{type} arguments in different orders.
3393 You should only use @code{gdbarch_value_to_register} with registers for which
3394 the @code{gdbarch_convert_register_p} function returns a non-zero value.
3397 @node Frame Interpretation
3398 @section Frame Interpretation
3400 @node Inferior Call Setup
3401 @section Inferior Call Setup
3403 @node Compiler Characteristics
3404 @section Compiler Characteristics
3406 @node Target Conditionals
3407 @section Target Conditionals
3409 This section describes the macros and functions that you can use to define the
3414 @item CORE_ADDR gdbarch_addr_bits_remove (@var{gdbarch}, @var{addr})
3415 @findex gdbarch_addr_bits_remove
3416 If a raw machine instruction address includes any bits that are not
3417 really part of the address, then this function is used to zero those bits in
3418 @var{addr}. This is only used for addresses of instructions, and even then not
3421 For example, the two low-order bits of the PC on the Hewlett-Packard PA
3422 2.0 architecture contain the privilege level of the corresponding
3423 instruction. Since instructions must always be aligned on four-byte
3424 boundaries, the processor masks out these bits to generate the actual
3425 address of the instruction. @code{gdbarch_addr_bits_remove} would then for
3426 example look like that:
3428 arch_addr_bits_remove (CORE_ADDR addr)
3430 return (addr &= ~0x3);
3434 @item int address_class_name_to_type_flags (@var{gdbarch}, @var{name}, @var{type_flags_ptr})
3435 @findex address_class_name_to_type_flags
3436 If @var{name} is a valid address class qualifier name, set the @code{int}
3437 referenced by @var{type_flags_ptr} to the mask representing the qualifier
3438 and return 1. If @var{name} is not a valid address class qualifier name,
3441 The value for @var{type_flags_ptr} should be one of
3442 @code{TYPE_FLAG_ADDRESS_CLASS_1}, @code{TYPE_FLAG_ADDRESS_CLASS_2}, or
3443 possibly some combination of these values or'd together.
3444 @xref{Target Architecture Definition, , Address Classes}.
3446 @item int address_class_name_to_type_flags_p (@var{gdbarch})
3447 @findex address_class_name_to_type_flags_p
3448 Predicate which indicates whether @code{address_class_name_to_type_flags}
3451 @item int gdbarch_address_class_type_flags (@var{gdbarch}, @var{byte_size}, @var{dwarf2_addr_class})
3452 @findex gdbarch_address_class_type_flags
3453 Given a pointers byte size (as described by the debug information) and
3454 the possible @code{DW_AT_address_class} value, return the type flags
3455 used by @value{GDBN} to represent this address class. The value
3456 returned should be one of @code{TYPE_FLAG_ADDRESS_CLASS_1},
3457 @code{TYPE_FLAG_ADDRESS_CLASS_2}, or possibly some combination of these
3458 values or'd together.
3459 @xref{Target Architecture Definition, , Address Classes}.
3461 @item int gdbarch_address_class_type_flags_p (@var{gdbarch})
3462 @findex gdbarch_address_class_type_flags_p
3463 Predicate which indicates whether @code{gdbarch_address_class_type_flags_p} has
3466 @item const char *gdbarch_address_class_type_flags_to_name (@var{gdbarch}, @var{type_flags})
3467 @findex gdbarch_address_class_type_flags_to_name
3468 Return the name of the address class qualifier associated with the type
3469 flags given by @var{type_flags}.
3471 @item int gdbarch_address_class_type_flags_to_name_p (@var{gdbarch})
3472 @findex gdbarch_address_class_type_flags_to_name_p
3473 Predicate which indicates whether @code{gdbarch_address_class_type_flags_to_name} has been defined.
3474 @xref{Target Architecture Definition, , Address Classes}.
3476 @item void gdbarch_address_to_pointer (@var{gdbarch}, @var{type}, @var{buf}, @var{addr})
3477 @findex gdbarch_address_to_pointer
3478 Store in @var{buf} a pointer of type @var{type} representing the address
3479 @var{addr}, in the appropriate format for the current architecture.
3480 This function may safely assume that @var{type} is either a pointer or a
3481 C@t{++} reference type.
3482 @xref{Target Architecture Definition, , Pointers Are Not Always Addresses}.
3484 @item int gdbarch_believe_pcc_promotion (@var{gdbarch})
3485 @findex gdbarch_believe_pcc_promotion
3486 Used to notify if the compiler promotes a @code{short} or @code{char}
3487 parameter to an @code{int}, but still reports the parameter as its
3488 original type, rather than the promoted type.
3490 @item gdbarch_bits_big_endian (@var{gdbarch})
3491 @findex gdbarch_bits_big_endian
3492 This is used if the numbering of bits in the targets does @strong{not} match
3493 the endianness of the target byte order. A value of 1 means that the bits
3494 are numbered in a big-endian bit order, 0 means little-endian.
3496 @item set_gdbarch_bits_big_endian (@var{gdbarch}, @var{bits_big_endian})
3497 @findex set_gdbarch_bits_big_endian
3498 Calling set_gdbarch_bits_big_endian with a value of 1 indicates that the
3499 bits in the target are numbered in a big-endian bit order, 0 indicates
3504 This is the character array initializer for the bit pattern to put into
3505 memory where a breakpoint is set. Although it's common to use a trap
3506 instruction for a breakpoint, it's not required; for instance, the bit
3507 pattern could be an invalid instruction. The breakpoint must be no
3508 longer than the shortest instruction of the architecture.
3510 @code{BREAKPOINT} has been deprecated in favor of
3511 @code{gdbarch_breakpoint_from_pc}.
3513 @item BIG_BREAKPOINT
3514 @itemx LITTLE_BREAKPOINT
3515 @findex LITTLE_BREAKPOINT
3516 @findex BIG_BREAKPOINT
3517 Similar to BREAKPOINT, but used for bi-endian targets.
3519 @code{BIG_BREAKPOINT} and @code{LITTLE_BREAKPOINT} have been deprecated in
3520 favor of @code{gdbarch_breakpoint_from_pc}.
3522 @item const gdb_byte *gdbarch_breakpoint_from_pc (@var{gdbarch}, @var{pcptr}, @var{lenptr})
3523 @findex gdbarch_breakpoint_from_pc
3524 @anchor{gdbarch_breakpoint_from_pc} Use the program counter to determine the
3525 contents and size of a breakpoint instruction. It returns a pointer to
3526 a static string of bytes that encode a breakpoint instruction, stores the
3527 length of the string to @code{*@var{lenptr}}, and adjusts the program
3528 counter (if necessary) to point to the actual memory location where the
3529 breakpoint should be inserted. May return @code{NULL} to indicate that
3530 software breakpoints are not supported.
3532 Although it is common to use a trap instruction for a breakpoint, it's
3533 not required; for instance, the bit pattern could be an invalid
3534 instruction. The breakpoint must be no longer than the shortest
3535 instruction of the architecture.
3537 Provided breakpoint bytes can be also used by @code{bp_loc_is_permanent} to
3538 detect permanent breakpoints. @code{gdbarch_breakpoint_from_pc} should return
3539 an unchanged memory copy if it was called for a location with permanent
3540 breakpoint as some architectures use breakpoint instructions containing
3541 arbitrary parameter value.
3543 Replaces all the other @var{BREAKPOINT} macros.
3545 @item int gdbarch_memory_insert_breakpoint (@var{gdbarch}, @var{bp_tgt})
3546 @itemx gdbarch_memory_remove_breakpoint (@var{gdbarch}, @var{bp_tgt})
3547 @findex gdbarch_memory_remove_breakpoint
3548 @findex gdbarch_memory_insert_breakpoint
3549 Insert or remove memory based breakpoints. Reasonable defaults
3550 (@code{default_memory_insert_breakpoint} and
3551 @code{default_memory_remove_breakpoint} respectively) have been
3552 provided so that it is not necessary to set these for most
3553 architectures. Architectures which may want to set
3554 @code{gdbarch_memory_insert_breakpoint} and @code{gdbarch_memory_remove_breakpoint} will likely have instructions that are oddly sized or are not stored in a
3555 conventional manner.
3557 It may also be desirable (from an efficiency standpoint) to define
3558 custom breakpoint insertion and removal routines if
3559 @code{gdbarch_breakpoint_from_pc} needs to read the target's memory for some
3562 @item CORE_ADDR gdbarch_adjust_breakpoint_address (@var{gdbarch}, @var{bpaddr})
3563 @findex gdbarch_adjust_breakpoint_address
3564 @cindex breakpoint address adjusted
3565 Given an address at which a breakpoint is desired, return a breakpoint
3566 address adjusted to account for architectural constraints on
3567 breakpoint placement. This method is not needed by most targets.
3569 The FR-V target (see @file{frv-tdep.c}) requires this method.
3570 The FR-V is a VLIW architecture in which a number of RISC-like
3571 instructions are grouped (packed) together into an aggregate
3572 instruction or instruction bundle. When the processor executes
3573 one of these bundles, the component instructions are executed
3576 In the course of optimization, the compiler may group instructions
3577 from distinct source statements into the same bundle. The line number
3578 information associated with one of the latter statements will likely
3579 refer to some instruction other than the first one in the bundle. So,
3580 if the user attempts to place a breakpoint on one of these latter
3581 statements, @value{GDBN} must be careful to @emph{not} place the break
3582 instruction on any instruction other than the first one in the bundle.
3583 (Remember though that the instructions within a bundle execute
3584 in parallel, so the @emph{first} instruction is the instruction
3585 at the lowest address and has nothing to do with execution order.)
3587 The FR-V's @code{gdbarch_adjust_breakpoint_address} method will adjust a
3588 breakpoint's address by scanning backwards for the beginning of
3589 the bundle, returning the address of the bundle.
3591 Since the adjustment of a breakpoint may significantly alter a user's
3592 expectation, @value{GDBN} prints a warning when an adjusted breakpoint
3593 is initially set and each time that that breakpoint is hit.
3595 @item int gdbarch_call_dummy_location (@var{gdbarch})
3596 @findex gdbarch_call_dummy_location
3597 See the file @file{inferior.h}.
3599 This method has been replaced by @code{gdbarch_push_dummy_code}
3600 (@pxref{gdbarch_push_dummy_code}).
3602 @item int gdbarch_cannot_fetch_register (@var{gdbarch}, @var{regum})
3603 @findex gdbarch_cannot_fetch_register
3604 This function should return nonzero if @var{regno} cannot be fetched
3605 from an inferior process.
3607 @item int gdbarch_cannot_store_register (@var{gdbarch}, @var{regnum})
3608 @findex gdbarch_cannot_store_register
3609 This function should return nonzero if @var{regno} should not be
3610 written to the target. This is often the case for program counters,
3611 status words, and other special registers. This function returns 0 as
3612 default so that @value{GDBN} will assume that all registers may be written.
3614 @item int gdbarch_convert_register_p (@var{gdbarch}, @var{regnum}, struct type *@var{type})
3615 @findex gdbarch_convert_register_p
3616 Return non-zero if register @var{regnum} represents data values of type
3617 @var{type} in a non-standard form.
3618 @xref{Target Architecture Definition, , Using Different Register and Memory Data Representations}.
3620 @item int gdbarch_fp0_regnum (@var{gdbarch})
3621 @findex gdbarch_fp0_regnum
3622 This function returns the number of the first floating point register,
3623 if the machine has such registers. Otherwise, it returns -1.
3625 @item CORE_ADDR gdbarch_decr_pc_after_break (@var{gdbarch})
3626 @findex gdbarch_decr_pc_after_break
3627 This function shall return the amount by which to decrement the PC after the
3628 program encounters a breakpoint. This is often the number of bytes in
3629 @code{BREAKPOINT}, though not always. For most targets this value will be 0.
3631 @item DISABLE_UNSETTABLE_BREAK (@var{addr})
3632 @findex DISABLE_UNSETTABLE_BREAK
3633 If defined, this should evaluate to 1 if @var{addr} is in a shared
3634 library in which breakpoints cannot be set and so should be disabled.
3636 @item void gdbarch_print_float_info (@var{gdbarch}, @var{file}, @var{frame}, @var{args})
3637 @findex gdbarch_print_float_info
3638 If defined, then the @samp{info float} command will print information about
3639 the processor's floating point unit.
3641 @item void gdbarch_print_registers_info (@var{gdbarch}, @var{frame}, @var{regnum}, @var{all})
3642 @findex gdbarch_print_registers_info
3643 If defined, pretty print the value of the register @var{regnum} for the
3644 specified @var{frame}. If the value of @var{regnum} is -1, pretty print
3645 either all registers (@var{all} is non zero) or a select subset of
3646 registers (@var{all} is zero).
3648 The default method prints one register per line, and if @var{all} is
3649 zero omits floating-point registers.
3651 @item int gdbarch_print_vector_info (@var{gdbarch}, @var{file}, @var{frame}, @var{args})
3652 @findex gdbarch_print_vector_info
3653 If defined, then the @samp{info vector} command will call this function
3654 to print information about the processor's vector unit.
3656 By default, the @samp{info vector} command will print all vector
3657 registers (the register's type having the vector attribute).
3659 @item int gdbarch_dwarf2_reg_to_regnum (@var{gdbarch}, @var{dwarf2_regnr})
3660 @findex gdbarch_dwarf2_reg_to_regnum
3661 Convert DWARF2 register number @var{dwarf2_regnr} into @value{GDBN} regnum.
3662 If not defined, no conversion will be performed.
3664 @item int gdbarch_ecoff_reg_to_regnum (@var{gdbarch}, @var{ecoff_regnr})
3665 @findex gdbarch_ecoff_reg_to_regnum
3666 Convert ECOFF register number @var{ecoff_regnr} into @value{GDBN} regnum. If
3667 not defined, no conversion will be performed.
3669 @item CORE_ADDR frame_align (@var{gdbarch}, @var{address})
3670 @anchor{frame_align}
3672 Define this to adjust @var{address} so that it meets the alignment
3673 requirements for the start of a new stack frame. A stack frame's
3674 alignment requirements are typically stronger than a target processors
3675 stack alignment requirements.
3677 This function is used to ensure that, when creating a dummy frame, both
3678 the initial stack pointer and (if needed) the address of the return
3679 value are correctly aligned.
3681 This function always adjusts the address in the direction of stack
3684 By default, no frame based stack alignment is performed.
3686 @item int gdbarch_frame_red_zone_size (@var{gdbarch})
3687 @findex gdbarch_frame_red_zone_size
3688 The number of bytes, beyond the innermost-stack-address, reserved by the
3689 @sc{abi}. A function is permitted to use this scratch area (instead of
3690 allocating extra stack space).
3692 When performing an inferior function call, to ensure that it does not
3693 modify this area, @value{GDBN} adjusts the innermost-stack-address by
3694 @var{gdbarch_frame_red_zone_size} bytes before pushing parameters onto the
3697 By default, zero bytes are allocated. The value must be aligned
3698 (@pxref{frame_align}).
3700 The @sc{amd64} (nee x86-64) @sc{abi} documentation refers to the
3701 @emph{red zone} when describing this scratch area.
3704 @code{FRAME_FIND_SAVED_REGS} is deprecated.
3706 @item int gdbarch_frame_num_args (@var{gdbarch}, @var{frame})
3707 @findex gdbarch_frame_num_args
3708 For the frame described by @var{frame} return the number of arguments that
3709 are being passed. If the number of arguments is not known, return
3712 @item CORE_ADDR gdbarch_unwind_pc (@var{next_frame})
3713 @findex gdbarch_unwind_pc
3714 @anchor{gdbarch_unwind_pc} Return the instruction address, in
3715 @var{next_frame}'s caller, at which execution will resume after
3716 @var{next_frame} returns. This is commonly referred to as the return address.
3718 The implementation, which must be frame agnostic (work with any frame),
3719 is typically no more than:
3723 pc = frame_unwind_register_unsigned (next_frame, S390_PC_REGNUM);
3724 return gdbarch_addr_bits_remove (gdbarch, pc);
3729 @item CORE_ADDR gdbarch_unwind_sp (@var{gdbarch}, @var{next_frame})
3730 @findex gdbarch_unwind_sp
3731 @anchor{gdbarch_unwind_sp} Return the frame's inner most stack address. This is
3732 commonly referred to as the frame's @dfn{stack pointer}.
3734 The implementation, which must be frame agnostic (work with any frame),
3735 is typically no more than:
3739 sp = frame_unwind_register_unsigned (next_frame, S390_SP_REGNUM);
3740 return gdbarch_addr_bits_remove (gdbarch, sp);
3744 @xref{TARGET_READ_SP}, which this method replaces.
3746 @item GCC_COMPILED_FLAG_SYMBOL
3747 @itemx GCC2_COMPILED_FLAG_SYMBOL
3748 @findex GCC2_COMPILED_FLAG_SYMBOL
3749 @findex GCC_COMPILED_FLAG_SYMBOL
3750 If defined, these are the names of the symbols that @value{GDBN} will
3751 look for to detect that GCC compiled the file. The default symbols
3752 are @code{gcc_compiled.} and @code{gcc2_compiled.},
3753 respectively. (Currently only defined for the Delta 68.)
