1 \input texinfo @c -*- texinfo -*-
2 @setfilename gdbint.info
4 @settitle @value{GDBN} Internals
6 @dircategory Software development
8 * Gdb-Internals: (gdbint). The GNU debugger's internals.
12 Copyright @copyright{} 1990, 1991, 1992, 1993, 1994, 1996, 1998, 1999,
13 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2008, 2009, 2010, 2011
14 Free Software Foundation, Inc.
15 Contributed by Cygnus Solutions. Written by John Gilmore.
16 Second Edition by Stan Shebs.
18 Permission is granted to copy, distribute and/or modify this document
19 under the terms of the GNU Free Documentation License, Version 1.3 or
20 any later version published by the Free Software Foundation; with no
21 Invariant Sections, with no Front-Cover Texts, and with no Back-Cover
22 Texts. A copy of the license is included in the section entitled ``GNU
23 Free Documentation License''.
27 This file documents the internals of the GNU debugger @value{GDBN}.
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::
89 * Versions and Branches::
90 * Start of New Year Procedure::
95 * GDB Observers:: @value{GDBN} Currently available observers
96 * GNU Free Documentation License:: The license for this documentation
109 @section Requirements
110 @cindex requirements for @value{GDBN}
112 Before diving into the internals, you should understand the formal
113 requirements and other expectations for @value{GDBN}. Although some
114 of these may seem obvious, there have been proposals for @value{GDBN}
115 that have run counter to these requirements.
117 First of all, @value{GDBN} is a debugger. It's not designed to be a
118 front panel for embedded systems. It's not a text editor. It's not a
119 shell. It's not a programming environment.
121 @value{GDBN} is an interactive tool. Although a batch mode is
122 available, @value{GDBN}'s primary role is to interact with a human
125 @value{GDBN} should be responsive to the user. A programmer hot on
126 the trail of a nasty bug, and operating under a looming deadline, is
127 going to be very impatient of everything, including the response time
128 to debugger commands.
130 @value{GDBN} should be relatively permissive, such as for expressions.
131 While the compiler should be picky (or have the option to be made
132 picky), since source code lives for a long time usually, the
133 programmer doing debugging shouldn't be spending time figuring out to
134 mollify the debugger.
136 @value{GDBN} will be called upon to deal with really large programs.
137 Executable sizes of 50 to 100 megabytes occur regularly, and we've
138 heard reports of programs approaching 1 gigabyte in size.
140 @value{GDBN} should be able to run everywhere. No other debugger is
141 available for even half as many configurations as @value{GDBN}
145 @section Contributors
147 The first edition of this document was written by John Gilmore of
148 Cygnus Solutions. The current second edition was written by Stan Shebs
149 of Cygnus Solutions, who continues to update the manual.
151 Over the years, many others have made additions and changes to this
152 document. This section attempts to record the significant contributors
153 to that effort. One of the virtues of free software is that everyone
154 is free to contribute to it; with regret, we cannot actually
155 acknowledge everyone here.
158 @emph{Plea:} This section has only been added relatively recently (four
159 years after publication of the second edition). Additions to this
160 section are particularly welcome. If you or your friends (or enemies,
161 to be evenhanded) have been unfairly omitted from this list, we would
162 like to add your names!
165 A document such as this relies on being kept up to date by numerous
166 small updates by contributing engineers as they make changes to the
167 code base. The file @file{ChangeLog} in the @value{GDBN} distribution
168 approximates a blow-by-blow account. The most prolific contributors to
169 this important, but low profile task are Andrew Cagney (responsible
170 for over half the entries), Daniel Jacobowitz, Mark Kettenis, Jim
171 Blandy and Eli Zaretskii.
173 Eli Zaretskii and Daniel Jacobowitz wrote the sections documenting
176 Jeremy Bennett updated the sections on initializing a new architecture
177 and register representation, and added the section on Frame Interpretation.
180 @node Overall Structure
182 @chapter Overall Structure
184 @value{GDBN} consists of three major subsystems: user interface,
185 symbol handling (the @dfn{symbol side}), and target system handling (the
188 The user interface consists of several actual interfaces, plus
191 The symbol side consists of object file readers, debugging info
192 interpreters, symbol table management, source language expression
193 parsing, type and value printing.
195 The target side consists of execution control, stack frame analysis, and
196 physical target manipulation.
198 The target side/symbol side division is not formal, and there are a
199 number of exceptions. For instance, core file support involves symbolic
200 elements (the basic core file reader is in BFD) and target elements (it
201 supplies the contents of memory and the values of registers). Instead,
202 this division is useful for understanding how the minor subsystems
205 @section The Symbol Side
207 The symbolic side of @value{GDBN} can be thought of as ``everything
208 you can do in @value{GDBN} without having a live program running''.
209 For instance, you can look at the types of variables, and evaluate
210 many kinds of expressions.
212 @section The Target Side
214 The target side of @value{GDBN} is the ``bits and bytes manipulator''.
215 Although it may make reference to symbolic info here and there, most
216 of the target side will run with only a stripped executable
217 available---or even no executable at all, in remote debugging cases.
219 Operations such as disassembly, stack frame crawls, and register
220 display, are able to work with no symbolic info at all. In some cases,
221 such as disassembly, @value{GDBN} will use symbolic info to present addresses
222 relative to symbols rather than as raw numbers, but it will work either
225 @section Configurations
229 @dfn{Host} refers to attributes of the system where @value{GDBN} runs.
230 @dfn{Target} refers to the system where the program being debugged
231 executes. In most cases they are the same machine, in which case a
232 third type of @dfn{Native} attributes come into play.
234 Defines and include files needed to build on the host are host
235 support. Examples are tty support, system defined types, host byte
236 order, host float format. These are all calculated by @code{autoconf}
237 when the debugger is built.
239 Defines and information needed to handle the target format are target
240 dependent. Examples are the stack frame format, instruction set,
241 breakpoint instruction, registers, and how to set up and tear down the stack
244 Information that is only needed when the host and target are the same,
245 is native dependent. One example is Unix child process support; if the
246 host and target are not the same, calling @code{fork} to start the target
247 process is a bad idea. The various macros needed for finding the
248 registers in the @code{upage}, running @code{ptrace}, and such are all
249 in the native-dependent files.
251 Another example of native-dependent code is support for features that
252 are really part of the target environment, but which require
253 @code{#include} files that are only available on the host system. Core
254 file handling and @code{setjmp} handling are two common cases.
256 When you want to make @value{GDBN} work as the traditional native debugger
257 on a system, you will need to supply both target and native information.
259 @section Source Tree Structure
260 @cindex @value{GDBN} source tree structure
262 The @value{GDBN} source directory has a mostly flat structure---there
263 are only a few subdirectories. A file's name usually gives a hint as
264 to what it does; for example, @file{stabsread.c} reads stabs,
265 @file{dwarf2read.c} reads @sc{DWARF 2}, etc.
267 Files that are related to some common task have names that share
268 common substrings. For example, @file{*-thread.c} files deal with
269 debugging threads on various platforms; @file{*read.c} files deal with
270 reading various kinds of symbol and object files; @file{inf*.c} files
271 deal with direct control of the @dfn{inferior program} (@value{GDBN}
272 parlance for the program being debugged).
274 There are several dozens of files in the @file{*-tdep.c} family.
275 @samp{tdep} stands for @dfn{target-dependent code}---each of these
276 files implements debug support for a specific target architecture
277 (sparc, mips, etc). Usually, only one of these will be used in a
278 specific @value{GDBN} configuration (sometimes two, closely related).
280 Similarly, there are many @file{*-nat.c} files, each one for native
281 debugging on a specific system (e.g., @file{sparc-linux-nat.c} is for
282 native debugging of Sparc machines running the Linux kernel).
284 The few subdirectories of the source tree are:
288 Code that implements @dfn{CLI}, the @value{GDBN} Command-Line
289 Interpreter. @xref{User Interface, Command Interpreter}.
292 Code for the @value{GDBN} remote server.
295 Code for Insight, the @value{GDBN} TK-based GUI front-end.
298 The @dfn{GDB/MI}, the @value{GDBN} Machine Interface interpreter.
301 Target signal translation code.
304 Code for @dfn{TUI}, the @value{GDBN} Text-mode full-screen User
305 Interface. @xref{User Interface, TUI}.
313 @value{GDBN} uses a number of debugging-specific algorithms. They are
314 often not very complicated, but get lost in the thicket of special
315 cases and real-world issues. This chapter describes the basic
316 algorithms and mentions some of the specific target definitions that
319 @section Prologue Analysis
321 @cindex prologue analysis
322 @cindex call frame information
323 @cindex CFI (call frame information)
324 To produce a backtrace and allow the user to manipulate older frames'
325 variables and arguments, @value{GDBN} needs to find the base addresses
326 of older frames, and discover where those frames' registers have been
327 saved. Since a frame's ``callee-saves'' registers get saved by
328 younger frames if and when they're reused, a frame's registers may be
329 scattered unpredictably across younger frames. This means that
330 changing the value of a register-allocated variable in an older frame
331 may actually entail writing to a save slot in some younger frame.
333 Modern versions of GCC emit Dwarf call frame information (``CFI''),
334 which describes how to find frame base addresses and saved registers.
335 But CFI is not always available, so as a fallback @value{GDBN} uses a
336 technique called @dfn{prologue analysis} to find frame sizes and saved
337 registers. A prologue analyzer disassembles the function's machine
338 code starting from its entry point, and looks for instructions that
339 allocate frame space, save the stack pointer in a frame pointer
340 register, save registers, and so on. Obviously, this can't be done
341 accurately in general, but it's tractable to do well enough to be very
342 helpful. Prologue analysis predates the GNU toolchain's support for
343 CFI; at one time, prologue analysis was the only mechanism
344 @value{GDBN} used for stack unwinding at all, when the function
345 calling conventions didn't specify a fixed frame layout.
347 In the olden days, function prologues were generated by hand-written,
348 target-specific code in GCC, and treated as opaque and untouchable by
349 optimizers. Looking at this code, it was usually straightforward to
350 write a prologue analyzer for @value{GDBN} that would accurately
351 understand all the prologues GCC would generate. However, over time
352 GCC became more aggressive about instruction scheduling, and began to
353 understand more about the semantics of the prologue instructions
354 themselves; in response, @value{GDBN}'s analyzers became more complex
355 and fragile. Keeping the prologue analyzers working as GCC (and the
356 instruction sets themselves) evolved became a substantial task.
358 @cindex @file{prologue-value.c}
359 @cindex abstract interpretation of function prologues
360 @cindex pseudo-evaluation of function prologues
361 To try to address this problem, the code in @file{prologue-value.h}
362 and @file{prologue-value.c} provides a general framework for writing
363 prologue analyzers that are simpler and more robust than ad-hoc
364 analyzers. When we analyze a prologue using the prologue-value
365 framework, we're really doing ``abstract interpretation'' or
366 ``pseudo-evaluation'': running the function's code in simulation, but
367 using conservative approximations of the values registers and memory
368 would hold when the code actually runs. For example, if our function
369 starts with the instruction:
372 addi r1, 42 # add 42 to r1
375 we don't know exactly what value will be in @code{r1} after executing
376 this instruction, but we do know it'll be 42 greater than its original
379 If we then see an instruction like:
382 addi r1, 22 # add 22 to r1
385 we still don't know what @code{r1's} value is, but again, we can say
386 it is now 64 greater than its original value.
388 If the next instruction were:
391 mov r2, r1 # set r2 to r1's value
394 then we can say that @code{r2's} value is now the original value of
397 It's common for prologues to save registers on the stack, so we'll
398 need to track the values of stack frame slots, as well as the
399 registers. So after an instruction like this:
405 then we'd know that the stack slot four bytes above the frame pointer
406 holds the original value of @code{r1} plus 64.
410 Of course, this can only go so far before it gets unreasonable. If we
411 wanted to be able to say anything about the value of @code{r1} after
415 xor r1, r3 # exclusive-or r1 and r3, place result in r1
418 then things would get pretty complex. But remember, we're just doing
419 a conservative approximation; if exclusive-or instructions aren't
420 relevant to prologues, we can just say @code{r1}'s value is now
421 ``unknown''. We can ignore things that are too complex, if that loss of
422 information is acceptable for our application.
424 So when we say ``conservative approximation'' here, what we mean is an
425 approximation that is either accurate, or marked ``unknown'', but
428 Using this framework, a prologue analyzer is simply an interpreter for
429 machine code, but one that uses conservative approximations for the
430 contents of registers and memory instead of actual values. Starting
431 from the function's entry point, you simulate instructions up to the
432 current PC, or an instruction that you don't know how to simulate.
433 Now you can examine the state of the registers and stack slots you've
439 To see how large your stack frame is, just check the value of the
440 stack pointer register; if it's the original value of the SP
441 minus a constant, then that constant is the stack frame's size.
442 If the SP's value has been marked as ``unknown'', then that means
443 the prologue has done something too complex for us to track, and
444 we don't know the frame size.
447 To see where we've saved the previous frame's registers, we just
448 search the values we've tracked --- stack slots, usually, but
449 registers, too, if you want --- for something equal to the register's
450 original value. If the calling conventions suggest a standard place
451 to save a given register, then we can check there first, but really,
452 anything that will get us back the original value will probably work.
455 This does take some work. But prologue analyzers aren't
456 quick-and-simple pattern patching to recognize a few fixed prologue
457 forms any more; they're big, hairy functions. Along with inferior
458 function calls, prologue analysis accounts for a substantial portion
459 of the time needed to stabilize a @value{GDBN} port. So it's
460 worthwhile to look for an approach that will be easier to understand
461 and maintain. In the approach described above:
466 It's easier to see that the analyzer is correct: you just see
467 whether the analyzer properly (albeit conservatively) simulates
468 the effect of each instruction.
471 It's easier to extend the analyzer: you can add support for new
472 instructions, and know that you haven't broken anything that
473 wasn't already broken before.
476 It's orthogonal: to gather new information, you don't need to
477 complicate the code for each instruction. As long as your domain
478 of conservative values is already detailed enough to tell you
479 what you need, then all the existing instruction simulations are
480 already gathering the right data for you.
484 The file @file{prologue-value.h} contains detailed comments explaining
485 the framework and how to use it.
488 @section Breakpoint Handling
491 In general, a breakpoint is a user-designated location in the program
492 where the user wants to regain control if program execution ever reaches
495 There are two main ways to implement breakpoints; either as ``hardware''
496 breakpoints or as ``software'' breakpoints.
498 @cindex hardware breakpoints
499 @cindex program counter
500 Hardware breakpoints are sometimes available as a builtin debugging
501 features with some chips. Typically these work by having dedicated
502 register into which the breakpoint address may be stored. If the PC
503 (shorthand for @dfn{program counter})
504 ever matches a value in a breakpoint registers, the CPU raises an
505 exception and reports it to @value{GDBN}.
507 Another possibility is when an emulator is in use; many emulators
508 include circuitry that watches the address lines coming out from the
509 processor, and force it to stop if the address matches a breakpoint's
512 A third possibility is that the target already has the ability to do
513 breakpoints somehow; for instance, a ROM monitor may do its own
514 software breakpoints. So although these are not literally ``hardware
515 breakpoints'', from @value{GDBN}'s point of view they work the same;
516 @value{GDBN} need not do anything more than set the breakpoint and wait
517 for something to happen.
519 Since they depend on hardware resources, hardware breakpoints may be
520 limited in number; when the user asks for more, @value{GDBN} will
521 start trying to set software breakpoints. (On some architectures,
522 notably the 32-bit x86 platforms, @value{GDBN} cannot always know
523 whether there's enough hardware resources to insert all the hardware
524 breakpoints and watchpoints. On those platforms, @value{GDBN} prints
525 an error message only when the program being debugged is continued.)
527 @cindex software breakpoints
528 Software breakpoints require @value{GDBN} to do somewhat more work.
529 The basic theory is that @value{GDBN} will replace a program
530 instruction with a trap, illegal divide, or some other instruction
531 that will cause an exception, and then when it's encountered,
532 @value{GDBN} will take the exception and stop the program. When the
533 user says to continue, @value{GDBN} will restore the original
534 instruction, single-step, re-insert the trap, and continue on.
536 Since it literally overwrites the program being tested, the program area
537 must be writable, so this technique won't work on programs in ROM. It
538 can also distort the behavior of programs that examine themselves,
539 although such a situation would be highly unusual.
541 Also, the software breakpoint instruction should be the smallest size of
542 instruction, so it doesn't overwrite an instruction that might be a jump
543 target, and cause disaster when the program jumps into the middle of the
544 breakpoint instruction. (Strictly speaking, the breakpoint must be no
545 larger than the smallest interval between instructions that may be jump
546 targets; perhaps there is an architecture where only even-numbered
547 instructions may jumped to.) Note that it's possible for an instruction
548 set not to have any instructions usable for a software breakpoint,
549 although in practice only the ARC has failed to define such an
552 Basic breakpoint object handling is in @file{breakpoint.c}. However,
553 much of the interesting breakpoint action is in @file{infrun.c}.
556 @cindex insert or remove software breakpoint
557 @findex target_remove_breakpoint
558 @findex target_insert_breakpoint
559 @item target_remove_breakpoint (@var{bp_tgt})
560 @itemx target_insert_breakpoint (@var{bp_tgt})
561 Insert or remove a software breakpoint at address
562 @code{@var{bp_tgt}->placed_address}. Returns zero for success,
563 non-zero for failure. On input, @var{bp_tgt} contains the address of the
564 breakpoint, and is otherwise initialized to zero. The fields of the
565 @code{struct bp_target_info} pointed to by @var{bp_tgt} are updated
566 to contain other information about the breakpoint on output. The field
567 @code{placed_address} may be updated if the breakpoint was placed at a
568 related address; the field @code{shadow_contents} contains the real
569 contents of the bytes where the breakpoint has been inserted,
570 if reading memory would return the breakpoint instead of the
571 underlying memory; the field @code{shadow_len} is the length of
572 memory cached in @code{shadow_contents}, if any; and the field
573 @code{placed_size} is optionally set and used by the target, if
574 it could differ from @code{shadow_len}.
576 For example, the remote target @samp{Z0} packet does not require
577 shadowing memory, so @code{shadow_len} is left at zero. However,
578 the length reported by @code{gdbarch_breakpoint_from_pc} is cached in
579 @code{placed_size}, so that a matching @samp{z0} packet can be
580 used to remove the breakpoint.
582 @cindex insert or remove hardware breakpoint
583 @findex target_remove_hw_breakpoint
584 @findex target_insert_hw_breakpoint
585 @item target_remove_hw_breakpoint (@var{bp_tgt})
586 @itemx target_insert_hw_breakpoint (@var{bp_tgt})
587 Insert or remove a hardware-assisted breakpoint at address
588 @code{@var{bp_tgt}->placed_address}. Returns zero for success,
589 non-zero for failure. See @code{target_insert_breakpoint} for
590 a description of the @code{struct bp_target_info} pointed to by
591 @var{bp_tgt}; the @code{shadow_contents} and
592 @code{shadow_len} members are not used for hardware breakpoints,
593 but @code{placed_size} may be.
596 @section Single Stepping
598 @section Signal Handling
600 @section Thread Handling
602 @section Inferior Function Calls
604 @section Longjmp Support
606 @cindex @code{longjmp} debugging
607 @value{GDBN} has support for figuring out that the target is doing a
608 @code{longjmp} and for stopping at the target of the jump, if we are
609 stepping. This is done with a few specialized internal breakpoints,
610 which are visible in the output of the @samp{maint info breakpoint}
613 @findex gdbarch_get_longjmp_target
614 To make this work, you need to define a function called
615 @code{gdbarch_get_longjmp_target}, which will examine the
616 @code{jmp_buf} structure and extract the @code{longjmp} target address.
617 Since @code{jmp_buf} is target specific and typically defined in a
618 target header not available to @value{GDBN}, you will need to
619 determine the offset of the PC manually and return that; many targets
620 define a @code{jb_pc_offset} field in the tdep structure to save the
621 value once calculated.
626 Watchpoints are a special kind of breakpoints (@pxref{Algorithms,
627 breakpoints}) which break when data is accessed rather than when some
628 instruction is executed. When you have data which changes without
629 your knowing what code does that, watchpoints are the silver bullet to
630 hunt down and kill such bugs.
632 @cindex hardware watchpoints
633 @cindex software watchpoints
634 Watchpoints can be either hardware-assisted or not; the latter type is
635 known as ``software watchpoints.'' @value{GDBN} always uses
636 hardware-assisted watchpoints if they are available, and falls back on
637 software watchpoints otherwise. Typical situations where @value{GDBN}
638 will use software watchpoints are:
642 The watched memory region is too large for the underlying hardware
643 watchpoint support. For example, each x86 debug register can watch up
644 to 4 bytes of memory, so trying to watch data structures whose size is
645 more than 16 bytes will cause @value{GDBN} to use software
649 The value of the expression to be watched depends on data held in
650 registers (as opposed to memory).
653 Too many different watchpoints requested. (On some architectures,
654 this situation is impossible to detect until the debugged program is
655 resumed.) Note that x86 debug registers are used both for hardware
656 breakpoints and for watchpoints, so setting too many hardware
657 breakpoints might cause watchpoint insertion to fail.
660 No hardware-assisted watchpoints provided by the target
664 Software watchpoints are very slow, since @value{GDBN} needs to
665 single-step the program being debugged and test the value of the
666 watched expression(s) after each instruction. The rest of this
667 section is mostly irrelevant for software watchpoints.
669 When the inferior stops, @value{GDBN} tries to establish, among other
670 possible reasons, whether it stopped due to a watchpoint being hit.
671 It first uses @code{STOPPED_BY_WATCHPOINT} to see if any watchpoint
672 was hit. If not, all watchpoint checking is skipped.
674 Then @value{GDBN} calls @code{target_stopped_data_address} exactly
675 once. This method returns the address of the watchpoint which
676 triggered, if the target can determine it. If the triggered address
677 is available, @value{GDBN} compares the address returned by this
678 method with each watched memory address in each active watchpoint.
679 For data-read and data-access watchpoints, @value{GDBN} announces
680 every watchpoint that watches the triggered address as being hit.
681 For this reason, data-read and data-access watchpoints
682 @emph{require} that the triggered address be available; if not, read
683 and access watchpoints will never be considered hit. For data-write
684 watchpoints, if the triggered address is available, @value{GDBN}
685 considers only those watchpoints which match that address;
686 otherwise, @value{GDBN} considers all data-write watchpoints. For
687 each data-write watchpoint that @value{GDBN} considers, it evaluates
688 the expression whose value is being watched, and tests whether the
689 watched value has changed. Watchpoints whose watched values have
690 changed are announced as hit.
692 @c FIXME move these to the main lists of target/native defns
694 @value{GDBN} uses several macros and primitives to support hardware
698 @findex TARGET_CAN_USE_HARDWARE_WATCHPOINT
699 @item TARGET_CAN_USE_HARDWARE_WATCHPOINT (@var{type}, @var{count}, @var{other})
700 Return the number of hardware watchpoints of type @var{type} that are
701 possible to be set. The value is positive if @var{count} watchpoints
702 of this type can be set, zero if setting watchpoints of this type is
703 not supported, and negative if @var{count} is more than the maximum
704 number of watchpoints of type @var{type} that can be set. @var{other}
705 is non-zero if other types of watchpoints are currently enabled (there
706 are architectures which cannot set watchpoints of different types at
709 @findex TARGET_REGION_OK_FOR_HW_WATCHPOINT
710 @item TARGET_REGION_OK_FOR_HW_WATCHPOINT (@var{addr}, @var{len})
711 Return non-zero if hardware watchpoints can be used to watch a region
712 whose address is @var{addr} and whose length in bytes is @var{len}.
714 @cindex insert or remove hardware watchpoint
715 @findex target_insert_watchpoint
716 @findex target_remove_watchpoint
717 @item target_insert_watchpoint (@var{addr}, @var{len}, @var{type})
718 @itemx target_remove_watchpoint (@var{addr}, @var{len}, @var{type})
719 Insert or remove a hardware watchpoint starting at @var{addr}, for
720 @var{len} bytes. @var{type} is the watchpoint type, one of the
721 possible values of the enumerated data type @code{target_hw_bp_type},
722 defined by @file{breakpoint.h} as follows:
725 enum target_hw_bp_type
727 hw_write = 0, /* Common (write) HW watchpoint */
728 hw_read = 1, /* Read HW watchpoint */
729 hw_access = 2, /* Access (read or write) HW watchpoint */
730 hw_execute = 3 /* Execute HW breakpoint */
735 These two macros should return 0 for success, non-zero for failure.
737 @findex target_stopped_data_address
738 @item target_stopped_data_address (@var{addr_p})
739 If the inferior has some watchpoint that triggered, place the address
740 associated with the watchpoint at the location pointed to by
741 @var{addr_p} and return non-zero. Otherwise, return zero. This
742 is required for data-read and data-access watchpoints. It is
743 not required for data-write watchpoints, but @value{GDBN} uses
744 it to improve handling of those also.
746 @value{GDBN} will only call this method once per watchpoint stop,
747 immediately after calling @code{STOPPED_BY_WATCHPOINT}. If the
748 target's watchpoint indication is sticky, i.e., stays set after
749 resuming, this method should clear it. For instance, the x86 debug
750 control register has sticky triggered flags.
752 @findex target_watchpoint_addr_within_range
753 @item target_watchpoint_addr_within_range (@var{target}, @var{addr}, @var{start}, @var{length})
754 Check whether @var{addr} (as returned by @code{target_stopped_data_address})
755 lies within the hardware-defined watchpoint region described by
756 @var{start} and @var{length}. This only needs to be provided if the
757 granularity of a watchpoint is greater than one byte, i.e., if the
758 watchpoint can also trigger on nearby addresses outside of the watched
761 @findex HAVE_STEPPABLE_WATCHPOINT
762 @item HAVE_STEPPABLE_WATCHPOINT
763 If defined to a non-zero value, it is not necessary to disable a
764 watchpoint to step over it. Like @code{gdbarch_have_nonsteppable_watchpoint},
765 this is usually set when watchpoints trigger at the instruction
766 which will perform an interesting read or write. It should be
767 set if there is a temporary disable bit which allows the processor
768 to step over the interesting instruction without raising the
769 watchpoint exception again.
771 @findex gdbarch_have_nonsteppable_watchpoint
772 @item int gdbarch_have_nonsteppable_watchpoint (@var{gdbarch})
773 If it returns a non-zero value, @value{GDBN} should disable a
774 watchpoint to step the inferior over it. This is usually set when
775 watchpoints trigger at the instruction which will perform an
776 interesting read or write.
778 @findex HAVE_CONTINUABLE_WATCHPOINT
779 @item HAVE_CONTINUABLE_WATCHPOINT
780 If defined to a non-zero value, it is possible to continue the
781 inferior after a watchpoint has been hit. This is usually set
782 when watchpoints trigger at the instruction following an interesting
785 @findex STOPPED_BY_WATCHPOINT
786 @item STOPPED_BY_WATCHPOINT (@var{wait_status})
787 Return non-zero if stopped by a watchpoint. @var{wait_status} is of
788 the type @code{struct target_waitstatus}, defined by @file{target.h}.
789 Normally, this macro is defined to invoke the function pointed to by
790 the @code{to_stopped_by_watchpoint} member of the structure (of the
791 type @code{target_ops}, defined on @file{target.h}) that describes the
792 target-specific operations; @code{to_stopped_by_watchpoint} ignores
793 the @var{wait_status} argument.
795 @value{GDBN} does not require the non-zero value returned by
796 @code{STOPPED_BY_WATCHPOINT} to be 100% correct, so if a target cannot
797 determine for sure whether the inferior stopped due to a watchpoint,
798 it could return non-zero ``just in case''.
801 @subsection Watchpoints and Threads
802 @cindex watchpoints, with threads
804 @value{GDBN} only supports process-wide watchpoints, which trigger
805 in all threads. @value{GDBN} uses the thread ID to make watchpoints
806 act as if they were thread-specific, but it cannot set hardware
807 watchpoints that only trigger in a specific thread. Therefore, even
808 if the target supports threads, per-thread debug registers, and
809 watchpoints which only affect a single thread, it should set the
810 per-thread debug registers for all threads to the same value. On
811 @sc{gnu}/Linux native targets, this is accomplished by using
812 @code{ALL_LWPS} in @code{target_insert_watchpoint} and
813 @code{target_remove_watchpoint} and by using
814 @code{linux_set_new_thread} to register a handler for newly created
817 @value{GDBN}'s @sc{gnu}/Linux support only reports a single event
818 at a time, although multiple events can trigger simultaneously for
819 multi-threaded programs. When multiple events occur, @file{linux-nat.c}
820 queues subsequent events and returns them the next time the program
821 is resumed. This means that @code{STOPPED_BY_WATCHPOINT} and
822 @code{target_stopped_data_address} only need to consult the current
823 thread's state---the thread indicated by @code{inferior_ptid}. If
824 two threads have hit watchpoints simultaneously, those routines
825 will be called a second time for the second thread.
827 @subsection x86 Watchpoints
828 @cindex x86 debug registers
829 @cindex watchpoints, on x86
831 The 32-bit Intel x86 (a.k.a.@: ia32) processors feature special debug
832 registers designed to facilitate debugging. @value{GDBN} provides a
833 generic library of functions that x86-based ports can use to implement
834 support for watchpoints and hardware-assisted breakpoints. This
835 subsection documents the x86 watchpoint facilities in @value{GDBN}.
837 (At present, the library functions read and write debug registers directly, and are
838 thus only available for native configurations.)
840 To use the generic x86 watchpoint support, a port should do the
844 @findex I386_USE_GENERIC_WATCHPOINTS
846 Define the macro @code{I386_USE_GENERIC_WATCHPOINTS} somewhere in the
847 target-dependent headers.
850 Include the @file{config/i386/nm-i386.h} header file @emph{after}
851 defining @code{I386_USE_GENERIC_WATCHPOINTS}.
854 Add @file{i386-nat.o} to the value of the Make variable
855 @code{NATDEPFILES} (@pxref{Native Debugging, NATDEPFILES}).
858 Provide implementations for the @code{I386_DR_LOW_*} macros described
859 below. Typically, each macro should call a target-specific function
860 which does the real work.
863 The x86 watchpoint support works by maintaining mirror images of the
864 debug registers. Values are copied between the mirror images and the
865 real debug registers via a set of macros which each target needs to
869 @findex I386_DR_LOW_SET_CONTROL
870 @item I386_DR_LOW_SET_CONTROL (@var{val})
871 Set the Debug Control (DR7) register to the value @var{val}.
873 @findex I386_DR_LOW_SET_ADDR
874 @item I386_DR_LOW_SET_ADDR (@var{idx}, @var{addr})
875 Put the address @var{addr} into the debug register number @var{idx}.
877 @findex I386_DR_LOW_RESET_ADDR
878 @item I386_DR_LOW_RESET_ADDR (@var{idx})
879 Reset (i.e.@: zero out) the address stored in the debug register
882 @findex I386_DR_LOW_GET_STATUS
883 @item I386_DR_LOW_GET_STATUS
884 Return the value of the Debug Status (DR6) register. This value is
885 used immediately after it is returned by
886 @code{I386_DR_LOW_GET_STATUS}, so as to support per-thread status
890 For each one of the 4 debug registers (whose indices are from 0 to 3)
891 that store addresses, a reference count is maintained by @value{GDBN},
892 to allow sharing of debug registers by several watchpoints. This
893 allows users to define several watchpoints that watch the same
894 expression, but with different conditions and/or commands, without
895 wasting debug registers which are in short supply. @value{GDBN}
896 maintains the reference counts internally, targets don't have to do
897 anything to use this feature.
899 The x86 debug registers can each watch a region that is 1, 2, or 4
900 bytes long. The ia32 architecture requires that each watched region
901 be appropriately aligned: 2-byte region on 2-byte boundary, 4-byte
902 region on 4-byte boundary. However, the x86 watchpoint support in
903 @value{GDBN} can watch unaligned regions and regions larger than 4
904 bytes (up to 16 bytes) by allocating several debug registers to watch
905 a single region. This allocation of several registers per a watched
906 region is also done automatically without target code intervention.
908 The generic x86 watchpoint support provides the following API for the
909 @value{GDBN}'s application code:
912 @findex i386_region_ok_for_watchpoint
913 @item i386_region_ok_for_watchpoint (@var{addr}, @var{len})
914 The macro @code{TARGET_REGION_OK_FOR_HW_WATCHPOINT} is set to call
915 this function. It counts the number of debug registers required to
916 watch a given region, and returns a non-zero value if that number is
917 less than 4, the number of debug registers available to x86
920 @findex i386_stopped_data_address
921 @item i386_stopped_data_address (@var{addr_p})
923 @code{target_stopped_data_address} is set to call this function.
925 function examines the breakpoint condition bits in the DR6 Debug
926 Status register, as returned by the @code{I386_DR_LOW_GET_STATUS}
927 macro, and returns the address associated with the first bit that is
930 @findex i386_stopped_by_watchpoint
931 @item i386_stopped_by_watchpoint (void)
932 The macro @code{STOPPED_BY_WATCHPOINT}
933 is set to call this function. The
934 argument passed to @code{STOPPED_BY_WATCHPOINT} is ignored. This
935 function examines the breakpoint condition bits in the DR6 Debug
936 Status register, as returned by the @code{I386_DR_LOW_GET_STATUS}
937 macro, and returns true if any bit is set. Otherwise, false is
940 @findex i386_insert_watchpoint
941 @findex i386_remove_watchpoint
942 @item i386_insert_watchpoint (@var{addr}, @var{len}, @var{type})
943 @itemx i386_remove_watchpoint (@var{addr}, @var{len}, @var{type})
944 Insert or remove a watchpoint. The macros
945 @code{target_insert_watchpoint} and @code{target_remove_watchpoint}
946 are set to call these functions. @code{i386_insert_watchpoint} first
947 looks for a debug register which is already set to watch the same
948 region for the same access types; if found, it just increments the
949 reference count of that debug register, thus implementing debug
950 register sharing between watchpoints. If no such register is found,
951 the function looks for a vacant debug register, sets its mirrored
952 value to @var{addr}, sets the mirrored value of DR7 Debug Control
953 register as appropriate for the @var{len} and @var{type} parameters,
954 and then passes the new values of the debug register and DR7 to the
955 inferior by calling @code{I386_DR_LOW_SET_ADDR} and
956 @code{I386_DR_LOW_SET_CONTROL}. If more than one debug register is
957 required to cover the given region, the above process is repeated for
960 @code{i386_remove_watchpoint} does the opposite: it resets the address
961 in the mirrored value of the debug register and its read/write and
962 length bits in the mirrored value of DR7, then passes these new
963 values to the inferior via @code{I386_DR_LOW_RESET_ADDR} and
964 @code{I386_DR_LOW_SET_CONTROL}. If a register is shared by several
965 watchpoints, each time a @code{i386_remove_watchpoint} is called, it
966 decrements the reference count, and only calls
967 @code{I386_DR_LOW_RESET_ADDR} and @code{I386_DR_LOW_SET_CONTROL} when
968 the count goes to zero.
970 @findex i386_insert_hw_breakpoint
971 @findex i386_remove_hw_breakpoint
972 @item i386_insert_hw_breakpoint (@var{bp_tgt})
973 @itemx i386_remove_hw_breakpoint (@var{bp_tgt})
974 These functions insert and remove hardware-assisted breakpoints. The
975 macros @code{target_insert_hw_breakpoint} and
976 @code{target_remove_hw_breakpoint} are set to call these functions.
977 The argument is a @code{struct bp_target_info *}, as described in
978 the documentation for @code{target_insert_breakpoint}.
979 These functions work like @code{i386_insert_watchpoint} and
980 @code{i386_remove_watchpoint}, respectively, except that they set up
981 the debug registers to watch instruction execution, and each
982 hardware-assisted breakpoint always requires exactly one debug
985 @findex i386_cleanup_dregs
986 @item i386_cleanup_dregs (void)
987 This function clears all the reference counts, addresses, and control
988 bits in the mirror images of the debug registers. It doesn't affect
989 the actual debug registers in the inferior process.
996 x86 processors support setting watchpoints on I/O reads or writes.
997 However, since no target supports this (as of March 2001), and since
998 @code{enum target_hw_bp_type} doesn't even have an enumeration for I/O
999 watchpoints, this feature is not yet available to @value{GDBN} running
1003 x86 processors can enable watchpoints locally, for the current task
1004 only, or globally, for all the tasks. For each debug register,
1005 there's a bit in the DR7 Debug Control register that determines
1006 whether the associated address is watched locally or globally. The
1007 current implementation of x86 watchpoint support in @value{GDBN}
1008 always sets watchpoints to be locally enabled, since global
1009 watchpoints might interfere with the underlying OS and are probably
1010 unavailable in many platforms.
1013 @section Checkpoints
1016 In the abstract, a checkpoint is a point in the execution history of
1017 the program, which the user may wish to return to at some later time.
1019 Internally, a checkpoint is a saved copy of the program state, including
1020 whatever information is required in order to restore the program to that
1021 state at a later time. This can be expected to include the state of
1022 registers and memory, and may include external state such as the state
1023 of open files and devices.
1025 There are a number of ways in which checkpoints may be implemented
1026 in gdb, e.g.@: as corefiles, as forked processes, and as some opaque
1027 method implemented on the target side.
1029 A corefile can be used to save an image of target memory and register
1030 state, which can in principle be restored later --- but corefiles do
1031 not typically include information about external entities such as
1032 open files. Currently this method is not implemented in gdb.
1034 A forked process can save the state of user memory and registers,
1035 as well as some subset of external (kernel) state. This method
1036 is used to implement checkpoints on Linux, and in principle might
1037 be used on other systems.
1039 Some targets, e.g.@: simulators, might have their own built-in
1040 method for saving checkpoints, and gdb might be able to take
1041 advantage of that capability without necessarily knowing any
1042 details of how it is done.
1045 @section Observing changes in @value{GDBN} internals
1046 @cindex observer pattern interface
1047 @cindex notifications about changes in internals
1049 In order to function properly, several modules need to be notified when
1050 some changes occur in the @value{GDBN} internals. Traditionally, these
1051 modules have relied on several paradigms, the most common ones being
1052 hooks and gdb-events. Unfortunately, none of these paradigms was
1053 versatile enough to become the standard notification mechanism in
1054 @value{GDBN}. The fact that they only supported one ``client'' was also
1055 a strong limitation.
1057 A new paradigm, based on the Observer pattern of the @cite{Design
1058 Patterns} book, has therefore been implemented. The goal was to provide
1059 a new interface overcoming the issues with the notification mechanisms
1060 previously available. This new interface needed to be strongly typed,
1061 easy to extend, and versatile enough to be used as the standard
1062 interface when adding new notifications.
1064 See @ref{GDB Observers} for a brief description of the observers
1065 currently implemented in GDB. The rationale for the current
1066 implementation is also briefly discussed.
1068 @node User Interface
1070 @chapter User Interface
1072 @value{GDBN} has several user interfaces, of which the traditional
1073 command-line interface is perhaps the most familiar.
1075 @section Command Interpreter
1077 @cindex command interpreter
1079 The command interpreter in @value{GDBN} is fairly simple. It is designed to
1080 allow for the set of commands to be augmented dynamically, and also
1081 has a recursive subcommand capability, where the first argument to
1082 a command may itself direct a lookup on a different command list.
1084 For instance, the @samp{set} command just starts a lookup on the
1085 @code{setlist} command list, while @samp{set thread} recurses
1086 to the @code{set_thread_cmd_list}.
1090 To add commands in general, use @code{add_cmd}. @code{add_com} adds to
1091 the main command list, and should be used for those commands. The usual
1092 place to add commands is in the @code{_initialize_@var{xyz}} routines at
1093 the ends of most source files.
1095 @findex add_setshow_cmd
1096 @findex add_setshow_cmd_full
1097 To add paired @samp{set} and @samp{show} commands, use
1098 @code{add_setshow_cmd} or @code{add_setshow_cmd_full}. The former is
1099 a slightly simpler interface which is useful when you don't need to
1100 further modify the new command structures, while the latter returns
1101 the new command structures for manipulation.
1103 @cindex deprecating commands
1104 @findex deprecate_cmd
1105 Before removing commands from the command set it is a good idea to
1106 deprecate them for some time. Use @code{deprecate_cmd} on commands or
1107 aliases to set the deprecated flag. @code{deprecate_cmd} takes a
1108 @code{struct cmd_list_element} as it's first argument. You can use the
1109 return value from @code{add_com} or @code{add_cmd} to deprecate the
1110 command immediately after it is created.
1112 The first time a command is used the user will be warned and offered a
1113 replacement (if one exists). Note that the replacement string passed to
1114 @code{deprecate_cmd} should be the full name of the command, i.e., the
1115 entire string the user should type at the command line.
1117 @anchor{UI-Independent Output}
1118 @section UI-Independent Output---the @code{ui_out} Functions
1119 @c This section is based on the documentation written by Fernando
1120 @c Nasser <fnasser@redhat.com>.
1122 @cindex @code{ui_out} functions
1123 The @code{ui_out} functions present an abstraction level for the
1124 @value{GDBN} output code. They hide the specifics of different user
1125 interfaces supported by @value{GDBN}, and thus free the programmer
1126 from the need to write several versions of the same code, one each for
1127 every UI, to produce output.
1129 @subsection Overview and Terminology
1131 In general, execution of each @value{GDBN} command produces some sort
1132 of output, and can even generate an input request.
1134 Output can be generated for the following purposes:
1138 to display a @emph{result} of an operation;
1141 to convey @emph{info} or produce side-effects of a requested
1145 to provide a @emph{notification} of an asynchronous event (including
1146 progress indication of a prolonged asynchronous operation);
1149 to display @emph{error messages} (including warnings);
1152 to show @emph{debug data};
1155 to @emph{query} or prompt a user for input (a special case).
1159 This section mainly concentrates on how to build result output,
1160 although some of it also applies to other kinds of output.
1162 Generation of output that displays the results of an operation
1163 involves one or more of the following:
1167 output of the actual data
1170 formatting the output as appropriate for console output, to make it
1171 easily readable by humans
1174 machine oriented formatting--a more terse formatting to allow for easy
1175 parsing by programs which read @value{GDBN}'s output
1178 annotation, whose purpose is to help legacy GUIs to identify interesting
1182 The @code{ui_out} routines take care of the first three aspects.
1183 Annotations are provided by separate annotation routines. Note that use
1184 of annotations for an interface between a GUI and @value{GDBN} is
1187 Output can be in the form of a single item, which we call a @dfn{field};
1188 a @dfn{list} consisting of identical fields; a @dfn{tuple} consisting of
1189 non-identical fields; or a @dfn{table}, which is a tuple consisting of a
1190 header and a body. In a BNF-like form:
1193 @item <table> @expansion{}
1194 @code{<header> <body>}
1195 @item <header> @expansion{}
1196 @code{@{ <column> @}}
1197 @item <column> @expansion{}
1198 @code{<width> <alignment> <title>}
1199 @item <body> @expansion{}
1204 @subsection General Conventions
1206 Most @code{ui_out} routines are of type @code{void}, the exceptions are
1207 @code{ui_out_stream_new} (which returns a pointer to the newly created
1208 object) and the @code{make_cleanup} routines.
1210 The first parameter is always the @code{ui_out} vector object, a pointer
1211 to a @code{struct ui_out}.
1213 The @var{format} parameter is like in @code{printf} family of functions.
1214 When it is present, there must also be a variable list of arguments
1215 sufficient used to satisfy the @code{%} specifiers in the supplied
1218 When a character string argument is not used in a @code{ui_out} function
1219 call, a @code{NULL} pointer has to be supplied instead.
1222 @subsection Table, Tuple and List Functions
1224 @cindex list output functions
1225 @cindex table output functions
1226 @cindex tuple output functions
1227 This section introduces @code{ui_out} routines for building lists,
1228 tuples and tables. The routines to output the actual data items
1229 (fields) are presented in the next section.
1231 To recap: A @dfn{tuple} is a sequence of @dfn{fields}, each field
1232 containing information about an object; a @dfn{list} is a sequence of
1233 fields where each field describes an identical object.
1235 Use the @dfn{table} functions when your output consists of a list of
1236 rows (tuples) and the console output should include a heading. Use this
1237 even when you are listing just one object but you still want the header.
1239 @cindex nesting level in @code{ui_out} functions
1240 Tables can not be nested. Tuples and lists can be nested up to a
1241 maximum of five levels.
1243 The overall structure of the table output code is something like this:
1258 Here is the description of table-, tuple- and list-related @code{ui_out}
1261 @deftypefun void ui_out_table_begin (struct ui_out *@var{uiout}, int @var{nbrofcols}, int @var{nr_rows}, const char *@var{tblid})
1262 The function @code{ui_out_table_begin} marks the beginning of the output
1263 of a table. It should always be called before any other @code{ui_out}
1264 function for a given table. @var{nbrofcols} is the number of columns in
1265 the table. @var{nr_rows} is the number of rows in the table.
1266 @var{tblid} is an optional string identifying the table. The string
1267 pointed to by @var{tblid} is copied by the implementation of
1268 @code{ui_out_table_begin}, so the application can free the string if it
1269 was @code{malloc}ed.
1271 The companion function @code{ui_out_table_end}, described below, marks
1272 the end of the table's output.
1275 @deftypefun void ui_out_table_header (struct ui_out *@var{uiout}, int @var{width}, enum ui_align @var{alignment}, const char *@var{colhdr})
1276 @code{ui_out_table_header} provides the header information for a single
1277 table column. You call this function several times, one each for every
1278 column of the table, after @code{ui_out_table_begin}, but before
1279 @code{ui_out_table_body}.
1281 The value of @var{width} gives the column width in characters. The
1282 value of @var{alignment} is one of @code{left}, @code{center}, and
1283 @code{right}, and it specifies how to align the header: left-justify,
1284 center, or right-justify it. @var{colhdr} points to a string that
1285 specifies the column header; the implementation copies that string, so
1286 column header strings in @code{malloc}ed storage can be freed after the
1290 @deftypefun void ui_out_table_body (struct ui_out *@var{uiout})
1291 This function delimits the table header from the table body.
1294 @deftypefun void ui_out_table_end (struct ui_out *@var{uiout})
1295 This function signals the end of a table's output. It should be called
1296 after the table body has been produced by the list and field output
1299 There should be exactly one call to @code{ui_out_table_end} for each
1300 call to @code{ui_out_table_begin}, otherwise the @code{ui_out} functions
1301 will signal an internal error.
1304 The output of the tuples that represent the table rows must follow the
1305 call to @code{ui_out_table_body} and precede the call to
1306 @code{ui_out_table_end}. You build a tuple by calling
1307 @code{ui_out_tuple_begin} and @code{ui_out_tuple_end}, with suitable
1308 calls to functions which actually output fields between them.
1310 @deftypefun void ui_out_tuple_begin (struct ui_out *@var{uiout}, const char *@var{id})
1311 This function marks the beginning of a tuple output. @var{id} points
1312 to an optional string that identifies the tuple; it is copied by the
1313 implementation, and so strings in @code{malloc}ed storage can be freed
1317 @deftypefun void ui_out_tuple_end (struct ui_out *@var{uiout})
1318 This function signals an end of a tuple output. There should be exactly
1319 one call to @code{ui_out_tuple_end} for each call to
1320 @code{ui_out_tuple_begin}, otherwise an internal @value{GDBN} error will
1324 @deftypefun {struct cleanup *} make_cleanup_ui_out_tuple_begin_end (struct ui_out *@var{uiout}, const char *@var{id})
1325 This function first opens the tuple and then establishes a cleanup
1326 (@pxref{Misc Guidelines, Cleanups}) to close the tuple.
1327 It provides a convenient and correct implementation of the
1328 non-portable@footnote{The function cast is not portable ISO C.} code sequence:
1330 struct cleanup *old_cleanup;
1331 ui_out_tuple_begin (uiout, "...");
1332 old_cleanup = make_cleanup ((void(*)(void *)) ui_out_tuple_end,
1337 @deftypefun void ui_out_list_begin (struct ui_out *@var{uiout}, const char *@var{id})
1338 This function marks the beginning of a list output. @var{id} points to
1339 an optional string that identifies the list; it is copied by the
1340 implementation, and so strings in @code{malloc}ed storage can be freed
1344 @deftypefun void ui_out_list_end (struct ui_out *@var{uiout})
1345 This function signals an end of a list output. There should be exactly
1346 one call to @code{ui_out_list_end} for each call to
1347 @code{ui_out_list_begin}, otherwise an internal @value{GDBN} error will
1351 @deftypefun {struct cleanup *} make_cleanup_ui_out_list_begin_end (struct ui_out *@var{uiout}, const char *@var{id})
1352 Similar to @code{make_cleanup_ui_out_tuple_begin_end}, this function
1353 opens a list and then establishes cleanup
1354 (@pxref{Misc Guidelines, Cleanups})
1355 that will close the list.
1358 @subsection Item Output Functions
1360 @cindex item output functions
1361 @cindex field output functions
1363 The functions described below produce output for the actual data
1364 items, or fields, which contain information about the object.
1366 Choose the appropriate function accordingly to your particular needs.
1368 @deftypefun void ui_out_field_fmt (struct ui_out *@var{uiout}, char *@var{fldname}, char *@var{format}, ...)
1369 This is the most general output function. It produces the
1370 representation of the data in the variable-length argument list
1371 according to formatting specifications in @var{format}, a
1372 @code{printf}-like format string. The optional argument @var{fldname}
1373 supplies the name of the field. The data items themselves are
1374 supplied as additional arguments after @var{format}.
1376 This generic function should be used only when it is not possible to
1377 use one of the specialized versions (see below).
1380 @deftypefun void ui_out_field_int (struct ui_out *@var{uiout}, const char *@var{fldname}, int @var{value})
1381 This function outputs a value of an @code{int} variable. It uses the
1382 @code{"%d"} output conversion specification. @var{fldname} specifies
1383 the name of the field.
1386 @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})
1387 This function outputs a value of an @code{int} variable. It differs from
1388 @code{ui_out_field_int} in that the caller specifies the desired @var{width} and @var{alignment} of the output.
1389 @var{fldname} specifies
1390 the name of the field.
1393 @deftypefun void ui_out_field_core_addr (struct ui_out *@var{uiout}, const char *@var{fldname}, struct gdbarch *@var{gdbarch}, CORE_ADDR @var{address})
1394 This function outputs an address as appropriate for @var{gdbarch}.
1397 @deftypefun void ui_out_field_string (struct ui_out *@var{uiout}, const char *@var{fldname}, const char *@var{string})
1398 This function outputs a string using the @code{"%s"} conversion
1402 Sometimes, there's a need to compose your output piece by piece using
1403 functions that operate on a stream, such as @code{value_print} or
1404 @code{fprintf_symbol_filtered}. These functions accept an argument of
1405 the type @code{struct ui_file *}, a pointer to a @code{ui_file} object
1406 used to store the data stream used for the output. When you use one
1407 of these functions, you need a way to pass their results stored in a
1408 @code{ui_file} object to the @code{ui_out} functions. To this end,
1409 you first create a @code{ui_stream} object by calling
1410 @code{ui_out_stream_new}, pass the @code{stream} member of that
1411 @code{ui_stream} object to @code{value_print} and similar functions,
1412 and finally call @code{ui_out_field_stream} to output the field you
1413 constructed. When the @code{ui_stream} object is no longer needed,
1414 you should destroy it and free its memory by calling
1415 @code{ui_out_stream_delete}.
1417 @deftypefun {struct ui_stream *} ui_out_stream_new (struct ui_out *@var{uiout})
1418 This function creates a new @code{ui_stream} object which uses the
1419 same output methods as the @code{ui_out} object whose pointer is
1420 passed in @var{uiout}. It returns a pointer to the newly created
1421 @code{ui_stream} object.
1424 @deftypefun void ui_out_stream_delete (struct ui_stream *@var{streambuf})
1425 This functions destroys a @code{ui_stream} object specified by
1429 @deftypefun void ui_out_field_stream (struct ui_out *@var{uiout}, const char *@var{fieldname}, struct ui_stream *@var{streambuf})
1430 This function consumes all the data accumulated in
1431 @code{streambuf->stream} and outputs it like
1432 @code{ui_out_field_string} does. After a call to
1433 @code{ui_out_field_stream}, the accumulated data no longer exists, but
1434 the stream is still valid and may be used for producing more fields.
1437 @strong{Important:} If there is any chance that your code could bail
1438 out before completing output generation and reaching the point where
1439 @code{ui_out_stream_delete} is called, it is necessary to set up a
1440 cleanup, to avoid leaking memory and other resources. Here's a
1441 skeleton code to do that:
1444 struct ui_stream *mybuf = ui_out_stream_new (uiout);
1445 struct cleanup *old = make_cleanup (ui_out_stream_delete, mybuf);
1450 If the function already has the old cleanup chain set (for other kinds
1451 of cleanups), you just have to add your cleanup to it:
1454 mybuf = ui_out_stream_new (uiout);
1455 make_cleanup (ui_out_stream_delete, mybuf);
1458 Note that with cleanups in place, you should not call
1459 @code{ui_out_stream_delete} directly, or you would attempt to free the
1462 @subsection Utility Output Functions
1464 @deftypefun void ui_out_field_skip (struct ui_out *@var{uiout}, const char *@var{fldname})
1465 This function skips a field in a table. Use it if you have to leave
1466 an empty field without disrupting the table alignment. The argument
1467 @var{fldname} specifies a name for the (missing) filed.
1470 @deftypefun void ui_out_text (struct ui_out *@var{uiout}, const char *@var{string})
1471 This function outputs the text in @var{string} in a way that makes it
1472 easy to be read by humans. For example, the console implementation of
1473 this method filters the text through a built-in pager, to prevent it
1474 from scrolling off the visible portion of the screen.
1476 Use this function for printing relatively long chunks of text around
1477 the actual field data: the text it produces is not aligned according
1478 to the table's format. Use @code{ui_out_field_string} to output a
1479 string field, and use @code{ui_out_message}, described below, to
1480 output short messages.
1483 @deftypefun void ui_out_spaces (struct ui_out *@var{uiout}, int @var{nspaces})
1484 This function outputs @var{nspaces} spaces. It is handy to align the
1485 text produced by @code{ui_out_text} with the rest of the table or
1489 @deftypefun void ui_out_message (struct ui_out *@var{uiout}, int @var{verbosity}, const char *@var{format}, ...)
1490 This function produces a formatted message, provided that the current
1491 verbosity level is at least as large as given by @var{verbosity}. The
1492 current verbosity level is specified by the user with the @samp{set
1493 verbositylevel} command.@footnote{As of this writing (April 2001),
1494 setting verbosity level is not yet implemented, and is always returned
1495 as zero. So calling @code{ui_out_message} with a @var{verbosity}
1496 argument more than zero will cause the message to never be printed.}
1499 @deftypefun void ui_out_wrap_hint (struct ui_out *@var{uiout}, char *@var{indent})
1500 This function gives the console output filter (a paging filter) a hint
1501 of where to break lines which are too long. Ignored for all other
1502 output consumers. @var{indent}, if non-@code{NULL}, is the string to
1503 be printed to indent the wrapped text on the next line; it must remain
1504 accessible until the next call to @code{ui_out_wrap_hint}, or until an
1505 explicit newline is produced by one of the other functions. If
1506 @var{indent} is @code{NULL}, the wrapped text will not be indented.
1509 @deftypefun void ui_out_flush (struct ui_out *@var{uiout})
1510 This function flushes whatever output has been accumulated so far, if
1511 the UI buffers output.
1515 @subsection Examples of Use of @code{ui_out} functions
1517 @cindex using @code{ui_out} functions
1518 @cindex @code{ui_out} functions, usage examples
1519 This section gives some practical examples of using the @code{ui_out}
1520 functions to generalize the old console-oriented code in
1521 @value{GDBN}. The examples all come from functions defined on the
1522 @file{breakpoints.c} file.
1524 This example, from the @code{breakpoint_1} function, shows how to
1527 The original code was:
1530 if (!found_a_breakpoint++)
1532 annotate_breakpoints_headers ();
1535 printf_filtered ("Num ");
1537 printf_filtered ("Type ");
1539 printf_filtered ("Disp ");
1541 printf_filtered ("Enb ");
1545 printf_filtered ("Address ");
1548 printf_filtered ("What\n");
1550 annotate_breakpoints_table ();
1554 Here's the new version:
1557 nr_printable_breakpoints = @dots{};
1560 ui_out_table_begin (ui, 6, nr_printable_breakpoints, "BreakpointTable");
1562 ui_out_table_begin (ui, 5, nr_printable_breakpoints, "BreakpointTable");
1564 if (nr_printable_breakpoints > 0)
1565 annotate_breakpoints_headers ();
1566 if (nr_printable_breakpoints > 0)
1568 ui_out_table_header (uiout, 3, ui_left, "number", "Num"); /* 1 */
1569 if (nr_printable_breakpoints > 0)
1571 ui_out_table_header (uiout, 14, ui_left, "type", "Type"); /* 2 */
1572 if (nr_printable_breakpoints > 0)
1574 ui_out_table_header (uiout, 4, ui_left, "disp", "Disp"); /* 3 */
1575 if (nr_printable_breakpoints > 0)
1577 ui_out_table_header (uiout, 3, ui_left, "enabled", "Enb"); /* 4 */
1580 if (nr_printable_breakpoints > 0)
1582 if (print_address_bits <= 32)
1583 ui_out_table_header (uiout, 10, ui_left, "addr", "Address");/* 5 */
1585 ui_out_table_header (uiout, 18, ui_left, "addr", "Address");/* 5 */
1587 if (nr_printable_breakpoints > 0)
1589 ui_out_table_header (uiout, 40, ui_noalign, "what", "What"); /* 6 */
1590 ui_out_table_body (uiout);
1591 if (nr_printable_breakpoints > 0)
1592 annotate_breakpoints_table ();
1595 This example, from the @code{print_one_breakpoint} function, shows how
1596 to produce the actual data for the table whose structure was defined
1597 in the above example. The original code was:
1602 printf_filtered ("%-3d ", b->number);
1604 if ((int)b->type > (sizeof(bptypes)/sizeof(bptypes[0]))
1605 || ((int) b->type != bptypes[(int) b->type].type))
1606 internal_error ("bptypes table does not describe type #%d.",
1608 printf_filtered ("%-14s ", bptypes[(int)b->type].description);
1610 printf_filtered ("%-4s ", bpdisps[(int)b->disposition]);
1612 printf_filtered ("%-3c ", bpenables[(int)b->enable]);
1616 This is the new version:
1620 ui_out_tuple_begin (uiout, "bkpt");
1622 ui_out_field_int (uiout, "number", b->number);
1624 if (((int) b->type > (sizeof (bptypes) / sizeof (bptypes[0])))
1625 || ((int) b->type != bptypes[(int) b->type].type))
1626 internal_error ("bptypes table does not describe type #%d.",
1628 ui_out_field_string (uiout, "type", bptypes[(int)b->type].description);
1630 ui_out_field_string (uiout, "disp", bpdisps[(int)b->disposition]);
1632 ui_out_field_fmt (uiout, "enabled", "%c", bpenables[(int)b->enable]);
1636 This example, also from @code{print_one_breakpoint}, shows how to
1637 produce a complicated output field using the @code{print_expression}
1638 functions which requires a stream to be passed. It also shows how to
1639 automate stream destruction with cleanups. The original code was:
1643 print_expression (b->exp, gdb_stdout);
1649 struct ui_stream *stb = ui_out_stream_new (uiout);
1650 struct cleanup *old_chain = make_cleanup_ui_out_stream_delete (stb);
1653 print_expression (b->exp, stb->stream);
1654 ui_out_field_stream (uiout, "what", local_stream);
1657 This example, also from @code{print_one_breakpoint}, shows how to use
1658 @code{ui_out_text} and @code{ui_out_field_string}. The original code
1663 if (b->dll_pathname == NULL)
1664 printf_filtered ("<any library> ");
1666 printf_filtered ("library \"%s\" ", b->dll_pathname);
1673 if (b->dll_pathname == NULL)
1675 ui_out_field_string (uiout, "what", "<any library>");
1676 ui_out_spaces (uiout, 1);
1680 ui_out_text (uiout, "library \"");
1681 ui_out_field_string (uiout, "what", b->dll_pathname);
1682 ui_out_text (uiout, "\" ");
1686 The following example from @code{print_one_breakpoint} shows how to
1687 use @code{ui_out_field_int} and @code{ui_out_spaces}. The original
1692 if (b->forked_inferior_pid != 0)
1693 printf_filtered ("process %d ", b->forked_inferior_pid);
1700 if (b->forked_inferior_pid != 0)
1702 ui_out_text (uiout, "process ");
1703 ui_out_field_int (uiout, "what", b->forked_inferior_pid);
1704 ui_out_spaces (uiout, 1);
1708 Here's an example of using @code{ui_out_field_string}. The original
1713 if (b->exec_pathname != NULL)
1714 printf_filtered ("program \"%s\" ", b->exec_pathname);
1721 if (b->exec_pathname != NULL)
1723 ui_out_text (uiout, "program \"");
1724 ui_out_field_string (uiout, "what", b->exec_pathname);
1725 ui_out_text (uiout, "\" ");
1729 Finally, here's an example of printing an address. The original code:
1733 printf_filtered ("%s ",
1734 hex_string_custom ((unsigned long) b->address, 8));
1741 ui_out_field_core_addr (uiout, "Address", b->address);
1745 @section Console Printing
1754 @cindex @code{libgdb}
1755 @code{libgdb} 1.0 was an abortive project of years ago. The theory was
1756 to provide an API to @value{GDBN}'s functionality.
1759 @cindex @code{libgdb}
1760 @code{libgdb} 2.0 is an ongoing effort to update @value{GDBN} so that is
1761 better able to support graphical and other environments.
1763 Since @code{libgdb} development is on-going, its architecture is still
1764 evolving. The following components have so far been identified:
1768 Observer - @file{gdb-events.h}.
1770 Builder - @file{ui-out.h}
1772 Event Loop - @file{event-loop.h}
1774 Library - @file{gdb.h}
1777 The model that ties these components together is described below.
1779 @section The @code{libgdb} Model
1781 A client of @code{libgdb} interacts with the library in two ways.
1785 As an observer (using @file{gdb-events}) receiving notifications from
1786 @code{libgdb} of any internal state changes (break point changes, run
1789 As a client querying @code{libgdb} (using the @file{ui-out} builder) to
1790 obtain various status values from @value{GDBN}.
1793 Since @code{libgdb} could have multiple clients (e.g., a GUI supporting
1794 the existing @value{GDBN} CLI), those clients must co-operate when
1795 controlling @code{libgdb}. In particular, a client must ensure that
1796 @code{libgdb} is idle (i.e.@: no other client is using @code{libgdb})
1797 before responding to a @file{gdb-event} by making a query.
1799 @section CLI support
1801 At present @value{GDBN}'s CLI is very much entangled in with the core of
1802 @code{libgdb}. Consequently, a client wishing to include the CLI in
1803 their interface needs to carefully co-ordinate its own and the CLI's
1806 It is suggested that the client set @code{libgdb} up to be bi-modal
1807 (alternate between CLI and client query modes). The notes below sketch
1812 The client registers itself as an observer of @code{libgdb}.
1814 The client create and install @code{cli-out} builder using its own
1815 versions of the @code{ui-file} @code{gdb_stderr}, @code{gdb_stdtarg} and
1816 @code{gdb_stdout} streams.