3755 @item gdbarch_get_longjmp_target
3756 @findex gdbarch_get_longjmp_target
3757 This function determines the target PC address that @code{longjmp}
3758 will jump to, assuming that we have just stopped at a @code{longjmp}
3759 breakpoint. It takes a @code{CORE_ADDR *} as argument, and stores the
3760 target PC value through this pointer. It examines the current state
3761 of the machine as needed, typically by using a manually-determined
3762 offset into the @code{jmp_buf}. (While we might like to get the offset
3763 from the target's @file{jmpbuf.h}, that header file cannot be assumed
3764 to be available when building a cross-debugger.)
3766 @item DEPRECATED_IBM6000_TARGET
3767 @findex DEPRECATED_IBM6000_TARGET
3768 Shows that we are configured for an IBM RS/6000 system. This
3769 conditional should be eliminated (FIXME) and replaced by
3770 feature-specific macros. It was introduced in haste and we are
3771 repenting at leisure.
3773 @item I386_USE_GENERIC_WATCHPOINTS
3774 An x86-based target can define this to use the generic x86 watchpoint
3775 support; see @ref{Algorithms, I386_USE_GENERIC_WATCHPOINTS}.
3777 @item int gdbarch_inner_than (@var{gdbarch}, @var{lhs}, @var{rhs})
3778 @findex gdbarch_inner_than
3779 Returns non-zero if stack address @var{lhs} is inner than (nearer to the
3780 stack top) stack address @var{rhs}. Let the function return
3781 @w{@code{lhs < rhs}} if the target's stack grows downward in memory, or
3782 @w{@code{lhs > rsh}} if the stack grows upward.
3784 @item gdbarch_in_function_epilogue_p (@var{gdbarch}, @var{addr})
3785 @findex gdbarch_in_function_epilogue_p
3786 Returns non-zero if the given @var{addr} is in the epilogue of a function.
3787 The epilogue of a function is defined as the part of a function where
3788 the stack frame of the function already has been destroyed up to the
3789 final `return from function call' instruction.
3791 @item int gdbarch_in_solib_return_trampoline (@var{gdbarch}, @var{pc}, @var{name})
3792 @findex gdbarch_in_solib_return_trampoline
3793 Define this function to return nonzero if the program is stopped in the
3794 trampoline that returns from a shared library.
3796 @item target_so_ops.in_dynsym_resolve_code (@var{pc})
3797 @findex in_dynsym_resolve_code
3798 Define this to return nonzero if the program is stopped in the
3801 @item SKIP_SOLIB_RESOLVER (@var{pc})
3802 @findex SKIP_SOLIB_RESOLVER
3803 Define this to evaluate to the (nonzero) address at which execution
3804 should continue to get past the dynamic linker's symbol resolution
3805 function. A zero value indicates that it is not important or necessary
3806 to set a breakpoint to get through the dynamic linker and that single
3807 stepping will suffice.
3809 @item CORE_ADDR gdbarch_integer_to_address (@var{gdbarch}, @var{type}, @var{buf})
3810 @findex gdbarch_integer_to_address
3811 @cindex converting integers to addresses
3812 Define this when the architecture needs to handle non-pointer to address
3813 conversions specially. Converts that value to an address according to
3814 the current architectures conventions.
3816 @emph{Pragmatics: When the user copies a well defined expression from
3817 their source code and passes it, as a parameter, to @value{GDBN}'s
3818 @code{print} command, they should get the same value as would have been
3819 computed by the target program. Any deviation from this rule can cause
3820 major confusion and annoyance, and needs to be justified carefully. In
3821 other words, @value{GDBN} doesn't really have the freedom to do these
3822 conversions in clever and useful ways. It has, however, been pointed
3823 out that users aren't complaining about how @value{GDBN} casts integers
3824 to pointers; they are complaining that they can't take an address from a
3825 disassembly listing and give it to @code{x/i}. Adding an architecture
3826 method like @code{gdbarch_integer_to_address} certainly makes it possible for
3827 @value{GDBN} to ``get it right'' in all circumstances.}
3829 @xref{Target Architecture Definition, , Pointers Are Not Always
3832 @item CORE_ADDR gdbarch_pointer_to_address (@var{gdbarch}, @var{type}, @var{buf})
3833 @findex gdbarch_pointer_to_address
3834 Assume that @var{buf} holds a pointer of type @var{type}, in the
3835 appropriate format for the current architecture. Return the byte
3836 address the pointer refers to.
3837 @xref{Target Architecture Definition, , Pointers Are Not Always Addresses}.
3839 @item void gdbarch_register_to_value(@var{gdbarch}, @var{frame}, @var{regnum}, @var{type}, @var{fur})
3840 @findex gdbarch_register_to_value
3841 Convert the raw contents of register @var{regnum} into a value of type
3843 @xref{Target Architecture Definition, , Using Different Register and Memory Data Representations}.
3845 @item register_reggroup_p (@var{gdbarch}, @var{regnum}, @var{reggroup})
3846 @findex register_reggroup_p
3847 @cindex register groups
3848 Return non-zero if register @var{regnum} is a member of the register
3849 group @var{reggroup}.
3851 By default, registers are grouped as follows:
3854 @item float_reggroup
3855 Any register with a valid name and a floating-point type.
3856 @item vector_reggroup
3857 Any register with a valid name and a vector type.
3858 @item general_reggroup
3859 Any register with a valid name and a type other than vector or
3860 floating-point. @samp{float_reggroup}.
3862 @itemx restore_reggroup
3864 Any register with a valid name.
3867 @item struct type *register_type (@var{gdbarch}, @var{reg})
3868 @findex register_type
3869 If defined, return the type of register @var{reg}.
3870 @xref{Target Architecture Definition, , Raw and Virtual Register
3873 @item REGISTER_CONVERT_TO_VIRTUAL(@var{reg}, @var{type}, @var{from}, @var{to})
3874 @findex REGISTER_CONVERT_TO_VIRTUAL
3875 Convert the value of register @var{reg} from its raw form to its virtual
3877 @xref{Target Architecture Definition, , Raw and Virtual Register Representations}.
3879 @item REGISTER_CONVERT_TO_RAW(@var{type}, @var{reg}, @var{from}, @var{to})
3880 @findex REGISTER_CONVERT_TO_RAW
3881 Convert the value of register @var{reg} from its virtual form to its raw
3883 @xref{Target Architecture Definition, , Raw and Virtual Register Representations}.
3885 @item const struct regset *regset_from_core_section (struct gdbarch * @var{gdbarch}, const char * @var{sect_name}, size_t @var{sect_size})
3886 @findex regset_from_core_section
3887 Return the appropriate register set for a core file section with name
3888 @var{sect_name} and size @var{sect_size}.
3890 @item SOFTWARE_SINGLE_STEP_P()
3891 @findex SOFTWARE_SINGLE_STEP_P
3892 Define this as 1 if the target does not have a hardware single-step
3893 mechanism. The macro @code{SOFTWARE_SINGLE_STEP} must also be defined.
3895 @item SOFTWARE_SINGLE_STEP(@var{signal}, @var{insert_breakpoints_p})
3896 @findex SOFTWARE_SINGLE_STEP
3897 A function that inserts or removes (depending on
3898 @var{insert_breakpoints_p}) breakpoints at each possible destinations of
3899 the next instruction. See @file{sparc-tdep.c} and @file{rs6000-tdep.c}
3902 @item set_gdbarch_sofun_address_maybe_missing (@var{gdbarch}, @var{set})
3903 @findex set_gdbarch_sofun_address_maybe_missing
3904 Somebody clever observed that, the more actual addresses you have in the
3905 debug information, the more time the linker has to spend relocating
3906 them. So whenever there's some other way the debugger could find the
3907 address it needs, you should omit it from the debug info, to make
3910 Calling @code{set_gdbarch_sofun_address_maybe_missing} with a non-zero
3911 argument @var{set} indicates that a particular set of hacks of this sort
3912 are in use, affecting @code{N_SO} and @code{N_FUN} entries in stabs-format
3913 debugging information. @code{N_SO} stabs mark the beginning and ending
3914 addresses of compilation units in the text segment. @code{N_FUN} stabs
3915 mark the starts and ends of functions.
3917 In this case, @value{GDBN} assumes two things:
3921 @code{N_FUN} stabs have an address of zero. Instead of using those
3922 addresses, you should find the address where the function starts by
3923 taking the function name from the stab, and then looking that up in the
3924 minsyms (the linker/assembler symbol table). In other words, the stab
3925 has the name, and the linker/assembler symbol table is the only place
3926 that carries the address.
3929 @code{N_SO} stabs have an address of zero, too. You just look at the
3930 @code{N_FUN} stabs that appear before and after the @code{N_SO} stab, and
3931 guess the starting and ending addresses of the compilation unit from them.
3934 @item int gdbarch_pc_regnum (@var{gdbarch})
3935 @findex gdbarch_pc_regnum
3936 If the program counter is kept in a register, then let this function return
3937 the number (greater than or equal to zero) of that register.
3939 This should only need to be defined if @code{gdbarch_read_pc} and
3940 @code{gdbarch_write_pc} are not defined.
3942 @item int gdbarch_stabs_argument_has_addr (@var{gdbarch}, @var{type})
3943 @findex gdbarch_stabs_argument_has_addr
3944 @anchor{gdbarch_stabs_argument_has_addr} Define this function to return
3945 nonzero if a function argument of type @var{type} is passed by reference
3948 @item PROCESS_LINENUMBER_HOOK
3949 @findex PROCESS_LINENUMBER_HOOK
3950 A hook defined for XCOFF reading.
3952 @item gdbarch_ps_regnum (@var{gdbarch}
3953 @findex gdbarch_ps_regnum
3954 If defined, this function returns the number of the processor status
3956 (This definition is only used in generic code when parsing "$ps".)
3958 @item CORE_ADDR gdbarch_push_dummy_call (@var{gdbarch}, @var{function}, @var{regcache}, @var{bp_addr}, @var{nargs}, @var{args}, @var{sp}, @var{struct_return}, @var{struct_addr})
3959 @findex gdbarch_push_dummy_call
3960 @anchor{gdbarch_push_dummy_call} Define this to push the dummy frame's call to
3961 the inferior function onto the stack. In addition to pushing @var{nargs}, the
3962 code should push @var{struct_addr} (when @var{struct_return} is non-zero), and
3963 the return address (@var{bp_addr}).
3965 @var{function} is a pointer to a @code{struct value}; on architectures that use
3966 function descriptors, this contains the function descriptor value.
3968 Returns the updated top-of-stack pointer.
3970 @item CORE_ADDR gdbarch_push_dummy_code (@var{gdbarch}, @var{sp}, @var{funaddr}, @var{using_gcc}, @var{args}, @var{nargs}, @var{value_type}, @var{real_pc}, @var{bp_addr}, @var{regcache})
3971 @findex gdbarch_push_dummy_code
3972 @anchor{gdbarch_push_dummy_code} Given a stack based call dummy, push the
3973 instruction sequence (including space for a breakpoint) to which the
3974 called function should return.
3976 Set @var{bp_addr} to the address at which the breakpoint instruction
3977 should be inserted, @var{real_pc} to the resume address when starting
3978 the call sequence, and return the updated inner-most stack address.
3980 By default, the stack is grown sufficient to hold a frame-aligned
3981 (@pxref{frame_align}) breakpoint, @var{bp_addr} is set to the address
3982 reserved for that breakpoint, and @var{real_pc} set to @var{funaddr}.
3984 This method replaces @w{@code{gdbarch_call_dummy_location (@var{gdbarch})}}.
3986 @item const char *gdbarch_register_name (@var{gdbarch}, @var{regnr})
3987 @findex gdbarch_register_name
3988 Return the name of register @var{regnr} as a string. May return @code{NULL}
3989 to indicate that @var{regnr} is not a valid register.
3991 @item int gdbarch_sdb_reg_to_regnum (@var{gdbarch}, @var{sdb_regnr})
3992 @findex gdbarch_sdb_reg_to_regnum
3993 Use this function to convert sdb register @var{sdb_regnr} into @value{GDBN}
3994 regnum. If not defined, no conversion will be done.
3996 @item enum return_value_convention gdbarch_return_value (struct gdbarch *@var{gdbarch}, struct type *@var{valtype}, struct regcache *@var{regcache}, void *@var{readbuf}, const void *@var{writebuf})
3997 @findex gdbarch_return_value
3998 @anchor{gdbarch_return_value} Given a function with a return-value of
3999 type @var{rettype}, return which return-value convention that function
4002 @value{GDBN} currently recognizes two function return-value conventions:
4003 @code{RETURN_VALUE_REGISTER_CONVENTION} where the return value is found
4004 in registers; and @code{RETURN_VALUE_STRUCT_CONVENTION} where the return
4005 value is found in memory and the address of that memory location is
4006 passed in as the function's first parameter.
4008 If the register convention is being used, and @var{writebuf} is
4009 non-@code{NULL}, also copy the return-value in @var{writebuf} into
4012 If the register convention is being used, and @var{readbuf} is
4013 non-@code{NULL}, also copy the return value from @var{regcache} into
4014 @var{readbuf} (@var{regcache} contains a copy of the registers from the
4015 just returned function).
4017 @emph{Maintainer note: This method replaces separate predicate, extract,
4018 store methods. By having only one method, the logic needed to determine
4019 the return-value convention need only be implemented in one place. If
4020 @value{GDBN} were written in an @sc{oo} language, this method would
4021 instead return an object that knew how to perform the register
4022 return-value extract and store.}
4024 @emph{Maintainer note: This method does not take a @var{gcc_p}
4025 parameter, and such a parameter should not be added. If an architecture
4026 that requires per-compiler or per-function information be identified,
4027 then the replacement of @var{rettype} with @code{struct value}
4028 @var{function} should be pursued.}
4030 @emph{Maintainer note: The @var{regcache} parameter limits this methods
4031 to the inner most frame. While replacing @var{regcache} with a
4032 @code{struct frame_info} @var{frame} parameter would remove that
4033 limitation there has yet to be a demonstrated need for such a change.}
4035 @item void gdbarch_skip_permanent_breakpoint (@var{gdbarch}, @var{regcache})
4036 @findex gdbarch_skip_permanent_breakpoint
4037 Advance the inferior's PC past a permanent breakpoint. @value{GDBN} normally
4038 steps over a breakpoint by removing it, stepping one instruction, and
4039 re-inserting the breakpoint. However, permanent breakpoints are
4040 hardwired into the inferior, and can't be removed, so this strategy
4041 doesn't work. Calling @code{gdbarch_skip_permanent_breakpoint} adjusts the
4042 processor's state so that execution will resume just after the breakpoint.
4043 This function does the right thing even when the breakpoint is in the delay slot
4044 of a branch or jump.
4046 @item CORE_ADDR gdbarch_skip_prologue (@var{gdbarch}, @var{ip})
4047 @findex gdbarch_skip_prologue
4048 A function that returns the address of the ``real'' code beyond the
4049 function entry prologue found at @var{ip}.
4051 @item CORE_ADDR gdbarch_skip_trampoline_code (@var{gdbarch}, @var{frame}, @var{pc})
4052 @findex gdbarch_skip_trampoline_code
4053 If the target machine has trampoline code that sits between callers and
4054 the functions being called, then define this function to return a new PC
4055 that is at the start of the real function.
4057 @item int gdbarch_sp_regnum (@var{gdbarch})
4058 @findex gdbarch_sp_regnum
4059 If the stack-pointer is kept in a register, then use this function to return
4060 the number (greater than or equal to zero) of that register, or -1 if
4061 there is no such register.
4063 @item int gdbarch_deprecated_fp_regnum (@var{gdbarch})
4064 @findex gdbarch_deprecated_fp_regnum
4065 If the frame pointer is in a register, use this function to return the
4066 number of that register.
4068 @item int gdbarch_stab_reg_to_regnum (@var{gdbarch}, @var{stab_regnr})
4069 @findex gdbarch_stab_reg_to_regnum
4070 Use this function to convert stab register @var{stab_regnr} into @value{GDBN}
4071 regnum. If not defined, no conversion will be done.
4073 @item SYMBOL_RELOADING_DEFAULT
4074 @findex SYMBOL_RELOADING_DEFAULT
4075 The default value of the ``symbol-reloading'' variable. (Never defined in
4078 @item TARGET_CHAR_BIT
4079 @findex TARGET_CHAR_BIT
4080 Number of bits in a char; defaults to 8.
4082 @item int gdbarch_char_signed (@var{gdbarch})
4083 @findex gdbarch_char_signed
4084 Non-zero if @code{char} is normally signed on this architecture; zero if
4085 it should be unsigned.
4087 The ISO C standard requires the compiler to treat @code{char} as
4088 equivalent to either @code{signed char} or @code{unsigned char}; any
4089 character in the standard execution set is supposed to be positive.
4090 Most compilers treat @code{char} as signed, but @code{char} is unsigned
4091 on the IBM S/390, RS6000, and PowerPC targets.