1818 The client creates a separate custom @code{ui-out} builder that is only
1819 used while making direct queries to @code{libgdb}.
1822 When the client receives input intended for the CLI, it simply passes it
1823 along. Since the @code{cli-out} builder is installed by default, all
1824 the CLI output in response to that command is routed (pronounced rooted)
1825 through to the client controlled @code{gdb_stdout} et.@: al.@: streams.
1826 At the same time, the client is kept abreast of internal changes by
1827 virtue of being a @code{libgdb} observer.
1829 The only restriction on the client is that it must wait until
1830 @code{libgdb} becomes idle before initiating any queries (using the
1831 client's custom builder).
1833 @section @code{libgdb} components
1835 @subheading Observer - @file{gdb-events.h}
1836 @file{gdb-events} provides the client with a very raw mechanism that can
1837 be used to implement an observer. At present it only allows for one
1838 observer and that observer must, internally, handle the need to delay
1839 the processing of any event notifications until after @code{libgdb} has
1840 finished the current command.
1842 @subheading Builder - @file{ui-out.h}
1843 @file{ui-out} provides the infrastructure necessary for a client to
1844 create a builder. That builder is then passed down to @code{libgdb}
1845 when doing any queries.
1847 @subheading Event Loop - @file{event-loop.h}
1848 @c There could be an entire section on the event-loop
1849 @file{event-loop}, currently non-re-entrant, provides a simple event
1850 loop. A client would need to either plug its self into this loop or,
1851 implement a new event-loop that @value{GDBN} would use.
1853 The event-loop will eventually be made re-entrant. This is so that
1854 @value{GDBN} can better handle the problem of some commands blocking
1855 instead of returning.
1857 @subheading Library - @file{gdb.h}
1858 @file{libgdb} is the most obvious component of this system. It provides
1859 the query interface. Each function is parameterized by a @code{ui-out}
1860 builder. The result of the query is constructed using that builder
1861 before the query function returns.
1868 @cindex @code{value} structure
1869 @value{GDBN} uses @code{struct value}, or @dfn{values}, as an internal
1870 abstraction for the representation of a variety of inferior objects
1871 and @value{GDBN} convenience objects.
1873 Values have an associated @code{struct type}, that describes a virtual
1874 view of the raw data or object stored in or accessed through the
1877 A value is in addition discriminated by its lvalue-ness, given its
1878 @code{enum lval_type} enumeration type:
1880 @cindex @code{lval_type} enumeration, for values.
1882 @item @code{not_lval}
1883 This value is not an lval. It can't be assigned to.
1885 @item @code{lval_memory}
1886 This value represents an object in memory.
1888 @item @code{lval_register}
1889 This value represents an object that lives in a register.
1891 @item @code{lval_internalvar}
1892 Represents the value of an internal variable.
1894 @item @code{lval_internalvar_component}
1895 Represents part of a @value{GDBN} internal variable. E.g., a
1898 @cindex computed values
1899 @item @code{lval_computed}
1900 These are ``computed'' values. They allow creating specialized value
1901 objects for specific purposes, all abstracted away from the core value
1902 support code. The creator of such a value writes specialized
1903 functions to handle the reading and writing to/from the value's
1904 backend data, and optionally, a ``copy operator'' and a
1907 Pointers to these functions are stored in a @code{struct lval_funcs}
1908 instance (declared in @file{value.h}), and passed to the
1909 @code{allocate_computed_value} function, as in the example below.
1913 nil_value_read (struct value *v)
1915 /* This callback reads data from some backend, and stores it in V.
1916 In this case, we always read null data. You'll want to fill in
1917 something more interesting. */
1919 memset (value_contents_all_raw (v),
1921 TYPE_LENGTH (value_type (v)));
1925 nil_value_write (struct value *v, struct value *fromval)
1927 /* Takes the data from FROMVAL and stores it in the backend of V. */
1929 to_oblivion (value_contents_all_raw (fromval),
1931 TYPE_LENGTH (value_type (fromval)));
1934 static struct lval_funcs nil_value_funcs =
1941 make_nil_value (void)
1946 type = make_nils_type ();
1947 v = allocate_computed_value (type, &nil_value_funcs, NULL);
1953 See the implementation of the @code{$_siginfo} convenience variable in
1954 @file{infrun.c} as a real example use of lval_computed.
1959 @chapter Stack Frames
1962 @cindex call stack frame
1963 A frame is a construct that @value{GDBN} uses to keep track of calling
1964 and called functions.
1966 @cindex unwind frame
1967 @value{GDBN}'s frame model, a fresh design, was implemented with the
1968 need to support @sc{dwarf}'s Call Frame Information in mind. In fact,
1969 the term ``unwind'' is taken directly from that specification.
1970 Developers wishing to learn more about unwinders, are encouraged to
1971 read the @sc{dwarf} specification, available from
1972 @url{http://www.dwarfstd.org}.
1974 @findex frame_register_unwind
1975 @findex get_frame_register
1976 @value{GDBN}'s model is that you find a frame's registers by
1977 ``unwinding'' them from the next younger frame. That is,
1978 @samp{get_frame_register} which returns the value of a register in
1979 frame #1 (the next-to-youngest frame), is implemented by calling frame
1980 #0's @code{frame_register_unwind} (the youngest frame). But then the
1981 obvious question is: how do you access the registers of the youngest
1984 @cindex sentinel frame
1985 @findex get_frame_type
1986 @vindex SENTINEL_FRAME
1987 To answer this question, @value{GDBN} has the @dfn{sentinel} frame, the
1988 ``-1st'' frame. Unwinding registers from the sentinel frame gives you
1989 the current values of the youngest real frame's registers. If @var{f}
1990 is a sentinel frame, then @code{get_frame_type (@var{f}) @equiv{}
1993 @section Selecting an Unwinder
1995 @findex frame_unwind_prepend_unwinder
1996 @findex frame_unwind_append_unwinder
1997 The architecture registers a list of frame unwinders (@code{struct
1998 frame_unwind}), using the functions
1999 @code{frame_unwind_prepend_unwinder} and
2000 @code{frame_unwind_append_unwinder}. Each unwinder includes a
2001 sniffer. Whenever @value{GDBN} needs to unwind a frame (to fetch the
2002 previous frame's registers or the current frame's ID), it calls
2003 registered sniffers in order to find one which recognizes the frame.
2004 The first time a sniffer returns non-zero, the corresponding unwinder
2005 is assigned to the frame.
2007 @section Unwinding the Frame ID
2010 Every frame has an associated ID, of type @code{struct frame_id}.
2011 The ID includes the stack base and function start address for
2012 the frame. The ID persists through the entire life of the frame,
2013 including while other called frames are running; it is used to
2014 locate an appropriate @code{struct frame_info} from the cache.
2016 Every time the inferior stops, and at various other times, the frame
2017 cache is flushed. Because of this, parts of @value{GDBN} which need
2018 to keep track of individual frames cannot use pointers to @code{struct
2019 frame_info}. A frame ID provides a stable reference to a frame, even
2020 when the unwinder must be run again to generate a new @code{struct
2021 frame_info} for the same frame.
2023 The frame's unwinder's @code{this_id} method is called to find the ID.
2024 Note that this is different from register unwinding, where the next
2025 frame's @code{prev_register} is called to unwind this frame's
2028 Both stack base and function address are required to identify the
2029 frame, because a recursive function has the same function address for
2030 two consecutive frames and a leaf function may have the same stack
2031 address as its caller. On some platforms, a third address is part of
2032 the ID to further disambiguate frames---for instance, on IA-64
2033 the separate register stack address is included in the ID.
2035 An invalid frame ID (@code{outer_frame_id}) returned from the
2036 @code{this_id} method means to stop unwinding after this frame.
2038 @code{null_frame_id} is another invalid frame ID which should be used
2039 when there is no frame. For instance, certain breakpoints are attached
2040 to a specific frame, and that frame is identified through its frame ID
2041 (we use this to implement the "finish" command). Using
2042 @code{null_frame_id} as the frame ID for a given breakpoint means
2043 that the breakpoint is not specific to any frame. The @code{this_id}
2044 method should never return @code{null_frame_id}.
2046 @section Unwinding Registers
2048 Each unwinder includes a @code{prev_register} method. This method
2049 takes a frame, an associated cache pointer, and a register number.
2050 It returns a @code{struct value *} describing the requested register,
2051 as saved by this frame. This is the value of the register that is
2052 current in this frame's caller.
2054 The returned value must have the same type as the register. It may
2055 have any lvalue type. In most circumstances one of these routines
2056 will generate the appropriate value:
2059 @item frame_unwind_got_optimized
2060 @findex frame_unwind_got_optimized
2061 This register was not saved.
2063 @item frame_unwind_got_register
2064 @findex frame_unwind_got_register
2065 This register was copied into another register in this frame. This
2066 is also used for unchanged registers; they are ``copied'' into the
2069 @item frame_unwind_got_memory
2070 @findex frame_unwind_got_memory
2071 This register was saved in memory.
2073 @item frame_unwind_got_constant
2074 @findex frame_unwind_got_constant
2075 This register was not saved, but the unwinder can compute the previous
2076 value some other way.
2078 @item frame_unwind_got_address
2079 @findex frame_unwind_got_address
2080 Same as @code{frame_unwind_got_constant}, except that the value is a target
2081 address. This is frequently used for the stack pointer, which is not
2082 explicitly saved but has a known offset from this frame's stack
2083 pointer. For architectures with a flat unified address space, this is
2084 generally the same as @code{frame_unwind_got_constant}.
2087 @node Symbol Handling
2089 @chapter Symbol Handling
2091 Symbols are a key part of @value{GDBN}'s operation. Symbols include
2092 variables, functions, and types.
2094 Symbol information for a large program can be truly massive, and
2095 reading of symbol information is one of the major performance
2096 bottlenecks in @value{GDBN}; it can take many minutes to process it
2097 all. Studies have shown that nearly all the time spent is
2098 computational, rather than file reading.
2100 One of the ways for @value{GDBN} to provide a good user experience is
2101 to start up quickly, taking no more than a few seconds. It is simply
2102 not possible to process all of a program's debugging info in that
2103 time, and so we attempt to handle symbols incrementally. For instance,
2104 we create @dfn{partial symbol tables} consisting of only selected
2105 symbols, and only expand them to full symbol tables when necessary.
2107 @section Symbol Reading
2109 @cindex symbol reading
2110 @cindex reading of symbols
2111 @cindex symbol files
2112 @value{GDBN} reads symbols from @dfn{symbol files}. The usual symbol
2113 file is the file containing the program which @value{GDBN} is
2114 debugging. @value{GDBN} can be directed to use a different file for
2115 symbols (with the @samp{symbol-file} command), and it can also read
2116 more symbols via the @samp{add-file} and @samp{load} commands. In
2117 addition, it may bring in more symbols while loading shared
2120 @findex find_sym_fns
2121 Symbol files are initially opened by code in @file{symfile.c} using
2122 the BFD library (@pxref{Support Libraries}). BFD identifies the type
2123 of the file by examining its header. @code{find_sym_fns} then uses
2124 this identification to locate a set of symbol-reading functions.
2126 @findex add_symtab_fns
2127 @cindex @code{sym_fns} structure
2128 @cindex adding a symbol-reading module
2129 Symbol-reading modules identify themselves to @value{GDBN} by calling
2130 @code{add_symtab_fns} during their module initialization. The argument
2131 to @code{add_symtab_fns} is a @code{struct sym_fns} which contains the
2132 name (or name prefix) of the symbol format, the length of the prefix,
2133 and pointers to four functions. These functions are called at various
2134 times to process symbol files whose identification matches the specified
2137 The functions supplied by each module are:
2140 @item @var{xyz}_symfile_init(struct sym_fns *sf)
2142 @cindex secondary symbol file
2143 Called from @code{symbol_file_add} when we are about to read a new
2144 symbol file. This function should clean up any internal state (possibly
2145 resulting from half-read previous files, for example) and prepare to
2146 read a new symbol file. Note that the symbol file which we are reading
2147 might be a new ``main'' symbol file, or might be a secondary symbol file
2148 whose symbols are being added to the existing symbol table.
2150 The argument to @code{@var{xyz}_symfile_init} is a newly allocated
2151 @code{struct sym_fns} whose @code{bfd} field contains the BFD for the
2152 new symbol file being read. Its @code{private} field has been zeroed,
2153 and can be modified as desired. Typically, a struct of private
2154 information will be @code{malloc}'d, and a pointer to it will be placed
2155 in the @code{private} field.
2157 There is no result from @code{@var{xyz}_symfile_init}, but it can call
2158 @code{error} if it detects an unavoidable problem.
2160 @item @var{xyz}_new_init()
2162 Called from @code{symbol_file_add} when discarding existing symbols.
2163 This function needs only handle the symbol-reading module's internal
2164 state; the symbol table data structures visible to the rest of
2165 @value{GDBN} will be discarded by @code{symbol_file_add}. It has no
2166 arguments and no result. It may be called after
2167 @code{@var{xyz}_symfile_init}, if a new symbol table is being read, or
2168 may be called alone if all symbols are simply being discarded.
2170 @item @var{xyz}_symfile_read(struct sym_fns *sf, CORE_ADDR addr, int mainline)
2172 Called from @code{symbol_file_add} to actually read the symbols from a
2173 symbol-file into a set of psymtabs or symtabs.
2175 @code{sf} points to the @code{struct sym_fns} originally passed to
2176 @code{@var{xyz}_sym_init} for possible initialization. @code{addr} is
2177 the offset between the file's specified start address and its true
2178 address in memory. @code{mainline} is 1 if this is the main symbol
2179 table being read, and 0 if a secondary symbol file (e.g., shared library
2180 or dynamically loaded file) is being read.@refill
2183 In addition, if a symbol-reading module creates psymtabs when
2184 @var{xyz}_symfile_read is called, these psymtabs will contain a pointer
2185 to a function @code{@var{xyz}_psymtab_to_symtab}, which can be called
2186 from any point in the @value{GDBN} symbol-handling code.
2189 @item @var{xyz}_psymtab_to_symtab (struct partial_symtab *pst)
2191 Called from @code{psymtab_to_symtab} (or the @code{PSYMTAB_TO_SYMTAB} macro) if
2192 the psymtab has not already been read in and had its @code{pst->symtab}
2193 pointer set. The argument is the psymtab to be fleshed-out into a
2194 symtab. Upon return, @code{pst->readin} should have been set to 1, and
2195 @code{pst->symtab} should contain a pointer to the new corresponding symtab, or
2196 zero if there were no symbols in that part of the symbol file.
2199 @section Partial Symbol Tables
2201 @value{GDBN} has three types of symbol tables:
2204 @cindex full symbol table
2207 Full symbol tables (@dfn{symtabs}). These contain the main
2208 information about symbols and addresses.
2212 Partial symbol tables (@dfn{psymtabs}). These contain enough
2213 information to know when to read the corresponding part of the full
2216 @cindex minimal symbol table
2219 Minimal symbol tables (@dfn{msymtabs}). These contain information
2220 gleaned from non-debugging symbols.
2223 @cindex partial symbol table
2224 This section describes partial symbol tables.
2226 A psymtab is constructed by doing a very quick pass over an executable
2227 file's debugging information. Small amounts of information are
2228 extracted---enough to identify which parts of the symbol table will
2229 need to be re-read and fully digested later, when the user needs the
2230 information. The speed of this pass causes @value{GDBN} to start up very
2231 quickly. Later, as the detailed rereading occurs, it occurs in small
2232 pieces, at various times, and the delay therefrom is mostly invisible to
2234 @c (@xref{Symbol Reading}.)
2236 The symbols that show up in a file's psymtab should be, roughly, those
2237 visible to the debugger's user when the program is not running code from
2238 that file. These include external symbols and types, static symbols and
2239 types, and @code{enum} values declared at file scope.
2241 The psymtab also contains the range of instruction addresses that the
2242 full symbol table would represent.
2244 @cindex finding a symbol
2245 @cindex symbol lookup
2246 The idea is that there are only two ways for the user (or much of the
2247 code in the debugger) to reference a symbol:
2250 @findex find_pc_function
2251 @findex find_pc_line
2253 By its address (e.g., execution stops at some address which is inside a
2254 function in this file). The address will be noticed to be in the
2255 range of this psymtab, and the full symtab will be read in.
2256 @code{find_pc_function}, @code{find_pc_line}, and other
2257 @code{find_pc_@dots{}} functions handle this.
2259 @cindex lookup_symbol
2262 (e.g., the user asks to print a variable, or set a breakpoint on a
2263 function). Global names and file-scope names will be found in the
2264 psymtab, which will cause the symtab to be pulled in. Local names will
2265 have to be qualified by a global name, or a file-scope name, in which
2266 case we will have already read in the symtab as we evaluated the
2267 qualifier. Or, a local symbol can be referenced when we are ``in'' a
2268 local scope, in which case the first case applies. @code{lookup_symbol}
2269 does most of the work here.
2272 The only reason that psymtabs exist is to cause a symtab to be read in
2273 at the right moment. Any symbol that can be elided from a psymtab,
2274 while still causing that to happen, should not appear in it. Since
2275 psymtabs don't have the idea of scope, you can't put local symbols in
2276 them anyway. Psymtabs don't have the idea of the type of a symbol,
2277 either, so types need not appear, unless they will be referenced by
2280 It is a bug for @value{GDBN} to behave one way when only a psymtab has
2281 been read, and another way if the corresponding symtab has been read
2282 in. Such bugs are typically caused by a psymtab that does not contain
2283 all the visible symbols, or which has the wrong instruction address
2286 The psymtab for a particular section of a symbol file (objfile) could be
2287 thrown away after the symtab has been read in. The symtab should always
2288 be searched before the psymtab, so the psymtab will never be used (in a
2289 bug-free environment). Currently, psymtabs are allocated on an obstack,
2290 and all the psymbols themselves are allocated in a pair of large arrays
2291 on an obstack, so there is little to be gained by trying to free them
2292 unless you want to do a lot more work.
2294 Whether or not psymtabs are created depends on the objfile's symbol
2295 reader. The core of @value{GDBN} hides the details of partial symbols
2296 and partial symbol tables behind a set of function pointers known as
2297 the @dfn{quick symbol functions}. These are documented in
2302 @unnumberedsubsec Fundamental Types (e.g., @code{FT_VOID}, @code{FT_BOOLEAN}).
2304 @cindex fundamental types
2305 These are the fundamental types that @value{GDBN} uses internally. Fundamental
2306 types from the various debugging formats (stabs, ELF, etc) are mapped
2307 into one of these. They are basically a union of all fundamental types
2308 that @value{GDBN} knows about for all the languages that @value{GDBN}
2311 @unnumberedsubsec Type Codes (e.g., @code{TYPE_CODE_PTR}, @code{TYPE_CODE_ARRAY}).
2314 Each time @value{GDBN} builds an internal type, it marks it with one
2315 of these types. The type may be a fundamental type, such as
2316 @code{TYPE_CODE_INT}, or a derived type, such as @code{TYPE_CODE_PTR}
2317 which is a pointer to another type. Typically, several @code{FT_*}
2318 types map to one @code{TYPE_CODE_*} type, and are distinguished by
2319 other members of the type struct, such as whether the type is signed
2320 or unsigned, and how many bits it uses.
2322 @unnumberedsubsec Builtin Types (e.g., @code{builtin_type_void}, @code{builtin_type_char}).
2324 These are instances of type structs that roughly correspond to
2325 fundamental types and are created as global types for @value{GDBN} to
2326 use for various ugly historical reasons. We eventually want to
2327 eliminate these. Note for example that @code{builtin_type_int}
2328 initialized in @file{gdbtypes.c} is basically the same as a
2329 @code{TYPE_CODE_INT} type that is initialized in @file{c-lang.c} for
2330 an @code{FT_INTEGER} fundamental type. The difference is that the
2331 @code{builtin_type} is not associated with any particular objfile, and
2332 only one instance exists, while @file{c-lang.c} builds as many
2333 @code{TYPE_CODE_INT} types as needed, with each one associated with
2334 some particular objfile.
2336 @section Object File Formats
2337 @cindex object file formats
2341 @cindex @code{a.out} format
2342 The @code{a.out} format is the original file format for Unix. It
2343 consists of three sections: @code{text}, @code{data}, and @code{bss},
2344 which are for program code, initialized data, and uninitialized data,
2347 The @code{a.out} format is so simple that it doesn't have any reserved
2348 place for debugging information. (Hey, the original Unix hackers used
2349 @samp{adb}, which is a machine-language debugger!) The only debugging
2350 format for @code{a.out} is stabs, which is encoded as a set of normal
2351 symbols with distinctive attributes.
2353 The basic @code{a.out} reader is in @file{dbxread.c}.
2358 The COFF format was introduced with System V Release 3 (SVR3) Unix.
2359 COFF files may have multiple sections, each prefixed by a header. The
2360 number of sections is limited.
2362 The COFF specification includes support for debugging. Although this
2363 was a step forward, the debugging information was woefully limited.
2364 For instance, it was not possible to represent code that came from an
2365 included file. GNU's COFF-using configs often use stabs-type info,
2366 encapsulated in special sections.
2368 The COFF reader is in @file{coffread.c}.
2372 @cindex ECOFF format
2373 ECOFF is an extended COFF originally introduced for Mips and Alpha
2376 The basic ECOFF reader is in @file{mipsread.c}.
2380 @cindex XCOFF format
2381 The IBM RS/6000 running AIX uses an object file format called XCOFF.
2382 The COFF sections, symbols, and line numbers are used, but debugging
2383 symbols are @code{dbx}-style stabs whose strings are located in the
2384 @code{.debug} section (rather than the string table). For more
2385 information, see @ref{Top,,,stabs,The Stabs Debugging Format}.
2387 The shared library scheme has a clean interface for figuring out what
2388 shared libraries are in use, but the catch is that everything which
2389 refers to addresses (symbol tables and breakpoints at least) needs to be
2390 relocated for both shared libraries and the main executable. At least
2391 using the standard mechanism this can only be done once the program has
2392 been run (or the core file has been read).
2396 @cindex PE-COFF format
2397 Windows 95 and NT use the PE (@dfn{Portable Executable}) format for their
2398 executables. PE is basically COFF with additional headers.
2400 While BFD includes special PE support, @value{GDBN} needs only the basic
2406 The ELF format came with System V Release 4 (SVR4) Unix. ELF is
2407 similar to COFF in being organized into a number of sections, but it
2408 removes many of COFF's limitations. Debugging info may be either stabs
2409 encapsulated in ELF sections, or more commonly these days, DWARF.
2411 The basic ELF reader is in @file{elfread.c}.
2416 SOM is HP's object file and debug format (not to be confused with IBM's
2417 SOM, which is a cross-language ABI).
2419 The SOM reader is in @file{somread.c}.
2421 @section Debugging File Formats
2423 This section describes characteristics of debugging information that
2424 are independent of the object file format.
2428 @cindex stabs debugging info
2429 @code{stabs} started out as special symbols within the @code{a.out}
2430 format. Since then, it has been encapsulated into other file
2431 formats, such as COFF and ELF.
2433 While @file{dbxread.c} does some of the basic stab processing,
2434 including for encapsulated versions, @file{stabsread.c} does
2439 @cindex COFF debugging info
2440 The basic COFF definition includes debugging information. The level
2441 of support is minimal and non-extensible, and is not often used.
2443 @subsection Mips debug (Third Eye)
2445 @cindex ECOFF debugging info
2446 ECOFF includes a definition of a special debug format.
2448 The file @file{mdebugread.c} implements reading for this format.
2450 @c mention DWARF 1 as a formerly-supported format
2454 @cindex DWARF 2 debugging info
2455 DWARF 2 is an improved but incompatible version of DWARF 1.
2457 The DWARF 2 reader is in @file{dwarf2read.c}.
2459 @subsection Compressed DWARF 2
2461 @cindex Compressed DWARF 2 debugging info
2462 Compressed DWARF 2 is not technically a separate debugging format, but
2463 merely DWARF 2 debug information that has been compressed. In this
2464 format, every object-file section holding DWARF 2 debugging
2465 information is compressed and prepended with a header. (The section
2466 is also typically renamed, so a section called @code{.debug_info} in a
2467 DWARF 2 binary would be called @code{.zdebug_info} in a compressed
2468 DWARF 2 binary.) The header is 12 bytes long:
2472 4 bytes: the literal string ``ZLIB''
2474 8 bytes: the uncompressed size of the section, in big-endian byte
2478 The same reader is used for both compressed an normal DWARF 2 info.
2479 Section decompression is done in @code{zlib_decompress_section} in
2480 @file{dwarf2read.c}.
2484 @cindex DWARF 3 debugging info
2485 DWARF 3 is an improved version of DWARF 2.
2489 @cindex SOM debugging info
2490 Like COFF, the SOM definition includes debugging information.
2492 @section Adding a New Symbol Reader to @value{GDBN}
2494 @cindex adding debugging info reader
2495 If you are using an existing object file format (@code{a.out}, COFF, ELF, etc),
2496 there is probably little to be done.
2498 If you need to add a new object file format, you must first add it to
2499 BFD. This is beyond the scope of this document.
2501 You must then arrange for the BFD code to provide access to the
2502 debugging symbols. Generally @value{GDBN} will have to call swapping
2503 routines from BFD and a few other BFD internal routines to locate the
2504 debugging information. As much as possible, @value{GDBN} should not
2505 depend on the BFD internal data structures.
2507 For some targets (e.g., COFF), there is a special transfer vector used
2508 to call swapping routines, since the external data structures on various
2509 platforms have different sizes and layouts. Specialized routines that
2510 will only ever be implemented by one object file format may be called
2511 directly. This interface should be described in a file
2512 @file{bfd/lib@var{xyz}.h}, which is included by @value{GDBN}.
2514 @section Memory Management for Symbol Files
2516 Most memory associated with a loaded symbol file is stored on
2517 its @code{objfile_obstack}. This includes symbols, types,
2518 namespace data, and other information produced by the symbol readers.
2520 Because this data lives on the objfile's obstack, it is automatically
2521 released when the objfile is unloaded or reloaded. Therefore one
2522 objfile must not reference symbol or type data from another objfile;
2523 they could be unloaded at different times.
2525 User convenience variables, et cetera, have associated types. Normally
2526 these types live in the associated objfile. However, when the objfile
2527 is unloaded, those types are deep copied to global memory, so that
2528 the values of the user variables and history items are not lost.
2531 @node Language Support
2533 @chapter Language Support
2535 @cindex language support
2536 @value{GDBN}'s language support is mainly driven by the symbol reader,
2537 although it is possible for the user to set the source language
2540 @value{GDBN} chooses the source language by looking at the extension
2541 of the file recorded in the debug info; @file{.c} means C, @file{.f}
2542 means Fortran, etc. It may also use a special-purpose language
2543 identifier if the debug format supports it, like with DWARF.
2545 @section Adding a Source Language to @value{GDBN}
2547 @cindex adding source language
2548 To add other languages to @value{GDBN}'s expression parser, follow the
2552 @item Create the expression parser.
2554 @cindex expression parser
2555 This should reside in a file @file{@var{lang}-exp.y}. Routines for
2556 building parsed expressions into a @code{union exp_element} list are in
2559 @cindex language parser
2560 Since we can't depend upon everyone having Bison, and YACC produces
2561 parsers that define a bunch of global names, the following lines
2562 @strong{must} be included at the top of the YACC parser, to prevent the
2563 various parsers from defining the same global names:
2566 #define yyparse @var{lang}_parse
2567 #define yylex @var{lang}_lex
2568 #define yyerror @var{lang}_error
2569 #define yylval @var{lang}_lval
2570 #define yychar @var{lang}_char
2571 #define yydebug @var{lang}_debug
2572 #define yypact @var{lang}_pact
2573 #define yyr1 @var{lang}_r1
2574 #define yyr2 @var{lang}_r2
2575 #define yydef @var{lang}_def
2576 #define yychk @var{lang}_chk
2577 #define yypgo @var{lang}_pgo
2578 #define yyact @var{lang}_act
2579 #define yyexca @var{lang}_exca
2580 #define yyerrflag @var{lang}_errflag
2581 #define yynerrs @var{lang}_nerrs
2584 At the bottom of your parser, define a @code{struct language_defn} and
2585 initialize it with the right values for your language. Define an
2586 @code{initialize_@var{lang}} routine and have it call
2587 @samp{add_language(@var{lang}_language_defn)} to tell the rest of @value{GDBN}
2588 that your language exists. You'll need some other supporting variables
2589 and functions, which will be used via pointers from your
2590 @code{@var{lang}_language_defn}. See the declaration of @code{struct
2591 language_defn} in @file{language.h}, and the other @file{*-exp.y} files,
2592 for more information.
2594 @item Add any evaluation routines, if necessary
2596 @cindex expression evaluation routines
2597 @findex evaluate_subexp
2598 @findex prefixify_subexp
2599 @findex length_of_subexp
2600 If you need new opcodes (that represent the operations of the language),
2601 add them to the enumerated type in @file{expression.h}. Add support
2602 code for these operations in the @code{evaluate_subexp} function
2603 defined in the file @file{eval.c}. Add cases
2604 for new opcodes in two functions from @file{parse.c}:
2605 @code{prefixify_subexp} and @code{length_of_subexp}. These compute
2606 the number of @code{exp_element}s that a given operation takes up.
2608 @item Update some existing code
2610 Add an enumerated identifier for your language to the enumerated type
2611 @code{enum language} in @file{defs.h}.
2613 Update the routines in @file{language.c} so your language is included.
2614 These routines include type predicates and such, which (in some cases)
2615 are language dependent. If your language does not appear in the switch
2616 statement, an error is reported.
2618 @vindex current_language
2619 Also included in @file{language.c} is the code that updates the variable
2620 @code{current_language}, and the routines that translate the
2621 @code{language_@var{lang}} enumerated identifier into a printable
2624 @findex _initialize_language
2625 Update the function @code{_initialize_language} to include your
2626 language. This function picks the default language upon startup, so is
2627 dependent upon which languages that @value{GDBN} is built for.
2629 @findex allocate_symtab
2630 Update @code{allocate_symtab} in @file{symfile.c} and/or symbol-reading
2631 code so that the language of each symtab (source file) is set properly.
2632 This is used to determine the language to use at each stack frame level.
2633 Currently, the language is set based upon the extension of the source
2634 file. If the language can be better inferred from the symbol
2635 information, please set the language of the symtab in the symbol-reading
2638 @findex print_subexp
2639 @findex op_print_tab
2640 Add helper code to @code{print_subexp} (in @file{expprint.c}) to handle any new
2641 expression opcodes you have added to @file{expression.h}. Also, add the
2642 printed representations of your operators to @code{op_print_tab}.
2644 @item Add a place of call
2647 Add a call to @code{@var{lang}_parse()} and @code{@var{lang}_error} in
2648 @code{parse_exp_1} (defined in @file{parse.c}).
2650 @item Edit @file{Makefile.in}
2652 Add dependencies in @file{Makefile.in}. Make sure you update the macro
2653 variables such as @code{HFILES} and @code{OBJS}, otherwise your code may
2654 not get linked in, or, worse yet, it may not get @code{tar}red into the
2659 @node Host Definition
2661 @chapter Host Definition
2663 With the advent of Autoconf, it's rarely necessary to have host
2664 definition machinery anymore. The following information is provided,
2665 mainly, as an historical reference.
2667 @section Adding a New Host
2669 @cindex adding a new host
2670 @cindex host, adding
2671 @value{GDBN}'s host configuration support normally happens via Autoconf.
2672 New host-specific definitions should not be needed. Older hosts
2673 @value{GDBN} still use the host-specific definitions and files listed
2674 below, but these mostly exist for historical reasons, and will
2675 eventually disappear.