4093 @item int gdbarch_double_bit (@var{gdbarch})
4094 @findex gdbarch_double_bit
4095 Number of bits in a double float; defaults to @w{@code{8 * TARGET_CHAR_BIT}}.
4097 @item int gdbarch_float_bit (@var{gdbarch})
4098 @findex gdbarch_float_bit
4099 Number of bits in a float; defaults to @w{@code{4 * TARGET_CHAR_BIT}}.
4101 @item int gdbarch_int_bit (@var{gdbarch})
4102 @findex gdbarch_int_bit
4103 Number of bits in an integer; defaults to @w{@code{4 * TARGET_CHAR_BIT}}.
4105 @item int gdbarch_long_bit (@var{gdbarch})
4106 @findex gdbarch_long_bit
4107 Number of bits in a long integer; defaults to @w{@code{4 * TARGET_CHAR_BIT}}.
4109 @item int gdbarch_long_double_bit (@var{gdbarch})
4110 @findex gdbarch_long_double_bit
4111 Number of bits in a long double float;
4112 defaults to @w{@code{2 * gdbarch_double_bit (@var{gdbarch})}}.
4114 @item int gdbarch_long_long_bit (@var{gdbarch})
4115 @findex gdbarch_long_long_bit
4116 Number of bits in a long long integer; defaults to
4117 @w{@code{2 * gdbarch_long_bit (@var{gdbarch})}}.
4119 @item int gdbarch_ptr_bit (@var{gdbarch})
4120 @findex gdbarch_ptr_bit
4121 Number of bits in a pointer; defaults to
4122 @w{@code{gdbarch_int_bit (@var{gdbarch})}}.
4124 @item int gdbarch_short_bit (@var{gdbarch})
4125 @findex gdbarch_short_bit
4126 Number of bits in a short integer; defaults to @w{@code{2 * TARGET_CHAR_BIT}}.
4128 @item CORE_ADDR gdbarch_read_pc (@var{gdbarch}, @var{regcache})
4129 @findex gdbarch_read_pc
4130 @itemx gdbarch_write_pc (@var{gdbarch}, @var{regcache}, @var{val})
4131 @findex gdbarch_write_pc
4132 @anchor{gdbarch_write_pc}
4133 @itemx TARGET_READ_SP
4134 @findex TARGET_READ_SP
4135 @itemx TARGET_READ_FP
4136 @findex TARGET_READ_FP
4137 @findex gdbarch_read_pc
4138 @findex gdbarch_write_pc
4141 @anchor{TARGET_READ_SP} These change the behavior of @code{gdbarch_read_pc},
4142 @code{gdbarch_write_pc}, and @code{read_sp}. For most targets, these may be
4143 left undefined. @value{GDBN} will call the read and write register
4144 functions with the relevant @code{_REGNUM} argument.
4146 These macros and functions are useful when a target keeps one of these
4147 registers in a hard to get at place; for example, part in a segment register
4148 and part in an ordinary register.
4150 @xref{gdbarch_unwind_sp}, which replaces @code{TARGET_READ_SP}.
4152 @item void gdbarch_virtual_frame_pointer (@var{gdbarch}, @var{pc}, @var{frame_regnum}, @var{frame_offset})
4153 @findex gdbarch_virtual_frame_pointer
4154 Returns a @code{(@var{register}, @var{offset})} pair representing the virtual
4155 frame pointer in use at the code address @var{pc}. If virtual frame
4156 pointers are not used, a default definition simply returns
4157 @code{gdbarch_deprecated_fp_regnum} (or @code{gdbarch_sp_regnum}, if
4158 no frame pointer is defined), with an offset of zero.
4160 @c need to explain virtual frame pointers, they are recorded in agent expressions
4163 @item TARGET_HAS_HARDWARE_WATCHPOINTS
4164 If non-zero, the target has support for hardware-assisted
4165 watchpoints. @xref{Algorithms, watchpoints}, for more details and
4166 other related macros.
4168 @item int gdbarch_print_insn (@var{gdbarch}, @var{vma}, @var{info})
4169 @findex gdbarch_print_insn
4170 This is the function used by @value{GDBN} to print an assembly
4171 instruction. It prints the instruction at address @var{vma} in
4172 debugged memory and returns the length of the instruction, in bytes.
4173 This usually points to a function in the @code{opcodes} library
4174 (@pxref{Support Libraries, ,Opcodes}). @var{info} is a structure (of
4175 type @code{disassemble_info}) defined in the header file
4176 @file{include/dis-asm.h}, and used to pass information to the
4177 instruction decoding routine.
4179 @item frame_id gdbarch_dummy_id (@var{gdbarch}, @var{frame})
4180 @findex gdbarch_dummy_id
4181 @anchor{gdbarch_dummy_id} Given @var{frame} return a @w{@code{struct
4182 frame_id}} that uniquely identifies an inferior function call's dummy
4183 frame. The value returned must match the dummy frame stack value
4184 previously saved by @code{call_function_by_hand}.
4186 @item void gdbarch_value_to_register (@var{gdbarch}, @var{frame}, @var{type}, @var{buf})
4187 @findex gdbarch_value_to_register
4188 Convert a value of type @var{type} into the raw contents of a register.
4189 @xref{Target Architecture Definition, , Using Different Register and Memory Data Representations}.
4193 Motorola M68K target conditionals.
4197 Define this to be the 4-bit location of the breakpoint trap vector. If
4198 not defined, it will default to @code{0xf}.
4200 @item REMOTE_BPT_VECTOR
4201 Defaults to @code{1}.
4205 @node Adding a New Target
4206 @section Adding a New Target
4208 @cindex adding a target
4209 The following files add a target to @value{GDBN}:
4212 @cindex target dependent files
4214 @item gdb/@var{ttt}-tdep.c
4215 Contains any miscellaneous code required for this target machine. On
4216 some machines it doesn't exist at all.
4218 @item gdb/@var{arch}-tdep.c
4219 @itemx gdb/@var{arch}-tdep.h
4220 This is required to describe the basic layout of the target machine's
4221 processor chip (registers, stack, etc.). It can be shared among many
4222 targets that use the same processor architecture.
4226 (Target header files such as
4227 @file{gdb/config/@var{arch}/tm-@var{ttt}.h},
4228 @file{gdb/config/@var{arch}/tm-@var{arch}.h}, and
4229 @file{config/tm-@var{os}.h} are no longer used.)
4231 @node Target Descriptions
4232 @chapter Target Descriptions
4233 @cindex target descriptions
4235 The target architecture definition (@pxref{Target Architecture Definition})
4236 contains @value{GDBN}'s hard-coded knowledge about an architecture. For
4237 some platforms, it is handy to have more flexible knowledge about a specific
4238 instance of the architecture---for instance, a processor or development board.
4239 @dfn{Target descriptions} provide a mechanism for the user to tell @value{GDBN}
4240 more about what their target supports, or for the target to tell @value{GDBN}
4243 For details on writing, automatically supplying, and manually selecting
4244 target descriptions, see @ref{Target Descriptions, , , gdb,
4245 Debugging with @value{GDBN}}. This section will cover some related
4246 topics about the @value{GDBN} internals.
4249 * Target Descriptions Implementation::
4250 * Adding Target Described Register Support::
4253 @node Target Descriptions Implementation
4254 @section Target Descriptions Implementation
4255 @cindex target descriptions, implementation
4257 Before @value{GDBN} connects to a new target, or runs a new program on
4258 an existing target, it discards any existing target description and
4259 reverts to a default gdbarch. Then, after connecting, it looks for a
4260 new target description by calling @code{target_find_description}.
4262 A description may come from a user specified file (XML), the remote
4263 @samp{qXfer:features:read} packet (also XML), or from any custom
4264 @code{to_read_description} routine in the target vector. For instance,
4265 the remote target supports guessing whether a MIPS target is 32-bit or
4266 64-bit based on the size of the @samp{g} packet.
4268 If any target description is found, @value{GDBN} creates a new gdbarch
4269 incorporating the description by calling @code{gdbarch_update_p}. Any
4270 @samp{<architecture>} element is handled first, to determine which
4271 architecture's gdbarch initialization routine is called to create the
4272 new architecture. Then the initialization routine is called, and has
4273 a chance to adjust the constructed architecture based on the contents
4274 of the target description. For instance, it can recognize any
4275 properties set by a @code{to_read_description} routine. Also
4276 see @ref{Adding Target Described Register Support}.
4278 @node Adding Target Described Register Support
4279 @section Adding Target Described Register Support
4280 @cindex target descriptions, adding register support
4282 Target descriptions can report additional registers specific to an
4283 instance of the target. But it takes a little work in the architecture
4284 specific routines to support this.
4286 A target description must either have no registers or a complete
4287 set---this avoids complexity in trying to merge standard registers
4288 with the target defined registers. It is the architecture's
4289 responsibility to validate that a description with registers has
4290 everything it needs. To keep architecture code simple, the same
4291 mechanism is used to assign fixed internal register numbers to
4294 If @code{tdesc_has_registers} returns 1, the description contains
4295 registers. The architecture's @code{gdbarch_init} routine should:
4300 Call @code{tdesc_data_alloc} to allocate storage, early, before
4301 searching for a matching gdbarch or allocating a new one.
4304 Use @code{tdesc_find_feature} to locate standard features by name.
4307 Use @code{tdesc_numbered_register} and @code{tdesc_numbered_register_choices}
4308 to locate the expected registers in the standard features.
4311 Return @code{NULL} if a required feature is missing, or if any standard
4312 feature is missing expected registers. This will produce a warning that
4313 the description was incomplete.
4316 Free the allocated data before returning, unless @code{tdesc_use_registers}
4320 Call @code{set_gdbarch_num_regs} as usual, with a number higher than any
4321 fixed number passed to @code{tdesc_numbered_register}.
4324 Call @code{tdesc_use_registers} after creating a new gdbarch, before
4329 After @code{tdesc_use_registers} has been called, the architecture's
4330 @code{register_name}, @code{register_type}, and @code{register_reggroup_p}
4331 routines will not be called; that information will be taken from
4332 the target description. @code{num_regs} may be increased to account
4333 for any additional registers in the description.
4335 Pseudo-registers require some extra care:
4340 Using @code{tdesc_numbered_register} allows the architecture to give
4341 constant register numbers to standard architectural registers, e.g.@:
4342 as an @code{enum} in @file{@var{arch}-tdep.h}. But because
4343 pseudo-registers are always numbered above @code{num_regs},
4344 which may be increased by the description, constant numbers
4345 can not be used for pseudos. They must be numbered relative to
4346 @code{num_regs} instead.
4349 The description will not describe pseudo-registers, so the
4350 architecture must call @code{set_tdesc_pseudo_register_name},
4351 @code{set_tdesc_pseudo_register_type}, and
4352 @code{set_tdesc_pseudo_register_reggroup_p} to supply routines
4353 describing pseudo registers. These routines will be passed
4354 internal register numbers, so the same routines used for the
4355 gdbarch equivalents are usually suitable.
4360 @node Target Vector Definition
4362 @chapter Target Vector Definition
4363 @cindex target vector
4365 The target vector defines the interface between @value{GDBN}'s
4366 abstract handling of target systems, and the nitty-gritty code that
4367 actually exercises control over a process or a serial port.
4368 @value{GDBN} includes some 30-40 different target vectors; however,
4369 each configuration of @value{GDBN} includes only a few of them.
4372 * Managing Execution State::
4373 * Existing Targets::
4376 @node Managing Execution State
4377 @section Managing Execution State
4378 @cindex execution state
4380 A target vector can be completely inactive (not pushed on the target
4381 stack), active but not running (pushed, but not connected to a fully
4382 manifested inferior), or completely active (pushed, with an accessible
4383 inferior). Most targets are only completely inactive or completely
4384 active, but some support persistent connections to a target even
4385 when the target has exited or not yet started.
4387 For example, connecting to the simulator using @code{target sim} does
4388 not create a running program. Neither registers nor memory are
4389 accessible until @code{run}. Similarly, after @code{kill}, the
4390 program can not continue executing. But in both cases @value{GDBN}
4391 remains connected to the simulator, and target-specific commands
4392 are directed to the simulator.
4394 A target which only supports complete activation should push itself
4395 onto the stack in its @code{to_open} routine (by calling
4396 @code{push_target}), and unpush itself from the stack in its
4397 @code{to_mourn_inferior} routine (by calling @code{unpush_target}).
4399 A target which supports both partial and complete activation should
4400 still call @code{push_target} in @code{to_open}, but not call
4401 @code{unpush_target} in @code{to_mourn_inferior}. Instead, it should
4402 call either @code{target_mark_running} or @code{target_mark_exited}
4403 in its @code{to_open}, depending on whether the target is fully active
4404 after connection. It should also call @code{target_mark_running} any
4405 time the inferior becomes fully active (e.g.@: in
4406 @code{to_create_inferior} and @code{to_attach}), and
4407 @code{target_mark_exited} when the inferior becomes inactive (in
4408 @code{to_mourn_inferior}). The target should also make sure to call
4409 @code{target_mourn_inferior} from its @code{to_kill}, to return the
4410 target to inactive state.
4412 @node Existing Targets
4413 @section Existing Targets
4416 @subsection File Targets
4418 Both executables and core files have target vectors.
4420 @subsection Standard Protocol and Remote Stubs
4422 @value{GDBN}'s file @file{remote.c} talks a serial protocol to code
4423 that runs in the target system. @value{GDBN} provides several sample
4424 @dfn{stubs} that can be integrated into target programs or operating
4425 systems for this purpose; they are named @file{@var{cpu}-stub.c}. Many
4426 operating systems, embedded targets, emulators, and simulators already
4427 have a GDB stub built into them, and maintenance of the remote
4428 protocol must be careful to preserve compatibility.
4430 The @value{GDBN} user's manual describes how to put such a stub into
4431 your target code. What follows is a discussion of integrating the
4432 SPARC stub into a complicated operating system (rather than a simple
4433 program), by Stu Grossman, the author of this stub.
4435 The trap handling code in the stub assumes the following upon entry to
4440 %l1 and %l2 contain pc and npc respectively at the time of the trap;
4446 you are in the correct trap window.
4449 As long as your trap handler can guarantee those conditions, then there
4450 is no reason why you shouldn't be able to ``share'' traps with the stub.
4451 The stub has no requirement that it be jumped to directly from the
4452 hardware trap vector. That is why it calls @code{exceptionHandler()},
4453 which is provided by the external environment. For instance, this could
4454 set up the hardware traps to actually execute code which calls the stub
4455 first, and then transfers to its own trap handler.
4457 For the most point, there probably won't be much of an issue with
4458 ``sharing'' traps, as the traps we use are usually not used by the kernel,
4459 and often indicate unrecoverable error conditions. Anyway, this is all
4460 controlled by a table, and is trivial to modify. The most important
4461 trap for us is for @code{ta 1}. Without that, we can't single step or
4462 do breakpoints. Everything else is unnecessary for the proper operation
4463 of the debugger/stub.
4465 From reading the stub, it's probably not obvious how breakpoints work.
4466 They are simply done by deposit/examine operations from @value{GDBN}.
4468 @subsection ROM Monitor Interface
4470 @subsection Custom Protocols
4472 @subsection Transport Layer
4474 @subsection Builtin Simulator
4477 @node Native Debugging
4479 @chapter Native Debugging
4480 @cindex native debugging
4482 Several files control @value{GDBN}'s configuration for native support:
4486 @item gdb/config/@var{arch}/@var{xyz}.mh
4487 Specifies Makefile fragments needed by a @emph{native} configuration on
4488 machine @var{xyz}. In particular, this lists the required
4489 native-dependent object files, by defining @samp{NATDEPFILES=@dots{}}.
4490 Also specifies the header file which describes native support on
4491 @var{xyz}, by defining @samp{NAT_FILE= nm-@var{xyz}.h}. You can also
4492 define @samp{NAT_CFLAGS}, @samp{NAT_ADD_FILES}, @samp{NAT_CLIBS},
4493 @samp{NAT_CDEPS}, @samp{NAT_GENERATED_FILES}, etc.; see @file{Makefile.in}.
4495 @emph{Maintainer's note: The @file{.mh} suffix is because this file
4496 originally contained @file{Makefile} fragments for hosting @value{GDBN}
4497 on machine @var{xyz}. While the file is no longer used for this
4498 purpose, the @file{.mh} suffix remains. Perhaps someone will
4499 eventually rename these fragments so that they have a @file{.mn}
4502 @item gdb/config/@var{arch}/nm-@var{xyz}.h
4503 (@file{nm.h} is a link to this file, created by @code{configure}). Contains C
4504 macro definitions describing the native system environment, such as
4505 child process control and core file support.
4507 @item gdb/@var{xyz}-nat.c
4508 Contains any miscellaneous C code required for this native support of
4509 this machine. On some machines it doesn't exist at all.
4512 There are some ``generic'' versions of routines that can be used by
4513 various systems. These can be customized in various ways by macros
4514 defined in your @file{nm-@var{xyz}.h} file. If these routines work for
4515 the @var{xyz} host, you can just include the generic file's name (with
4516 @samp{.o}, not @samp{.c}) in @code{NATDEPFILES}.
4518 Otherwise, if your machine needs custom support routines, you will need
4519 to write routines that perform the same functions as the generic file.
4520 Put them into @file{@var{xyz}-nat.c}, and put @file{@var{xyz}-nat.o}
4521 into @code{NATDEPFILES}.
4525 This contains the @emph{target_ops vector} that supports Unix child
4526 processes on systems which use ptrace and wait to control the child.
4529 This contains the @emph{target_ops vector} that supports Unix child
4530 processes on systems which use /proc to control the child.