2678 @item gdb/config/@var{arch}/@var{xyz}.mh
2679 This file is a Makefile fragment that once contained both host and
2680 native configuration information (@pxref{Native Debugging}) for the
2681 machine @var{xyz}. The host configuration information is now handled
2684 Host configuration information included definitions for @code{CC},
2685 @code{SYSV_DEFINE}, @code{XM_CFLAGS}, @code{XM_ADD_FILES},
2686 @code{XM_CLIBS}, @code{XM_CDEPS}, etc.; see @file{Makefile.in}.
2688 New host-only configurations do not need this file.
2692 (Files named @file{gdb/config/@var{arch}/xm-@var{xyz}.h} were once
2693 used to define host-specific macros, but were no longer needed and
2694 have all been removed.)
2696 @subheading Generic Host Support Files
2698 @cindex generic host support
2699 There are some ``generic'' versions of routines that can be used by
2703 @cindex remote debugging support
2704 @cindex serial line support
2706 This contains serial line support for Unix systems. It is included by
2707 default on all Unix-like hosts.
2710 This contains serial pipe support for Unix systems. It is included by
2711 default on all Unix-like hosts.
2714 This contains serial line support for 32-bit programs running under
2715 Windows using MinGW.
2718 This contains serial line support for 32-bit programs running under DOS,
2719 using the DJGPP (a.k.a.@: GO32) execution environment.
2721 @cindex TCP remote support
2723 This contains generic TCP support using sockets. It is included by
2724 default on all Unix-like hosts and with MinGW.
2727 @section Host Conditionals
2729 When @value{GDBN} is configured and compiled, various macros are
2730 defined or left undefined, to control compilation based on the
2731 attributes of the host system. While formerly they could be set in
2732 host-specific header files, at present they can be changed only by
2733 setting @code{CFLAGS} when building, or by editing the source code.
2735 These macros and their meanings (or if the meaning is not documented
2736 here, then one of the source files where they are used is indicated)
2740 @item @value{GDBN}INIT_FILENAME
2741 The default name of @value{GDBN}'s initialization file (normally
2744 @item SIGWINCH_HANDLER
2745 If your host defines @code{SIGWINCH}, you can define this to be the name
2746 of a function to be called if @code{SIGWINCH} is received.
2748 @item SIGWINCH_HANDLER_BODY
2749 Define this to expand into code that will define the function named by
2750 the expansion of @code{SIGWINCH_HANDLER}.
2752 @item CRLF_SOURCE_FILES
2753 @cindex DOS text files
2754 Define this if host files use @code{\r\n} rather than @code{\n} as a
2755 line terminator. This will cause source file listings to omit @code{\r}
2756 characters when printing and it will allow @code{\r\n} line endings of files
2757 which are ``sourced'' by gdb. It must be possible to open files in binary
2758 mode using @code{O_BINARY} or, for fopen, @code{"rb"}.
2760 @item DEFAULT_PROMPT
2762 The default value of the prompt string (normally @code{"(gdb) "}).
2765 @cindex terminal device
2766 The name of the generic TTY device, defaults to @code{"/dev/tty"}.
2769 Substitute for isatty, if not available.
2772 Define this if binary files are opened the same way as text files.
2774 @item CC_HAS_LONG_LONG
2775 @cindex @code{long long} data type
2776 Define this if the host C compiler supports @code{long long}. This is set
2777 by the @code{configure} script.
2779 @item PRINTF_HAS_LONG_LONG
2780 Define this if the host can handle printing of long long integers via
2781 the printf format conversion specifier @code{ll}. This is set by the
2782 @code{configure} script.
2784 @item LSEEK_NOT_LINEAR
2785 Define this if @code{lseek (n)} does not necessarily move to byte number
2786 @code{n} in the file. This is only used when reading source files. It
2787 is normally faster to define @code{CRLF_SOURCE_FILES} when possible.
2790 Define this to help placate @code{lint} in some situations.
2793 Define this to override the defaults of @code{__volatile__} or
2798 @node Target Architecture Definition
2800 @chapter Target Architecture Definition
2802 @cindex target architecture definition
2803 @value{GDBN}'s target architecture defines what sort of
2804 machine-language programs @value{GDBN} can work with, and how it works
2807 The target architecture object is implemented as the C structure
2808 @code{struct gdbarch *}. The structure, and its methods, are generated
2809 using the Bourne shell script @file{gdbarch.sh}.
2812 * OS ABI Variant Handling::
2813 * Initialize New Architecture::
2814 * Registers and Memory::
2815 * Pointers and Addresses::
2817 * Register Representation::
2818 * Frame Interpretation::
2819 * Inferior Call Setup::
2820 * Adding support for debugging core files::
2821 * Defining Other Architecture Features::
2822 * Adding a New Target::
2825 @node OS ABI Variant Handling
2826 @section Operating System ABI Variant Handling
2827 @cindex OS ABI variants
2829 @value{GDBN} provides a mechanism for handling variations in OS
2830 ABIs. An OS ABI variant may have influence over any number of
2831 variables in the target architecture definition. There are two major
2832 components in the OS ABI mechanism: sniffers and handlers.
2834 A @dfn{sniffer} examines a file matching a BFD architecture/flavour pair
2835 (the architecture may be wildcarded) in an attempt to determine the
2836 OS ABI of that file. Sniffers with a wildcarded architecture are considered
2837 to be @dfn{generic}, while sniffers for a specific architecture are
2838 considered to be @dfn{specific}. A match from a specific sniffer
2839 overrides a match from a generic sniffer. Multiple sniffers for an
2840 architecture/flavour may exist, in order to differentiate between two
2841 different operating systems which use the same basic file format. The
2842 OS ABI framework provides a generic sniffer for ELF-format files which
2843 examines the @code{EI_OSABI} field of the ELF header, as well as note
2844 sections known to be used by several operating systems.
2846 @cindex fine-tuning @code{gdbarch} structure
2847 A @dfn{handler} is used to fine-tune the @code{gdbarch} structure for the
2848 selected OS ABI. There may be only one handler for a given OS ABI
2849 for each BFD architecture.
2851 The following OS ABI variants are defined in @file{defs.h}:
2855 @findex GDB_OSABI_UNINITIALIZED
2856 @item GDB_OSABI_UNINITIALIZED
2857 Used for struct gdbarch_info if ABI is still uninitialized.
2859 @findex GDB_OSABI_UNKNOWN
2860 @item GDB_OSABI_UNKNOWN
2861 The ABI of the inferior is unknown. The default @code{gdbarch}
2862 settings for the architecture will be used.
2864 @findex GDB_OSABI_SVR4
2865 @item GDB_OSABI_SVR4
2866 UNIX System V Release 4.
2868 @findex GDB_OSABI_HURD
2869 @item GDB_OSABI_HURD
2870 GNU using the Hurd kernel.
2872 @findex GDB_OSABI_SOLARIS
2873 @item GDB_OSABI_SOLARIS
2876 @findex GDB_OSABI_OSF1
2877 @item GDB_OSABI_OSF1
2878 OSF/1, including Digital UNIX and Compaq Tru64 UNIX.
2880 @findex GDB_OSABI_LINUX
2881 @item GDB_OSABI_LINUX
2882 GNU using the Linux kernel.
2884 @findex GDB_OSABI_FREEBSD_AOUT
2885 @item GDB_OSABI_FREEBSD_AOUT
2886 FreeBSD using the @code{a.out} executable format.
2888 @findex GDB_OSABI_FREEBSD_ELF
2889 @item GDB_OSABI_FREEBSD_ELF
2890 FreeBSD using the ELF executable format.
2892 @findex GDB_OSABI_NETBSD_AOUT
2893 @item GDB_OSABI_NETBSD_AOUT
2894 NetBSD using the @code{a.out} executable format.
2896 @findex GDB_OSABI_NETBSD_ELF
2897 @item GDB_OSABI_NETBSD_ELF
2898 NetBSD using the ELF executable format.
2900 @findex GDB_OSABI_OPENBSD_ELF
2901 @item GDB_OSABI_OPENBSD_ELF
2902 OpenBSD using the ELF executable format.
2904 @findex GDB_OSABI_WINCE
2905 @item GDB_OSABI_WINCE
2908 @findex GDB_OSABI_GO32
2909 @item GDB_OSABI_GO32
2912 @findex GDB_OSABI_IRIX
2913 @item GDB_OSABI_IRIX
2916 @findex GDB_OSABI_INTERIX
2917 @item GDB_OSABI_INTERIX
2918 Interix (Posix layer for MS-Windows systems).
2920 @findex GDB_OSABI_HPUX_ELF
2921 @item GDB_OSABI_HPUX_ELF
2922 HP/UX using the ELF executable format.
2924 @findex GDB_OSABI_HPUX_SOM
2925 @item GDB_OSABI_HPUX_SOM
2926 HP/UX using the SOM executable format.
2928 @findex GDB_OSABI_QNXNTO
2929 @item GDB_OSABI_QNXNTO
2932 @findex GDB_OSABI_CYGWIN
2933 @item GDB_OSABI_CYGWIN
2936 @findex GDB_OSABI_AIX
2942 Here are the functions that make up the OS ABI framework:
2944 @deftypefun {const char *} gdbarch_osabi_name (enum gdb_osabi @var{osabi})
2945 Return the name of the OS ABI corresponding to @var{osabi}.
2948 @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}))
2949 Register the OS ABI handler specified by @var{init_osabi} for the
2950 architecture, machine type and OS ABI specified by @var{arch},
2951 @var{machine} and @var{osabi}. In most cases, a value of zero for the
2952 machine type, which implies the architecture's default machine type,
2956 @deftypefun void gdbarch_register_osabi_sniffer (enum bfd_architecture @var{arch}, enum bfd_flavour @var{flavour}, enum gdb_osabi (*@var{sniffer})(bfd *@var{abfd}))
2957 Register the OS ABI file sniffer specified by @var{sniffer} for the
2958 BFD architecture/flavour pair specified by @var{arch} and @var{flavour}.
2959 If @var{arch} is @code{bfd_arch_unknown}, the sniffer is considered to
2960 be generic, and is allowed to examine @var{flavour}-flavoured files for
2964 @deftypefun {enum gdb_osabi} gdbarch_lookup_osabi (bfd *@var{abfd})
2965 Examine the file described by @var{abfd} to determine its OS ABI.
2966 The value @code{GDB_OSABI_UNKNOWN} is returned if the OS ABI cannot
2970 @deftypefun void gdbarch_init_osabi (struct gdbarch info @var{info}, struct gdbarch *@var{gdbarch}, enum gdb_osabi @var{osabi})
2971 Invoke the OS ABI handler corresponding to @var{osabi} to fine-tune the
2972 @code{gdbarch} structure specified by @var{gdbarch}. If a handler
2973 corresponding to @var{osabi} has not been registered for @var{gdbarch}'s
2974 architecture, a warning will be issued and the debugging session will continue
2975 with the defaults already established for @var{gdbarch}.
2978 @deftypefun void generic_elf_osabi_sniff_abi_tag_sections (bfd *@var{abfd}, asection *@var{sect}, void *@var{obj})
2979 Helper routine for ELF file sniffers. Examine the file described by
2980 @var{abfd} and look at ABI tag note sections to determine the OS ABI
2981 from the note. This function should be called via
2982 @code{bfd_map_over_sections}.
2985 @node Initialize New Architecture
2986 @section Initializing a New Architecture
2989 * How an Architecture is Represented::
2990 * Looking Up an Existing Architecture::
2991 * Creating a New Architecture::
2994 @node How an Architecture is Represented
2995 @subsection How an Architecture is Represented
2996 @cindex architecture representation
2997 @cindex representation of architecture
2999 Each @code{gdbarch} is associated with a single @sc{bfd} architecture,
3000 via a @code{bfd_arch_@var{arch}} in the @code{bfd_architecture}
3001 enumeration. The @code{gdbarch} is registered by a call to
3002 @code{register_gdbarch_init}, usually from the file's
3003 @code{_initialize_@var{filename}} routine, which will be automatically
3004 called during @value{GDBN} startup. The arguments are a @sc{bfd}
3005 architecture constant and an initialization function.
3007 @findex _initialize_@var{arch}_tdep
3008 @cindex @file{@var{arch}-tdep.c}
3009 A @value{GDBN} description for a new architecture, @var{arch} is created by
3010 defining a global function @code{_initialize_@var{arch}_tdep}, by
3011 convention in the source file @file{@var{arch}-tdep.c}. For example,
3012 in the case of the OpenRISC 1000, this function is called
3013 @code{_initialize_or1k_tdep} and is found in the file
3016 @cindex @file{configure.tgt}
3017 @cindex @code{gdbarch}
3018 @findex gdbarch_register
3019 The resulting object files containing the implementation of the
3020 @code{_initialize_@var{arch}_tdep} function are specified in the @value{GDBN}
3021 @file{configure.tgt} file, which includes a large case statement
3022 pattern matching against the @code{--target} option of the
3023 @code{configure} script. The new @code{struct gdbarch} is created
3024 within the @code{_initialize_@var{arch}_tdep} function by calling
3025 @code{gdbarch_register}:
3028 void gdbarch_register (enum bfd_architecture @var{architecture},
3029 gdbarch_init_ftype *@var{init_func},
3030 gdbarch_dump_tdep_ftype *@var{tdep_dump_func});
3033 The @var{architecture} will identify the unique @sc{bfd} to be
3034 associated with this @code{gdbarch}. The @var{init_func} funciton is
3035 called to create and return the new @code{struct gdbarch}. The
3036 @var{tdep_dump_func} function will dump the target specific details
3037 associated with this architecture.
3039 For example the function @code{_initialize_or1k_tdep} creates its
3040 architecture for 32-bit OpenRISC 1000 architectures by calling:
3043 gdbarch_register (bfd_arch_or32, or1k_gdbarch_init, or1k_dump_tdep);
3046 @node Looking Up an Existing Architecture
3047 @subsection Looking Up an Existing Architecture
3048 @cindex @code{gdbarch} lookup
3050 The initialization function has this prototype:
3053 static struct gdbarch *
3054 @var{arch}_gdbarch_init (struct gdbarch_info @var{info},
3055 struct gdbarch_list *@var{arches})
3058 The @var{info} argument contains parameters used to select the correct
3059 architecture, and @var{arches} is a list of architectures which
3060 have already been created with the same @code{bfd_arch_@var{arch}}
3063 The initialization function should first make sure that @var{info}
3064 is acceptable, and return @code{NULL} if it is not. Then, it should
3065 search through @var{arches} for an exact match to @var{info}, and
3066 return one if found. Lastly, if no exact match was found, it should
3067 create a new architecture based on @var{info} and return it.
3069 @findex gdbarch_list_lookup_by_info
3070 @cindex @code{gdbarch_info}
3071 The lookup is done using @code{gdbarch_list_lookup_by_info}. It is
3072 passed the list of existing architectures, @var{arches}, and the
3073 @code{struct gdbarch_info}, @var{info}, and returns the first matching
3074 architecture it finds, or @code{NULL} if none are found. If an
3075 architecture is found it can be returned as the result from the
3076 initialization function, otherwise a new @code{struct gdbach} will need
3079 The struct gdbarch_info has the following components:
3084 const struct bfd_arch_info *bfd_arch_info;
3087 struct gdbarch_tdep_info *tdep_info;
3088 enum gdb_osabi osabi;
3089 const struct target_desc *target_desc;
3093 @vindex bfd_arch_info
3094 The @code{bfd_arch_info} member holds the key details about the
3095 architecture. The @code{byte_order} member is a value in an
3096 enumeration indicating the endianism. The @code{abfd} member is a
3097 pointer to the full @sc{bfd}, the @code{tdep_info} member is
3098 additional custom target specific information, @code{osabi} identifies
3099 which (if any) of a number of operating specific ABIs are used by this
3100 architecture and the @code{target_desc} member is a set of name-value
3101 pairs with information about register usage in this target.
3103 When the @code{struct gdbarch} initialization function is called, not
3104 all the fields are provided---only those which can be deduced from the
3105 @sc{bfd}. The @code{struct gdbarch_info}, @var{info} is used as a
3106 look-up key with the list of existing architectures, @var{arches} to
3107 see if a suitable architecture already exists. The @var{tdep_info},
3108 @var{osabi} and @var{target_desc} fields may be added before this
3109 lookup to refine the search.
3111 Only information in @var{info} should be used to choose the new
3112 architecture. Historically, @var{info} could be sparse, and
3113 defaults would be collected from the first element on @var{arches}.
3114 However, @value{GDBN} now fills in @var{info} more thoroughly,
3115 so new @code{gdbarch} initialization functions should not take
3116 defaults from @var{arches}.
3118 @node Creating a New Architecture
3119 @subsection Creating a New Architecture
3120 @cindex @code{struct gdbarch} creation
3122 @findex gdbarch_alloc
3123 @cindex @code{gdbarch_tdep} when allocating new @code{gdbarch}
3124 If no architecture is found, then a new architecture must be created,
3125 by calling @code{gdbarch_alloc} using the supplied @code{@w{struct
3126 gdbarch_info}} and any additional custom target specific
3127 information in a @code{struct gdbarch_tdep}. The prototype for
3128 @code{gdbarch_alloc} is:
3131 struct gdbarch *gdbarch_alloc (const struct gdbarch_info *@var{info},
3132 struct gdbarch_tdep *@var{tdep});
3135 @cindex @code{set_gdbarch} functions
3136 @cindex @code{gdbarch} accessor functions
3137 The newly created struct gdbarch must then be populated. Although
3138 there are default values, in most cases they are not what is
3141 For each element, @var{X}, there is are a pair of corresponding accessor
3142 functions, one to set the value of that element,
3143 @code{set_gdbarch_@var{X}}, the second to either get the value of an
3144 element (if it is a variable) or to apply the element (if it is a
3145 function), @code{gdbarch_@var{X}}. Note that both accessor functions
3146 take a pointer to the @code{@w{struct gdbarch}} as first
3147 argument. Populating the new @code{gdbarch} should use the
3148 @code{set_gdbarch} functions.
3150 The following sections identify the main elements that should be set
3151 in this way. This is not the complete list, but represents the
3152 functions and elements that must commonly be specified for a new
3153 architecture. Many of the functions and variables are described in the
3154 header file @file{gdbarch.h}.
3156 This is the main work in defining a new architecture. Implementing the
3157 set of functions to populate the @code{struct gdbarch}.
3159 @cindex @code{gdbarch_tdep} definition
3160 @code{struct gdbarch_tdep} is not defined within @value{GDBN}---it is up
3161 to the user to define this struct if it is needed to hold custom target
3162 information that is not covered by the standard @code{@w{struct
3163 gdbarch}}. For example with the OpenRISC 1000 architecture it is used to
3164 hold the number of matchpoints available in the target (along with other
3167 If there is no additional target specific information, it can be set to
3170 @node Registers and Memory
3171 @section Registers and Memory
3173 @value{GDBN}'s model of the target machine is rather simple.
3174 @value{GDBN} assumes the machine includes a bank of registers and a
3175 block of memory. Each register may have a different size.
3177 @value{GDBN} does not have a magical way to match up with the
3178 compiler's idea of which registers are which; however, it is critical
3179 that they do match up accurately. The only way to make this work is
3180 to get accurate information about the order that the compiler uses,
3181 and to reflect that in the @code{gdbarch_register_name} and related functions.
3183 @value{GDBN} can handle big-endian, little-endian, and bi-endian architectures.
3185 @node Pointers and Addresses
3186 @section Pointers Are Not Always Addresses
3187 @cindex pointer representation
3188 @cindex address representation
3189 @cindex word-addressed machines
3190 @cindex separate data and code address spaces
3191 @cindex spaces, separate data and code address
3192 @cindex address spaces, separate data and code
3193 @cindex code pointers, word-addressed
3194 @cindex converting between pointers and addresses
3195 @cindex D10V addresses
3197 On almost all 32-bit architectures, the representation of a pointer is
3198 indistinguishable from the representation of some fixed-length number
3199 whose value is the byte address of the object pointed to. On such
3200 machines, the words ``pointer'' and ``address'' can be used interchangeably.
3201 However, architectures with smaller word sizes are often cramped for
3202 address space, so they may choose a pointer representation that breaks this
3203 identity, and allows a larger code address space.
3205 @c D10V is gone from sources - more current example?
3207 For example, the Renesas D10V is a 16-bit VLIW processor whose
3208 instructions are 32 bits long@footnote{Some D10V instructions are
3209 actually pairs of 16-bit sub-instructions. However, since you can't
3210 jump into the middle of such a pair, code addresses can only refer to
3211 full 32 bit instructions, which is what matters in this explanation.}.
3212 If the D10V used ordinary byte addresses to refer to code locations,
3213 then the processor would only be able to address 64kb of instructions.
3214 However, since instructions must be aligned on four-byte boundaries, the
3215 low two bits of any valid instruction's byte address are always
3216 zero---byte addresses waste two bits. So instead of byte addresses,
3217 the D10V uses word addresses---byte addresses shifted right two bits---to
3218 refer to code. Thus, the D10V can use 16-bit words to address 256kb of
3221 However, this means that code pointers and data pointers have different
3222 forms on the D10V. The 16-bit word @code{0xC020} refers to byte address
3223 @code{0xC020} when used as a data address, but refers to byte address
3224 @code{0x30080} when used as a code address.
3226 (The D10V also uses separate code and data address spaces, which also
3227 affects the correspondence between pointers and addresses, but we're
3228 going to ignore that here; this example is already too long.)
3230 To cope with architectures like this---the D10V is not the only
3231 one!---@value{GDBN} tries to distinguish between @dfn{addresses}, which are
3232 byte numbers, and @dfn{pointers}, which are the target's representation
3233 of an address of a particular type of data. In the example above,
3234 @code{0xC020} is the pointer, which refers to one of the addresses
3235 @code{0xC020} or @code{0x30080}, depending on the type imposed upon it.
3236 @value{GDBN} provides functions for turning a pointer into an address
3237 and vice versa, in the appropriate way for the current architecture.
3239 Unfortunately, since addresses and pointers are identical on almost all
3240 processors, this distinction tends to bit-rot pretty quickly. Thus,
3241 each time you port @value{GDBN} to an architecture which does
3242 distinguish between pointers and addresses, you'll probably need to
3243 clean up some architecture-independent code.
3245 Here are functions which convert between pointers and addresses:
3247 @deftypefun CORE_ADDR extract_typed_address (void *@var{buf}, struct type *@var{type})
3248 Treat the bytes at @var{buf} as a pointer or reference of type
3249 @var{type}, and return the address it represents, in a manner
3250 appropriate for the current architecture. This yields an address
3251 @value{GDBN} can use to read target memory, disassemble, etc. Note that
3252 @var{buf} refers to a buffer in @value{GDBN}'s memory, not the
3255 For example, if the current architecture is the Intel x86, this function
3256 extracts a little-endian integer of the appropriate length from
3257 @var{buf} and returns it. However, if the current architecture is the
3258 D10V, this function will return a 16-bit integer extracted from
3259 @var{buf}, multiplied by four if @var{type} is a pointer to a function.
3261 If @var{type} is not a pointer or reference type, then this function
3262 will signal an internal error.
3265 @deftypefun CORE_ADDR store_typed_address (void *@var{buf}, struct type *@var{type}, CORE_ADDR @var{addr})
3266 Store the address @var{addr} in @var{buf}, in the proper format for a
3267 pointer of type @var{type} in the current architecture. Note that
3268 @var{buf} refers to a buffer in @value{GDBN}'s memory, not the
3271 For example, if the current architecture is the Intel x86, this function
3272 stores @var{addr} unmodified as a little-endian integer of the
3273 appropriate length in @var{buf}. However, if the current architecture
3274 is the D10V, this function divides @var{addr} by four if @var{type} is
3275 a pointer to a function, and then stores it in @var{buf}.
3277 If @var{type} is not a pointer or reference type, then this function
3278 will signal an internal error.
3281 @deftypefun CORE_ADDR value_as_address (struct value *@var{val})
3282 Assuming that @var{val} is a pointer, return the address it represents,
3283 as appropriate for the current architecture.
3285 This function actually works on integral values, as well as pointers.
3286 For pointers, it performs architecture-specific conversions as
3287 described above for @code{extract_typed_address}.
3290 @deftypefun CORE_ADDR value_from_pointer (struct type *@var{type}, CORE_ADDR @var{addr})
3291 Create and return a value representing a pointer of type @var{type} to
3292 the address @var{addr}, as appropriate for the current architecture.
3293 This function performs architecture-specific conversions as described
3294 above for @code{store_typed_address}.
3297 Here are two functions which architectures can define to indicate the
3298 relationship between pointers and addresses. These have default
3299 definitions, appropriate for architectures on which all pointers are
3300 simple unsigned byte addresses.
3302 @deftypefun CORE_ADDR gdbarch_pointer_to_address (struct gdbarch *@var{gdbarch}, struct type *@var{type}, char *@var{buf})
3303 Assume that @var{buf} holds a pointer of type @var{type}, in the
3304 appropriate format for the current architecture. Return the byte
3305 address the pointer refers to.
3307 This function may safely assume that @var{type} is either a pointer or a
3308 C@t{++} reference type.
3311 @deftypefun void gdbarch_address_to_pointer (struct gdbarch *@var{gdbarch}, struct type *@var{type}, char *@var{buf}, CORE_ADDR @var{addr})
3312 Store in @var{buf} a pointer of type @var{type} representing the address
3313 @var{addr}, in the appropriate format for the current architecture.
3315 This function may safely assume that @var{type} is either a pointer or a
3316 C@t{++} reference type.
3319 @node Address Classes
3320 @section Address Classes
3321 @cindex address classes
3322 @cindex DW_AT_byte_size
3323 @cindex DW_AT_address_class
3325 Sometimes information about different kinds of addresses is available
3326 via the debug information. For example, some programming environments
3327 define addresses of several different sizes. If the debug information
3328 distinguishes these kinds of address classes through either the size
3329 info (e.g, @code{DW_AT_byte_size} in @w{DWARF 2}) or through an explicit
3330 address class attribute (e.g, @code{DW_AT_address_class} in @w{DWARF 2}), the
3331 following macros should be defined in order to disambiguate these
3332 types within @value{GDBN} as well as provide the added information to
3333 a @value{GDBN} user when printing type expressions.
3335 @deftypefun int gdbarch_address_class_type_flags (struct gdbarch *@var{gdbarch}, int @var{byte_size}, int @var{dwarf2_addr_class})
3336 Returns the type flags needed to construct a pointer type whose size
3337 is @var{byte_size} and whose address class is @var{dwarf2_addr_class}.
3338 This function is normally called from within a symbol reader. See
3339 @file{dwarf2read.c}.
3342 @deftypefun {char *} gdbarch_address_class_type_flags_to_name (struct gdbarch *@var{gdbarch}, int @var{type_flags})
3343 Given the type flags representing an address class qualifier, return
3346 @deftypefun int gdbarch_address_class_name_to_type_flags (struct gdbarch *@var{gdbarch}, int @var{name}, int *@var{type_flags_ptr})
3347 Given an address qualifier name, set the @code{int} referenced by @var{type_flags_ptr} to the type flags
3348 for that address class qualifier.
3351 Since the need for address classes is rather rare, none of
3352 the address class functions are defined by default. Predicate
3353 functions are provided to detect when they are defined.
3355 Consider a hypothetical architecture in which addresses are normally
3356 32-bits wide, but 16-bit addresses are also supported. Furthermore,
3357 suppose that the @w{DWARF 2} information for this architecture simply
3358 uses a @code{DW_AT_byte_size} value of 2 to indicate the use of one
3359 of these "short" pointers. The following functions could be defined
3360 to implement the address class functions:
3363 somearch_address_class_type_flags (int byte_size,
3364 int dwarf2_addr_class)
3367 return TYPE_FLAG_ADDRESS_CLASS_1;
3373 somearch_address_class_type_flags_to_name (int type_flags)
3375 if (type_flags & TYPE_FLAG_ADDRESS_CLASS_1)
3382 somearch_address_class_name_to_type_flags (char *name,
3383 int *type_flags_ptr)
3385 if (strcmp (name, "short") == 0)
3387 *type_flags_ptr = TYPE_FLAG_ADDRESS_CLASS_1;
3395 The qualifier @code{@@short} is used in @value{GDBN}'s type expressions
3396 to indicate the presence of one of these ``short'' pointers. For
3397 example if the debug information indicates that @code{short_ptr_var} is
3398 one of these short pointers, @value{GDBN} might show the following
3402 (gdb) ptype short_ptr_var
3403 type = int * @@short
3407 @node Register Representation
3408 @section Register Representation
3411 * Raw and Cooked Registers::
3412 * Register Architecture Functions & Variables::
3413 * Register Information Functions::
3414 * Register and Memory Data::
3415 * Register Caching::
3418 @node Raw and Cooked Registers
3419 @subsection Raw and Cooked Registers
3420 @cindex raw register representation
3421 @cindex cooked register representation
3422 @cindex representations, raw and cooked registers
3424 @value{GDBN} considers registers to be a set with members numbered
3425 linearly from 0 upwards. The first part of that set corresponds to real
3426 physical registers, the second part to any @dfn{pseudo-registers}.
3427 Pseudo-registers have no independent physical existence, but are useful
3428 representations of information within the architecture. For example the
3429 OpenRISC 1000 architecture has up to 32 general purpose registers, which
3430 are typically represented as 32-bit (or 64-bit) integers. However the
3431 GPRs are also used as operands to the floating point operations, and it
3432 could be convenient to define a set of pseudo-registers, to show the
3433 GPRs represented as floating point values.
3435 For any architecture, the implementer will decide on a mapping from
3436 hardware to @value{GDBN} register numbers. The registers corresponding to real
3437 hardware are referred to as @dfn{raw} registers, the remaining registers are
3438 @dfn{pseudo-registers}. The total register set (raw and pseudo) is called
3439 the @dfn{cooked} register set.
3442 @node Register Architecture Functions & Variables
3443 @subsection Functions and Variables Specifying the Register Architecture
3444 @cindex @code{gdbarch} register architecture functions
3446 These @code{struct gdbarch} functions and variables specify the number
3447 and type of registers in the architecture.
3449 @deftypefn {Architecture Function} CORE_ADDR read_pc (struct regcache *@var{regcache})
3451 @deftypefn {Architecture Function} void write_pc (struct regcache *@var{regcache}, CORE_ADDR @var{val})
3453 Read or write the program counter. The default value of both
3454 functions is @code{NULL} (no function available). If the program
3455 counter is just an ordinary register, it can be specified in
3456 @code{struct gdbarch} instead (see @code{pc_regnum} below) and it will
3457 be read or written using the standard routines to access registers. This
3458 function need only be specified if the program counter is not an
3461 Any register information can be obtained using the supplied register
3462 cache, @var{regcache}. @xref{Register Caching, , Register Caching}.
3466 @deftypefn {Architecture Function} void pseudo_register_read (struct gdbarch *@var{gdbarch}, struct regcache *@var{regcache}, int @var{regnum}, const gdb_byte *@var{buf})
3468 @deftypefn {Architecture Function} void pseudo_register_write (struct gdbarch *@var{gdbarch}, struct regcache *@var{regcache}, int @var{regnum}, const gdb_byte *@var{buf})
3470 These functions should be defined if there are any pseudo-registers.
3471 The default value is @code{NULL}. @var{regnum} is the number of the
3472 register to read or write (which will be a @dfn{cooked} register
3473 number) and @var{buf} is the buffer where the value read will be
3474 placed, or from which the value to be written will be taken. The
3475 value in the buffer may be converted to or from a signed or unsigned
3476 integral value using one of the utility functions (@pxref{Register and
3477 Memory Data, , Using Different Register and Memory Data
3480 The access should be for the specified architecture,
3481 @var{gdbarch}. Any register information can be obtained using the
3482 supplied register cache, @var{regcache}. @xref{Register Caching, ,
3487 @deftypevr {Architecture Variable} int sp_regnum
3489 @cindex stack pointer
3492 This specifies the register holding the stack pointer, which may be a
3493 raw or pseudo-register. It defaults to -1 (not defined), but it is an
3494 error for it not to be defined.
3496 The value of the stack pointer register can be accessed withing
3497 @value{GDBN} as the variable @kbd{$sp}.
3501 @deftypevr {Architecture Variable} int pc_regnum
3503 @cindex program counter
3506 This specifies the register holding the program counter, which may be a
3507 raw or pseudo-register. It defaults to -1 (not defined). If
3508 @code{pc_regnum} is not defined, then the functions @code{read_pc} and
3509 @code{write_pc} (see above) must be defined.
3511 The value of the program counter (whether defined as a register, or
3512 through @code{read_pc} and @code{write_pc}) can be accessed withing
3513 @value{GDBN} as the variable @kbd{$pc}.
3517 @deftypevr {Architecture Variable} int ps_regnum
3519 @cindex processor status register
3520 @cindex status register
3523 This specifies the register holding the processor status (often called
3524 the status register), which may be a raw or pseudo-register. It
3525 defaults to -1 (not defined).
3527 If defined, the value of this register can be accessed withing
3528 @value{GDBN} as the variable @kbd{$ps}.
3532 @deftypevr {Architecture Variable} int fp0_regnum
3534 @cindex first floating point register
3536 This specifies the first floating point register. It defaults to
3537 0. @code{fp0_regnum} is not needed unless the target offers support
3542 @node Register Information Functions
3543 @subsection Functions Giving Register Information
3544 @cindex @code{gdbarch} register information functions
3546 These functions return information about registers.
3548 @deftypefn {Architecture Function} {const char *} register_name (struct gdbarch *@var{gdbarch}, int @var{regnum})
3550 This function should convert a register number (raw or pseudo) to a
3551 register name (as a C @code{const char *}). This is used both to
3552 determine the name of a register for output and to work out the meaning
3553 of any register names used as input. The function may also return
3554 @code{NULL}, to indicate that @var{regnum} is not a valid register.
3556 For example with the OpenRISC 1000, @value{GDBN} registers 0-31 are the
3557 General Purpose Registers, register 32 is the program counter and
3558 register 33 is the supervision register (i.e.@: the processor status
3559 register), which map to the strings @code{"gpr00"} through
3560 @code{"gpr31"}, @code{"pc"} and @code{"sr"} respectively. This means
3561 that the @value{GDBN} command @kbd{print $gpr5} should print the value of
3562 the OR1K general purpose register 5@footnote{
3563 @cindex frame pointer
3565 Historically, @value{GDBN} always had a concept of a frame pointer
3566 register, which could be accessed via the @value{GDBN} variable,
3567 @kbd{$fp}. That concept is now deprecated, recognizing that not all
3568 architectures have a frame pointer. However if an architecture does
3569 have a frame pointer register, and defines a register or
3570 pseudo-register with the name @code{"fp"}, then that register will be
3571 used as the value of the @kbd{$fp} variable.}.
3573 The default value for this function is @code{NULL}, meaning
3574 undefined. It should always be defined.
3576 The access should be for the specified architecture, @var{gdbarch}.
3580 @deftypefn {Architecture Function} {struct type *} register_type (struct gdbarch *@var{gdbarch}, int @var{regnum})
3582 Given a register number, this function identifies the type of data it
3583 may be holding, specified as a @code{struct type}. @value{GDBN} allows
3584 creation of arbitrary types, but a number of built in types are
3585 provided (@code{builtin_type_void}, @code{builtin_type_int32} etc),
3586 together with functions to derive types from these.
3588 Typically the program counter will have a type of ``pointer to
3589 function'' (it points to code), the frame pointer and stack pointer
3590 will have types of ``pointer to void'' (they point to data on the stack)
3591 and all other integer registers will have a type of 32-bit integer or
3594 This information guides the formatting when displaying register
3595 information. The default value is @code{NULL} meaning no information is
3596 available to guide formatting when displaying registers.
3600 @deftypefn {Architecture Function} void print_registers_info (struct gdbarch *@var{gdbarch}, struct ui_file *@var{file}, struct frame_info *@var{frame}, int @var{regnum}, int @var{all})
3602 Define this function to print out one or all of the registers for the
3603 @value{GDBN} @kbd{info registers} command. The default value is the
3604 function @code{default_print_registers_info}, which uses the register
3605 type information (see @code{register_type} above) to determine how each
3606 register should be printed. Define a custom version of this function
3607 for fuller control over how the registers are displayed.
3609 The access should be for the specified architecture, @var{gdbarch},
3610 with output to the file specified by the User Interface
3611 Independent Output file handle, @var{file} (@pxref{UI-Independent
3612 Output, , UI-Independent Output---the @code{ui_out}
3615 The registers should show their values in the frame specified by
3616 @var{frame}. If @var{regnum} is -1 and @var{all} is zero, then all
3617 the ``significant'' registers should be shown (the implementer should
3618 decide which registers are ``significant''). Otherwise only the value of
3619 the register specified by @var{regnum} should be output. If
3620 @var{regnum} is -1 and @var{all} is non-zero (true), then the value of
3621 all registers should be shown.
3623 By default @code{default_print_registers_info} prints one register per
3624 line, and if @var{all} is zero omits floating-point registers.
3628 @deftypefn {Architecture Function} void print_float_info (struct gdbarch *@var{gdbarch}, struct ui_file *@var{file}, struct frame_info *@var{frame}, const char *@var{args})
3630 Define this function to provide output about the floating point unit and
3631 registers for the @value{GDBN} @kbd{info float} command respectively.
3632 The default value is @code{NULL} (not defined), meaning no information
3635 The @var{gdbarch} and @var{file} and @var{frame} arguments have the same
3636 meaning as in the @code{print_registers_info} function above. The string
3637 @var{args} contains any supplementary arguments to the @kbd{info float}
3640 Define this function if the target supports floating point operations.
3644 @deftypefn {Architecture Function} void print_vector_info (struct gdbarch *@var{gdbarch}, struct ui_file *@var{file}, struct frame_info *@var{frame}, const char *@var{args})
3646 Define this function to provide output about the vector unit and
3647 registers for the @value{GDBN} @kbd{info vector} command respectively.
3648 The default value is @code{NULL} (not defined), meaning no information
3651 The @var{gdbarch}, @var{file} and @var{frame} arguments have the
3652 same meaning as in the @code{print_registers_info} function above. The
3653 string @var{args} contains any supplementary arguments to the @kbd{info
3656 Define this function if the target supports vector operations.
3660 @deftypefn {Architecture Function} int register_reggroup_p (struct gdbarch *@var{gdbarch}, int @var{regnum}, struct reggroup *@var{group})
3662 @value{GDBN} groups registers into different categories (general,
3663 vector, floating point etc). This function, given a register,
3664 @var{regnum}, and group, @var{group}, returns 1 (true) if the register
3665 is in the group and 0 (false) otherwise.
3667 The information should be for the specified architecture,
3670 The default value is the function @code{default_register_reggroup_p}
3671 which will do a reasonable job based on the type of the register (see
3672 the function @code{register_type} above), with groups for general
3673 purpose registers, floating point registers, vector registers and raw
3674 (i.e not pseudo) registers.
3678 @node Register and Memory Data
3679 @subsection Using Different Register and Memory Data Representations
3680 @cindex register representation
3681 @cindex memory representation
3682 @cindex representations, register and memory
3683 @cindex register data formats, converting
3684 @cindex @code{struct value}, converting register contents to
3686 Some architectures have different representations of data objects,
3687 depending whether the object is held in a register or memory. For
3693 The Alpha architecture can represent 32 bit integer values in
3694 floating-point registers.
3697 The x86 architecture supports 80-bit floating-point registers. The
3698 @code{long double} data type occupies 96 bits in memory but only 80
3699 bits when stored in a register.
3703 In general, the register representation of a data type is determined by
3704 the architecture, or @value{GDBN}'s interface to the architecture, while
3705 the memory representation is determined by the Application Binary
3708 For almost all data types on almost all architectures, the two
3709 representations are identical, and no special handling is needed.
3710 However, they do occasionally differ. An architecture may define the
3711 following @code{struct gdbarch} functions to request conversions
3712 between the register and memory representations of a data type:
3714 @deftypefn {Architecture Function} int gdbarch_convert_register_p (struct gdbarch *@var{gdbarch}, int @var{reg})
3716 Return non-zero (true) if the representation of a data value stored in
3717 this register may be different to the representation of that same data
3718 value when stored in memory. The default value is @code{NULL}
3721 If this function is defined and returns non-zero, the @code{struct
3722 gdbarch} functions @code{gdbarch_register_to_value} and
3723 @code{gdbarch_value_to_register} (see below) should be used to perform
3724 any necessary conversion.
3726 If defined, this function should return zero for the register's native
3727 type, when no conversion is necessary.
3730 @deftypefn {Architecture Function} void gdbarch_register_to_value (struct gdbarch *@var{gdbarch}, int @var{reg}, struct type *@var{type}, char *@var{from}, char *@var{to})
3732 Convert the value of register number @var{reg} to a data object of
3733 type @var{type}. The buffer at @var{from} holds the register's value
3734 in raw format; the converted value should be placed in the buffer at
3738 @emph{Note:} @code{gdbarch_register_to_value} and
3739 @code{gdbarch_value_to_register} take their @var{reg} and @var{type}
3740 arguments in different orders.
3743 @code{gdbarch_register_to_value} should only be used with registers
3744 for which the @code{gdbarch_convert_register_p} function returns a
3749 @deftypefn {Architecture Function} void gdbarch_value_to_register (struct gdbarch *@var{gdbarch}, struct type *@var{type}, int @var{reg}, char *@var{from}, char *@var{to})
3751 Convert a data value of type @var{type} to register number @var{reg}'
3755 @emph{Note:} @code{gdbarch_register_to_value} and
3756 @code{gdbarch_value_to_register} take their @var{reg} and @var{type}
3757 arguments in different orders.
3760 @code{gdbarch_value_to_register} should only be used with registers
3761 for which the @code{gdbarch_convert_register_p} function returns a
3766 @node Register Caching
3767 @subsection Register Caching
3768 @cindex register caching
3770 Caching of registers is used, so that the target does not need to be
3771 accessed and reanalyzed multiple times for each register in
3772 circumstances where the register value cannot have changed.
3774 @cindex @code{struct regcache}
3775 @value{GDBN} provides @code{struct regcache}, associated with a
3776 particular @code{struct gdbarch} to hold the cached values of the raw
3777 registers. A set of functions is provided to access both the raw
3778 registers (with @code{raw} in their name) and the full set of cooked
3779 registers (with @code{cooked} in their name). Functions are provided
3780 to ensure the register cache is kept synchronized with the values of
3781 the actual registers in the target.
3783 Accessing registers through the @code{struct regcache} routines will
3784 ensure that the appropriate @code{struct gdbarch} functions are called
3785 when necessary to access the underlying target architecture. In general
3786 users should use the @dfn{cooked} functions, since these will map to the
3787 @dfn{raw} functions automatically as appropriate.
3789 @findex regcache_cooked_read
3790 @findex regcache_cooked_write
3791 @cindex @code{gdb_byte}
3792 @findex regcache_cooked_read_signed
3793 @findex regcache_cooked_read_unsigned
3794 @findex regcache_cooked_write_signed
3795 @findex regcache_cooked_write_unsigned
3796 The two key functions are @code{regcache_cooked_read} and
3797 @code{regcache_cooked_write} which read or write a register from or to
3798 a byte buffer (type @code{gdb_byte *}). For convenience the wrapper
3799 functions @code{regcache_cooked_read_signed},
3800 @code{regcache_cooked_read_unsigned},
3801 @code{regcache_cooked_write_signed} and
3802 @code{regcache_cooked_write_unsigned} are provided, which read or
3803 write the value using the buffer and convert to or from an integral
3804 value as appropriate.
3806 @node Frame Interpretation
3807 @section Frame Interpretation
3810 * All About Stack Frames::
3811 * Frame Handling Terminology::
3813 * Functions and Variable to Analyze Frames::
3814 * Functions to Access Frame Data::
3815 * Analyzing Stacks---Frame Sniffers::
3818 @node All About Stack Frames
3819 @subsection All About Stack Frames
3821 @value{GDBN} needs to understand the stack on which local (automatic)
3822 variables are stored. The area of the stack containing all the local
3823 variables for a function invocation is known as the @dfn{stack frame}
3824 for that function (or colloquially just as the @dfn{frame}). In turn the
3825 function that called the function will have its stack frame, and so on
3826 back through the chain of functions that have been called.
3828 Almost all architectures have one register dedicated to point to the
3829 end of the stack (the @dfn{stack pointer}). Many have a second register
3830 which points to the start of the currently active stack frame (the
3831 @dfn{frame pointer}). The specific arrangements for an architecture are
3832 a key part of the ABI.
3834 A diagram helps to explain this. Here is a simple program to compute
3847 return n * fact (n - 1);
3855 for (i = 0; i < 10; i++)
3858 printf ("%d! = %d\n", i, f);
3863 Consider the state of the stack when the code reaches line 6 after the
3864 main program has called @code{fact@w{ }(3)}. The chain of function
3865 calls will be @code{main ()}, @code{fact@w{ }(3)}, @code{fact@w{
3866 }(2)}, @code{@w{fact (1)}} and @code{fact@w{ }(0)}.
3868 In this illustration the stack is falling (as used for example by the
3869 OpenRISC 1000 ABI). The stack pointer (SP) is at the end of the stack
3870 (lowest address) and the frame pointer (FP) is at the highest address
3871 in the current stack frame. The following diagram shows how the stack
3874 @center @image{stack_frame,14cm}
3876 In each stack frame, offset 0 from the stack pointer is the frame
3877 pointer of the previous frame and offset 4 (this is illustrating a
3878 32-bit architecture) from the stack pointer is the return address.
3879 Local variables are indexed from the frame pointer, with negative
3880 indexes. In the function @code{fact}, offset -4 from the frame
3881 pointer is the argument @var{n}. In the @code{main} function, offset
3882 -4 from the frame pointer is the local variable @var{i} and offset -8
3883 from the frame pointer is the local variable @var{f}@footnote{This is
3884 a simplified example for illustrative purposes only. Good optimizing
3885 compilers would not put anything on the stack for such simple
3886 functions. Indeed they might eliminate the recursion and use of the
3889 It is very easy to get confused when examining stacks. @value{GDBN}
3890 has terminology it uses rigorously throughout. The stack frame of the
3891 function currently executing, or where execution stopped is numbered
3892 zero. In this example frame #0 is the stack frame of the call to
3893 @code{fact@w{ }(0)}. The stack frame of its calling function
3894 (@code{fact@w{ }(1)} in this case) is numbered #1 and so on back
3895 through the chain of calls.
3897 The main @value{GDBN} data structure describing frames is
3898 @code{@w{struct frame_info}}. It is not used directly, but only via
3899 its accessor functions. @code{frame_info} includes information about
3900 the registers in the frame and a pointer to the code of the function
3901 with which the frame is associated. The entire stack is represented as
3902 a linked list of @code{frame_info} structs.
3904 @node Frame Handling Terminology
3905 @subsection Frame Handling Terminology
3907 It is easy to get confused when referencing stack frames. @value{GDBN}
3908 uses some precise terminology.
3914 @cindex stack frame, definition of THIS frame
3915 @cindex frame, definition of THIS frame
3916 @dfn{THIS} frame is the frame currently under consideration.
3920 @cindex stack frame, definition of NEXT frame
3921 @cindex frame, definition of NEXT frame
3922 The @dfn{NEXT} frame, also sometimes called the inner or newer frame is the
3923 frame of the function called by the function of THIS frame.
3926 @cindex PREVIOUS frame
3927 @cindex stack frame, definition of PREVIOUS frame
3928 @cindex frame, definition of PREVIOUS frame
3929 The @dfn{PREVIOUS} frame, also sometimes called the outer or older frame is
3930 the frame of the function which called the function of THIS frame.
3934 So in the example in the previous section (@pxref{All About Stack
3935 Frames, , All About Stack Frames}), if THIS frame is #3 (the call to
3936 @code{fact@w{ }(3)}), the NEXT frame is frame #2 (the call to
3937 @code{fact@w{ }(2)}) and the PREVIOUS frame is frame #4 (the call to
3938 @code{main@w{ }()}).
3940 @cindex innermost frame
3941 @cindex stack frame, definition of innermost frame
3942 @cindex frame, definition of innermost frame
3943 The @dfn{innermost} frame is the frame of the current executing
3944 function, or where the program stopped, in this example, in the middle
3945 of the call to @code{@w{fact (0))}}. It is always numbered frame #0.
3947 @cindex base of a frame
3948 @cindex stack frame, definition of base of a frame
3949 @cindex frame, definition of base of a frame
3950 The @dfn{base} of a frame is the address immediately before the start
3951 of the NEXT frame. For a stack which grows down in memory (a
3952 @dfn{falling} stack) this will be the lowest address and for a stack
3953 which grows up in memory (a @dfn{rising} stack) this will be the
3954 highest address in the frame.
3956 @value{GDBN} functions to analyze the stack are typically given a
3957 pointer to the NEXT frame to determine information about THIS
3958 frame. Information about THIS frame includes data on where the
3959 registers of the PREVIOUS frame are stored in this stack frame. In
3960 this example the frame pointer of the PREVIOUS frame is stored at
3961 offset 0 from the stack pointer of THIS frame.
3964 @cindex stack frame, definition of unwinding
3965 @cindex frame, definition of unwinding
3966 The process whereby a function is given a pointer to the NEXT
3967 frame to work out information about THIS frame is referred to as
3968 @dfn{unwinding}. The @value{GDBN} functions involved in this typically
3969 include unwind in their name.
3972 @cindex stack frame, definition of sniffing
3973 @cindex frame, definition of sniffing
3974 The process of analyzing a target to determine the information that
3975 should go in struct frame_info is called @dfn{sniffing}. The functions
3976 that carry this out are called sniffers and typically include sniffer
3977 in their name. More than one sniffer may be required to extract all
3978 the information for a particular frame.
3980 @cindex sentinel frame
3981 @cindex stack frame, definition of sentinel frame
3982 @cindex frame, definition of sentinel frame
3983 Because so many functions work using the NEXT frame, there is an issue
3984 about addressing the innermost frame---it has no NEXT frame. To solve
3985 this @value{GDBN} creates a dummy frame #-1, known as the
3986 @dfn{sentinel} frame.
3988 @node Prologue Caches
3989 @subsection Prologue Caches
3991 @cindex function prologue
3992 @cindex prologue of a function
3993 All the frame sniffing functions typically examine the code at the
3994 start of the corresponding function, to determine the state of
3995 registers. The ABI will save old values and set new values of key
3996 registers at the start of each function in what is known as the
3997 function @dfn{prologue}.
3999 @cindex prologue cache
4000 For any particular stack frame this data does not change, so all the
4001 standard unwinding functions, in addition to receiving a pointer to
4002 the NEXT frame as their first argument, receive a pointer to a
4003 @dfn{prologue cache} as their second argument. This can be used to store
4004 values associated with a particular frame, for reuse on subsequent
4005 calls involving the same frame.
4007 It is up to the user to define the structure used (it is a
4008 @code{void@w{ }*} pointer) and arrange allocation and deallocation of
4009 storage. However for general use, @value{GDBN} provides
4010 @code{@w{struct trad_frame_cache}}, with a set of accessor
4011 routines. This structure holds the stack and code address of
4012 THIS frame, the base address of the frame, a pointer to the
4013 struct @code{frame_info} for the NEXT frame and details of
4014 where the registers of the PREVIOUS frame may be found in THIS
4017 Typically the first time any sniffer function is called with NEXT
4018 frame, the prologue sniffer for THIS frame will be @code{NULL}. The
4019 sniffer will analyze the frame, allocate a prologue cache structure
4020 and populate it. Subsequent calls using the same NEXT frame will
4021 pass in this prologue cache, so the data can be returned with no
4022 additional analysis.
4024 @node Functions and Variable to Analyze Frames
4025 @subsection Functions and Variable to Analyze Frames
4027 These struct @code{gdbarch} functions and variable should be defined
4028 to provide analysis of the stack frame and allow it to be adjusted as
4031 @deftypefn {Architecture Function} CORE_ADDR skip_prologue (struct gdbarch *@var{gdbarch}, CORE_ADDR @var{pc})
4033 The prologue of a function is the code at the beginning of the
4034 function which sets up the stack frame, saves the return address
4035 etc. The code representing the behavior of the function starts after
4038 This function skips past the prologue of a function if the program
4039 counter, @var{pc}, is within the prologue of a function. The result is
4040 the program counter immediately after the prologue. With modern
4041 optimizing compilers, this may be a far from trivial exercise. However
4042 the required information may be within the binary as DWARF2 debugging
4043 information, making the job much easier.
4045 The default value is @code{NULL} (not defined). This function should always
4046 be provided, but can take advantage of DWARF2 debugging information,
4047 if that is available.
4051 @deftypefn {Architecture Function} int inner_than (CORE_ADDR @var{lhs}, CORE_ADDR @var{rhs})
4052 @findex core_addr_lessthan
4053 @findex core_addr_greaterthan
4055 Given two frame or stack pointers, return non-zero (true) if the first
4056 represents the @dfn{inner} stack frame and 0 (false) otherwise. This
4057 is used to determine whether the target has a stack which grows up in
4058 memory (rising stack) or grows down in memory (falling stack).
4059 @xref{All About Stack Frames, , All About Stack Frames}, for an
4060 explanation of @dfn{inner} frames.
4062 The default value of this function is @code{NULL} and it should always
4063 be defined. However for almost all architectures one of the built-in
4064 functions can be used: @code{core_addr_lessthan} (for stacks growing
4065 down in memory) or @code{core_addr_greaterthan} (for stacks growing up
4070 @anchor{frame_align}
4071 @deftypefn {Architecture Function} CORE_ADDR frame_align (struct gdbarch *@var{gdbarch}, CORE_ADDR @var{address})
4075 The architecture may have constraints on how its frames are
4076 aligned. For example the OpenRISC 1000 ABI requires stack frames to be
4077 double-word aligned, but 32-bit versions of the architecture allocate
4078 single-word values to the stack. Thus extra padding may be needed at
4079 the end of a stack frame.
4081 Given a proposed address for the stack pointer, this function
4082 returns a suitably aligned address (by expanding the stack frame).
4084 The default value is @code{NULL} (undefined). This function should be defined
4085 for any architecture where it is possible the stack could become
4086 misaligned. The utility functions @code{align_down} (for falling
4087 stacks) and @code{align_up} (for rising stacks) will facilitate the
4088 implementation of this function.
4092 @deftypevr {Architecture Variable} int frame_red_zone_size
4094 Some ABIs reserve space beyond the end of the stack for use by leaf
4095 functions without prologue or epilogue or by exception handlers (for
4096 example the OpenRISC 1000).
4098 This is known as a @dfn{red zone} (AMD terminology). The @sc{amd64}
4099 (nee x86-64) ABI documentation refers to the @dfn{red zone} when
4100 describing this scratch area.
4102 The default value is 0. Set this field if the architecture has such a
4103 red zone. The value must be aligned as required by the ABI (see
4104 @code{frame_align} above for an explanation of stack frame alignment).
4108 @node Functions to Access Frame Data
4109 @subsection Functions to Access Frame Data
4111 These functions provide access to key registers and arguments in the
4114 @deftypefn {Architecture Function} CORE_ADDR unwind_pc (struct gdbarch *@var{gdbarch}, struct frame_info *@var{next_frame})
4116 This function is given a pointer to the NEXT stack frame (@pxref{All
4117 About Stack Frames, , All About Stack Frames}, for how frames are
4118 represented) and returns the value of the program counter in the
4119 PREVIOUS frame (i.e.@: the frame of the function that called THIS
4120 one). This is commonly referred to as the @dfn{return address}.
4122 The implementation, which must be frame agnostic (work with any frame),
4123 is typically no more than:
4127 pc = frame_unwind_register_unsigned (next_frame, @var{ARCH}_PC_REGNUM);
4128 return gdbarch_addr_bits_remove (gdbarch, pc);
4133 @deftypefn {Architecture Function} CORE_ADDR unwind_sp (struct gdbarch *@var{gdbarch}, struct frame_info *@var{next_frame})
4135 This function is given a pointer to the NEXT stack frame
4136 (@pxref{All About Stack Frames, , All About Stack Frames} for how
4137 frames are represented) and returns the value of the stack pointer in
4138 the PREVIOUS frame (i.e.@: the frame of the function that called
4141 The implementation, which must be frame agnostic (work with any frame),
4142 is typically no more than:
4146 sp = frame_unwind_register_unsigned (next_frame, @var{ARCH}_SP_REGNUM);
4147 return gdbarch_addr_bits_remove (gdbarch, sp);
4152 @deftypefn {Architecture Function} int frame_num_args (struct gdbarch *@var{gdbarch}, struct frame_info *@var{this_frame})
4154 This function is given a pointer to THIS stack frame (@pxref{All
4155 About Stack Frames, , All About Stack Frames} for how frames are
4156 represented), and returns the number of arguments that are being
4157 passed, or -1 if not known.
4159 The default value is @code{NULL} (undefined), in which case the number of
4160 arguments passed on any stack frame is always unknown. For many
4161 architectures this will be a suitable default.
4165 @node Analyzing Stacks---Frame Sniffers
4166 @subsection Analyzing Stacks---Frame Sniffers
4168 When a program stops, @value{GDBN} needs to construct the chain of
4169 struct @code{frame_info} representing the state of the stack using
4170 appropriate @dfn{sniffers}.
4172 Each architecture requires appropriate sniffers, but they do not form
4173 entries in @code{@w{struct gdbarch}}, since more than one sniffer may
4174 be required and a sniffer may be suitable for more than one
4175 @code{@w{struct gdbarch}}. Instead sniffers are associated with
4176 architectures using the following functions.
4181 @findex frame_unwind_append_sniffer
4182 @code{frame_unwind_append_sniffer} is used to add a new sniffer to
4183 analyze THIS frame when given a pointer to the NEXT frame.
4186 @findex frame_base_append_sniffer
4187 @code{frame_base_append_sniffer} is used to add a new sniffer
4188 which can determine information about the base of a stack frame.
4191 @findex frame_base_set_default
4192 @code{frame_base_set_default} is used to specify the default base
4197 These functions all take a reference to @code{@w{struct gdbarch}}, so
4198 they are associated with a specific architecture. They are usually
4199 called in the @code{gdbarch} initialization function, after the
4200 @code{gdbarch} struct has been set up. Unless a default has been set, the
4201 most recently appended sniffer will be tried first.
4203 The main frame unwinding sniffer (as set by
4204 @code{frame_unwind_append_sniffer)} returns a structure specifying
4205 a set of sniffing functions:
4207 @cindex @code{frame_unwind}
4211 enum frame_type type;
4212 frame_this_id_ftype *this_id;
4213 frame_prev_register_ftype *prev_register;
4214 const struct frame_data *unwind_data;
4215 frame_sniffer_ftype *sniffer;
4216 frame_prev_pc_ftype *prev_pc;
4217 frame_dealloc_cache_ftype *dealloc_cache;
4221 The @code{type} field indicates the type of frame this sniffer can
4222 handle: normal, dummy (@pxref{Functions Creating Dummy Frames, ,
4223 Functions Creating Dummy Frames}), signal handler or sentinel. Signal
4224 handlers sometimes have their own simplified stack structure for
4225 efficiency, so may need their own handlers.
4227 The @code{unwind_data} field holds additional information which may be
4228 relevant to particular types of frame. For example it may hold
4229 additional information for signal handler frames.
4231 The remaining fields define functions that yield different types of
4232 information when given a pointer to the NEXT stack frame. Not all
4233 functions need be provided. If an entry is @code{NULL}, the next sniffer will
4239 @code{this_id} determines the stack pointer and function (code
4240 entry point) for THIS stack frame.
4243 @code{prev_register} determines where the values of registers for
4244 the PREVIOUS stack frame are stored in THIS stack frame.
4247 @code{sniffer} takes a look at THIS frame's registers to
4248 determine if this is the appropriate unwinder.
4251 @code{prev_pc} determines the program counter for THIS
4252 frame. Only needed if the program counter is not an ordinary register
4253 (@pxref{Register Architecture Functions & Variables,
4254 , Functions and Variables Specifying the Register Architecture}).
4257 @code{dealloc_cache} frees any additional memory associated with
4258 the prologue cache for this frame (@pxref{Prologue Caches, , Prologue
4263 In general it is only the @code{this_id} and @code{prev_register}
4264 fields that need be defined for custom sniffers.
4266 The frame base sniffer is much simpler. It is a @code{@w{struct
4267 frame_base}}, which refers to the corresponding @code{frame_unwind}
4268 struct and whose fields refer to functions yielding various addresses
4271 @cindex @code{frame_base}
4275 const struct frame_unwind *unwind;
4276 frame_this_base_ftype *this_base;
4277 frame_this_locals_ftype *this_locals;
4278 frame_this_args_ftype *this_args;
4282 All the functions referred to take a pointer to the NEXT frame as
4283 argument. The function referred to by @code{this_base} returns the
4284 base address of THIS frame, the function referred to by
4285 @code{this_locals} returns the base address of local variables in THIS
4286 frame and the function referred to by @code{this_args} returns the
4287 base address of the function arguments in this frame.
4289 As described above, the base address of a frame is the address
4290 immediately before the start of the NEXT frame. For a falling
4291 stack, this is the lowest address in the frame and for a rising stack
4292 it is the highest address in the frame. For most architectures the
4293 same address is also the base address for local variables and
4294 arguments, in which case the same function can be used for all three
4295 entries@footnote{It is worth noting that if it cannot be determined in any
4296 other way (for example by there being a register with the name
4297 @code{"fp"}), then the result of the @code{this_base} function will be
4298 used as the value of the frame pointer variable @kbd{$fp} in
4299 @value{GDBN}. This is very often not correct (for example with the
4300 OpenRISC 1000, this value is the stack pointer, @kbd{$sp}). In this
4301 case a register (raw or pseudo) with the name @code{"fp"} should be
4302 defined. It will be used in preference as the value of @kbd{$fp}.}.
4304 @node Inferior Call Setup
4305 @section Inferior Call Setup
4306 @cindex calls to the inferior
4309 * About Dummy Frames::
4310 * Functions Creating Dummy Frames::
4313 @node About Dummy Frames
4314 @subsection About Dummy Frames
4315 @cindex dummy frames
4317 @value{GDBN} can call functions in the target code (for example by
4318 using the @kbd{call} or @kbd{print} commands). These functions may be
4319 breakpointed, and it is essential that if a function does hit a
4320 breakpoint, commands like @kbd{backtrace} work correctly.
4322 This is achieved by making the stack look as though the function had
4323 been called from the point where @value{GDBN} had previously stopped.
4324 This requires that @value{GDBN} can set up stack frames appropriate for
4325 such function calls.
4327 @node Functions Creating Dummy Frames
4328 @subsection Functions Creating Dummy Frames
4330 The following functions provide the functionality to set up such
4331 @dfn{dummy} stack frames.
4333 @deftypefn {Architecture Function} CORE_ADDR push_dummy_call (struct gdbarch *@var{gdbarch}, struct value *@var{function}, struct regcache *@var{regcache}, CORE_ADDR @var{bp_addr}, int @var{nargs}, struct value **@var{args}, CORE_ADDR @var{sp}, int @var{struct_return}, CORE_ADDR @var{struct_addr})
4335 This function sets up a dummy stack frame for the function about to be
4336 called. @code{push_dummy_call} is given the arguments to be passed
4337 and must copy them into registers or push them on to the stack as
4338 appropriate for the ABI.