4533 This does the low-level grunge that uses Unix system calls to do a ``fork
4534 and exec'' to start up a child process.
4537 This is the low level interface to inferior processes for systems using
4538 the Unix @code{ptrace} call in a vanilla way.
4541 @section Native core file Support
4542 @cindex native core files
4545 @findex fetch_core_registers
4546 @item core-aout.c::fetch_core_registers()
4547 Support for reading registers out of a core file. This routine calls
4548 @code{register_addr()}, see below. Now that BFD is used to read core
4549 files, virtually all machines should use @code{core-aout.c}, and should
4550 just provide @code{fetch_core_registers} in @code{@var{xyz}-nat.c} (or
4551 @code{REGISTER_U_ADDR} in @code{nm-@var{xyz}.h}).
4553 @item core-aout.c::register_addr()
4554 If your @code{nm-@var{xyz}.h} file defines the macro
4555 @code{REGISTER_U_ADDR(addr, blockend, regno)}, it should be defined to
4556 set @code{addr} to the offset within the @samp{user} struct of @value{GDBN}
4557 register number @code{regno}. @code{blockend} is the offset within the
4558 ``upage'' of @code{u.u_ar0}. If @code{REGISTER_U_ADDR} is defined,
4559 @file{core-aout.c} will define the @code{register_addr()} function and
4560 use the macro in it. If you do not define @code{REGISTER_U_ADDR}, but
4561 you are using the standard @code{fetch_core_registers()}, you will need
4562 to define your own version of @code{register_addr()}, put it into your
4563 @code{@var{xyz}-nat.c} file, and be sure @code{@var{xyz}-nat.o} is in
4564 the @code{NATDEPFILES} list. If you have your own
4565 @code{fetch_core_registers()}, you may not need a separate
4566 @code{register_addr()}. Many custom @code{fetch_core_registers()}
4567 implementations simply locate the registers themselves.@refill
4570 When making @value{GDBN} run native on a new operating system, to make it
4571 possible to debug core files, you will need to either write specific
4572 code for parsing your OS's core files, or customize
4573 @file{bfd/trad-core.c}. First, use whatever @code{#include} files your
4574 machine uses to define the struct of registers that is accessible
4575 (possibly in the u-area) in a core file (rather than
4576 @file{machine/reg.h}), and an include file that defines whatever header
4577 exists on a core file (e.g., the u-area or a @code{struct core}). Then
4578 modify @code{trad_unix_core_file_p} to use these values to set up the
4579 section information for the data segment, stack segment, any other
4580 segments in the core file (perhaps shared library contents or control
4581 information), ``registers'' segment, and if there are two discontiguous
4582 sets of registers (e.g., integer and float), the ``reg2'' segment. This
4583 section information basically delimits areas in the core file in a
4584 standard way, which the section-reading routines in BFD know how to seek
4587 Then back in @value{GDBN}, you need a matching routine called
4588 @code{fetch_core_registers}. If you can use the generic one, it's in
4589 @file{core-aout.c}; if not, it's in your @file{@var{xyz}-nat.c} file.
4590 It will be passed a char pointer to the entire ``registers'' segment,
4591 its length, and a zero; or a char pointer to the entire ``regs2''
4592 segment, its length, and a 2. The routine should suck out the supplied
4593 register values and install them into @value{GDBN}'s ``registers'' array.
4595 If your system uses @file{/proc} to control processes, and uses ELF
4596 format core files, then you may be able to use the same routines for
4597 reading the registers out of processes and out of core files.
4605 @section shared libraries
4607 @section Native Conditionals
4608 @cindex native conditionals
4610 When @value{GDBN} is configured and compiled, various macros are
4611 defined or left undefined, to control compilation when the host and
4612 target systems are the same. These macros should be defined (or left
4613 undefined) in @file{nm-@var{system}.h}.
4617 @item I386_USE_GENERIC_WATCHPOINTS
4618 An x86-based machine can define this to use the generic x86 watchpoint
4619 support; see @ref{Algorithms, I386_USE_GENERIC_WATCHPOINTS}.
4622 @findex PROC_NAME_FMT
4623 Defines the format for the name of a @file{/proc} device. Should be
4624 defined in @file{nm.h} @emph{only} in order to override the default
4625 definition in @file{procfs.c}.
4627 @item SOLIB_ADD (@var{filename}, @var{from_tty}, @var{targ}, @var{readsyms})
4629 Define this to expand into an expression that will cause the symbols in
4630 @var{filename} to be added to @value{GDBN}'s symbol table. If
4631 @var{readsyms} is zero symbols are not read but any necessary low level
4632 processing for @var{filename} is still done.
4634 @item SOLIB_CREATE_INFERIOR_HOOK
4635 @findex SOLIB_CREATE_INFERIOR_HOOK
4636 Define this to expand into any shared-library-relocation code that you
4637 want to be run just after the child process has been forked.
4639 @item START_INFERIOR_TRAPS_EXPECTED
4640 @findex START_INFERIOR_TRAPS_EXPECTED
4641 When starting an inferior, @value{GDBN} normally expects to trap
4643 the shell execs, and once when the program itself execs. If the actual
4644 number of traps is something other than 2, then define this macro to
4645 expand into the number expected.
4649 @node Support Libraries
4651 @chapter Support Libraries
4656 BFD provides support for @value{GDBN} in several ways:
4659 @item identifying executable and core files
4660 BFD will identify a variety of file types, including a.out, coff, and
4661 several variants thereof, as well as several kinds of core files.
4663 @item access to sections of files
4664 BFD parses the file headers to determine the names, virtual addresses,
4665 sizes, and file locations of all the various named sections in files
4666 (such as the text section or the data section). @value{GDBN} simply
4667 calls BFD to read or write section @var{x} at byte offset @var{y} for
4670 @item specialized core file support
4671 BFD provides routines to determine the failing command name stored in a
4672 core file, the signal with which the program failed, and whether a core
4673 file matches (i.e.@: could be a core dump of) a particular executable
4676 @item locating the symbol information
4677 @value{GDBN} uses an internal interface of BFD to determine where to find the
4678 symbol information in an executable file or symbol-file. @value{GDBN} itself
4679 handles the reading of symbols, since BFD does not ``understand'' debug
4680 symbols, but @value{GDBN} uses BFD's cached information to find the symbols,
4685 @cindex opcodes library
4687 The opcodes library provides @value{GDBN}'s disassembler. (It's a separate
4688 library because it's also used in binutils, for @file{objdump}).
4691 @cindex readline library
4692 The @code{readline} library provides a set of functions for use by applications
4693 that allow users to edit command lines as they are typed in.
4696 @cindex @code{libiberty} library
4698 The @code{libiberty} library provides a set of functions and features
4699 that integrate and improve on functionality found in modern operating
4700 systems. Broadly speaking, such features can be divided into three
4701 groups: supplemental functions (functions that may be missing in some
4702 environments and operating systems), replacement functions (providing
4703 a uniform and easier to use interface for commonly used standard
4704 functions), and extensions (which provide additional functionality
4705 beyond standard functions).
4707 @value{GDBN} uses various features provided by the @code{libiberty}
4708 library, for instance the C@t{++} demangler, the @acronym{IEEE}
4709 floating format support functions, the input options parser
4710 @samp{getopt}, the @samp{obstack} extension, and other functions.
4712 @subsection @code{obstacks} in @value{GDBN}
4713 @cindex @code{obstacks}
4715 The obstack mechanism provides a convenient way to allocate and free
4716 chunks of memory. Each obstack is a pool of memory that is managed
4717 like a stack. Objects (of any nature, size and alignment) are
4718 allocated and freed in a @acronym{LIFO} fashion on an obstack (see
4719 @code{libiberty}'s documentation for a more detailed explanation of
4722 The most noticeable use of the @code{obstacks} in @value{GDBN} is in
4723 object files. There is an obstack associated with each internal
4724 representation of an object file. Lots of things get allocated on
4725 these @code{obstacks}: dictionary entries, blocks, blockvectors,
4726 symbols, minimal symbols, types, vectors of fundamental types, class
4727 fields of types, object files section lists, object files section
4728 offset lists, line tables, symbol tables, partial symbol tables,
4729 string tables, symbol table private data, macros tables, debug
4730 information sections and entries, import and export lists (som),
4731 unwind information (hppa), dwarf2 location expressions data. Plus
4732 various strings such as directory names strings, debug format strings,
4735 An essential and convenient property of all data on @code{obstacks} is
4736 that memory for it gets allocated (with @code{obstack_alloc}) at
4737 various times during a debugging session, but it is released all at
4738 once using the @code{obstack_free} function. The @code{obstack_free}
4739 function takes a pointer to where in the stack it must start the
4740 deletion from (much like the cleanup chains have a pointer to where to
4741 start the cleanups). Because of the stack like structure of the
4742 @code{obstacks}, this allows to free only a top portion of the
4743 obstack. There are a few instances in @value{GDBN} where such thing
4744 happens. Calls to @code{obstack_free} are done after some local data
4745 is allocated to the obstack. Only the local data is deleted from the
4746 obstack. Of course this assumes that nothing between the
4747 @code{obstack_alloc} and the @code{obstack_free} allocates anything
4748 else on the same obstack. For this reason it is best and safest to
4749 use temporary @code{obstacks}.
4751 Releasing the whole obstack is also not safe per se. It is safe only
4752 under the condition that we know the @code{obstacks} memory is no
4753 longer needed. In @value{GDBN} we get rid of the @code{obstacks} only
4754 when we get rid of the whole objfile(s), for instance upon reading a
4758 @cindex regular expressions library
4769 @item SIGN_EXTEND_CHAR
4771 @item SWITCH_ENUM_BUG
4780 @section Array Containers
4781 @cindex Array Containers
4784 Often it is necessary to manipulate a dynamic array of a set of
4785 objects. C forces some bookkeeping on this, which can get cumbersome
4786 and repetitive. The @file{vec.h} file contains macros for defining
4787 and using a typesafe vector type. The functions defined will be
4788 inlined when compiling, and so the abstraction cost should be zero.
4789 Domain checks are added to detect programming errors.
4791 An example use would be an array of symbols or section information.
4792 The array can be grown as symbols are read in (or preallocated), and
4793 the accessor macros provided keep care of all the necessary
4794 bookkeeping. Because the arrays are type safe, there is no danger of
4795 accidentally mixing up the contents. Think of these as C++ templates,
4796 but implemented in C.
4798 Because of the different behavior of structure objects, scalar objects
4799 and of pointers, there are three flavors of vector, one for each of
4800 these variants. Both the structure object and pointer variants pass
4801 pointers to objects around --- in the former case the pointers are
4802 stored into the vector and in the latter case the pointers are
4803 dereferenced and the objects copied into the vector. The scalar
4804 object variant is suitable for @code{int}-like objects, and the vector
4805 elements are returned by value.
4807 There are both @code{index} and @code{iterate} accessors. The iterator
4808 returns a boolean iteration condition and updates the iteration
4809 variable passed by reference. Because the iterator will be inlined,
4810 the address-of can be optimized away.
4812 The vectors are implemented using the trailing array idiom, thus they
4813 are not resizeable without changing the address of the vector object
4814 itself. This means you cannot have variables or fields of vector type
4815 --- always use a pointer to a vector. The one exception is the final
4816 field of a structure, which could be a vector type. You will have to
4817 use the @code{embedded_size} & @code{embedded_init} calls to create
4818 such objects, and they will probably not be resizeable (so don't use
4819 the @dfn{safe} allocation variants). The trailing array idiom is used
4820 (rather than a pointer to an array of data), because, if we allow
4821 @code{NULL} to also represent an empty vector, empty vectors occupy
4822 minimal space in the structure containing them.
4824 Each operation that increases the number of active elements is
4825 available in @dfn{quick} and @dfn{safe} variants. The former presumes
4826 that there is sufficient allocated space for the operation to succeed
4827 (it dies if there is not). The latter will reallocate the vector, if
4828 needed. Reallocation causes an exponential increase in vector size.
4829 If you know you will be adding N elements, it would be more efficient
4830 to use the reserve operation before adding the elements with the
4831 @dfn{quick} operation. This will ensure there are at least as many
4832 elements as you ask for, it will exponentially increase if there are
4833 too few spare slots. If you want reserve a specific number of slots,
4834 but do not want the exponential increase (for instance, you know this
4835 is the last allocation), use a negative number for reservation. You
4836 can also create a vector of a specific size from the get go.
4838 You should prefer the push and pop operations, as they append and
4839 remove from the end of the vector. If you need to remove several items
4840 in one go, use the truncate operation. The insert and remove
4841 operations allow you to change elements in the middle of the vector.
4842 There are two remove operations, one which preserves the element
4843 ordering @code{ordered_remove}, and one which does not
4844 @code{unordered_remove}. The latter function copies the end element
4845 into the removed slot, rather than invoke a memmove operation. The
4846 @code{lower_bound} function will determine where to place an item in
4847 the array using insert that will maintain sorted order.
4849 If you need to directly manipulate a vector, then the @code{address}
4850 accessor will return the address of the start of the vector. Also the
4851 @code{space} predicate will tell you whether there is spare capacity in the
4852 vector. You will not normally need to use these two functions.
4854 Vector types are defined using a
4855 @code{DEF_VEC_@{O,P,I@}(@var{typename})} macro. Variables of vector
4856 type are declared using a @code{VEC(@var{typename})} macro. The
4857 characters @code{O}, @code{P} and @code{I} indicate whether
4858 @var{typename} is an object (@code{O}), pointer (@code{P}) or integral
4859 (@code{I}) type. Be careful to pick the correct one, as you'll get an
4860 awkward and inefficient API if you use the wrong one. There is a
4861 check, which results in a compile-time warning, for the @code{P} and
4862 @code{I} versions, but there is no check for the @code{O} versions, as
4863 that is not possible in plain C.
4865 An example of their use would be,
4868 DEF_VEC_P(tree); // non-managed tree vector.
4871 VEC(tree) *v; // A (pointer to) a vector of tree pointers.
4874 struct my_struct *s;
4876 if (VEC_length(tree, s->v)) @{ we have some contents @}
4877 VEC_safe_push(tree, s->v, decl); // append some decl onto the end
4878 for (ix = 0; VEC_iterate(tree, s->v, ix, elt); ix++)
4879 @{ do something with elt @}
4883 The @file{vec.h} file provides details on how to invoke the various
4884 accessors provided. They are enumerated here:
4888 Return the number of items in the array,
4891 Return true if the array has no elements.
4895 Return the last or arbitrary item in the array.
4898 Access an array element and indicate whether the array has been
4903 Create and destroy an array.
4905 @item VEC_embedded_size
4906 @itemx VEC_embedded_init
4907 Helpers for embedding an array as the final element of another struct.
4913 Return the amount of free space in an array.
4916 Ensure a certain amount of free space.
4918 @item VEC_quick_push
4919 @itemx VEC_safe_push
4920 Append to an array, either assuming the space is available, or making
4924 Remove the last item from an array.
4927 Remove several items from the end of an array.
4930 Add several items to the end of an array.
4933 Overwrite an item in the array.
4935 @item VEC_quick_insert
4936 @itemx VEC_safe_insert
4937 Insert an item into the middle of the array. Either the space must
4938 already exist, or the space is created.
4940 @item VEC_ordered_remove
4941 @itemx VEC_unordered_remove
4942 Remove an item from the array, preserving order or not.
4944 @item VEC_block_remove
4945 Remove a set of items from the array.
4948 Provide the address of the first element.
4950 @item VEC_lower_bound
4951 Binary search the array.
4961 This chapter covers topics that are lower-level than the major
4962 algorithms of @value{GDBN}.
4967 Cleanups are a structured way to deal with things that need to be done
4970 When your code does something (e.g., @code{xmalloc} some memory, or
4971 @code{open} a file) that needs to be undone later (e.g., @code{xfree}
4972 the memory or @code{close} the file), it can make a cleanup. The
4973 cleanup will be done at some future point: when the command is finished
4974 and control returns to the top level; when an error occurs and the stack
4975 is unwound; or when your code decides it's time to explicitly perform
4976 cleanups. Alternatively you can elect to discard the cleanups you
4982 @item struct cleanup *@var{old_chain};
4983 Declare a variable which will hold a cleanup chain handle.
4985 @findex make_cleanup
4986 @item @var{old_chain} = make_cleanup (@var{function}, @var{arg});
4987 Make a cleanup which will cause @var{function} to be called with
4988 @var{arg} (a @code{char *}) later. The result, @var{old_chain}, is a
4989 handle that can later be passed to @code{do_cleanups} or
4990 @code{discard_cleanups}. Unless you are going to call
4991 @code{do_cleanups} or @code{discard_cleanups}, you can ignore the result
4992 from @code{make_cleanup}.
4995 @item do_cleanups (@var{old_chain});
4996 Do all cleanups added to the chain since the corresponding
4997 @code{make_cleanup} call was made.
4999 @findex discard_cleanups
5000 @item discard_cleanups (@var{old_chain});
5001 Same as @code{do_cleanups} except that it just removes the cleanups from
5002 the chain and does not call the specified functions.
5005 Cleanups are implemented as a chain. The handle returned by
5006 @code{make_cleanups} includes the cleanup passed to the call and any
5007 later cleanups appended to the chain (but not yet discarded or
5011 make_cleanup (a, 0);
5013 struct cleanup *old = make_cleanup (b, 0);
5021 will call @code{c()} and @code{b()} but will not call @code{a()}. The
5022 cleanup that calls @code{a()} will remain in the cleanup chain, and will
5023 be done later unless otherwise discarded.@refill
5025 Your function should explicitly do or discard the cleanups it creates.
5026 Failing to do this leads to non-deterministic behavior since the caller
5027 will arbitrarily do or discard your functions cleanups. This need leads
5028 to two common cleanup styles.
5030 The first style is try/finally. Before it exits, your code-block calls
5031 @code{do_cleanups} with the old cleanup chain and thus ensures that your
5032 code-block's cleanups are always performed. For instance, the following
5033 code-segment avoids a memory leak problem (even when @code{error} is
5034 called and a forced stack unwind occurs) by ensuring that the
5035 @code{xfree} will always be called:
5038 struct cleanup *old = make_cleanup (null_cleanup, 0);
5039 data = xmalloc (sizeof blah);
5040 make_cleanup (xfree, data);
5045 The second style is try/except. Before it exits, your code-block calls
5046 @code{discard_cleanups} with the old cleanup chain and thus ensures that
5047 any created cleanups are not performed. For instance, the following
5048 code segment, ensures that the file will be closed but only if there is
5052 FILE *file = fopen ("afile", "r");
5053 struct cleanup *old = make_cleanup (close_file, file);
5055 discard_cleanups (old);
5059 Some functions, e.g., @code{fputs_filtered()} or @code{error()}, specify
5060 that they ``should not be called when cleanups are not in place''. This
5061 means that any actions you need to reverse in the case of an error or
5062 interruption must be on the cleanup chain before you call these
5063 functions, since they might never return to your code (they
5064 @samp{longjmp} instead).
5066 @section Per-architecture module data
5067 @cindex per-architecture module data
5068 @cindex multi-arch data
5069 @cindex data-pointer, per-architecture/per-module
5071 The multi-arch framework includes a mechanism for adding module
5072 specific per-architecture data-pointers to the @code{struct gdbarch}
5073 architecture object.
5075 A module registers one or more per-architecture data-pointers using:
5077 @deftypefun struct gdbarch_data *gdbarch_data_register_pre_init (gdbarch_data_pre_init_ftype *@var{pre_init})
5078 @var{pre_init} is used to, on-demand, allocate an initial value for a
5079 per-architecture data-pointer using the architecture's obstack (passed
5080 in as a parameter). Since @var{pre_init} can be called during
5081 architecture creation, it is not parameterized with the architecture.
5082 and must not call modules that use per-architecture data.
5085 @deftypefun struct gdbarch_data *gdbarch_data_register_post_init (gdbarch_data_post_init_ftype *@var{post_init})
5086 @var{post_init} is used to obtain an initial value for a
5087 per-architecture data-pointer @emph{after}. Since @var{post_init} is
5088 always called after architecture creation, it both receives the fully
5089 initialized architecture and is free to call modules that use
5090 per-architecture data (care needs to be taken to ensure that those
5091 other modules do not try to call back to this module as that will
5092 create in cycles in the initialization call graph).
5095 These functions return a @code{struct gdbarch_data} that is used to
5096 identify the per-architecture data-pointer added for that module.
5098 The per-architecture data-pointer is accessed using the function:
5100 @deftypefun void *gdbarch_data (struct gdbarch *@var{gdbarch}, struct gdbarch_data *@var{data_handle})
5101 Given the architecture @var{arch} and module data handle
5102 @var{data_handle} (returned by @code{gdbarch_data_register_pre_init}
5103 or @code{gdbarch_data_register_post_init}), this function returns the
5104 current value of the per-architecture data-pointer. If the data
5105 pointer is @code{NULL}, it is first initialized by calling the
5106 corresponding @var{pre_init} or @var{post_init} method.
5109 The examples below assume the following definitions:
5112 struct nozel @{ int total; @};
5113 static struct gdbarch_data *nozel_handle;
5116 A module can extend the architecture vector, adding additional
5117 per-architecture data, using the @var{pre_init} method. The module's
5118 per-architecture data is then initialized during architecture
5121 In the below, the module's per-architecture @emph{nozel} is added. An
5122 architecture can specify its nozel by calling @code{set_gdbarch_nozel}
5123 from @code{gdbarch_init}.
5127 nozel_pre_init (struct obstack *obstack)
5129 struct nozel *data = OBSTACK_ZALLOC (obstack, struct nozel);
5136 set_gdbarch_nozel (struct gdbarch *gdbarch, int total)
5138 struct nozel *data = gdbarch_data (gdbarch, nozel_handle);
5139 data->total = nozel;
5143 A module can on-demand create architecture dependant data structures
5144 using @code{post_init}.
5146 In the below, the nozel's total is computed on-demand by
5147 @code{nozel_post_init} using information obtained from the
5152 nozel_post_init (struct gdbarch *gdbarch)
5154 struct nozel *data = GDBARCH_OBSTACK_ZALLOC (gdbarch, struct nozel);
5155 nozel->total = gdbarch@dots{} (gdbarch);
5162 nozel_total (struct gdbarch *gdbarch)
5164 struct nozel *data = gdbarch_data (gdbarch, nozel_handle);
5169 @section Wrapping Output Lines
5170 @cindex line wrap in output
5173 Output that goes through @code{printf_filtered} or @code{fputs_filtered}
5174 or @code{fputs_demangled} needs only to have calls to @code{wrap_here}
5175 added in places that would be good breaking points. The utility
5176 routines will take care of actually wrapping if the line width is
5179 The argument to @code{wrap_here} is an indentation string which is
5180 printed @emph{only} if the line breaks there. This argument is saved
5181 away and used later. It must remain valid until the next call to
5182 @code{wrap_here} or until a newline has been printed through the
5183 @code{*_filtered} functions. Don't pass in a local variable and then
5186 It is usually best to call @code{wrap_here} after printing a comma or
5187 space. If you call it before printing a space, make sure that your
5188 indentation properly accounts for the leading space that will print if
5189 the line wraps there.
5191 Any function or set of functions that produce filtered output must
5192 finish by printing a newline, to flush the wrap buffer, before switching
5193 to unfiltered (@code{printf}) output. Symbol reading routines that
5194 print warnings are a good example.
5196 @section @value{GDBN} Coding Standards
5197 @cindex coding standards
5199 @value{GDBN} follows the GNU coding standards, as described in
5200 @file{etc/standards.texi}. This file is also available for anonymous
5201 FTP from GNU archive sites. @value{GDBN} takes a strict interpretation
5202 of the standard; in general, when the GNU standard recommends a practice
5203 but does not require it, @value{GDBN} requires it.
5205 @value{GDBN} follows an additional set of coding standards specific to
5206 @value{GDBN}, as described in the following sections.
5211 @value{GDBN} assumes an ISO/IEC 9899:1990 (a.k.a.@: ISO C90) compliant
5214 @value{GDBN} does not assume an ISO C or POSIX compliant C library.
5217 @subsection Memory Management
5219 @value{GDBN} does not use the functions @code{malloc}, @code{realloc},
5220 @code{calloc}, @code{free} and @code{asprintf}.
5222 @value{GDBN} uses the functions @code{xmalloc}, @code{xrealloc} and
5223 @code{xcalloc} when allocating memory. Unlike @code{malloc} et.al.@:
5224 these functions do not return when the memory pool is empty. Instead,
5225 they unwind the stack using cleanups. These functions return
5226 @code{NULL} when requested to allocate a chunk of memory of size zero.
5228 @emph{Pragmatics: By using these functions, the need to check every
5229 memory allocation is removed. These functions provide portable
5232 @value{GDBN} does not use the function @code{free}.
5234 @value{GDBN} uses the function @code{xfree} to return memory to the
5235 memory pool. Consistent with ISO-C, this function ignores a request to
5236 free a @code{NULL} pointer.
5238 @emph{Pragmatics: On some systems @code{free} fails when passed a
5239 @code{NULL} pointer.}
5241 @value{GDBN} can use the non-portable function @code{alloca} for the
5242 allocation of small temporary values (such as strings).
5244 @emph{Pragmatics: This function is very non-portable. Some systems
5245 restrict the memory being allocated to no more than a few kilobytes.}
5247 @value{GDBN} uses the string function @code{xstrdup} and the print
5248 function @code{xstrprintf}.
5250 @emph{Pragmatics: @code{asprintf} and @code{strdup} can fail. Print
5251 functions such as @code{sprintf} are very prone to buffer overflow
5255 @subsection Compiler Warnings
5256 @cindex compiler warnings
5258 With few exceptions, developers should avoid the configuration option
5259 @samp{--disable-werror} when building @value{GDBN}. The exceptions
5260 are listed in the file @file{gdb/MAINTAINERS}. The default, when
5261 building with @sc{gcc}, is @samp{--enable-werror}.
5263 This option causes @value{GDBN} (when built using GCC) to be compiled
5264 with a carefully selected list of compiler warning flags. Any warnings
5265 from those flags are treated as errors.
5267 The current list of warning flags includes:
5271 Recommended @sc{gcc} warnings.
5273 @item -Wdeclaration-after-statement
5275 @sc{gcc} 3.x (and later) and @sc{c99} allow declarations mixed with
5276 code, but @sc{gcc} 2.x and @sc{c89} do not.
5278 @item -Wpointer-arith
5280 @item -Wformat-nonliteral
5281 Non-literal format strings, with a few exceptions, are bugs - they
5282 might contain unintended user-supplied format specifiers.
5283 Since @value{GDBN} uses the @code{format printf} attribute on all
5284 @code{printf} like functions this checks not just @code{printf} calls
5285 but also calls to functions such as @code{fprintf_unfiltered}.
5287 @item -Wno-pointer-sign
5288 In version 4.0, GCC began warning about pointer argument passing or
5289 assignment even when the source and destination differed only in
5290 signedness. However, most @value{GDBN} code doesn't distinguish
5291 carefully between @code{char} and @code{unsigned char}. In early 2006
5292 the @value{GDBN} developers decided correcting these warnings wasn't
5293 worth the time it would take.
5295 @item -Wno-unused-parameter
5296 Due to the way that @value{GDBN} is implemented many functions have
5297 unused parameters. Consequently this warning is avoided. The macro
5298 @code{ATTRIBUTE_UNUSED} is not used as it leads to false negatives ---
5299 it is not an error to have @code{ATTRIBUTE_UNUSED} on a parameter that
5304 @itemx -Wno-char-subscripts
5305 These are warnings which might be useful for @value{GDBN}, but are
5306 currently too noisy to enable with @samp{-Werror}.
5310 @subsection Formatting
5312 @cindex source code formatting
5313 The standard GNU recommendations for formatting must be followed
5316 A function declaration should not have its name in column zero. A
5317 function definition should have its name in column zero.
5321 static void foo (void);
5329 @emph{Pragmatics: This simplifies scripting. Function definitions can
5330 be found using @samp{^function-name}.}
5332 There must be a space between a function or macro name and the opening
5333 parenthesis of its argument list (except for macro definitions, as
5334 required by C). There must not be a space after an open paren/bracket
5335 or before a close paren/bracket.
5337 While additional whitespace is generally helpful for reading, do not use
5338 more than one blank line to separate blocks, and avoid adding whitespace
5339 after the end of a program line (as of 1/99, some 600 lines had
5340 whitespace after the semicolon). Excess whitespace causes difficulties
5341 for @code{diff} and @code{patch} utilities.
5343 Pointers are declared using the traditional K&R C style:
5357 @subsection Comments
5359 @cindex comment formatting
5360 The standard GNU requirements on comments must be followed strictly.
5362 Block comments must appear in the following form, with no @code{/*}- or
5363 @code{*/}-only lines, and no leading @code{*}:
5366 /* Wait for control to return from inferior to debugger. If inferior
5367 gets a signal, we may decide to start it up again instead of
5368 returning. That is why there is a loop in this function. When
5369 this function actually returns it means the inferior should be left
5370 stopped and @value{GDBN} should read more commands. */
5373 (Note that this format is encouraged by Emacs; tabbing for a multi-line
5374 comment works correctly, and @kbd{M-q} fills the block consistently.)
5376 Put a blank line between the block comments preceding function or
5377 variable definitions, and the definition itself.
5379 In general, put function-body comments on lines by themselves, rather
5380 than trying to fit them into the 20 characters left at the end of a
5381 line, since either the comment or the code will inevitably get longer
5382 than will fit, and then somebody will have to move it anyhow.
5386 @cindex C data types
5387 Code must not depend on the sizes of C data types, the format of the
5388 host's floating point numbers, the alignment of anything, or the order
5389 of evaluation of expressions.
5391 @cindex function usage
5392 Use functions freely. There are only a handful of compute-bound areas
5393 in @value{GDBN} that might be affected by the overhead of a function
5394 call, mainly in symbol reading. Most of @value{GDBN}'s performance is
5395 limited by the target interface (whether serial line or system call).
5397 However, use functions with moderation. A thousand one-line functions
5398 are just as hard to understand as a single thousand-line function.
5400 @emph{Macros are bad, M'kay.}
5401 (But if you have to use a macro, make sure that the macro arguments are
5402 protected with parentheses.)
5406 Declarations like @samp{struct foo *} should be used in preference to
5407 declarations like @samp{typedef struct foo @{ @dots{} @} *foo_ptr}.
5410 @subsection Function Prototypes
5411 @cindex function prototypes
5413 Prototypes must be used when both @emph{declaring} and @emph{defining}
5414 a function. Prototypes for @value{GDBN} functions must include both the
5415 argument type and name, with the name matching that used in the actual
5416 function definition.
5418 All external functions should have a declaration in a header file that
5419 callers include, except for @code{_initialize_*} functions, which must
5420 be external so that @file{init.c} construction works, but shouldn't be
5421 visible to random source files.
5423 Where a source file needs a forward declaration of a static function,
5424 that declaration must appear in a block near the top of the source file.
5427 @subsection Internal Error Recovery
5429 During its execution, @value{GDBN} can encounter two types of errors.
5430 User errors and internal errors. User errors include not only a user
5431 entering an incorrect command but also problems arising from corrupt
5432 object files and system errors when interacting with the target.
5433 Internal errors include situations where @value{GDBN} has detected, at
5434 run time, a corrupt or erroneous situation.
5436 When reporting an internal error, @value{GDBN} uses
5437 @code{internal_error} and @code{gdb_assert}.
5439 @value{GDBN} must not call @code{abort} or @code{assert}.
5441 @emph{Pragmatics: There is no @code{internal_warning} function. Either
5442 the code detected a user error, recovered from it and issued a
5443 @code{warning} or the code failed to correctly recover from the user
5444 error and issued an @code{internal_error}.}
5446 @subsection File Names
5448 Any file used when building the core of @value{GDBN} must be in lower
5449 case. Any file used when building the core of @value{GDBN} must be 8.3
5450 unique. These requirements apply to both source and generated files.
5452 @emph{Pragmatics: The core of @value{GDBN} must be buildable on many
5453 platforms including DJGPP and MacOS/HFS. Every time an unfriendly file
5454 is introduced to the build process both @file{Makefile.in} and
5455 @file{configure.in} need to be modified accordingly. Compare the
5456 convoluted conversion process needed to transform @file{COPYING} into
5457 @file{copying.c} with the conversion needed to transform
5458 @file{version.in} into @file{version.c}.}
5460 Any file non 8.3 compliant file (that is not used when building the core
5461 of @value{GDBN}) must be added to @file{gdb/config/djgpp/fnchange.lst}.
5463 @emph{Pragmatics: This is clearly a compromise.}
5465 When @value{GDBN} has a local version of a system header file (ex
5466 @file{string.h}) the file name based on the POSIX header prefixed with
5467 @file{gdb_} (@file{gdb_string.h}). These headers should be relatively
5468 independent: they should use only macros defined by @file{configure},
5469 the compiler, or the host; they should include only system headers; they
5470 should refer only to system types. They may be shared between multiple
5471 programs, e.g.@: @value{GDBN} and @sc{gdbserver}.
5473 For other files @samp{-} is used as the separator.
5476 @subsection Include Files
5478 A @file{.c} file should include @file{defs.h} first.
5480 A @file{.c} file should directly include the @code{.h} file of every
5481 declaration and/or definition it directly refers to. It cannot rely on
5484 A @file{.h} file should directly include the @code{.h} file of every
5485 declaration and/or definition it directly refers to. It cannot rely on
5486 indirect inclusion. Exception: The file @file{defs.h} does not need to
5487 be directly included.
5489 An external declaration should only appear in one include file.
5491 An external declaration should never appear in a @code{.c} file.
5492 Exception: a declaration for the @code{_initialize} function that
5493 pacifies @option{-Wmissing-declaration}.
5495 A @code{typedef} definition should only appear in one include file.
5497 An opaque @code{struct} declaration can appear in multiple @file{.h}
5498 files. Where possible, a @file{.h} file should use an opaque
5499 @code{struct} declaration instead of an include.