4340 @var{function} is a pointer to the function
4341 that will be called and @var{regcache} the register cache from which
4342 values should be obtained. @var{bp_addr} is the address to which the
4343 function should return (which is breakpointed, so @value{GDBN} can
4344 regain control, hence the name). @var{nargs} is the number of
4345 arguments to pass and @var{args} an array containing the argument
4346 values. @var{struct_return} is non-zero (true) if the function returns
4347 a structure, and if so @var{struct_addr} is the address in which the
4348 structure should be returned.
4350 After calling this function, @value{GDBN} will pass control to the
4351 target at the address of the function, which will find the stack and
4352 registers set up just as expected.
4354 The default value of this function is @code{NULL} (undefined). If the
4355 function is not defined, then @value{GDBN} will not allow the user to
4356 call functions within the target being debugged.
4360 @deftypefn {Architecture Function} {struct frame_id} unwind_dummy_id (struct gdbarch *@var{gdbarch}, struct frame_info *@var{next_frame})
4362 This is the inverse of @code{push_dummy_call} which restores the stack
4363 pointer and program counter after a call to evaluate a function using
4364 a dummy stack frame. The result is a @code{@w{struct frame_id}}, which
4365 contains the value of the stack pointer and program counter to be
4368 The NEXT frame pointer is provided as argument,
4369 @var{next_frame}. THIS frame is the frame of the dummy function,
4370 which can be unwound, to yield the required stack pointer and program
4371 counter from the PREVIOUS frame.
4373 The default value is @code{NULL} (undefined). If @code{push_dummy_call} is
4374 defined, then this function should also be defined.
4378 @deftypefn {Architecture Function} CORE_ADDR push_dummy_code (struct gdbarch *@var{gdbarch}, CORE_ADDR @var{sp}, CORE_ADDR @var{funaddr}, struct value **@var{args}, int @var{nargs}, struct type *@var{value_type}, CORE_ADDR *@var{real_pc}, CORE_ADDR *@var{bp_addr}, struct regcache *@var{regcache})
4380 If this function is not defined (its default value is @code{NULL}), a dummy
4381 call will use the entry point of the currently loaded code on the
4382 target as its return address. A temporary breakpoint will be set
4383 there, so the location must be writable and have room for a
4386 It is possible that this default is not suitable. It might not be
4387 writable (in ROM possibly), or the ABI might require code to be
4388 executed on return from a call to unwind the stack before the
4389 breakpoint is encountered.
4391 If either of these is the case, then push_dummy_code should be defined
4392 to push an instruction sequence onto the end of the stack to which the
4393 dummy call should return.
4395 The arguments are essentially the same as those to
4396 @code{push_dummy_call}. However the function is provided with the
4397 type of the function result, @var{value_type}, @var{bp_addr} is used
4398 to return a value (the address at which the breakpoint instruction
4399 should be inserted) and @var{real pc} is used to specify the resume
4400 address when starting the call sequence. The function should return
4401 the updated innermost stack address.
4404 @emph{Note:} This does require that code in the stack can be executed.
4405 Some Harvard architectures may not allow this.
4410 @node Adding support for debugging core files
4411 @section Adding support for debugging core files
4414 The prerequisite for adding core file support in @value{GDBN} is to have
4415 core file support in BFD.
4417 Once BFD support is available, writing the apropriate
4418 @code{regset_from_core_section} architecture function should be all
4419 that is needed in order to add support for core files in @value{GDBN}.
4421 @node Defining Other Architecture Features
4422 @section Defining Other Architecture Features
4424 This section describes other functions and values in @code{gdbarch},
4425 together with some useful macros, that you can use to define the
4426 target architecture.
4430 @item CORE_ADDR gdbarch_addr_bits_remove (@var{gdbarch}, @var{addr})
4431 @findex gdbarch_addr_bits_remove
4432 If a raw machine instruction address includes any bits that are not
4433 really part of the address, then this function is used to zero those bits in
4434 @var{addr}. This is only used for addresses of instructions, and even then not
4437 For example, the two low-order bits of the PC on the Hewlett-Packard PA
4438 2.0 architecture contain the privilege level of the corresponding
4439 instruction. Since instructions must always be aligned on four-byte
4440 boundaries, the processor masks out these bits to generate the actual
4441 address of the instruction. @code{gdbarch_addr_bits_remove} would then for
4442 example look like that:
4444 arch_addr_bits_remove (CORE_ADDR addr)
4446 return (addr &= ~0x3);
4450 @item int address_class_name_to_type_flags (@var{gdbarch}, @var{name}, @var{type_flags_ptr})
4451 @findex address_class_name_to_type_flags
4452 If @var{name} is a valid address class qualifier name, set the @code{int}
4453 referenced by @var{type_flags_ptr} to the mask representing the qualifier
4454 and return 1. If @var{name} is not a valid address class qualifier name,
4457 The value for @var{type_flags_ptr} should be one of
4458 @code{TYPE_FLAG_ADDRESS_CLASS_1}, @code{TYPE_FLAG_ADDRESS_CLASS_2}, or
4459 possibly some combination of these values or'd together.
4460 @xref{Target Architecture Definition, , Address Classes}.
4462 @item int address_class_name_to_type_flags_p (@var{gdbarch})
4463 @findex address_class_name_to_type_flags_p
4464 Predicate which indicates whether @code{address_class_name_to_type_flags}
4467 @item int gdbarch_address_class_type_flags (@var{gdbarch}, @var{byte_size}, @var{dwarf2_addr_class})
4468 @findex gdbarch_address_class_type_flags
4469 Given a pointers byte size (as described by the debug information) and
4470 the possible @code{DW_AT_address_class} value, return the type flags
4471 used by @value{GDBN} to represent this address class. The value
4472 returned should be one of @code{TYPE_FLAG_ADDRESS_CLASS_1},
4473 @code{TYPE_FLAG_ADDRESS_CLASS_2}, or possibly some combination of these
4474 values or'd together.
4475 @xref{Target Architecture Definition, , Address Classes}.
4477 @item int gdbarch_address_class_type_flags_p (@var{gdbarch})
4478 @findex gdbarch_address_class_type_flags_p
4479 Predicate which indicates whether @code{gdbarch_address_class_type_flags_p} has
4482 @item const char *gdbarch_address_class_type_flags_to_name (@var{gdbarch}, @var{type_flags})
4483 @findex gdbarch_address_class_type_flags_to_name
4484 Return the name of the address class qualifier associated with the type
4485 flags given by @var{type_flags}.
4487 @item int gdbarch_address_class_type_flags_to_name_p (@var{gdbarch})
4488 @findex gdbarch_address_class_type_flags_to_name_p
4489 Predicate which indicates whether @code{gdbarch_address_class_type_flags_to_name} has been defined.
4490 @xref{Target Architecture Definition, , Address Classes}.
4492 @item void gdbarch_address_to_pointer (@var{gdbarch}, @var{type}, @var{buf}, @var{addr})
4493 @findex gdbarch_address_to_pointer
4494 Store in @var{buf} a pointer of type @var{type} representing the address
4495 @var{addr}, in the appropriate format for the current architecture.
4496 This function may safely assume that @var{type} is either a pointer or a
4497 C@t{++} reference type.
4498 @xref{Target Architecture Definition, , Pointers Are Not Always Addresses}.
4500 @item int gdbarch_believe_pcc_promotion (@var{gdbarch})
4501 @findex gdbarch_believe_pcc_promotion
4502 Used to notify if the compiler promotes a @code{short} or @code{char}
4503 parameter to an @code{int}, but still reports the parameter as its
4504 original type, rather than the promoted type.
4506 @item gdbarch_bits_big_endian (@var{gdbarch})
4507 @findex gdbarch_bits_big_endian
4508 This is used if the numbering of bits in the targets does @strong{not} match
4509 the endianism of the target byte order. A value of 1 means that the bits
4510 are numbered in a big-endian bit order, 0 means little-endian.
4512 @item set_gdbarch_bits_big_endian (@var{gdbarch}, @var{bits_big_endian})
4513 @findex set_gdbarch_bits_big_endian
4514 Calling set_gdbarch_bits_big_endian with a value of 1 indicates that the
4515 bits in the target are numbered in a big-endian bit order, 0 indicates
4520 This is the character array initializer for the bit pattern to put into
4521 memory where a breakpoint is set. Although it's common to use a trap
4522 instruction for a breakpoint, it's not required; for instance, the bit
4523 pattern could be an invalid instruction. The breakpoint must be no
4524 longer than the shortest instruction of the architecture.
4526 @code{BREAKPOINT} has been deprecated in favor of
4527 @code{gdbarch_breakpoint_from_pc}.
4529 @item BIG_BREAKPOINT
4530 @itemx LITTLE_BREAKPOINT
4531 @findex LITTLE_BREAKPOINT
4532 @findex BIG_BREAKPOINT
4533 Similar to BREAKPOINT, but used for bi-endian targets.
4535 @code{BIG_BREAKPOINT} and @code{LITTLE_BREAKPOINT} have been deprecated in
4536 favor of @code{gdbarch_breakpoint_from_pc}.
4538 @item const gdb_byte *gdbarch_breakpoint_from_pc (@var{gdbarch}, @var{pcptr}, @var{lenptr})
4539 @findex gdbarch_breakpoint_from_pc
4540 @anchor{gdbarch_breakpoint_from_pc} Use the program counter to determine the
4541 contents and size of a breakpoint instruction. It returns a pointer to
4542 a static string of bytes that encode a breakpoint instruction, stores the
4543 length of the string to @code{*@var{lenptr}}, and adjusts the program
4544 counter (if necessary) to point to the actual memory location where the
4545 breakpoint should be inserted. May return @code{NULL} to indicate that
4546 software breakpoints are not supported.
4548 Although it is common to use a trap instruction for a breakpoint, it's
4549 not required; for instance, the bit pattern could be an invalid
4550 instruction. The breakpoint must be no longer than the shortest
4551 instruction of the architecture.
4553 Provided breakpoint bytes can be also used by @code{bp_loc_is_permanent} to
4554 detect permanent breakpoints. @code{gdbarch_breakpoint_from_pc} should return
4555 an unchanged memory copy if it was called for a location with permanent
4556 breakpoint as some architectures use breakpoint instructions containing
4557 arbitrary parameter value.
4559 Replaces all the other @var{BREAKPOINT} macros.
4561 @item int gdbarch_memory_insert_breakpoint (@var{gdbarch}, @var{bp_tgt})
4562 @itemx gdbarch_memory_remove_breakpoint (@var{gdbarch}, @var{bp_tgt})
4563 @findex gdbarch_memory_remove_breakpoint
4564 @findex gdbarch_memory_insert_breakpoint
4565 Insert or remove memory based breakpoints. Reasonable defaults
4566 (@code{default_memory_insert_breakpoint} and
4567 @code{default_memory_remove_breakpoint} respectively) have been
4568 provided so that it is not necessary to set these for most
4569 architectures. Architectures which may want to set
4570 @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
4571 conventional manner.
4573 It may also be desirable (from an efficiency standpoint) to define
4574 custom breakpoint insertion and removal routines if
4575 @code{gdbarch_breakpoint_from_pc} needs to read the target's memory for some
4578 @item CORE_ADDR gdbarch_adjust_breakpoint_address (@var{gdbarch}, @var{bpaddr})
4579 @findex gdbarch_adjust_breakpoint_address
4580 @cindex breakpoint address adjusted
4581 Given an address at which a breakpoint is desired, return a breakpoint
4582 address adjusted to account for architectural constraints on
4583 breakpoint placement. This method is not needed by most targets.
4585 The FR-V target (see @file{frv-tdep.c}) requires this method.
4586 The FR-V is a VLIW architecture in which a number of RISC-like
4587 instructions are grouped (packed) together into an aggregate
4588 instruction or instruction bundle. When the processor executes
4589 one of these bundles, the component instructions are executed
4592 In the course of optimization, the compiler may group instructions
4593 from distinct source statements into the same bundle. The line number
4594 information associated with one of the latter statements will likely
4595 refer to some instruction other than the first one in the bundle. So,
4596 if the user attempts to place a breakpoint on one of these latter
4597 statements, @value{GDBN} must be careful to @emph{not} place the break
4598 instruction on any instruction other than the first one in the bundle.
4599 (Remember though that the instructions within a bundle execute
4600 in parallel, so the @emph{first} instruction is the instruction
4601 at the lowest address and has nothing to do with execution order.)
4603 The FR-V's @code{gdbarch_adjust_breakpoint_address} method will adjust a
4604 breakpoint's address by scanning backwards for the beginning of
4605 the bundle, returning the address of the bundle.
4607 Since the adjustment of a breakpoint may significantly alter a user's
4608 expectation, @value{GDBN} prints a warning when an adjusted breakpoint
4609 is initially set and each time that that breakpoint is hit.
4611 @item int gdbarch_call_dummy_location (@var{gdbarch})
4612 @findex gdbarch_call_dummy_location
4613 See the file @file{inferior.h}.
4615 This method has been replaced by @code{gdbarch_push_dummy_code}
4616 (@pxref{gdbarch_push_dummy_code}).
4618 @item int gdbarch_cannot_fetch_register (@var{gdbarch}, @var{regum})
4619 @findex gdbarch_cannot_fetch_register
4620 This function should return nonzero if @var{regno} cannot be fetched
4621 from an inferior process.
4623 @item int gdbarch_cannot_store_register (@var{gdbarch}, @var{regnum})
4624 @findex gdbarch_cannot_store_register
4625 This function should return nonzero if @var{regno} should not be
4626 written to the target. This is often the case for program counters,
4627 status words, and other special registers. This function returns 0 as
4628 default so that @value{GDBN} will assume that all registers may be written.
4630 @item int gdbarch_convert_register_p (@var{gdbarch}, @var{regnum}, struct type *@var{type})
4631 @findex gdbarch_convert_register_p
4632 Return non-zero if register @var{regnum} represents data values of type
4633 @var{type} in a non-standard form.
4634 @xref{Target Architecture Definition, , Using Different Register and Memory Data Representations}.
4636 @item int gdbarch_fp0_regnum (@var{gdbarch})
4637 @findex gdbarch_fp0_regnum
4638 This function returns the number of the first floating point register,
4639 if the machine has such registers. Otherwise, it returns -1.
4641 @item CORE_ADDR gdbarch_decr_pc_after_break (@var{gdbarch})
4642 @findex gdbarch_decr_pc_after_break
4643 This function shall return the amount by which to decrement the PC after the
4644 program encounters a breakpoint. This is often the number of bytes in
4645 @code{BREAKPOINT}, though not always. For most targets this value will be 0.
4647 @item DISABLE_UNSETTABLE_BREAK (@var{addr})
4648 @findex DISABLE_UNSETTABLE_BREAK
4649 If defined, this should evaluate to 1 if @var{addr} is in a shared
4650 library in which breakpoints cannot be set and so should be disabled.
4652 @item int gdbarch_dwarf2_reg_to_regnum (@var{gdbarch}, @var{dwarf2_regnr})
4653 @findex gdbarch_dwarf2_reg_to_regnum
4654 Convert DWARF2 register number @var{dwarf2_regnr} into @value{GDBN} regnum.
4655 If not defined, no conversion will be performed.
4657 @item int gdbarch_ecoff_reg_to_regnum (@var{gdbarch}, @var{ecoff_regnr})
4658 @findex gdbarch_ecoff_reg_to_regnum
4659 Convert ECOFF register number @var{ecoff_regnr} into @value{GDBN} regnum. If
4660 not defined, no conversion will be performed.
4662 @item GCC_COMPILED_FLAG_SYMBOL
4663 @itemx GCC2_COMPILED_FLAG_SYMBOL
4664 @findex GCC2_COMPILED_FLAG_SYMBOL
4665 @findex GCC_COMPILED_FLAG_SYMBOL
4666 If defined, these are the names of the symbols that @value{GDBN} will
4667 look for to detect that GCC compiled the file. The default symbols
4668 are @code{gcc_compiled.} and @code{gcc2_compiled.},
4669 respectively. (Currently only defined for the Delta 68.)
4671 @item gdbarch_get_longjmp_target
4672 @findex gdbarch_get_longjmp_target
4673 This function determines the target PC address that @code{longjmp}
4674 will jump to, assuming that we have just stopped at a @code{longjmp}
4675 breakpoint. It takes a @code{CORE_ADDR *} as argument, and stores the
4676 target PC value through this pointer. It examines the current state
4677 of the machine as needed, typically by using a manually-determined
4678 offset into the @code{jmp_buf}. (While we might like to get the offset
4679 from the target's @file{jmpbuf.h}, that header file cannot be assumed
4680 to be available when building a cross-debugger.)
4682 @item DEPRECATED_IBM6000_TARGET
4683 @findex DEPRECATED_IBM6000_TARGET
4684 Shows that we are configured for an IBM RS/6000 system. This
4685 conditional should be eliminated (FIXME) and replaced by
4686 feature-specific macros. It was introduced in haste and we are
4687 repenting at leisure.
4689 @item I386_USE_GENERIC_WATCHPOINTS
4690 An x86-based target can define this to use the generic x86 watchpoint
4691 support; see @ref{Algorithms, I386_USE_GENERIC_WATCHPOINTS}.
4693 @item gdbarch_in_function_epilogue_p (@var{gdbarch}, @var{addr})
4694 @findex gdbarch_in_function_epilogue_p
4695 Returns non-zero if the given @var{addr} is in the epilogue of a function.
4696 The epilogue of a function is defined as the part of a function where
4697 the stack frame of the function already has been destroyed up to the
4698 final `return from function call' instruction.
4700 @item int gdbarch_in_solib_return_trampoline (@var{gdbarch}, @var{pc}, @var{name})
4701 @findex gdbarch_in_solib_return_trampoline
4702 Define this function to return nonzero if the program is stopped in the
4703 trampoline that returns from a shared library.
4705 @item target_so_ops.in_dynsym_resolve_code (@var{pc})
4706 @findex in_dynsym_resolve_code
4707 Define this to return nonzero if the program is stopped in the
4710 @item SKIP_SOLIB_RESOLVER (@var{pc})
4711 @findex SKIP_SOLIB_RESOLVER
4712 Define this to evaluate to the (nonzero) address at which execution
4713 should continue to get past the dynamic linker's symbol resolution
4714 function. A zero value indicates that it is not important or necessary
4715 to set a breakpoint to get through the dynamic linker and that single
4716 stepping will suffice.
4718 @item CORE_ADDR gdbarch_integer_to_address (@var{gdbarch}, @var{type}, @var{buf})
4719 @findex gdbarch_integer_to_address
4720 @cindex converting integers to addresses
4721 Define this when the architecture needs to handle non-pointer to address
4722 conversions specially. Converts that value to an address according to
4723 the current architectures conventions.
4725 @emph{Pragmatics: When the user copies a well defined expression from
4726 their source code and passes it, as a parameter, to @value{GDBN}'s
4727 @code{print} command, they should get the same value as would have been
4728 computed by the target program. Any deviation from this rule can cause
4729 major confusion and annoyance, and needs to be justified carefully. In
4730 other words, @value{GDBN} doesn't really have the freedom to do these
4731 conversions in clever and useful ways. It has, however, been pointed
4732 out that users aren't complaining about how @value{GDBN} casts integers
4733 to pointers; they are complaining that they can't take an address from a
4734 disassembly listing and give it to @code{x/i}. Adding an architecture
4735 method like @code{gdbarch_integer_to_address} certainly makes it possible for
4736 @value{GDBN} to ``get it right'' in all circumstances.}
4738 @xref{Target Architecture Definition, , Pointers Are Not Always
4741 @item CORE_ADDR gdbarch_pointer_to_address (@var{gdbarch}, @var{type}, @var{buf})
4742 @findex gdbarch_pointer_to_address
4743 Assume that @var{buf} holds a pointer of type @var{type}, in the
4744 appropriate format for the current architecture. Return the byte
4745 address the pointer refers to.
4746 @xref{Target Architecture Definition, , Pointers Are Not Always Addresses}.
4748 @item void gdbarch_register_to_value(@var{gdbarch}, @var{frame}, @var{regnum}, @var{type}, @var{fur})
4749 @findex gdbarch_register_to_value
4750 Convert the raw contents of register @var{regnum} into a value of type
4752 @xref{Target Architecture Definition, , Using Different Register and Memory Data Representations}.
4754 @item REGISTER_CONVERT_TO_VIRTUAL(@var{reg}, @var{type}, @var{from}, @var{to})
4755 @findex REGISTER_CONVERT_TO_VIRTUAL
4756 Convert the value of register @var{reg} from its raw form to its virtual
4758 @xref{Target Architecture Definition, , Raw and Virtual Register Representations}.
4760 @item REGISTER_CONVERT_TO_RAW(@var{type}, @var{reg}, @var{from}, @var{to})
4761 @findex REGISTER_CONVERT_TO_RAW
4762 Convert the value of register @var{reg} from its virtual form to its raw
4764 @xref{Target Architecture Definition, , Raw and Virtual Register Representations}.
4766 @item const struct regset *regset_from_core_section (struct gdbarch * @var{gdbarch}, const char * @var{sect_name}, size_t @var{sect_size})
4767 @findex regset_from_core_section
4768 Return the appropriate register set for a core file section with name
4769 @var{sect_name} and size @var{sect_size}.
4771 @item SOFTWARE_SINGLE_STEP_P()
4772 @findex SOFTWARE_SINGLE_STEP_P
4773 Define this as 1 if the target does not have a hardware single-step
4774 mechanism. The macro @code{SOFTWARE_SINGLE_STEP} must also be defined.
4776 @item SOFTWARE_SINGLE_STEP(@var{signal}, @var{insert_breakpoints_p})
4777 @findex SOFTWARE_SINGLE_STEP
4778 A function that inserts or removes (depending on
4779 @var{insert_breakpoints_p}) breakpoints at each possible destinations of
4780 the next instruction. See @file{sparc-tdep.c} and @file{rs6000-tdep.c}
4783 @item set_gdbarch_sofun_address_maybe_missing (@var{gdbarch}, @var{set})
4784 @findex set_gdbarch_sofun_address_maybe_missing
4785 Somebody clever observed that, the more actual addresses you have in the
4786 debug information, the more time the linker has to spend relocating
4787 them. So whenever there's some other way the debugger could find the
4788 address it needs, you should omit it from the debug info, to make
4791 Calling @code{set_gdbarch_sofun_address_maybe_missing} with a non-zero
4792 argument @var{set} indicates that a particular set of hacks of this sort
4793 are in use, affecting @code{N_SO} and @code{N_FUN} entries in stabs-format
4794 debugging information. @code{N_SO} stabs mark the beginning and ending
4795 addresses of compilation units in the text segment. @code{N_FUN} stabs
4796 mark the starts and ends of functions.
4798 In this case, @value{GDBN} assumes two things:
4802 @code{N_FUN} stabs have an address of zero. Instead of using those
4803 addresses, you should find the address where the function starts by
4804 taking the function name from the stab, and then looking that up in the
4805 minsyms (the linker/assembler symbol table). In other words, the stab
4806 has the name, and the linker/assembler symbol table is the only place
4807 that carries the address.
4810 @code{N_SO} stabs have an address of zero, too. You just look at the
4811 @code{N_FUN} stabs that appear before and after the @code{N_SO} stab, and
4812 guess the starting and ending addresses of the compilation unit from them.
4815 @item int gdbarch_stabs_argument_has_addr (@var{gdbarch}, @var{type})
4816 @findex gdbarch_stabs_argument_has_addr
4817 @anchor{gdbarch_stabs_argument_has_addr} Define this function to return
4818 nonzero if a function argument of type @var{type} is passed by reference
4821 @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})
4822 @findex gdbarch_push_dummy_call
4823 @anchor{gdbarch_push_dummy_call} Define this to push the dummy frame's call to
4824 the inferior function onto the stack. In addition to pushing @var{nargs}, the
4825 code should push @var{struct_addr} (when @var{struct_return} is non-zero), and
4826 the return address (@var{bp_addr}).
4828 @var{function} is a pointer to a @code{struct value}; on architectures that use
4829 function descriptors, this contains the function descriptor value.
4831 Returns the updated top-of-stack pointer.
4833 @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})
4834 @findex gdbarch_push_dummy_code
4835 @anchor{gdbarch_push_dummy_code} Given a stack based call dummy, push the
4836 instruction sequence (including space for a breakpoint) to which the
4837 called function should return.
4839 Set @var{bp_addr} to the address at which the breakpoint instruction
4840 should be inserted, @var{real_pc} to the resume address when starting
4841 the call sequence, and return the updated inner-most stack address.
4843 By default, the stack is grown sufficient to hold a frame-aligned
4844 (@pxref{frame_align}) breakpoint, @var{bp_addr} is set to the address
4845 reserved for that breakpoint, and @var{real_pc} set to @var{funaddr}.
4847 This method replaces @w{@code{gdbarch_call_dummy_location (@var{gdbarch})}}.
4849 @item int gdbarch_sdb_reg_to_regnum (@var{gdbarch}, @var{sdb_regnr})
4850 @findex gdbarch_sdb_reg_to_regnum
4851 Use this function to convert sdb register @var{sdb_regnr} into @value{GDBN}
4852 regnum. If not defined, no conversion will be done.
4854 @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})
4855 @findex gdbarch_return_value
4856 @anchor{gdbarch_return_value} Given a function with a return-value of
4857 type @var{rettype}, return which return-value convention that function
4860 @value{GDBN} currently recognizes two function return-value conventions:
4861 @code{RETURN_VALUE_REGISTER_CONVENTION} where the return value is found
4862 in registers; and @code{RETURN_VALUE_STRUCT_CONVENTION} where the return
4863 value is found in memory and the address of that memory location is
4864 passed in as the function's first parameter.
4866 If the register convention is being used, and @var{writebuf} is
4867 non-@code{NULL}, also copy the return-value in @var{writebuf} into
4870 If the register convention is being used, and @var{readbuf} is
4871 non-@code{NULL}, also copy the return value from @var{regcache} into
4872 @var{readbuf} (@var{regcache} contains a copy of the registers from the
4873 just returned function).
4875 @emph{Maintainer note: This method replaces separate predicate, extract,
4876 store methods. By having only one method, the logic needed to determine
4877 the return-value convention need only be implemented in one place. If
4878 @value{GDBN} were written in an @sc{oo} language, this method would
4879 instead return an object that knew how to perform the register
4880 return-value extract and store.}
4882 @emph{Maintainer note: This method does not take a @var{gcc_p}
4883 parameter, and such a parameter should not be added. If an architecture
4884 that requires per-compiler or per-function information be identified,
4885 then the replacement of @var{rettype} with @code{struct value}
4886 @var{function} should be pursued.}
4888 @emph{Maintainer note: The @var{regcache} parameter limits this methods
4889 to the inner most frame. While replacing @var{regcache} with a
4890 @code{struct frame_info} @var{frame} parameter would remove that
4891 limitation there has yet to be a demonstrated need for such a change.}
4893 @item void gdbarch_skip_permanent_breakpoint (@var{gdbarch}, @var{regcache})
4894 @findex gdbarch_skip_permanent_breakpoint
4895 Advance the inferior's PC past a permanent breakpoint. @value{GDBN} normally
4896 steps over a breakpoint by removing it, stepping one instruction, and
4897 re-inserting the breakpoint. However, permanent breakpoints are
4898 hardwired into the inferior, and can't be removed, so this strategy
4899 doesn't work. Calling @code{gdbarch_skip_permanent_breakpoint} adjusts the
4900 processor's state so that execution will resume just after the breakpoint.
4901 This function does the right thing even when the breakpoint is in the delay slot
4902 of a branch or jump.
4904 @item CORE_ADDR gdbarch_skip_trampoline_code (@var{gdbarch}, @var{frame}, @var{pc})
4905 @findex gdbarch_skip_trampoline_code
4906 If the target machine has trampoline code that sits between callers and
4907 the functions being called, then define this function to return a new PC
4908 that is at the start of the real function.
4910 @item int gdbarch_deprecated_fp_regnum (@var{gdbarch})
4911 @findex gdbarch_deprecated_fp_regnum
4912 If the frame pointer is in a register, use this function to return the
4913 number of that register.
4915 @item int gdbarch_stab_reg_to_regnum (@var{gdbarch}, @var{stab_regnr})
4916 @findex gdbarch_stab_reg_to_regnum
4917 Use this function to convert stab register @var{stab_regnr} into @value{GDBN}
4918 regnum. If not defined, no conversion will be done.
4920 @item SYMBOL_RELOADING_DEFAULT
4921 @findex SYMBOL_RELOADING_DEFAULT
4922 The default value of the ``symbol-reloading'' variable. (Never defined in
4925 @item TARGET_CHAR_BIT
4926 @findex TARGET_CHAR_BIT
4927 Number of bits in a char; defaults to 8.
4929 @item int gdbarch_char_signed (@var{gdbarch})
4930 @findex gdbarch_char_signed
4931 Non-zero if @code{char} is normally signed on this architecture; zero if
4932 it should be unsigned.
4934 The ISO C standard requires the compiler to treat @code{char} as
4935 equivalent to either @code{signed char} or @code{unsigned char}; any
4936 character in the standard execution set is supposed to be positive.
4937 Most compilers treat @code{char} as signed, but @code{char} is unsigned
4938 on the IBM S/390, RS6000, and PowerPC targets.
4940 @item int gdbarch_double_bit (@var{gdbarch})
4941 @findex gdbarch_double_bit
4942 Number of bits in a double float; defaults to @w{@code{8 * TARGET_CHAR_BIT}}.
4944 @item int gdbarch_float_bit (@var{gdbarch})
4945 @findex gdbarch_float_bit
4946 Number of bits in a float; defaults to @w{@code{4 * TARGET_CHAR_BIT}}.
4948 @item int gdbarch_int_bit (@var{gdbarch})
4949 @findex gdbarch_int_bit
4950 Number of bits in an integer; defaults to @w{@code{4 * TARGET_CHAR_BIT}}.
4952 @item int gdbarch_long_bit (@var{gdbarch})
4953 @findex gdbarch_long_bit
4954 Number of bits in a long integer; defaults to @w{@code{4 * TARGET_CHAR_BIT}}.
4956 @item int gdbarch_long_double_bit (@var{gdbarch})
4957 @findex gdbarch_long_double_bit
4958 Number of bits in a long double float;
4959 defaults to @w{@code{2 * gdbarch_double_bit (@var{gdbarch})}}.
4961 @item int gdbarch_long_long_bit (@var{gdbarch})
4962 @findex gdbarch_long_long_bit
4963 Number of bits in a long long integer; defaults to
4964 @w{@code{2 * gdbarch_long_bit (@var{gdbarch})}}.
4966 @item int gdbarch_ptr_bit (@var{gdbarch})
4967 @findex gdbarch_ptr_bit
4968 Number of bits in a pointer; defaults to
4969 @w{@code{gdbarch_int_bit (@var{gdbarch})}}.
4971 @item int gdbarch_short_bit (@var{gdbarch})
4972 @findex gdbarch_short_bit
4973 Number of bits in a short integer; defaults to @w{@code{2 * TARGET_CHAR_BIT}}.
4975 @item void gdbarch_virtual_frame_pointer (@var{gdbarch}, @var{pc}, @var{frame_regnum}, @var{frame_offset})
4976 @findex gdbarch_virtual_frame_pointer
4977 Returns a @code{(@var{register}, @var{offset})} pair representing the virtual
4978 frame pointer in use at the code address @var{pc}. If virtual frame
4979 pointers are not used, a default definition simply returns
4980 @code{gdbarch_deprecated_fp_regnum} (or @code{gdbarch_sp_regnum}, if
4981 no frame pointer is defined), with an offset of zero.
4983 @c need to explain virtual frame pointers, they are recorded in agent
4984 @c expressions for tracepoints
4986 @item TARGET_HAS_HARDWARE_WATCHPOINTS
4987 If non-zero, the target has support for hardware-assisted
4988 watchpoints. @xref{Algorithms, watchpoints}, for more details and
4989 other related macros.