5501 All @file{.h} files should be wrapped in:
5504 #ifndef INCLUDE_FILE_NAME_H
5505 #define INCLUDE_FILE_NAME_H
5511 @subsection Clean Design and Portable Implementation
5514 In addition to getting the syntax right, there's the little question of
5515 semantics. Some things are done in certain ways in @value{GDBN} because long
5516 experience has shown that the more obvious ways caused various kinds of
5519 @cindex assumptions about targets
5520 You can't assume the byte order of anything that comes from a target
5521 (including @var{value}s, object files, and instructions). Such things
5522 must be byte-swapped using @code{SWAP_TARGET_AND_HOST} in
5523 @value{GDBN}, or one of the swap routines defined in @file{bfd.h},
5524 such as @code{bfd_get_32}.
5526 You can't assume that you know what interface is being used to talk to
5527 the target system. All references to the target must go through the
5528 current @code{target_ops} vector.
5530 You can't assume that the host and target machines are the same machine
5531 (except in the ``native'' support modules). In particular, you can't
5532 assume that the target machine's header files will be available on the
5533 host machine. Target code must bring along its own header files --
5534 written from scratch or explicitly donated by their owner, to avoid
5538 Insertion of new @code{#ifdef}'s will be frowned upon. It's much better
5539 to write the code portably than to conditionalize it for various
5542 @cindex system dependencies
5543 New @code{#ifdef}'s which test for specific compilers or manufacturers
5544 or operating systems are unacceptable. All @code{#ifdef}'s should test
5545 for features. The information about which configurations contain which
5546 features should be segregated into the configuration files. Experience
5547 has proven far too often that a feature unique to one particular system
5548 often creeps into other systems; and that a conditional based on some
5549 predefined macro for your current system will become worthless over
5550 time, as new versions of your system come out that behave differently
5551 with regard to this feature.
5553 Adding code that handles specific architectures, operating systems,
5554 target interfaces, or hosts, is not acceptable in generic code.
5556 @cindex portable file name handling
5557 @cindex file names, portability
5558 One particularly notorious area where system dependencies tend to
5559 creep in is handling of file names. The mainline @value{GDBN} code
5560 assumes Posix semantics of file names: absolute file names begin with
5561 a forward slash @file{/}, slashes are used to separate leading
5562 directories, case-sensitive file names. These assumptions are not
5563 necessarily true on non-Posix systems such as MS-Windows. To avoid
5564 system-dependent code where you need to take apart or construct a file
5565 name, use the following portable macros:
5568 @findex HAVE_DOS_BASED_FILE_SYSTEM
5569 @item HAVE_DOS_BASED_FILE_SYSTEM
5570 This preprocessing symbol is defined to a non-zero value on hosts
5571 whose filesystems belong to the MS-DOS/MS-Windows family. Use this
5572 symbol to write conditional code which should only be compiled for
5575 @findex IS_DIR_SEPARATOR
5576 @item IS_DIR_SEPARATOR (@var{c})
5577 Evaluates to a non-zero value if @var{c} is a directory separator
5578 character. On Unix and GNU/Linux systems, only a slash @file{/} is
5579 such a character, but on Windows, both @file{/} and @file{\} will
5582 @findex IS_ABSOLUTE_PATH
5583 @item IS_ABSOLUTE_PATH (@var{file})
5584 Evaluates to a non-zero value if @var{file} is an absolute file name.
5585 For Unix and GNU/Linux hosts, a name which begins with a slash
5586 @file{/} is absolute. On DOS and Windows, @file{d:/foo} and
5587 @file{x:\bar} are also absolute file names.
5589 @findex FILENAME_CMP
5590 @item FILENAME_CMP (@var{f1}, @var{f2})
5591 Calls a function which compares file names @var{f1} and @var{f2} as
5592 appropriate for the underlying host filesystem. For Posix systems,
5593 this simply calls @code{strcmp}; on case-insensitive filesystems it
5594 will call @code{strcasecmp} instead.
5596 @findex DIRNAME_SEPARATOR
5597 @item DIRNAME_SEPARATOR
5598 Evaluates to a character which separates directories in
5599 @code{PATH}-style lists, typically held in environment variables.
5600 This character is @samp{:} on Unix, @samp{;} on DOS and Windows.
5602 @findex SLASH_STRING
5604 This evaluates to a constant string you should use to produce an
5605 absolute filename from leading directories and the file's basename.
5606 @code{SLASH_STRING} is @code{"/"} on most systems, but might be
5607 @code{"\\"} for some Windows-based ports.
5610 In addition to using these macros, be sure to use portable library
5611 functions whenever possible. For example, to extract a directory or a
5612 basename part from a file name, use the @code{dirname} and
5613 @code{basename} library functions (available in @code{libiberty} for
5614 platforms which don't provide them), instead of searching for a slash
5615 with @code{strrchr}.
5617 Another way to generalize @value{GDBN} along a particular interface is with an
5618 attribute struct. For example, @value{GDBN} has been generalized to handle
5619 multiple kinds of remote interfaces---not by @code{#ifdef}s everywhere, but
5620 by defining the @code{target_ops} structure and having a current target (as
5621 well as a stack of targets below it, for memory references). Whenever
5622 something needs to be done that depends on which remote interface we are
5623 using, a flag in the current target_ops structure is tested (e.g.,
5624 @code{target_has_stack}), or a function is called through a pointer in the
5625 current target_ops structure. In this way, when a new remote interface
5626 is added, only one module needs to be touched---the one that actually
5627 implements the new remote interface. Other examples of
5628 attribute-structs are BFD access to multiple kinds of object file
5629 formats, or @value{GDBN}'s access to multiple source languages.
5631 Please avoid duplicating code. For example, in @value{GDBN} 3.x all
5632 the code interfacing between @code{ptrace} and the rest of
5633 @value{GDBN} was duplicated in @file{*-dep.c}, and so changing
5634 something was very painful. In @value{GDBN} 4.x, these have all been
5635 consolidated into @file{infptrace.c}. @file{infptrace.c} can deal
5636 with variations between systems the same way any system-independent
5637 file would (hooks, @code{#if defined}, etc.), and machines which are
5638 radically different don't need to use @file{infptrace.c} at all.
5640 All debugging code must be controllable using the @samp{set debug
5641 @var{module}} command. Do not use @code{printf} to print trace
5642 messages. Use @code{fprintf_unfiltered(gdb_stdlog, ...}. Do not use
5643 @code{#ifdef DEBUG}.
5648 @chapter Porting @value{GDBN}
5649 @cindex porting to new machines
5651 Most of the work in making @value{GDBN} compile on a new machine is in
5652 specifying the configuration of the machine. This is done in a
5653 dizzying variety of header files and configuration scripts, which we
5654 hope to make more sensible soon. Let's say your new host is called an
5655 @var{xyz} (e.g., @samp{sun4}), and its full three-part configuration
5656 name is @code{@var{arch}-@var{xvend}-@var{xos}} (e.g.,
5657 @samp{sparc-sun-sunos4}). In particular:
5661 In the top level directory, edit @file{config.sub} and add @var{arch},
5662 @var{xvend}, and @var{xos} to the lists of supported architectures,
5663 vendors, and operating systems near the bottom of the file. Also, add
5664 @var{xyz} as an alias that maps to
5665 @code{@var{arch}-@var{xvend}-@var{xos}}. You can test your changes by
5669 ./config.sub @var{xyz}
5676 ./config.sub @code{@var{arch}-@var{xvend}-@var{xos}}
5680 which should both respond with @code{@var{arch}-@var{xvend}-@var{xos}}
5681 and no error messages.
5684 You need to port BFD, if that hasn't been done already. Porting BFD is
5685 beyond the scope of this manual.
5688 To configure @value{GDBN} itself, edit @file{gdb/configure.host} to recognize
5689 your system and set @code{gdb_host} to @var{xyz}, and (unless your
5690 desired target is already available) also edit @file{gdb/configure.tgt},
5691 setting @code{gdb_target} to something appropriate (for instance,
5694 @emph{Maintainer's note: Work in progress. The file
5695 @file{gdb/configure.host} originally needed to be modified when either a
5696 new native target or a new host machine was being added to @value{GDBN}.
5697 Recent changes have removed this requirement. The file now only needs
5698 to be modified when adding a new native configuration. This will likely
5699 changed again in the future.}
5702 Finally, you'll need to specify and define @value{GDBN}'s host-, native-, and
5703 target-dependent @file{.h} and @file{.c} files used for your
5707 @node Versions and Branches
5708 @chapter Versions and Branches
5712 @value{GDBN}'s version is determined by the file
5713 @file{gdb/version.in} and takes one of the following forms:
5716 @item @var{major}.@var{minor}
5717 @itemx @var{major}.@var{minor}.@var{patchlevel}
5718 an official release (e.g., 6.2 or 6.2.1)
5719 @item @var{major}.@var{minor}.@var{patchlevel}.@var{YYYY}@var{MM}@var{DD}
5720 a snapshot taken at @var{YYYY}-@var{MM}-@var{DD}-gmt (e.g.,
5721 6.1.50.20020302, 6.1.90.20020304, or 6.1.0.20020308)
5722 @item @var{major}.@var{minor}.@var{patchlevel}.@var{YYYY}@var{MM}@var{DD}-cvs
5723 a @sc{cvs} check out drawn on @var{YYYY}-@var{MM}-@var{DD} (e.g.,
5724 6.1.50.20020302-cvs, 6.1.90.20020304-cvs, or 6.1.0.20020308-cvs)
5725 @item @var{major}.@var{minor}.@var{patchlevel}.@var{YYYY}@var{MM}@var{DD} (@var{vendor})
5726 a vendor specific release of @value{GDBN}, that while based on@*
5727 @var{major}.@var{minor}.@var{patchlevel}.@var{YYYY}@var{MM}@var{DD},
5728 may include additional changes
5731 @value{GDBN}'s mainline uses the @var{major} and @var{minor} version
5732 numbers from the most recent release branch, with a @var{patchlevel}
5733 of 50. At the time each new release branch is created, the mainline's
5734 @var{major} and @var{minor} version numbers are updated.
5736 @value{GDBN}'s release branch is similar. When the branch is cut, the
5737 @var{patchlevel} is changed from 50 to 90. As draft releases are
5738 drawn from the branch, the @var{patchlevel} is incremented. Once the
5739 first release (@var{major}.@var{minor}) has been made, the
5740 @var{patchlevel} is set to 0 and updates have an incremented
5743 For snapshots, and @sc{cvs} check outs, it is also possible to
5744 identify the @sc{cvs} origin:
5747 @item @var{major}.@var{minor}.50.@var{YYYY}@var{MM}@var{DD}
5748 drawn from the @sc{head} of mainline @sc{cvs} (e.g., 6.1.50.20020302)
5749 @item @var{major}.@var{minor}.90.@var{YYYY}@var{MM}@var{DD}
5750 @itemx @var{major}.@var{minor}.91.@var{YYYY}@var{MM}@var{DD} @dots{}
5751 drawn from a release branch prior to the release (e.g.,
5753 @item @var{major}.@var{minor}.0.@var{YYYY}@var{MM}@var{DD}
5754 @itemx @var{major}.@var{minor}.1.@var{YYYY}@var{MM}@var{DD} @dots{}
5755 drawn from a release branch after the release (e.g., 6.2.0.20020308)
5758 If the previous @value{GDBN} version is 6.1 and the current version is
5759 6.2, then, substituting 6 for @var{major} and 1 or 2 for @var{minor},
5760 here's an illustration of a typical sequence:
5767 +--------------------------.
5770 6.2.50.20020303-cvs 6.1.90 (draft #1)
5772 6.2.50.20020304-cvs 6.1.90.20020304-cvs
5774 6.2.50.20020305-cvs 6.1.91 (draft #2)
5776 6.2.50.20020306-cvs 6.1.91.20020306-cvs
5778 6.2.50.20020307-cvs 6.2 (release)
5780 6.2.50.20020308-cvs 6.2.0.20020308-cvs
5782 6.2.50.20020309-cvs 6.2.1 (update)
5784 6.2.50.20020310-cvs <branch closed>
5788 +--------------------------.
5791 6.3.50.20020312-cvs 6.2.90 (draft #1)
5795 @section Release Branches
5796 @cindex Release Branches
5798 @value{GDBN} draws a release series (6.2, 6.2.1, @dots{}) from a
5799 single release branch, and identifies that branch using the @sc{cvs}
5803 gdb_@var{major}_@var{minor}-@var{YYYY}@var{MM}@var{DD}-branchpoint
5804 gdb_@var{major}_@var{minor}-branch
5805 gdb_@var{major}_@var{minor}-@var{YYYY}@var{MM}@var{DD}-release
5808 @emph{Pragmatics: To help identify the date at which a branch or
5809 release is made, both the branchpoint and release tags include the
5810 date that they are cut (@var{YYYY}@var{MM}@var{DD}) in the tag. The
5811 branch tag, denoting the head of the branch, does not need this.}
5813 @section Vendor Branches
5814 @cindex vendor branches
5816 To avoid version conflicts, vendors are expected to modify the file
5817 @file{gdb/version.in} to include a vendor unique alphabetic identifier
5818 (an official @value{GDBN} release never uses alphabetic characters in
5819 its version identifier). E.g., @samp{6.2widgit2}, or @samp{6.2 (Widgit
5822 @section Experimental Branches
5823 @cindex experimental branches
5825 @subsection Guidelines
5827 @value{GDBN} permits the creation of branches, cut from the @sc{cvs}
5828 repository, for experimental development. Branches make it possible
5829 for developers to share preliminary work, and maintainers to examine
5830 significant new developments.
5832 The following are a set of guidelines for creating such branches:
5836 @item a branch has an owner
5837 The owner can set further policy for a branch, but may not change the
5838 ground rules. In particular, they can set a policy for commits (be it
5839 adding more reviewers or deciding who can commit).
5841 @item all commits are posted
5842 All changes committed to a branch shall also be posted to
5843 @email{gdb-patches@@sources.redhat.com, the @value{GDBN} patches
5844 mailing list}. While commentary on such changes are encouraged, people
5845 should remember that the changes only apply to a branch.
5847 @item all commits are covered by an assignment
5848 This ensures that all changes belong to the Free Software Foundation,
5849 and avoids the possibility that the branch may become contaminated.
5851 @item a branch is focused
5852 A focused branch has a single objective or goal, and does not contain
5853 unnecessary or irrelevant changes. Cleanups, where identified, being
5854 be pushed into the mainline as soon as possible.
5856 @item a branch tracks mainline
5857 This keeps the level of divergence under control. It also keeps the
5858 pressure on developers to push cleanups and other stuff into the
5861 @item a branch shall contain the entire @value{GDBN} module
5862 The @value{GDBN} module @code{gdb} should be specified when creating a
5863 branch (branches of individual files should be avoided). @xref{Tags}.
5865 @item a branch shall be branded using @file{version.in}
5866 The file @file{gdb/version.in} shall be modified so that it identifies
5867 the branch @var{owner} and branch @var{name}, e.g.,
5868 @samp{6.2.50.20030303_owner_name} or @samp{6.2 (Owner Name)}.
5875 To simplify the identification of @value{GDBN} branches, the following
5876 branch tagging convention is strongly recommended:
5880 @item @var{owner}_@var{name}-@var{YYYYMMDD}-branchpoint
5881 @itemx @var{owner}_@var{name}-@var{YYYYMMDD}-branch
5882 The branch point and corresponding branch tag. @var{YYYYMMDD} is the
5883 date that the branch was created. A branch is created using the
5884 sequence: @anchor{experimental branch tags}
5886 cvs rtag @var{owner}_@var{name}-@var{YYYYMMDD}-branchpoint gdb
5887 cvs rtag -b -r @var{owner}_@var{name}-@var{YYYYMMDD}-branchpoint \
5888 @var{owner}_@var{name}-@var{YYYYMMDD}-branch gdb
5891 @item @var{owner}_@var{name}-@var{yyyymmdd}-mergepoint
5892 The tagged point, on the mainline, that was used when merging the branch
5893 on @var{yyyymmdd}. To merge in all changes since the branch was cut,
5894 use a command sequence like:
5896 cvs rtag @var{owner}_@var{name}-@var{yyyymmdd}-mergepoint gdb
5898 -j@var{owner}_@var{name}-@var{YYYYMMDD}-branchpoint
5899 -j@var{owner}_@var{name}-@var{yyyymmdd}-mergepoint
5902 Similar sequences can be used to just merge in changes since the last
5908 For further information on @sc{cvs}, see
5909 @uref{http://www.gnu.org/software/cvs/, Concurrent Versions System}.
5911 @node Start of New Year Procedure
5912 @chapter Start of New Year Procedure
5913 @cindex new year procedure
5915 At the start of each new year, the following actions should be performed:
5919 Rotate the ChangeLog file
5921 The current @file{ChangeLog} file should be renamed into
5922 @file{ChangeLog-YYYY} where YYYY is the year that has just passed.
5923 A new @file{ChangeLog} file should be created, and its contents should
5924 contain a reference to the previous ChangeLog. The following should
5925 also be preserved at the end of the new ChangeLog, in order to provide
5926 the appropriate settings when editing this file with Emacs:
5932 version-control: never
5938 Add an entry for the newly created ChangeLog file (@file{ChangeLog-YYYY})
5939 in @file{gdb/config/djgpp/fnchange.lst}.