4991 @item int gdbarch_print_insn (@var{gdbarch}, @var{vma}, @var{info})
4992 @findex gdbarch_print_insn
4993 This is the function used by @value{GDBN} to print an assembly
4994 instruction. It prints the instruction at address @var{vma} in
4995 debugged memory and returns the length of the instruction, in bytes.
4996 This usually points to a function in the @code{opcodes} library
4997 (@pxref{Support Libraries, ,Opcodes}). @var{info} is a structure (of
4998 type @code{disassemble_info}) defined in the header file
4999 @file{include/dis-asm.h}, and used to pass information to the
5000 instruction decoding routine.
5002 @item frame_id gdbarch_dummy_id (@var{gdbarch}, @var{frame})
5003 @findex gdbarch_dummy_id
5004 @anchor{gdbarch_dummy_id} Given @var{frame} return a @w{@code{struct
5005 frame_id}} that uniquely identifies an inferior function call's dummy
5006 frame. The value returned must match the dummy frame stack value
5007 previously saved by @code{call_function_by_hand}.
5009 @item void gdbarch_value_to_register (@var{gdbarch}, @var{frame}, @var{type}, @var{buf})
5010 @findex gdbarch_value_to_register
5011 Convert a value of type @var{type} into the raw contents of a register.
5012 @xref{Target Architecture Definition, , Using Different Register and Memory Data Representations}.
5016 Motorola M68K target conditionals.
5020 Define this to be the 4-bit location of the breakpoint trap vector. If
5021 not defined, it will default to @code{0xf}.
5023 @item REMOTE_BPT_VECTOR
5024 Defaults to @code{1}.
5028 @node Adding a New Target
5029 @section Adding a New Target
5031 @cindex adding a target
5032 The following files add a target to @value{GDBN}:
5035 @cindex target dependent files
5037 @item gdb/@var{ttt}-tdep.c
5038 Contains any miscellaneous code required for this target machine. On
5039 some machines it doesn't exist at all.
5041 @item gdb/@var{arch}-tdep.c
5042 @itemx gdb/@var{arch}-tdep.h
5043 This is required to describe the basic layout of the target machine's
5044 processor chip (registers, stack, etc.). It can be shared among many
5045 targets that use the same processor architecture.
5049 (Target header files such as
5050 @file{gdb/config/@var{arch}/tm-@var{ttt}.h},
5051 @file{gdb/config/@var{arch}/tm-@var{arch}.h}, and
5052 @file{config/tm-@var{os}.h} are no longer used.)
5054 @findex _initialize_@var{arch}_tdep
5055 A @value{GDBN} description for a new architecture, arch is created by
5056 defining a global function @code{_initialize_@var{arch}_tdep}, by
5057 convention in the source file @file{@var{arch}-tdep.c}. For
5058 example, in the case of the OpenRISC 1000, this function is called
5059 @code{_initialize_or1k_tdep} and is found in the file
5062 The object file resulting from compiling this source file, which will
5063 contain the implementation of the
5064 @code{_initialize_@var{arch}_tdep} function is specified in the
5065 @value{GDBN} @file{configure.tgt} file, which includes a large case
5066 statement pattern matching against the @code{--target} option of the
5067 @kbd{configure} script.
5070 @emph{Note:} If the architecture requires multiple source files, the
5071 corresponding binaries should be included in
5072 @file{configure.tgt}. However if there are header files, the
5073 dependencies on these will not be picked up from the entries in
5074 @file{configure.tgt}. The @file{Makefile.in} file will need extending to
5075 show these dependencies.
5078 @findex gdbarch_register
5079 A new struct gdbarch, defining the new architecture, is created within
5080 the @code{_initialize_@var{arch}_tdep} function by calling
5081 @code{gdbarch_register}:
5084 void gdbarch_register (enum bfd_architecture architecture,
5085 gdbarch_init_ftype *init_func,
5086 gdbarch_dump_tdep_ftype *tdep_dump_func);
5089 This function has been described fully in an earlier
5090 section. @xref{How an Architecture is Represented, , How an
5091 Architecture is Represented}.
5093 The new @code{@w{struct gdbarch}} should contain implementations of
5094 the necessary functions (described in the previous sections) to
5095 describe the basic layout of the target machine's processor chip
5096 (registers, stack, etc.). It can be shared among many targets that use
5097 the same processor architecture.
5099 @node Target Descriptions
5100 @chapter Target Descriptions
5101 @cindex target descriptions
5103 The target architecture definition (@pxref{Target Architecture Definition})
5104 contains @value{GDBN}'s hard-coded knowledge about an architecture. For
5105 some platforms, it is handy to have more flexible knowledge about a specific
5106 instance of the architecture---for instance, a processor or development board.
5107 @dfn{Target descriptions} provide a mechanism for the user to tell @value{GDBN}
5108 more about what their target supports, or for the target to tell @value{GDBN}
5111 For details on writing, automatically supplying, and manually selecting
5112 target descriptions, see @ref{Target Descriptions, , , gdb,
5113 Debugging with @value{GDBN}}. This section will cover some related
5114 topics about the @value{GDBN} internals.
5117 * Target Descriptions Implementation::
5118 * Adding Target Described Register Support::
5121 @node Target Descriptions Implementation
5122 @section Target Descriptions Implementation
5123 @cindex target descriptions, implementation
5125 Before @value{GDBN} connects to a new target, or runs a new program on
5126 an existing target, it discards any existing target description and
5127 reverts to a default gdbarch. Then, after connecting, it looks for a
5128 new target description by calling @code{target_find_description}.
5130 A description may come from a user specified file (XML), the remote
5131 @samp{qXfer:features:read} packet (also XML), or from any custom
5132 @code{to_read_description} routine in the target vector. For instance,
5133 the remote target supports guessing whether a MIPS target is 32-bit or
5134 64-bit based on the size of the @samp{g} packet.
5136 If any target description is found, @value{GDBN} creates a new gdbarch
5137 incorporating the description by calling @code{gdbarch_update_p}. Any
5138 @samp{<architecture>} element is handled first, to determine which
5139 architecture's gdbarch initialization routine is called to create the
5140 new architecture. Then the initialization routine is called, and has
5141 a chance to adjust the constructed architecture based on the contents
5142 of the target description. For instance, it can recognize any
5143 properties set by a @code{to_read_description} routine. Also
5144 see @ref{Adding Target Described Register Support}.
5146 @node Adding Target Described Register Support
5147 @section Adding Target Described Register Support
5148 @cindex target descriptions, adding register support
5150 Target descriptions can report additional registers specific to an
5151 instance of the target. But it takes a little work in the architecture
5152 specific routines to support this.
5154 A target description must either have no registers or a complete
5155 set---this avoids complexity in trying to merge standard registers
5156 with the target defined registers. It is the architecture's
5157 responsibility to validate that a description with registers has
5158 everything it needs. To keep architecture code simple, the same
5159 mechanism is used to assign fixed internal register numbers to
5162 If @code{tdesc_has_registers} returns 1, the description contains
5163 registers. The architecture's @code{gdbarch_init} routine should:
5168 Call @code{tdesc_data_alloc} to allocate storage, early, before
5169 searching for a matching gdbarch or allocating a new one.
5172 Use @code{tdesc_find_feature} to locate standard features by name.
5175 Use @code{tdesc_numbered_register} and @code{tdesc_numbered_register_choices}
5176 to locate the expected registers in the standard features.
5179 Return @code{NULL} if a required feature is missing, or if any standard
5180 feature is missing expected registers. This will produce a warning that
5181 the description was incomplete.
5184 Free the allocated data before returning, unless @code{tdesc_use_registers}
5188 Call @code{set_gdbarch_num_regs} as usual, with a number higher than any
5189 fixed number passed to @code{tdesc_numbered_register}.
5192 Call @code{tdesc_use_registers} after creating a new gdbarch, before
5197 After @code{tdesc_use_registers} has been called, the architecture's
5198 @code{register_name}, @code{register_type}, and @code{register_reggroup_p}
5199 routines will not be called; that information will be taken from
5200 the target description. @code{num_regs} may be increased to account
5201 for any additional registers in the description.
5203 Pseudo-registers require some extra care:
5208 Using @code{tdesc_numbered_register} allows the architecture to give
5209 constant register numbers to standard architectural registers, e.g.@:
5210 as an @code{enum} in @file{@var{arch}-tdep.h}. But because
5211 pseudo-registers are always numbered above @code{num_regs},
5212 which may be increased by the description, constant numbers
5213 can not be used for pseudos. They must be numbered relative to
5214 @code{num_regs} instead.
5217 The description will not describe pseudo-registers, so the
5218 architecture must call @code{set_tdesc_pseudo_register_name},
5219 @code{set_tdesc_pseudo_register_type}, and
5220 @code{set_tdesc_pseudo_register_reggroup_p} to supply routines
5221 describing pseudo registers. These routines will be passed
5222 internal register numbers, so the same routines used for the
5223 gdbarch equivalents are usually suitable.
5228 @node Target Vector Definition
5230 @chapter Target Vector Definition
5231 @cindex target vector
5233 The target vector defines the interface between @value{GDBN}'s
5234 abstract handling of target systems, and the nitty-gritty code that
5235 actually exercises control over a process or a serial port.
5236 @value{GDBN} includes some 30-40 different target vectors; however,
5237 each configuration of @value{GDBN} includes only a few of them.
5240 * Managing Execution State::
5241 * Existing Targets::
5244 @node Managing Execution State
5245 @section Managing Execution State
5246 @cindex execution state
5248 A target vector can be completely inactive (not pushed on the target
5249 stack), active but not running (pushed, but not connected to a fully
5250 manifested inferior), or completely active (pushed, with an accessible
5251 inferior). Most targets are only completely inactive or completely
5252 active, but some support persistent connections to a target even
5253 when the target has exited or not yet started.
5255 For example, connecting to the simulator using @code{target sim} does
5256 not create a running program. Neither registers nor memory are
5257 accessible until @code{run}. Similarly, after @code{kill}, the
5258 program can not continue executing. But in both cases @value{GDBN}
5259 remains connected to the simulator, and target-specific commands
5260 are directed to the simulator.
5262 A target which only supports complete activation should push itself
5263 onto the stack in its @code{to_open} routine (by calling
5264 @code{push_target}), and unpush itself from the stack in its
5265 @code{to_mourn_inferior} routine (by calling @code{unpush_target}).
5267 A target which supports both partial and complete activation should
5268 still call @code{push_target} in @code{to_open}, but not call
5269 @code{unpush_target} in @code{to_mourn_inferior}. Instead, it should
5270 call either @code{target_mark_running} or @code{target_mark_exited}
5271 in its @code{to_open}, depending on whether the target is fully active
5272 after connection. It should also call @code{target_mark_running} any
5273 time the inferior becomes fully active (e.g.@: in
5274 @code{to_create_inferior} and @code{to_attach}), and
5275 @code{target_mark_exited} when the inferior becomes inactive (in
5276 @code{to_mourn_inferior}). The target should also make sure to call
5277 @code{target_mourn_inferior} from its @code{to_kill}, to return the
5278 target to inactive state.
5280 @node Existing Targets
5281 @section Existing Targets
5284 @subsection File Targets
5286 Both executables and core files have target vectors.
5288 @subsection Standard Protocol and Remote Stubs
5290 @value{GDBN}'s file @file{remote.c} talks a serial protocol to code that
5291 runs in the target system. @value{GDBN} provides several sample
5292 @dfn{stubs} that can be integrated into target programs or operating
5293 systems for this purpose; they are named @file{@var{cpu}-stub.c}. Many
5294 operating systems, embedded targets, emulators, and simulators already
5295 have a @value{GDBN} stub built into them, and maintenance of the remote
5296 protocol must be careful to preserve compatibility.
5298 The @value{GDBN} user's manual describes how to put such a stub into
5299 your target code. What follows is a discussion of integrating the
5300 SPARC stub into a complicated operating system (rather than a simple
5301 program), by Stu Grossman, the author of this stub.
5303 The trap handling code in the stub assumes the following upon entry to
5308 %l1 and %l2 contain pc and npc respectively at the time of the trap;
5314 you are in the correct trap window.
5317 As long as your trap handler can guarantee those conditions, then there
5318 is no reason why you shouldn't be able to ``share'' traps with the stub.
5319 The stub has no requirement that it be jumped to directly from the
5320 hardware trap vector. That is why it calls @code{exceptionHandler()},
5321 which is provided by the external environment. For instance, this could
5322 set up the hardware traps to actually execute code which calls the stub
5323 first, and then transfers to its own trap handler.
5325 For the most point, there probably won't be much of an issue with
5326 ``sharing'' traps, as the traps we use are usually not used by the kernel,
5327 and often indicate unrecoverable error conditions. Anyway, this is all
5328 controlled by a table, and is trivial to modify. The most important
5329 trap for us is for @code{ta 1}. Without that, we can't single step or
5330 do breakpoints. Everything else is unnecessary for the proper operation
5331 of the debugger/stub.
5333 From reading the stub, it's probably not obvious how breakpoints work.
5334 They are simply done by deposit/examine operations from @value{GDBN}.
5336 @subsection ROM Monitor Interface
5338 @subsection Custom Protocols
5340 @subsection Transport Layer
5342 @subsection Builtin Simulator
5345 @node Native Debugging
5347 @chapter Native Debugging
5348 @cindex native debugging
5350 Several files control @value{GDBN}'s configuration for native support:
5354 @item gdb/config/@var{arch}/@var{xyz}.mh
5355 Specifies Makefile fragments needed by a @emph{native} configuration on
5356 machine @var{xyz}. In particular, this lists the required
5357 native-dependent object files, by defining @samp{NATDEPFILES=@dots{}}.
5358 Also specifies the header file which describes native support on
5359 @var{xyz}, by defining @samp{NAT_FILE= nm-@var{xyz}.h}. You can also
5360 define @samp{NAT_CFLAGS}, @samp{NAT_ADD_FILES}, @samp{NAT_CLIBS},
5361 @samp{NAT_CDEPS}, @samp{NAT_GENERATED_FILES}, etc.; see @file{Makefile.in}.
5363 @emph{Maintainer's note: The @file{.mh} suffix is because this file
5364 originally contained @file{Makefile} fragments for hosting @value{GDBN}
5365 on machine @var{xyz}. While the file is no longer used for this
5366 purpose, the @file{.mh} suffix remains. Perhaps someone will
5367 eventually rename these fragments so that they have a @file{.mn}
5370 @item gdb/config/@var{arch}/nm-@var{xyz}.h
5371 (@file{nm.h} is a link to this file, created by @code{configure}). Contains C
5372 macro definitions describing the native system environment, such as
5373 child process control and core file support.
5375 @item gdb/@var{xyz}-nat.c
5376 Contains any miscellaneous C code required for this native support of
5377 this machine. On some machines it doesn't exist at all.
5380 There are some ``generic'' versions of routines that can be used by
5381 various systems. These can be customized in various ways by macros
5382 defined in your @file{nm-@var{xyz}.h} file. If these routines work for
5383 the @var{xyz} host, you can just include the generic file's name (with
5384 @samp{.o}, not @samp{.c}) in @code{NATDEPFILES}.
5386 Otherwise, if your machine needs custom support routines, you will need
5387 to write routines that perform the same functions as the generic file.
5388 Put them into @file{@var{xyz}-nat.c}, and put @file{@var{xyz}-nat.o}
5389 into @code{NATDEPFILES}.
5393 This contains the @emph{target_ops vector} that supports Unix child
5394 processes on systems which use ptrace and wait to control the child.
5397 This contains the @emph{target_ops vector} that supports Unix child
5398 processes on systems which use /proc to control the child.
5401 This does the low-level grunge that uses Unix system calls to do a ``fork
5402 and exec'' to start up a child process.
5405 This is the low level interface to inferior processes for systems using
5406 the Unix @code{ptrace} call in a vanilla way.
5415 @section shared libraries
5417 @section Native Conditionals
5418 @cindex native conditionals
5420 When @value{GDBN} is configured and compiled, various macros are
5421 defined or left undefined, to control compilation when the host and
5422 target systems are the same. These macros should be defined (or left
5423 undefined) in @file{nm-@var{system}.h}.
5427 @item I386_USE_GENERIC_WATCHPOINTS
5428 An x86-based machine can define this to use the generic x86 watchpoint
5429 support; see @ref{Algorithms, I386_USE_GENERIC_WATCHPOINTS}.
5431 @item SOLIB_ADD (@var{filename}, @var{from_tty}, @var{targ}, @var{readsyms})
5433 Define this to expand into an expression that will cause the symbols in
5434 @var{filename} to be added to @value{GDBN}'s symbol table. If
5435 @var{readsyms} is zero symbols are not read but any necessary low level
5436 processing for @var{filename} is still done.
5438 @item SOLIB_CREATE_INFERIOR_HOOK
5439 @findex SOLIB_CREATE_INFERIOR_HOOK
5440 Define this to expand into any shared-library-relocation code that you
5441 want to be run just after the child process has been forked.
5443 @item START_INFERIOR_TRAPS_EXPECTED
5444 @findex START_INFERIOR_TRAPS_EXPECTED
5445 When starting an inferior, @value{GDBN} normally expects to trap
5447 the shell execs, and once when the program itself execs. If the actual
5448 number of traps is something other than 2, then define this macro to
5449 expand into the number expected.
5453 @node Support Libraries
5455 @chapter Support Libraries
5460 BFD provides support for @value{GDBN} in several ways:
5463 @item identifying executable and core files
5464 BFD will identify a variety of file types, including a.out, coff, and
5465 several variants thereof, as well as several kinds of core files.
5467 @item access to sections of files
5468 BFD parses the file headers to determine the names, virtual addresses,
5469 sizes, and file locations of all the various named sections in files
5470 (such as the text section or the data section). @value{GDBN} simply
5471 calls BFD to read or write section @var{x} at byte offset @var{y} for
5474 @item specialized core file support
5475 BFD provides routines to determine the failing command name stored in a
5476 core file, the signal with which the program failed, and whether a core
5477 file matches (i.e.@: could be a core dump of) a particular executable
5480 @item locating the symbol information
5481 @value{GDBN} uses an internal interface of BFD to determine where to find the
5482 symbol information in an executable file or symbol-file. @value{GDBN} itself
5483 handles the reading of symbols, since BFD does not ``understand'' debug
5484 symbols, but @value{GDBN} uses BFD's cached information to find the symbols,
5489 @cindex opcodes library
5491 The opcodes library provides @value{GDBN}'s disassembler. (It's a separate
5492 library because it's also used in binutils, for @file{objdump}).
5495 @cindex readline library
5496 The @code{readline} library provides a set of functions for use by applications
5497 that allow users to edit command lines as they are typed in.
5500 @cindex @code{libiberty} library
5502 The @code{libiberty} library provides a set of functions and features
5503 that integrate and improve on functionality found in modern operating
5504 systems. Broadly speaking, such features can be divided into three
5505 groups: supplemental functions (functions that may be missing in some
5506 environments and operating systems), replacement functions (providing
5507 a uniform and easier to use interface for commonly used standard
5508 functions), and extensions (which provide additional functionality
5509 beyond standard functions).
5511 @value{GDBN} uses various features provided by the @code{libiberty}
5512 library, for instance the C@t{++} demangler, the @acronym{IEEE}
5513 floating format support functions, the input options parser
5514 @samp{getopt}, the @samp{obstack} extension, and other functions.
5516 @subsection @code{obstacks} in @value{GDBN}
5517 @cindex @code{obstacks}
5519 The obstack mechanism provides a convenient way to allocate and free
5520 chunks of memory. Each obstack is a pool of memory that is managed
5521 like a stack. Objects (of any nature, size and alignment) are
5522 allocated and freed in a @acronym{LIFO} fashion on an obstack (see
5523 @code{libiberty}'s documentation for a more detailed explanation of
5526 The most noticeable use of the @code{obstacks} in @value{GDBN} is in
5527 object files. There is an obstack associated with each internal
5528 representation of an object file. Lots of things get allocated on
5529 these @code{obstacks}: dictionary entries, blocks, blockvectors,
5530 symbols, minimal symbols, types, vectors of fundamental types, class
5531 fields of types, object files section lists, object files section
5532 offset lists, line tables, symbol tables, partial symbol tables,
5533 string tables, symbol table private data, macros tables, debug
5534 information sections and entries, import and export lists (som),
5535 unwind information (hppa), dwarf2 location expressions data. Plus
5536 various strings such as directory names strings, debug format strings,
5539 An essential and convenient property of all data on @code{obstacks} is
5540 that memory for it gets allocated (with @code{obstack_alloc}) at
5541 various times during a debugging session, but it is released all at
5542 once using the @code{obstack_free} function. The @code{obstack_free}
5543 function takes a pointer to where in the stack it must start the
5544 deletion from (much like the cleanup chains have a pointer to where to
5545 start the cleanups). Because of the stack like structure of the
5546 @code{obstacks}, this allows to free only a top portion of the
5547 obstack. There are a few instances in @value{GDBN} where such thing
5548 happens. Calls to @code{obstack_free} are done after some local data
5549 is allocated to the obstack. Only the local data is deleted from the
5550 obstack. Of course this assumes that nothing between the
5551 @code{obstack_alloc} and the @code{obstack_free} allocates anything
5552 else on the same obstack. For this reason it is best and safest to
5553 use temporary @code{obstacks}.
5555 Releasing the whole obstack is also not safe per se. It is safe only
5556 under the condition that we know the @code{obstacks} memory is no
5557 longer needed. In @value{GDBN} we get rid of the @code{obstacks} only
5558 when we get rid of the whole objfile(s), for instance upon reading a
5562 @cindex regular expressions library
5573 @item SIGN_EXTEND_CHAR
5575 @item SWITCH_ENUM_BUG
5584 @section Array Containers
5585 @cindex Array Containers
5588 Often it is necessary to manipulate a dynamic array of a set of
5589 objects. C forces some bookkeeping on this, which can get cumbersome
5590 and repetitive. The @file{vec.h} file contains macros for defining
5591 and using a typesafe vector type. The functions defined will be
5592 inlined when compiling, and so the abstraction cost should be zero.
5593 Domain checks are added to detect programming errors.
5595 An example use would be an array of symbols or section information.
5596 The array can be grown as symbols are read in (or preallocated), and
5597 the accessor macros provided keep care of all the necessary
5598 bookkeeping. Because the arrays are type safe, there is no danger of
5599 accidentally mixing up the contents. Think of these as C++ templates,
5600 but implemented in C.
5602 Because of the different behavior of structure objects, scalar objects
5603 and of pointers, there are three flavors of vector, one for each of
5604 these variants. Both the structure object and pointer variants pass
5605 pointers to objects around --- in the former case the pointers are
5606 stored into the vector and in the latter case the pointers are
5607 dereferenced and the objects copied into the vector. The scalar
5608 object variant is suitable for @code{int}-like objects, and the vector
5609 elements are returned by value.
5611 There are both @code{index} and @code{iterate} accessors. The iterator
5612 returns a boolean iteration condition and updates the iteration
5613 variable passed by reference. Because the iterator will be inlined,
5614 the address-of can be optimized away.
5616 The vectors are implemented using the trailing array idiom, thus they
5617 are not resizeable without changing the address of the vector object
5618 itself. This means you cannot have variables or fields of vector type
5619 --- always use a pointer to a vector. The one exception is the final
5620 field of a structure, which could be a vector type. You will have to
5621 use the @code{embedded_size} & @code{embedded_init} calls to create
5622 such objects, and they will probably not be resizeable (so don't use
5623 the @dfn{safe} allocation variants). The trailing array idiom is used
5624 (rather than a pointer to an array of data), because, if we allow
5625 @code{NULL} to also represent an empty vector, empty vectors occupy
5626 minimal space in the structure containing them.
5628 Each operation that increases the number of active elements is
5629 available in @dfn{quick} and @dfn{safe} variants. The former presumes
5630 that there is sufficient allocated space for the operation to succeed
5631 (it dies if there is not). The latter will reallocate the vector, if
5632 needed. Reallocation causes an exponential increase in vector size.
5633 If you know you will be adding N elements, it would be more efficient
5634 to use the reserve operation before adding the elements with the
5635 @dfn{quick} operation. This will ensure there are at least as many
5636 elements as you ask for, it will exponentially increase if there are
5637 too few spare slots. If you want reserve a specific number of slots,
5638 but do not want the exponential increase (for instance, you know this
5639 is the last allocation), use a negative number for reservation. You
5640 can also create a vector of a specific size from the get go.
5642 You should prefer the push and pop operations, as they append and
5643 remove from the end of the vector. If you need to remove several items
5644 in one go, use the truncate operation. The insert and remove
5645 operations allow you to change elements in the middle of the vector.
5646 There are two remove operations, one which preserves the element
5647 ordering @code{ordered_remove}, and one which does not
5648 @code{unordered_remove}. The latter function copies the end element
5649 into the removed slot, rather than invoke a memmove operation. The
5650 @code{lower_bound} function will determine where to place an item in
5651 the array using insert that will maintain sorted order.
5653 If you need to directly manipulate a vector, then the @code{address}
5654 accessor will return the address of the start of the vector. Also the
5655 @code{space} predicate will tell you whether there is spare capacity in the
5656 vector. You will not normally need to use these two functions.
5658 Vector types are defined using a
5659 @code{DEF_VEC_@{O,P,I@}(@var{typename})} macro. Variables of vector
5660 type are declared using a @code{VEC(@var{typename})} macro. The
5661 characters @code{O}, @code{P} and @code{I} indicate whether
5662 @var{typename} is an object (@code{O}), pointer (@code{P}) or integral
5663 (@code{I}) type. Be careful to pick the correct one, as you'll get an
5664 awkward and inefficient API if you use the wrong one. There is a
5665 check, which results in a compile-time warning, for the @code{P} and
5666 @code{I} versions, but there is no check for the @code{O} versions, as
5667 that is not possible in plain C.
5669 An example of their use would be,
5672 DEF_VEC_P(tree); // non-managed tree vector.
5675 VEC(tree) *v; // A (pointer to) a vector of tree pointers.
5678 struct my_struct *s;
5680 if (VEC_length(tree, s->v)) @{ we have some contents @}
5681 VEC_safe_push(tree, s->v, decl); // append some decl onto the end
5682 for (ix = 0; VEC_iterate(tree, s->v, ix, elt); ix++)
5683 @{ do something with elt @}
5687 The @file{vec.h} file provides details on how to invoke the various
5688 accessors provided. They are enumerated here:
5692 Return the number of items in the array,
5695 Return true if the array has no elements.
5699 Return the last or arbitrary item in the array.
5702 Access an array element and indicate whether the array has been
5707 Create and destroy an array.
5709 @item VEC_embedded_size
5710 @itemx VEC_embedded_init
5711 Helpers for embedding an array as the final element of another struct.
5717 Return the amount of free space in an array.
5720 Ensure a certain amount of free space.
5722 @item VEC_quick_push
5723 @itemx VEC_safe_push
5724 Append to an array, either assuming the space is available, or making
5728 Remove the last item from an array.
5731 Remove several items from the end of an array.
5734 Add several items to the end of an array.
5737 Overwrite an item in the array.
5739 @item VEC_quick_insert
5740 @itemx VEC_safe_insert
5741 Insert an item into the middle of the array. Either the space must
5742 already exist, or the space is created.
5744 @item VEC_ordered_remove
5745 @itemx VEC_unordered_remove
5746 Remove an item from the array, preserving order or not.
5748 @item VEC_block_remove
5749 Remove a set of items from the array.
5752 Provide the address of the first element.
5754 @item VEC_lower_bound
5755 Binary search the array.
5761 @node Coding Standards
5763 @chapter Coding Standards
5764 @cindex coding standards
5766 @section @value{GDBN} C Coding Standards
5768 @value{GDBN} follows the GNU coding standards, as described in
5769 @file{etc/standards.texi}. This file is also available for anonymous
5770 FTP from GNU archive sites. @value{GDBN} takes a strict interpretation
5771 of the standard; in general, when the GNU standard recommends a practice
5772 but does not require it, @value{GDBN} requires it.
5774 @value{GDBN} follows an additional set of coding standards specific to
5775 @value{GDBN}, as described in the following sections.
5779 @value{GDBN} assumes an ISO/IEC 9899:1990 (a.k.a.@: ISO C90) compliant
5782 @value{GDBN} does not assume an ISO C or POSIX compliant C library.
5784 @subsection Formatting
5786 @cindex source code formatting
5787 The standard GNU recommendations for formatting must be followed
5788 strictly. Any @value{GDBN}-specific deviation from GNU
5789 recomendations is described below.
5791 A function declaration should not have its name in column zero. A
5792 function definition should have its name in column zero.
5796 static void foo (void);
5804 @emph{Pragmatics: This simplifies scripting. Function definitions can
5805 be found using @samp{^function-name}.}
5807 There must be a space between a function or macro name and the opening
5808 parenthesis of its argument list (except for macro definitions, as
5809 required by C). There must not be a space after an open paren/bracket
5810 or before a close paren/bracket.
5812 While additional whitespace is generally helpful for reading, do not use
5813 more than one blank line to separate blocks, and avoid adding whitespace
5814 after the end of a program line (as of 1/99, some 600 lines had
5815 whitespace after the semicolon). Excess whitespace causes difficulties
5816 for @code{diff} and @code{patch} utilities.
5818 Pointers are declared using the traditional K&R C style:
5832 In addition, whitespace around casts and unary operators should follow
5833 the following guidelines:
5835 @multitable @columnfractions .2 .2 .8
5836 @item Use... @tab ...instead of @tab
5845 @item @code{(foo) x}
5850 @tab (pointer dereference)
5853 @subsection Comments
5855 @cindex comment formatting
5856 The standard GNU requirements on comments must be followed strictly.
5858 Block comments must appear in the following form, with no @code{/*}- or
5859 @code{*/}-only lines, and no leading @code{*}:
5862 /* Wait for control to return from inferior to debugger. If inferior
5863 gets a signal, we may decide to start it up again instead of
5864 returning. That is why there is a loop in this function. When
5865 this function actually returns it means the inferior should be left
5866 stopped and @value{GDBN} should read more commands. */
5869 (Note that this format is encouraged by Emacs; tabbing for a multi-line
5870 comment works correctly, and @kbd{M-q} fills the block consistently.)
5872 Put a blank line between the block comments preceding function or
5873 variable definitions, and the definition itself.
5875 In general, put function-body comments on lines by themselves, rather
5876 than trying to fit them into the 20 characters left at the end of a
5877 line, since either the comment or the code will inevitably get longer
5878 than will fit, and then somebody will have to move it anyhow.
5882 @cindex C data types
5883 Code must not depend on the sizes of C data types, the format of the
5884 host's floating point numbers, the alignment of anything, or the order
5885 of evaluation of expressions.
5887 @cindex function usage
5888 Use functions freely. There are only a handful of compute-bound areas
5889 in @value{GDBN} that might be affected by the overhead of a function
5890 call, mainly in symbol reading. Most of @value{GDBN}'s performance is
5891 limited by the target interface (whether serial line or system call).
5893 However, use functions with moderation. A thousand one-line functions
5894 are just as hard to understand as a single thousand-line function.
5896 @emph{Macros are bad, M'kay.}
5897 (But if you have to use a macro, make sure that the macro arguments are
5898 protected with parentheses.)
5902 Declarations like @samp{struct foo *} should be used in preference to
5903 declarations like @samp{typedef struct foo @{ @dots{} @} *foo_ptr}.
5905 @subsection Function Prototypes
5906 @cindex function prototypes
5908 Prototypes must be used when both @emph{declaring} and @emph{defining}
5909 a function. Prototypes for @value{GDBN} functions must include both the
5910 argument type and name, with the name matching that used in the actual
5911 function definition.
5913 All external functions should have a declaration in a header file that
5914 callers include, except for @code{_initialize_*} functions, which must
5915 be external so that @file{init.c} construction works, but shouldn't be
5916 visible to random source files.