5942 Update the copyright year in the startup message
5944 Update the copyright year in:
5946 @item file @file{top.c}, function @code{print_gdb_version}
5947 @item file @file{gdbserver/server.c}, function @code{gdbserver_version}
5948 @item file @file{gdbserver/gdbreplay.c}, function @code{gdbreplay_version}
5952 Add the new year in the copyright notices of all source and documentation
5953 files. This can be done semi-automatically by running the @code{copyright.sh}
5954 script. This script requires Emacs 22 or later to be installed.
5960 @chapter Releasing @value{GDBN}
5961 @cindex making a new release of gdb
5963 @section Branch Commit Policy
5965 The branch commit policy is pretty slack. @value{GDBN} releases 5.0,
5966 5.1 and 5.2 all used the below:
5970 The @file{gdb/MAINTAINERS} file still holds.
5972 Don't fix something on the branch unless/until it is also fixed in the
5973 trunk. If this isn't possible, mentioning it in the @file{gdb/PROBLEMS}
5974 file is better than committing a hack.
5976 When considering a patch for the branch, suggested criteria include:
5977 Does it fix a build? Does it fix the sequence @kbd{break main; run}
5978 when debugging a static binary?
5980 The further a change is from the core of @value{GDBN}, the less likely
5981 the change will worry anyone (e.g., target specific code).
5983 Only post a proposal to change the core of @value{GDBN} after you've
5984 sent individual bribes to all the people listed in the
5985 @file{MAINTAINERS} file @t{;-)}
5988 @emph{Pragmatics: Provided updates are restricted to non-core
5989 functionality there is little chance that a broken change will be fatal.
5990 This means that changes such as adding a new architectures or (within
5991 reason) support for a new host are considered acceptable.}
5994 @section Obsoleting code
5996 Before anything else, poke the other developers (and around the source
5997 code) to see if there is anything that can be removed from @value{GDBN}
5998 (an old target, an unused file).
6000 Obsolete code is identified by adding an @code{OBSOLETE} prefix to every
6001 line. Doing this means that it is easy to identify something that has
6002 been obsoleted when greping through the sources.
6004 The process is done in stages --- this is mainly to ensure that the
6005 wider @value{GDBN} community has a reasonable opportunity to respond.
6006 Remember, everything on the Internet takes a week.
6010 Post the proposal on @email{gdb@@sources.redhat.com, the GDB mailing
6011 list} Creating a bug report to track the task's state, is also highly
6016 Post the proposal on @email{gdb-announce@@sources.redhat.com, the GDB
6017 Announcement mailing list}.
6021 Go through and edit all relevant files and lines so that they are
6022 prefixed with the word @code{OBSOLETE}.
6024 Wait until the next GDB version, containing this obsolete code, has been
6027 Remove the obsolete code.
6031 @emph{Maintainer note: While removing old code is regrettable it is
6032 hopefully better for @value{GDBN}'s long term development. Firstly it
6033 helps the developers by removing code that is either no longer relevant
6034 or simply wrong. Secondly since it removes any history associated with
6035 the file (effectively clearing the slate) the developer has a much freer
6036 hand when it comes to fixing broken files.}
6040 @section Before the Branch
6042 The most important objective at this stage is to find and fix simple
6043 changes that become a pain to track once the branch is created. For
6044 instance, configuration problems that stop @value{GDBN} from even
6045 building. If you can't get the problem fixed, document it in the
6046 @file{gdb/PROBLEMS} file.
6048 @subheading Prompt for @file{gdb/NEWS}
6050 People always forget. Send a post reminding them but also if you know
6051 something interesting happened add it yourself. The @code{schedule}
6052 script will mention this in its e-mail.
6054 @subheading Review @file{gdb/README}
6056 Grab one of the nightly snapshots and then walk through the
6057 @file{gdb/README} looking for anything that can be improved. The
6058 @code{schedule} script will mention this in its e-mail.
6060 @subheading Refresh any imported files.
6062 A number of files are taken from external repositories. They include:
6066 @file{texinfo/texinfo.tex}
6068 @file{config.guess} et.@: al.@: (see the top-level @file{MAINTAINERS}
6071 @file{etc/standards.texi}, @file{etc/make-stds.texi}
6074 @subheading Check the ARI
6076 @uref{http://sources.redhat.com/gdb/ari,,A.R.I.} is an @code{awk} script
6077 (Awk Regression Index ;-) that checks for a number of errors and coding
6078 conventions. The checks include things like using @code{malloc} instead
6079 of @code{xmalloc} and file naming problems. There shouldn't be any
6082 @subsection Review the bug data base
6084 Close anything obviously fixed.
6086 @subsection Check all cross targets build
6088 The targets are listed in @file{gdb/MAINTAINERS}.
6091 @section Cut the Branch
6093 @subheading Create the branch
6098 $ V=`echo $v | sed 's/\./_/g'`
6099 $ D=`date -u +%Y-%m-%d`
6102 $ echo cvs -f -d :ext:sources.redhat.com:/cvs/src rtag \
6103 -D $D-gmt gdb_$V-$D-branchpoint insight
6104 cvs -f -d :ext:sources.redhat.com:/cvs/src rtag
6105 -D 2002-03-03-gmt gdb_5_2-2002-03-03-branchpoint insight
6108 $ echo cvs -f -d :ext:sources.redhat.com:/cvs/src rtag \
6109 -b -r gdb_$V-$D-branchpoint gdb_$V-branch insight
6110 cvs -f -d :ext:sources.redhat.com:/cvs/src rtag \
6111 -b -r gdb_5_2-2002-03-03-branchpoint gdb_5_2-branch insight
6119 By using @kbd{-D YYYY-MM-DD-gmt}, the branch is forced to an exact
6122 The trunk is first tagged so that the branch point can easily be found.
6124 Insight, which includes @value{GDBN}, is tagged at the same time.
6126 @file{version.in} gets bumped to avoid version number conflicts.
6128 The reading of @file{.cvsrc} is disabled using @file{-f}.
6131 @subheading Update @file{version.in}
6136 $ V=`echo $v | sed 's/\./_/g'`
6140 $ echo cvs -f -d :ext:sources.redhat.com:/cvs/src co \
6141 -r gdb_$V-branch src/gdb/version.in
6142 cvs -f -d :ext:sources.redhat.com:/cvs/src co
6143 -r gdb_5_2-branch src/gdb/version.in
6145 U src/gdb/version.in
6147 $ echo $u.90-0000-00-00-cvs > version.in
6149 5.1.90-0000-00-00-cvs
6150 $ cvs -f commit version.in
6155 @file{0000-00-00} is used as a date to pump prime the version.in update
6158 @file{.90} and the previous branch version are used as fairly arbitrary
6159 initial branch version number.
6163 @subheading Update the web and news pages
6167 @subheading Tweak cron to track the new branch
6169 The file @file{gdbadmin/cron/crontab} contains gdbadmin's cron table.
6170 This file needs to be updated so that:
6174 A daily timestamp is added to the file @file{version.in}.
6176 The new branch is included in the snapshot process.
6180 See the file @file{gdbadmin/cron/README} for how to install the updated
6183 The file @file{gdbadmin/ss/README} should also be reviewed to reflect
6184 any changes. That file is copied to both the branch/ and current/
6185 snapshot directories.
6188 @subheading Update the NEWS and README files
6190 The @file{NEWS} file needs to be updated so that on the branch it refers
6191 to @emph{changes in the current release} while on the trunk it also
6192 refers to @emph{changes since the current release}.
6194 The @file{README} file needs to be updated so that it refers to the
6197 @subheading Post the branch info
6199 Send an announcement to the mailing lists:
6203 @email{gdb-announce@@sources.redhat.com, GDB Announcement mailing list}
6205 @email{gdb@@sources.redhat.com, GDB Discussion mailing list} and
6206 @email{gdb-testers@@sources.redhat.com, GDB Testers mailing list}
6209 @emph{Pragmatics: The branch creation is sent to the announce list to
6210 ensure that people people not subscribed to the higher volume discussion
6213 The announcement should include:
6219 How to check out the branch using CVS.
6221 The date/number of weeks until the release.
6223 The branch commit policy still holds.
6226 @section Stabilize the branch
6228 Something goes here.
6230 @section Create a Release
6232 The process of creating and then making available a release is broken
6233 down into a number of stages. The first part addresses the technical
6234 process of creating a releasable tar ball. The later stages address the
6235 process of releasing that tar ball.
6237 When making a release candidate just the first section is needed.
6239 @subsection Create a release candidate
6241 The objective at this stage is to create a set of tar balls that can be
6242 made available as a formal release (or as a less formal release
6245 @subsubheading Freeze the branch
6247 Send out an e-mail notifying everyone that the branch is frozen to
6248 @email{gdb-patches@@sources.redhat.com}.
6250 @subsubheading Establish a few defaults.
6255 $ t=/sourceware/snapshot-tmp/gdbadmin-tmp
6257 /sourceware/snapshot-tmp/gdbadmin-tmp/gdb_5_2-branch/5.2
6261 /sourceware/snapshot-tmp/gdbadmin-tmp/gdb_5_2-branch/5.2
6263 /home/gdbadmin/bin/autoconf
6272 Check the @code{autoconf} version carefully. You want to be using the
6273 version taken from the @file{binutils} snapshot directory, which can be
6274 found at @uref{ftp://sources.redhat.com/pub/binutils/}. It is very
6275 unlikely that a system installed version of @code{autoconf} (e.g.,
6276 @file{/usr/bin/autoconf}) is correct.
6279 @subsubheading Check out the relevant modules:
6282 $ for m in gdb insight
6284 ( mkdir -p $m && cd $m && cvs -q -f -d /cvs/src co -P -r $b $m )
6294 The reading of @file{.cvsrc} is disabled (@file{-f}) so that there isn't
6295 any confusion between what is written here and what your local
6296 @code{cvs} really does.
6299 @subsubheading Update relevant files.
6305 Major releases get their comments added as part of the mainline. Minor
6306 releases should probably mention any significant bugs that were fixed.
6308 Don't forget to include the @file{ChangeLog} entry.
6311 $ emacs gdb/src/gdb/NEWS
6316 $ cp gdb/src/gdb/NEWS insight/src/gdb/NEWS
6317 $ cp gdb/src/gdb/ChangeLog insight/src/gdb/ChangeLog
6322 You'll need to update:
6334 $ emacs gdb/src/gdb/README
6339 $ cp gdb/src/gdb/README insight/src/gdb/README
6340 $ cp gdb/src/gdb/ChangeLog insight/src/gdb/ChangeLog
6343 @emph{Maintainer note: Hopefully the @file{README} file was reviewed
6344 before the initial branch was cut so just a simple substitute is needed
6347 @emph{Maintainer note: Other projects generate @file{README} and
6348 @file{INSTALL} from the core documentation. This might be worth
6351 @item gdb/version.in
6354 $ echo $v > gdb/src/gdb/version.in
6355 $ cat gdb/src/gdb/version.in
6357 $ emacs gdb/src/gdb/version.in
6360 ... Bump to version ...
6362 $ cp gdb/src/gdb/version.in insight/src/gdb/version.in
6363 $ cp gdb/src/gdb/ChangeLog insight/src/gdb/ChangeLog
6368 @subsubheading Do the dirty work
6370 This is identical to the process used to create the daily snapshot.
6373 $ for m in gdb insight
6375 ( cd $m/src && gmake -f src-release $m.tar )
6379 If the top level source directory does not have @file{src-release}
6380 (@value{GDBN} version 5.3.1 or earlier), try these commands instead:
6383 $ for m in gdb insight
6385 ( cd $m/src && gmake -f Makefile.in $m.tar )
6389 @subsubheading Check the source files
6391 You're looking for files that have mysteriously disappeared.
6392 @kbd{distclean} has the habit of deleting files it shouldn't. Watch out
6393 for the @file{version.in} update @kbd{cronjob}.
6396 $ ( cd gdb/src && cvs -f -q -n update )
6400 @dots{} lots of generated files @dots{}
6405 @dots{} lots of generated files @dots{}
6410 @emph{Don't worry about the @file{gdb.info-??} or
6411 @file{gdb/p-exp.tab.c}. They were generated (and yes @file{gdb.info-1}
6412 was also generated only something strange with CVS means that they
6413 didn't get suppressed). Fixing it would be nice though.}
6415 @subsubheading Create compressed versions of the release
6421 gdb/ gdb-5.2.tar insight/ insight-5.2.tar
6422 $ for m in gdb insight
6424 bzip2 -v -9 -c $m-$v.tar > $m-$v.tar.bz2
6425 gzip -v -9 -c $m-$v.tar > $m-$v.tar.gz
6435 A pipe such as @kbd{bunzip2 < xxx.bz2 | gzip -9 > xxx.gz} is not since,
6436 in that mode, @code{gzip} does not know the name of the file and, hence,
6437 can not include it in the compressed file. This is also why the release
6438 process runs @code{tar} and @code{bzip2} as separate passes.
6441 @subsection Sanity check the tar ball
6443 Pick a popular machine (Solaris/PPC?) and try the build on that.
6446 $ bunzip2 < gdb-5.2.tar.bz2 | tar xpf -
6451 $ ./gdb/gdb ./gdb/gdb
6455 Breakpoint 1 at 0x80732bc: file main.c, line 734.
6457 Starting program: /tmp/gdb-5.2/gdb/gdb
6459 Breakpoint 1, main (argc=1, argv=0xbffff8b4) at main.c:734
6460 734 catch_errors (captured_main, &args, "", RETURN_MASK_ALL);
6462 $1 = @{argc = 136426532, argv = 0x821b7f0@}
6466 @subsection Make a release candidate available
6468 If this is a release candidate then the only remaining steps are:
6472 Commit @file{version.in} and @file{ChangeLog}
6474 Tweak @file{version.in} (and @file{ChangeLog} to read
6475 @var{L}.@var{M}.@var{N}-0000-00-00-cvs so that the version update
6476 process can restart.
6478 Make the release candidate available in
6479 @uref{ftp://sources.redhat.com/pub/gdb/snapshots/branch}
6481 Notify the relevant mailing lists ( @email{gdb@@sources.redhat.com} and
6482 @email{gdb-testers@@sources.redhat.com} that the candidate is available.
6485 @subsection Make a formal release available
6487 (And you thought all that was required was to post an e-mail.)
6489 @subsubheading Install on sware
6491 Copy the new files to both the release and the old release directory:
6494 $ cp *.bz2 *.gz ~ftp/pub/gdb/old-releases/
6495 $ cp *.bz2 *.gz ~ftp/pub/gdb/releases
6499 Clean up the releases directory so that only the most recent releases
6500 are available (e.g. keep 5.2 and 5.2.1 but remove 5.1):
6503 $ cd ~ftp/pub/gdb/releases
6508 Update the file @file{README} and @file{.message} in the releases
6515 $ ln README .message
6518 @subsubheading Update the web pages.
6522 @item htdocs/download/ANNOUNCEMENT
6523 This file, which is posted as the official announcement, includes:
6526 General announcement.
6528 News. If making an @var{M}.@var{N}.1 release, retain the news from
6529 earlier @var{M}.@var{N} release.
6534 @item htdocs/index.html
6535 @itemx htdocs/news/index.html
6536 @itemx htdocs/download/index.html
6537 These files include:
6540 Announcement of the most recent release.
6542 News entry (remember to update both the top level and the news directory).
6544 These pages also need to be regenerate using @code{index.sh}.
6546 @item download/onlinedocs/
6547 You need to find the magic command that is used to generate the online
6548 docs from the @file{.tar.bz2}. The best way is to look in the output
6549 from one of the nightly @code{cron} jobs and then just edit accordingly.
6553 $ ~/ss/update-web-docs \
6554 ~ftp/pub/gdb/releases/gdb-5.2.tar.bz2 \
6556 /www/sourceware/htdocs/gdb/download/onlinedocs \
6561 Just like the online documentation. Something like:
6564 $ /bin/sh ~/ss/update-web-ari \
6565 ~ftp/pub/gdb/releases/gdb-5.2.tar.bz2 \
6567 /www/sourceware/htdocs/gdb/download/ari \
6573 @subsubheading Shadow the pages onto gnu
6575 Something goes here.
6578 @subsubheading Install the @value{GDBN} tar ball on GNU
6580 At the time of writing, the GNU machine was @kbd{gnudist.gnu.org} in
6581 @file{~ftp/gnu/gdb}.
6583 @subsubheading Make the @file{ANNOUNCEMENT}
6585 Post the @file{ANNOUNCEMENT} file you created above to:
6589 @email{gdb-announce@@sources.redhat.com, GDB Announcement mailing list}
6591 @email{info-gnu@@gnu.org, General GNU Announcement list} (but delay it a
6592 day or so to let things get out)
6594 @email{bug-gdb@@gnu.org, GDB Bug Report mailing list}
6599 The release is out but you're still not finished.
6601 @subsubheading Commit outstanding changes
6603 In particular you'll need to commit any changes to:
6607 @file{gdb/ChangeLog}
6609 @file{gdb/version.in}
6616 @subsubheading Tag the release
6621 $ d=`date -u +%Y-%m-%d`
6624 $ ( cd insight/src/gdb && cvs -f -q update )
6625 $ ( cd insight/src && cvs -f -q tag gdb_5_2-$d-release )
6628 Insight is used since that contains more of the release than
6631 @subsubheading Mention the release on the trunk
6633 Just put something in the @file{ChangeLog} so that the trunk also
6634 indicates when the release was made.
6636 @subsubheading Restart @file{gdb/version.in}
6638 If @file{gdb/version.in} does not contain an ISO date such as
6639 @kbd{2002-01-24} then the daily @code{cronjob} won't update it. Having
6640 committed all the release changes it can be set to
6641 @file{5.2.0_0000-00-00-cvs} which will restart things (yes the @kbd{_}
6642 is important - it affects the snapshot process).
6644 Don't forget the @file{ChangeLog}.
6646 @subsubheading Merge into trunk
6648 The files committed to the branch may also need changes merged into the
6651 @subsubheading Revise the release schedule
6653 Post a revised release schedule to @email{gdb@@sources.redhat.com, GDB
6654 Discussion List} with an updated announcement. The schedule can be
6655 generated by running:
6658 $ ~/ss/schedule `date +%s` schedule
6662 The first parameter is approximate date/time in seconds (from the epoch)
6663 of the most recent release.
6665 Also update the schedule @code{cronjob}.
6667 @section Post release
6669 Remove any @code{OBSOLETE} code.
6676 The testsuite is an important component of the @value{GDBN} package.
6677 While it is always worthwhile to encourage user testing, in practice
6678 this is rarely sufficient; users typically use only a small subset of
6679 the available commands, and it has proven all too common for a change
6680 to cause a significant regression that went unnoticed for some time.
6682 The @value{GDBN} testsuite uses the DejaGNU testing framework. The
6683 tests themselves are calls to various @code{Tcl} procs; the framework
6684 runs all the procs and summarizes the passes and fails.
6686 @section Using the Testsuite
6688 @cindex running the test suite
6689 To run the testsuite, simply go to the @value{GDBN} object directory (or to the
6690 testsuite's objdir) and type @code{make check}. This just sets up some
6691 environment variables and invokes DejaGNU's @code{runtest} script. While
6692 the testsuite is running, you'll get mentions of which test file is in use,
6693 and a mention of any unexpected passes or fails. When the testsuite is
6694 finished, you'll get a summary that looks like this:
6699 # of expected passes 6016
6700 # of unexpected failures 58
6701 # of unexpected successes 5
6702 # of expected failures 183
6703 # of unresolved testcases 3
6704 # of untested testcases 5
6707 To run a specific test script, type:
6709 make check RUNTESTFLAGS='@var{tests}'
6711 where @var{tests} is a list of test script file names, separated by
6714 The ideal test run consists of expected passes only; however, reality
6715 conspires to keep us from this ideal. Unexpected failures indicate
6716 real problems, whether in @value{GDBN} or in the testsuite. Expected
6717 failures are still failures, but ones which have been decided are too
6718 hard to deal with at the time; for instance, a test case might work
6719 everywhere except on AIX, and there is no prospect of the AIX case
6720 being fixed in the near future. Expected failures should not be added
6721 lightly, since you may be masking serious bugs in @value{GDBN}.
6722 Unexpected successes are expected fails that are passing for some
6723 reason, while unresolved and untested cases often indicate some minor
6724 catastrophe, such as the compiler being unable to deal with a test
6727 When making any significant change to @value{GDBN}, you should run the
6728 testsuite before and after the change, to confirm that there are no
6729 regressions. Note that truly complete testing would require that you
6730 run the testsuite with all supported configurations and a variety of
6731 compilers; however this is more than really necessary. In many cases
6732 testing with a single configuration is sufficient. Other useful
6733 options are to test one big-endian (Sparc) and one little-endian (x86)
6734 host, a cross config with a builtin simulator (powerpc-eabi,
6735 mips-elf), or a 64-bit host (Alpha).
6737 If you add new functionality to @value{GDBN}, please consider adding
6738 tests for it as well; this way future @value{GDBN} hackers can detect
6739 and fix their changes that break the functionality you added.
6740 Similarly, if you fix a bug that was not previously reported as a test
6741 failure, please add a test case for it. Some cases are extremely
6742 difficult to test, such as code that handles host OS failures or bugs
6743 in particular versions of compilers, and it's OK not to try to write
6744 tests for all of those.
6746 DejaGNU supports separate build, host, and target machines. However,
6747 some @value{GDBN} test scripts do not work if the build machine and
6748 the host machine are not the same. In such an environment, these scripts
6749 will give a result of ``UNRESOLVED'', like this:
6752 UNRESOLVED: gdb.base/example.exp: This test script does not work on a remote host.
6755 @section Testsuite Organization
6757 @cindex test suite organization
6758 The testsuite is entirely contained in @file{gdb/testsuite}. While the
6759 testsuite includes some makefiles and configury, these are very minimal,
6760 and used for little besides cleaning up, since the tests themselves
6761 handle the compilation of the programs that @value{GDBN} will run. The file
6762 @file{testsuite/lib/gdb.exp} contains common utility procs useful for
6763 all @value{GDBN} tests, while the directory @file{testsuite/config} contains
6764 configuration-specific files, typically used for special-purpose
6765 definitions of procs like @code{gdb_load} and @code{gdb_start}.
6767 The tests themselves are to be found in @file{testsuite/gdb.*} and
6768 subdirectories of those. The names of the test files must always end
6769 with @file{.exp}. DejaGNU collects the test files by wildcarding
6770 in the test directories, so both subdirectories and individual files
6771 get chosen and run in alphabetical order.
6773 The following table lists the main types of subdirectories and what they
6774 are for. Since DejaGNU finds test files no matter where they are
6775 located, and since each test file sets up its own compilation and
6776 execution environment, this organization is simply for convenience and
6781 This is the base testsuite. The tests in it should apply to all
6782 configurations of @value{GDBN} (but generic native-only tests may live here).
6783 The test programs should be in the subset of C that is valid K&R,
6784 ANSI/ISO, and C@t{++} (@code{#ifdef}s are allowed if necessary, for instance
6787 @item gdb.@var{lang}
6788 Language-specific tests for any language @var{lang} besides C. Examples are
6789 @file{gdb.cp} and @file{gdb.java}.
6791 @item gdb.@var{platform}
6792 Non-portable tests. The tests are specific to a specific configuration
6793 (host or target), such as HP-UX or eCos. Example is @file{gdb.hp}, for
6796 @item gdb.@var{compiler}
6797 Tests specific to a particular compiler. As of this writing (June
6798 1999), there aren't currently any groups of tests in this category that
6799 couldn't just as sensibly be made platform-specific, but one could
6800 imagine a @file{gdb.gcc}, for tests of @value{GDBN}'s handling of GCC
6803 @item gdb.@var{subsystem}
6804 Tests that exercise a specific @value{GDBN} subsystem in more depth. For
6805 instance, @file{gdb.disasm} exercises various disassemblers, while
6806 @file{gdb.stabs} tests pathways through the stabs symbol reader.
6809 @section Writing Tests
6810 @cindex writing tests
6812 In many areas, the @value{GDBN} tests are already quite comprehensive; you
6813 should be able to copy existing tests to handle new cases.
6815 You should try to use @code{gdb_test} whenever possible, since it
6816 includes cases to handle all the unexpected errors that might happen.
6817 However, it doesn't cost anything to add new test procedures; for
6818 instance, @file{gdb.base/exprs.exp} defines a @code{test_expr} that
6819 calls @code{gdb_test} multiple times.
6821 Only use @code{send_gdb} and @code{gdb_expect} when absolutely
6822 necessary. Even if @value{GDBN} has several valid responses to
6823 a command, you can use @code{gdb_test_multiple}. Like @code{gdb_test},
6824 @code{gdb_test_multiple} recognizes internal errors and unexpected
6827 Do not write tests which expect a literal tab character from @value{GDBN}.
6828 On some operating systems (e.g.@: OpenBSD) the TTY layer expands tabs to
6829 spaces, so by the time @value{GDBN}'s output reaches expect the tab is gone.
6831 The source language programs do @emph{not} need to be in a consistent
6832 style. Since @value{GDBN} is used to debug programs written in many different
6833 styles, it's worth having a mix of styles in the testsuite; for
6834 instance, some @value{GDBN} bugs involving the display of source lines would
6835 never manifest themselves if the programs used GNU coding style
6842 Check the @file{README} file, it often has useful information that does not
6843 appear anywhere else in the directory.
6846 * Getting Started:: Getting started working on @value{GDBN}
6847 * Debugging GDB:: Debugging @value{GDBN} with itself
6850 @node Getting Started,,, Hints
6852 @section Getting Started
6854 @value{GDBN} is a large and complicated program, and if you first starting to
6855 work on it, it can be hard to know where to start. Fortunately, if you
6856 know how to go about it, there are ways to figure out what is going on.
6858 This manual, the @value{GDBN} Internals manual, has information which applies
6859 generally to many parts of @value{GDBN}.
6861 Information about particular functions or data structures are located in
6862 comments with those functions or data structures. If you run across a
6863 function or a global variable which does not have a comment correctly
6864 explaining what is does, this can be thought of as a bug in @value{GDBN}; feel
6865 free to submit a bug report, with a suggested comment if you can figure
6866 out what the comment should say. If you find a comment which is
6867 actually wrong, be especially sure to report that.
6869 Comments explaining the function of macros defined in host, target, or
6870 native dependent files can be in several places. Sometimes they are
6871 repeated every place the macro is defined. Sometimes they are where the
6872 macro is used. Sometimes there is a header file which supplies a
6873 default definition of the macro, and the comment is there. This manual
6874 also documents all the available macros.
6875 @c (@pxref{Host Conditionals}, @pxref{Target
6876 @c Conditionals}, @pxref{Native Conditionals}, and @pxref{Obsolete
6879 Start with the header files. Once you have some idea of how
6880 @value{GDBN}'s internal symbol tables are stored (see @file{symtab.h},
6881 @file{gdbtypes.h}), you will find it much easier to understand the
6882 code which uses and creates those symbol tables.
6884 You may wish to process the information you are getting somehow, to
6885 enhance your understanding of it. Summarize it, translate it to another
6886 language, add some (perhaps trivial or non-useful) feature to @value{GDBN}, use
6887 the code to predict what a test case would do and write the test case
6888 and verify your prediction, etc. If you are reading code and your eyes
6889 are starting to glaze over, this is a sign you need to use a more active
6892 Once you have a part of @value{GDBN} to start with, you can find more
6893 specifically the part you are looking for by stepping through each
6894 function with the @code{next} command. Do not use @code{step} or you
6895 will quickly get distracted; when the function you are stepping through
6896 calls another function try only to get a big-picture understanding
6897 (perhaps using the comment at the beginning of the function being
6898 called) of what it does. This way you can identify which of the
6899 functions being called by the function you are stepping through is the
6900 one which you are interested in. You may need to examine the data
6901 structures generated at each stage, with reference to the comments in
6902 the header files explaining what the data structures are supposed to
6905 Of course, this same technique can be used if you are just reading the
6906 code, rather than actually stepping through it. The same general
6907 principle applies---when the code you are looking at calls something
6908 else, just try to understand generally what the code being called does,
6909 rather than worrying about all its details.
6911 @cindex command implementation
6912 A good place to start when tracking down some particular area is with
6913 a command which invokes that feature. Suppose you want to know how
6914 single-stepping works. As a @value{GDBN} user, you know that the
6915 @code{step} command invokes single-stepping. The command is invoked
6916 via command tables (see @file{command.h}); by convention the function
6917 which actually performs the command is formed by taking the name of
6918 the command and adding @samp{_command}, or in the case of an
6919 @code{info} subcommand, @samp{_info}. For example, the @code{step}
6920 command invokes the @code{step_command} function and the @code{info
6921 display} command invokes @code{display_info}. When this convention is
6922 not followed, you might have to use @code{grep} or @kbd{M-x
6923 tags-search} in emacs, or run @value{GDBN} on itself and set a
6924 breakpoint in @code{execute_command}.
6926 @cindex @code{bug-gdb} mailing list
6927 If all of the above fail, it may be appropriate to ask for information
6928 on @code{bug-gdb}. But @emph{never} post a generic question like ``I was
6929 wondering if anyone could give me some tips about understanding
6930 @value{GDBN}''---if we had some magic secret we would put it in this manual.
6931 Suggestions for improving the manual are always welcome, of course.
6933 @node Debugging GDB,,,Hints
6935 @section Debugging @value{GDBN} with itself
6936 @cindex debugging @value{GDBN}
6938 If @value{GDBN} is limping on your machine, this is the preferred way to get it
6939 fully functional. Be warned that in some ancient Unix systems, like
6940 Ultrix 4.2, a program can't be running in one process while it is being
6941 debugged in another. Rather than typing the command @kbd{@w{./gdb
6942 ./gdb}}, which works on Suns and such, you can copy @file{gdb} to
6943 @file{gdb2} and then type @kbd{@w{./gdb ./gdb2}}.
6945 When you run @value{GDBN} in the @value{GDBN} source directory, it will read a
6946 @file{.gdbinit} file that sets up some simple things to make debugging
6947 gdb easier. The @code{info} command, when executed without a subcommand
6948 in a @value{GDBN} being debugged by gdb, will pop you back up to the top level
6949 gdb. See @file{.gdbinit} for details.
6951 If you use emacs, you will probably want to do a @code{make TAGS} after
6952 you configure your distribution; this will put the machine dependent
6953 routines for your local machine where they will be accessed first by
6956 Also, make sure that you've either compiled @value{GDBN} with your local cc, or
6957 have run @code{fixincludes} if you are compiling with gcc.
6959 @section Submitting Patches
6961 @cindex submitting patches
6962 Thanks for thinking of offering your changes back to the community of
6963 @value{GDBN} users. In general we like to get well designed enhancements.
6964 Thanks also for checking in advance about the best way to transfer the
6967 The @value{GDBN} maintainers will only install ``cleanly designed'' patches.
6968 This manual summarizes what we believe to be clean design for @value{GDBN}.
6970 If the maintainers don't have time to put the patch in when it arrives,
6971 or if there is any question about a patch, it goes into a large queue
6972 with everyone else's patches and bug reports.
6974 @cindex legal papers for code contributions
6975 The legal issue is that to incorporate substantial changes requires a
6976 copyright assignment from you and/or your employer, granting ownership
6977 of the changes to the Free Software Foundation. You can get the
6978 standard documents for doing this by sending mail to @code{gnu@@gnu.org}
6979 and asking for it. We recommend that people write in "All programs
6980 owned by the Free Software Foundation" as "NAME OF PROGRAM", so that
6981 changes in many programs (not just @value{GDBN}, but GAS, Emacs, GCC,
6983 contributed with only one piece of legalese pushed through the
6984 bureaucracy and filed with the FSF. We can't start merging changes until
6985 this paperwork is received by the FSF (their rules, which we follow
6986 since we maintain it for them).
6988 Technically, the easiest way to receive changes is to receive each
6989 feature as a small context diff or unidiff, suitable for @code{patch}.
6990 Each message sent to me should include the changes to C code and
6991 header files for a single feature, plus @file{ChangeLog} entries for
6992 each directory where files were modified, and diffs for any changes
6993 needed to the manuals (@file{gdb/doc/gdb.texinfo} or
6994 @file{gdb/doc/gdbint.texinfo}). If there are a lot of changes for a
6995 single feature, they can be split down into multiple messages.
6997 In this way, if we read and like the feature, we can add it to the
6998 sources with a single patch command, do some testing, and check it in.
6999 If you leave out the @file{ChangeLog}, we have to write one. If you leave
7000 out the doc, we have to puzzle out what needs documenting. Etc., etc.
7002 The reason to send each change in a separate message is that we will not
7003 install some of the changes. They'll be returned to you with questions
7004 or comments. If we're doing our job correctly, the message back to you
7005 will say what you have to fix in order to make the change acceptable.
7006 The reason to have separate messages for separate features is so that
7007 the acceptable changes can be installed while one or more changes are
7008 being reworked. If multiple features are sent in a single message, we
7009 tend to not put in the effort to sort out the acceptable changes from
7010 the unacceptable, so none of the features get installed until all are
7013 If this sounds painful or authoritarian, well, it is. But we get a lot
7014 of bug reports and a lot of patches, and many of them don't get
7015 installed because we don't have the time to finish the job that the bug
7016 reporter or the contributor could have done. Patches that arrive
7017 complete, working, and well designed, tend to get installed on the day
7018 they arrive. The others go into a queue and get installed as time
7019 permits, which, since the maintainers have many demands to meet, may not
7020 be for quite some time.
7022 Please send patches directly to
7023 @email{gdb-patches@@sources.redhat.com, the @value{GDBN} maintainers}.
7025 @section Build Script
7027 @cindex build script
7029 The script @file{gdb_buildall.sh} builds @value{GDBN} with flag
7030 @option{--enable-targets=all} set. This builds @value{GDBN} with all supported
7031 targets activated. This helps testing @value{GDBN} when doing changes that
7032 affect more than one architecture and is much faster than using
7033 @file{gdb_mbuild.sh}.
7035 After building @value{GDBN} the script checks which architectures are
7036 supported and then switches the current architecture to each of those to get
7037 information about the architecture. The test results are stored in log files
7038 in the directory the script was called from.
7040 @include observer.texi