5918 Where a source file needs a forward declaration of a static function,
5919 that declaration must appear in a block near the top of the source file.
5921 @subsection File Names
5923 Any file used when building the core of @value{GDBN} must be in lower
5924 case. Any file used when building the core of @value{GDBN} must be 8.3
5925 unique. These requirements apply to both source and generated files.
5927 @emph{Pragmatics: The core of @value{GDBN} must be buildable on many
5928 platforms including DJGPP and MacOS/HFS. Every time an unfriendly file
5929 is introduced to the build process both @file{Makefile.in} and
5930 @file{configure.in} need to be modified accordingly. Compare the
5931 convoluted conversion process needed to transform @file{COPYING} into
5932 @file{copying.c} with the conversion needed to transform
5933 @file{version.in} into @file{version.c}.}
5935 Any file non 8.3 compliant file (that is not used when building the core
5936 of @value{GDBN}) must be added to @file{gdb/config/djgpp/fnchange.lst}.
5938 @emph{Pragmatics: This is clearly a compromise.}
5940 When @value{GDBN} has a local version of a system header file (ex
5941 @file{string.h}) the file name based on the POSIX header prefixed with
5942 @file{gdb_} (@file{gdb_string.h}). These headers should be relatively
5943 independent: they should use only macros defined by @file{configure},
5944 the compiler, or the host; they should include only system headers; they
5945 should refer only to system types. They may be shared between multiple
5946 programs, e.g.@: @value{GDBN} and @sc{gdbserver}.
5948 For other files @samp{-} is used as the separator.
5950 @subsection Include Files
5952 A @file{.c} file should include @file{defs.h} first.
5954 A @file{.c} file should directly include the @code{.h} file of every
5955 declaration and/or definition it directly refers to. It cannot rely on
5958 A @file{.h} file should directly include the @code{.h} file of every
5959 declaration and/or definition it directly refers to. It cannot rely on
5960 indirect inclusion. Exception: The file @file{defs.h} does not need to
5961 be directly included.
5963 An external declaration should only appear in one include file.
5965 An external declaration should never appear in a @code{.c} file.
5966 Exception: a declaration for the @code{_initialize} function that
5967 pacifies @option{-Wmissing-declaration}.
5969 A @code{typedef} definition should only appear in one include file.
5971 An opaque @code{struct} declaration can appear in multiple @file{.h}
5972 files. Where possible, a @file{.h} file should use an opaque
5973 @code{struct} declaration instead of an include.
5975 All @file{.h} files should be wrapped in:
5978 #ifndef INCLUDE_FILE_NAME_H
5979 #define INCLUDE_FILE_NAME_H
5984 @section @value{GDBN} Python Coding Standards
5986 @value{GDBN} follows the published @code{Python} coding standards in
5987 @uref{http://www.python.org/dev/peps/pep-0008/, @code{PEP008}}.
5989 In addition, the guidelines in the
5990 @uref{http://google-styleguide.googlecode.com/svn/trunk/pyguide.html,
5991 Google Python Style Guide} are also followed where they do not
5992 conflict with @code{PEP008}.
5994 @subsection @value{GDBN}-specific exceptions
5996 There are a few exceptions to the published standards.
5997 They exist mainly for consistency with the @code{C} standards.
5999 @c It is expected that there are a few more exceptions,
6000 @c so we use itemize here.
6005 Use @code{FIXME} instead of @code{TODO}.
6009 @node Misc Guidelines
6011 @chapter Misc Guidelines
6013 This chapter covers topics that are lower-level than the major
6014 algorithms of @value{GDBN}.
6019 Cleanups are a structured way to deal with things that need to be done
6022 When your code does something (e.g., @code{xmalloc} some memory, or
6023 @code{open} a file) that needs to be undone later (e.g., @code{xfree}
6024 the memory or @code{close} the file), it can make a cleanup. The
6025 cleanup will be done at some future point: when the command is finished
6026 and control returns to the top level; when an error occurs and the stack
6027 is unwound; or when your code decides it's time to explicitly perform
6028 cleanups. Alternatively you can elect to discard the cleanups you
6034 @item struct cleanup *@var{old_chain};
6035 Declare a variable which will hold a cleanup chain handle.
6037 @findex make_cleanup
6038 @item @var{old_chain} = make_cleanup (@var{function}, @var{arg});
6039 Make a cleanup which will cause @var{function} to be called with
6040 @var{arg} (a @code{char *}) later. The result, @var{old_chain}, is a
6041 handle that can later be passed to @code{do_cleanups} or
6042 @code{discard_cleanups}. Unless you are going to call
6043 @code{do_cleanups} or @code{discard_cleanups}, you can ignore the result
6044 from @code{make_cleanup}.
6047 @item do_cleanups (@var{old_chain});
6048 Do all cleanups added to the chain since the corresponding
6049 @code{make_cleanup} call was made.
6051 @findex discard_cleanups
6052 @item discard_cleanups (@var{old_chain});
6053 Same as @code{do_cleanups} except that it just removes the cleanups from
6054 the chain and does not call the specified functions.
6057 Cleanups are implemented as a chain. The handle returned by
6058 @code{make_cleanups} includes the cleanup passed to the call and any
6059 later cleanups appended to the chain (but not yet discarded or
6063 make_cleanup (a, 0);
6065 struct cleanup *old = make_cleanup (b, 0);
6073 will call @code{c()} and @code{b()} but will not call @code{a()}. The
6074 cleanup that calls @code{a()} will remain in the cleanup chain, and will
6075 be done later unless otherwise discarded.@refill
6077 Your function should explicitly do or discard the cleanups it creates.
6078 Failing to do this leads to non-deterministic behavior since the caller
6079 will arbitrarily do or discard your functions cleanups. This need leads
6080 to two common cleanup styles.
6082 The first style is try/finally. Before it exits, your code-block calls
6083 @code{do_cleanups} with the old cleanup chain and thus ensures that your
6084 code-block's cleanups are always performed. For instance, the following
6085 code-segment avoids a memory leak problem (even when @code{error} is
6086 called and a forced stack unwind occurs) by ensuring that the
6087 @code{xfree} will always be called:
6090 struct cleanup *old = make_cleanup (null_cleanup, 0);
6091 data = xmalloc (sizeof blah);
6092 make_cleanup (xfree, data);
6097 The second style is try/except. Before it exits, your code-block calls
6098 @code{discard_cleanups} with the old cleanup chain and thus ensures that
6099 any created cleanups are not performed. For instance, the following
6100 code segment, ensures that the file will be closed but only if there is
6104 FILE *file = fopen ("afile", "r");
6105 struct cleanup *old = make_cleanup (close_file, file);
6107 discard_cleanups (old);
6111 Some functions, e.g., @code{fputs_filtered()} or @code{error()}, specify
6112 that they ``should not be called when cleanups are not in place''. This
6113 means that any actions you need to reverse in the case of an error or
6114 interruption must be on the cleanup chain before you call these
6115 functions, since they might never return to your code (they
6116 @samp{longjmp} instead).
6118 @section Per-architecture module data
6119 @cindex per-architecture module data
6120 @cindex multi-arch data
6121 @cindex data-pointer, per-architecture/per-module
6123 The multi-arch framework includes a mechanism for adding module
6124 specific per-architecture data-pointers to the @code{struct gdbarch}
6125 architecture object.
6127 A module registers one or more per-architecture data-pointers using:
6129 @deftypefn {Architecture Function} {struct gdbarch_data *} gdbarch_data_register_pre_init (gdbarch_data_pre_init_ftype *@var{pre_init})
6130 @var{pre_init} is used to, on-demand, allocate an initial value for a
6131 per-architecture data-pointer using the architecture's obstack (passed
6132 in as a parameter). Since @var{pre_init} can be called during
6133 architecture creation, it is not parameterized with the architecture.
6134 and must not call modules that use per-architecture data.
6137 @deftypefn {Architecture Function} {struct gdbarch_data *} gdbarch_data_register_post_init (gdbarch_data_post_init_ftype *@var{post_init})
6138 @var{post_init} is used to obtain an initial value for a
6139 per-architecture data-pointer @emph{after}. Since @var{post_init} is
6140 always called after architecture creation, it both receives the fully
6141 initialized architecture and is free to call modules that use
6142 per-architecture data (care needs to be taken to ensure that those
6143 other modules do not try to call back to this module as that will
6144 create in cycles in the initialization call graph).
6147 These functions return a @code{struct gdbarch_data} that is used to
6148 identify the per-architecture data-pointer added for that module.
6150 The per-architecture data-pointer is accessed using the function:
6152 @deftypefn {Architecture Function} {void *} gdbarch_data (struct gdbarch *@var{gdbarch}, struct gdbarch_data *@var{data_handle})
6153 Given the architecture @var{arch} and module data handle
6154 @var{data_handle} (returned by @code{gdbarch_data_register_pre_init}
6155 or @code{gdbarch_data_register_post_init}), this function returns the
6156 current value of the per-architecture data-pointer. If the data
6157 pointer is @code{NULL}, it is first initialized by calling the
6158 corresponding @var{pre_init} or @var{post_init} method.
6161 The examples below assume the following definitions:
6164 struct nozel @{ int total; @};
6165 static struct gdbarch_data *nozel_handle;
6168 A module can extend the architecture vector, adding additional
6169 per-architecture data, using the @var{pre_init} method. The module's
6170 per-architecture data is then initialized during architecture
6173 In the below, the module's per-architecture @emph{nozel} is added. An
6174 architecture can specify its nozel by calling @code{set_gdbarch_nozel}
6175 from @code{gdbarch_init}.
6179 nozel_pre_init (struct obstack *obstack)
6181 struct nozel *data = OBSTACK_ZALLOC (obstack, struct nozel);
6188 set_gdbarch_nozel (struct gdbarch *gdbarch, int total)
6190 struct nozel *data = gdbarch_data (gdbarch, nozel_handle);
6191 data->total = nozel;
6195 A module can on-demand create architecture dependent data structures
6196 using @code{post_init}.
6198 In the below, the nozel's total is computed on-demand by
6199 @code{nozel_post_init} using information obtained from the
6204 nozel_post_init (struct gdbarch *gdbarch)
6206 struct nozel *data = GDBARCH_OBSTACK_ZALLOC (gdbarch, struct nozel);
6207 nozel->total = gdbarch@dots{} (gdbarch);
6214 nozel_total (struct gdbarch *gdbarch)
6216 struct nozel *data = gdbarch_data (gdbarch, nozel_handle);
6221 @section Wrapping Output Lines
6222 @cindex line wrap in output
6225 Output that goes through @code{printf_filtered} or @code{fputs_filtered}
6226 or @code{fputs_demangled} needs only to have calls to @code{wrap_here}
6227 added in places that would be good breaking points. The utility
6228 routines will take care of actually wrapping if the line width is
6231 The argument to @code{wrap_here} is an indentation string which is
6232 printed @emph{only} if the line breaks there. This argument is saved
6233 away and used later. It must remain valid until the next call to
6234 @code{wrap_here} or until a newline has been printed through the
6235 @code{*_filtered} functions. Don't pass in a local variable and then
6238 It is usually best to call @code{wrap_here} after printing a comma or
6239 space. If you call it before printing a space, make sure that your
6240 indentation properly accounts for the leading space that will print if
6241 the line wraps there.
6243 Any function or set of functions that produce filtered output must
6244 finish by printing a newline, to flush the wrap buffer, before switching
6245 to unfiltered (@code{printf}) output. Symbol reading routines that
6246 print warnings are a good example.
6248 @section Memory Management
6250 @value{GDBN} does not use the functions @code{malloc}, @code{realloc},
6251 @code{calloc}, @code{free} and @code{asprintf}.
6253 @value{GDBN} uses the functions @code{xmalloc}, @code{xrealloc} and
6254 @code{xcalloc} when allocating memory. Unlike @code{malloc} et.al.@:
6255 these functions do not return when the memory pool is empty. Instead,
6256 they unwind the stack using cleanups. These functions return
6257 @code{NULL} when requested to allocate a chunk of memory of size zero.
6259 @emph{Pragmatics: By using these functions, the need to check every
6260 memory allocation is removed. These functions provide portable
6263 @value{GDBN} does not use the function @code{free}.
6265 @value{GDBN} uses the function @code{xfree} to return memory to the
6266 memory pool. Consistent with ISO-C, this function ignores a request to
6267 free a @code{NULL} pointer.
6269 @emph{Pragmatics: On some systems @code{free} fails when passed a
6270 @code{NULL} pointer.}
6272 @value{GDBN} can use the non-portable function @code{alloca} for the
6273 allocation of small temporary values (such as strings).
6275 @emph{Pragmatics: This function is very non-portable. Some systems
6276 restrict the memory being allocated to no more than a few kilobytes.}
6278 @value{GDBN} uses the string function @code{xstrdup} and the print
6279 function @code{xstrprintf}.
6281 @emph{Pragmatics: @code{asprintf} and @code{strdup} can fail. Print
6282 functions such as @code{sprintf} are very prone to buffer overflow
6286 @section Compiler Warnings
6287 @cindex compiler warnings
6289 With few exceptions, developers should avoid the configuration option
6290 @samp{--disable-werror} when building @value{GDBN}. The exceptions
6291 are listed in the file @file{gdb/MAINTAINERS}. The default, when
6292 building with @sc{gcc}, is @samp{--enable-werror}.
6294 This option causes @value{GDBN} (when built using GCC) to be compiled
6295 with a carefully selected list of compiler warning flags. Any warnings
6296 from those flags are treated as errors.
6298 The current list of warning flags includes:
6302 Recommended @sc{gcc} warnings.
6304 @item -Wdeclaration-after-statement
6306 @sc{gcc} 3.x (and later) and @sc{c99} allow declarations mixed with
6307 code, but @sc{gcc} 2.x and @sc{c89} do not.
6309 @item -Wpointer-arith
6311 @item -Wformat-nonliteral
6312 Non-literal format strings, with a few exceptions, are bugs - they
6313 might contain unintended user-supplied format specifiers.
6314 Since @value{GDBN} uses the @code{format printf} attribute on all
6315 @code{printf} like functions this checks not just @code{printf} calls
6316 but also calls to functions such as @code{fprintf_unfiltered}.
6318 @item -Wno-pointer-sign
6319 In version 4.0, GCC began warning about pointer argument passing or
6320 assignment even when the source and destination differed only in
6321 signedness. However, most @value{GDBN} code doesn't distinguish
6322 carefully between @code{char} and @code{unsigned char}. In early 2006
6323 the @value{GDBN} developers decided correcting these warnings wasn't
6324 worth the time it would take.
6326 @item -Wno-unused-parameter
6327 Due to the way that @value{GDBN} is implemented many functions have
6328 unused parameters. Consequently this warning is avoided. The macro
6329 @code{ATTRIBUTE_UNUSED} is not used as it leads to false negatives ---
6330 it is not an error to have @code{ATTRIBUTE_UNUSED} on a parameter that
6335 @itemx -Wno-char-subscripts
6336 These are warnings which might be useful for @value{GDBN}, but are
6337 currently too noisy to enable with @samp{-Werror}.
6341 @section Internal Error Recovery
6343 During its execution, @value{GDBN} can encounter two types of errors.
6344 User errors and internal errors. User errors include not only a user
6345 entering an incorrect command but also problems arising from corrupt
6346 object files and system errors when interacting with the target.
6347 Internal errors include situations where @value{GDBN} has detected, at
6348 run time, a corrupt or erroneous situation.
6350 When reporting an internal error, @value{GDBN} uses
6351 @code{internal_error} and @code{gdb_assert}.
6353 @value{GDBN} must not call @code{abort} or @code{assert}.
6355 @emph{Pragmatics: There is no @code{internal_warning} function. Either
6356 the code detected a user error, recovered from it and issued a
6357 @code{warning} or the code failed to correctly recover from the user
6358 error and issued an @code{internal_error}.}
6360 @section Command Names
6362 GDB U/I commands are written @samp{foo-bar}, not @samp{foo_bar}.
6364 @section Clean Design and Portable Implementation
6367 In addition to getting the syntax right, there's the little question of
6368 semantics. Some things are done in certain ways in @value{GDBN} because long
6369 experience has shown that the more obvious ways caused various kinds of
6372 @cindex assumptions about targets
6373 You can't assume the byte order of anything that comes from a target
6374 (including @var{value}s, object files, and instructions). Such things
6375 must be byte-swapped using @code{SWAP_TARGET_AND_HOST} in
6376 @value{GDBN}, or one of the swap routines defined in @file{bfd.h},
6377 such as @code{bfd_get_32}.
6379 You can't assume that you know what interface is being used to talk to
6380 the target system. All references to the target must go through the
6381 current @code{target_ops} vector.
6383 You can't assume that the host and target machines are the same machine
6384 (except in the ``native'' support modules). In particular, you can't
6385 assume that the target machine's header files will be available on the
6386 host machine. Target code must bring along its own header files --
6387 written from scratch or explicitly donated by their owner, to avoid
6391 Insertion of new @code{#ifdef}'s will be frowned upon. It's much better
6392 to write the code portably than to conditionalize it for various
6395 @cindex system dependencies
6396 New @code{#ifdef}'s which test for specific compilers or manufacturers
6397 or operating systems are unacceptable. All @code{#ifdef}'s should test
6398 for features. The information about which configurations contain which
6399 features should be segregated into the configuration files. Experience
6400 has proven far too often that a feature unique to one particular system
6401 often creeps into other systems; and that a conditional based on some
6402 predefined macro for your current system will become worthless over
6403 time, as new versions of your system come out that behave differently
6404 with regard to this feature.
6406 Adding code that handles specific architectures, operating systems,
6407 target interfaces, or hosts, is not acceptable in generic code.
6409 @cindex portable file name handling
6410 @cindex file names, portability
6411 One particularly notorious area where system dependencies tend to
6412 creep in is handling of file names. The mainline @value{GDBN} code
6413 assumes Posix semantics of file names: absolute file names begin with
6414 a forward slash @file{/}, slashes are used to separate leading
6415 directories, case-sensitive file names. These assumptions are not
6416 necessarily true on non-Posix systems such as MS-Windows. To avoid
6417 system-dependent code where you need to take apart or construct a file
6418 name, use the following portable macros:
6421 @findex HAVE_DOS_BASED_FILE_SYSTEM
6422 @item HAVE_DOS_BASED_FILE_SYSTEM
6423 This preprocessing symbol is defined to a non-zero value on hosts
6424 whose filesystems belong to the MS-DOS/MS-Windows family. Use this
6425 symbol to write conditional code which should only be compiled for
6428 @findex IS_DIR_SEPARATOR
6429 @item IS_DIR_SEPARATOR (@var{c})
6430 Evaluates to a non-zero value if @var{c} is a directory separator
6431 character. On Unix and GNU/Linux systems, only a slash @file{/} is
6432 such a character, but on Windows, both @file{/} and @file{\} will
6435 @findex IS_ABSOLUTE_PATH
6436 @item IS_ABSOLUTE_PATH (@var{file})
6437 Evaluates to a non-zero value if @var{file} is an absolute file name.
6438 For Unix and GNU/Linux hosts, a name which begins with a slash
6439 @file{/} is absolute. On DOS and Windows, @file{d:/foo} and
6440 @file{x:\bar} are also absolute file names.
6442 @findex FILENAME_CMP
6443 @item FILENAME_CMP (@var{f1}, @var{f2})
6444 Calls a function which compares file names @var{f1} and @var{f2} as
6445 appropriate for the underlying host filesystem. For Posix systems,
6446 this simply calls @code{strcmp}; on case-insensitive filesystems it
6447 will call @code{strcasecmp} instead.
6449 @findex DIRNAME_SEPARATOR
6450 @item DIRNAME_SEPARATOR
6451 Evaluates to a character which separates directories in
6452 @code{PATH}-style lists, typically held in environment variables.
6453 This character is @samp{:} on Unix, @samp{;} on DOS and Windows.
6455 @findex SLASH_STRING
6457 This evaluates to a constant string you should use to produce an
6458 absolute filename from leading directories and the file's basename.
6459 @code{SLASH_STRING} is @code{"/"} on most systems, but might be
6460 @code{"\\"} for some Windows-based ports.
6463 In addition to using these macros, be sure to use portable library
6464 functions whenever possible. For example, to extract a directory or a
6465 basename part from a file name, use the @code{dirname} and
6466 @code{basename} library functions (available in @code{libiberty} for
6467 platforms which don't provide them), instead of searching for a slash
6468 with @code{strrchr}.
6470 Another way to generalize @value{GDBN} along a particular interface is with an
6471 attribute struct. For example, @value{GDBN} has been generalized to handle
6472 multiple kinds of remote interfaces---not by @code{#ifdef}s everywhere, but
6473 by defining the @code{target_ops} structure and having a current target (as
6474 well as a stack of targets below it, for memory references). Whenever
6475 something needs to be done that depends on which remote interface we are
6476 using, a flag in the current target_ops structure is tested (e.g.,
6477 @code{target_has_stack}), or a function is called through a pointer in the
6478 current target_ops structure. In this way, when a new remote interface
6479 is added, only one module needs to be touched---the one that actually
6480 implements the new remote interface. Other examples of
6481 attribute-structs are BFD access to multiple kinds of object file
6482 formats, or @value{GDBN}'s access to multiple source languages.
6484 Please avoid duplicating code. For example, in @value{GDBN} 3.x all
6485 the code interfacing between @code{ptrace} and the rest of
6486 @value{GDBN} was duplicated in @file{*-dep.c}, and so changing
6487 something was very painful. In @value{GDBN} 4.x, these have all been
6488 consolidated into @file{infptrace.c}. @file{infptrace.c} can deal
6489 with variations between systems the same way any system-independent
6490 file would (hooks, @code{#if defined}, etc.), and machines which are
6491 radically different don't need to use @file{infptrace.c} at all.
6493 All debugging code must be controllable using the @samp{set debug
6494 @var{module}} command. Do not use @code{printf} to print trace
6495 messages. Use @code{fprintf_unfiltered(gdb_stdlog, ...}. Do not use
6496 @code{#ifdef DEBUG}.
6500 @chapter Porting @value{GDBN}
6501 @cindex porting to new machines
6503 Most of the work in making @value{GDBN} compile on a new machine is in
6504 specifying the configuration of the machine. Porting a new
6505 architecture to @value{GDBN} can be broken into a number of steps.
6510 Ensure a @sc{bfd} exists for executables of the target architecture in
6511 the @file{bfd} directory. If one does not exist, create one by
6512 modifying an existing similar one.
6515 Implement a disassembler for the target architecture in the @file{opcodes}
6519 Define the target architecture in the @file{gdb} directory
6520 (@pxref{Adding a New Target, , Adding a New Target}). Add the pattern
6521 for the new target to @file{configure.tgt} with the names of the files
6522 that contain the code. By convention the target architecture
6523 definition for an architecture @var{arch} is placed in
6524 @file{@var{arch}-tdep.c}.
6526 Within @file{@var{arch}-tdep.c} define the function
6527 @code{_initialize_@var{arch}_tdep} which calls
6528 @code{gdbarch_register} to create the new @code{@w{struct
6529 gdbarch}} for the architecture.
6532 If a new remote target is needed, consider adding a new remote target
6533 by defining a function
6534 @code{_initialize_remote_@var{arch}}. However if at all possible
6535 use the @value{GDBN} @emph{Remote Serial Protocol} for this and implement
6536 the server side protocol independently with the target.
6539 If desired implement a simulator in the @file{sim} directory. This
6540 should create the library @file{libsim.a} implementing the interface
6541 in @file{remote-sim.h} (found in the @file{include} directory).
6544 Build and test. If desired, lobby the @sc{gdb} steering group to
6545 have the new port included in the main distribution!
6548 Add a description of the new architecture to the main @value{GDBN} user
6549 guide (@pxref{Configuration Specific Information, , Configuration
6550 Specific Information, gdb, Debugging with @value{GDBN}}).
6554 @node Versions and Branches
6555 @chapter Versions and Branches
6559 @value{GDBN}'s version is determined by the file
6560 @file{gdb/version.in} and takes one of the following forms:
6563 @item @var{major}.@var{minor}
6564 @itemx @var{major}.@var{minor}.@var{patchlevel}
6565 an official release (e.g., 6.2 or 6.2.1)
6566 @item @var{major}.@var{minor}.@var{patchlevel}.@var{YYYY}@var{MM}@var{DD}
6567 a snapshot taken at @var{YYYY}-@var{MM}-@var{DD}-gmt (e.g.,
6568 6.1.50.20020302, 6.1.90.20020304, or 6.1.0.20020308)
6569 @item @var{major}.@var{minor}.@var{patchlevel}.@var{YYYY}@var{MM}@var{DD}-cvs
6570 a @sc{cvs} check out drawn on @var{YYYY}-@var{MM}-@var{DD} (e.g.,
6571 6.1.50.20020302-cvs, 6.1.90.20020304-cvs, or 6.1.0.20020308-cvs)
6572 @item @var{major}.@var{minor}.@var{patchlevel}.@var{YYYY}@var{MM}@var{DD} (@var{vendor})
6573 a vendor specific release of @value{GDBN}, that while based on@*
6574 @var{major}.@var{minor}.@var{patchlevel}.@var{YYYY}@var{MM}@var{DD},
6575 may include additional changes
6578 @value{GDBN}'s mainline uses the @var{major} and @var{minor} version
6579 numbers from the most recent release branch, with a @var{patchlevel}
6580 of 50. At the time each new release branch is created, the mainline's
6581 @var{major} and @var{minor} version numbers are updated.
6583 @value{GDBN}'s release branch is similar. When the branch is cut, the
6584 @var{patchlevel} is changed from 50 to 90. As draft releases are
6585 drawn from the branch, the @var{patchlevel} is incremented. Once the
6586 first release (@var{major}.@var{minor}) has been made, the
6587 @var{patchlevel} is set to 0 and updates have an incremented
6590 For snapshots, and @sc{cvs} check outs, it is also possible to
6591 identify the @sc{cvs} origin:
6594 @item @var{major}.@var{minor}.50.@var{YYYY}@var{MM}@var{DD}
6595 drawn from the @sc{head} of mainline @sc{cvs} (e.g., 6.1.50.20020302)
6596 @item @var{major}.@var{minor}.90.@var{YYYY}@var{MM}@var{DD}
6597 @itemx @var{major}.@var{minor}.91.@var{YYYY}@var{MM}@var{DD} @dots{}
6598 drawn from a release branch prior to the release (e.g.,
6600 @item @var{major}.@var{minor}.0.@var{YYYY}@var{MM}@var{DD}
6601 @itemx @var{major}.@var{minor}.1.@var{YYYY}@var{MM}@var{DD} @dots{}
6602 drawn from a release branch after the release (e.g., 6.2.0.20020308)
6605 If the previous @value{GDBN} version is 6.1 and the current version is
6606 6.2, then, substituting 6 for @var{major} and 1 or 2 for @var{minor},
6607 here's an illustration of a typical sequence:
6614 +--------------------------.
6617 6.2.50.20020303-cvs 6.1.90 (draft #1)
6619 6.2.50.20020304-cvs 6.1.90.20020304-cvs
6621 6.2.50.20020305-cvs 6.1.91 (draft #2)
6623 6.2.50.20020306-cvs 6.1.91.20020306-cvs
6625 6.2.50.20020307-cvs 6.2 (release)
6627 6.2.50.20020308-cvs 6.2.0.20020308-cvs
6629 6.2.50.20020309-cvs 6.2.1 (update)
6631 6.2.50.20020310-cvs <branch closed>
6635 +--------------------------.
6638 6.3.50.20020312-cvs 6.2.90 (draft #1)
6642 @section Release Branches
6643 @cindex Release Branches
6645 @value{GDBN} draws a release series (6.2, 6.2.1, @dots{}) from a
6646 single release branch, and identifies that branch using the @sc{cvs}
6650 gdb_@var{major}_@var{minor}-@var{YYYY}@var{MM}@var{DD}-branchpoint
6651 gdb_@var{major}_@var{minor}-branch
6652 gdb_@var{major}_@var{minor}-@var{YYYY}@var{MM}@var{DD}-release
6655 @emph{Pragmatics: To help identify the date at which a branch or
6656 release is made, both the branchpoint and release tags include the
6657 date that they are cut (@var{YYYY}@var{MM}@var{DD}) in the tag. The
6658 branch tag, denoting the head of the branch, does not need this.}
6660 @section Vendor Branches
6661 @cindex vendor branches
6663 To avoid version conflicts, vendors are expected to modify the file
6664 @file{gdb/version.in} to include a vendor unique alphabetic identifier
6665 (an official @value{GDBN} release never uses alphabetic characters in
6666 its version identifier). E.g., @samp{6.2widgit2}, or @samp{6.2 (Widgit
6669 @section Experimental Branches
6670 @cindex experimental branches
6672 @subsection Guidelines
6674 @value{GDBN} permits the creation of branches, cut from the @sc{cvs}
6675 repository, for experimental development. Branches make it possible
6676 for developers to share preliminary work, and maintainers to examine
6677 significant new developments.
6679 The following are a set of guidelines for creating such branches:
6683 @item a branch has an owner
6684 The owner can set further policy for a branch, but may not change the
6685 ground rules. In particular, they can set a policy for commits (be it
6686 adding more reviewers or deciding who can commit).
6688 @item all commits are posted
6689 All changes committed to a branch shall also be posted to
6690 @email{gdb-patches@@sourceware.org, the @value{GDBN} patches
6691 mailing list}. While commentary on such changes are encouraged, people
6692 should remember that the changes only apply to a branch.
6694 @item all commits are covered by an assignment
6695 This ensures that all changes belong to the Free Software Foundation,
6696 and avoids the possibility that the branch may become contaminated.
6698 @item a branch is focused
6699 A focused branch has a single objective or goal, and does not contain
6700 unnecessary or irrelevant changes. Cleanups, where identified, being
6701 be pushed into the mainline as soon as possible.
6703 @item a branch tracks mainline
6704 This keeps the level of divergence under control. It also keeps the
6705 pressure on developers to push cleanups and other stuff into the
6708 @item a branch shall contain the entire @value{GDBN} module
6709 The @value{GDBN} module @code{gdb} should be specified when creating a
6710 branch (branches of individual files should be avoided). @xref{Tags}.
6712 @item a branch shall be branded using @file{version.in}
6713 The file @file{gdb/version.in} shall be modified so that it identifies
6714 the branch @var{owner} and branch @var{name}, e.g.,
6715 @samp{6.2.50.20030303_owner_name} or @samp{6.2 (Owner Name)}.
6722 To simplify the identification of @value{GDBN} branches, the following
6723 branch tagging convention is strongly recommended:
6727 @item @var{owner}_@var{name}-@var{YYYYMMDD}-branchpoint
6728 @itemx @var{owner}_@var{name}-@var{YYYYMMDD}-branch
6729 The branch point and corresponding branch tag. @var{YYYYMMDD} is the
6730 date that the branch was created. A branch is created using the
6731 sequence: @anchor{experimental branch tags}
6733 cvs rtag @var{owner}_@var{name}-@var{YYYYMMDD}-branchpoint gdb
6734 cvs rtag -b -r @var{owner}_@var{name}-@var{YYYYMMDD}-branchpoint \
6735 @var{owner}_@var{name}-@var{YYYYMMDD}-branch gdb
6738 @item @var{owner}_@var{name}-@var{yyyymmdd}-mergepoint
6739 The tagged point, on the mainline, that was used when merging the branch
6740 on @var{yyyymmdd}. To merge in all changes since the branch was cut,
6741 use a command sequence like:
6743 cvs rtag @var{owner}_@var{name}-@var{yyyymmdd}-mergepoint gdb
6745 -j@var{owner}_@var{name}-@var{YYYYMMDD}-branchpoint
6746 -j@var{owner}_@var{name}-@var{yyyymmdd}-mergepoint
6749 Similar sequences can be used to just merge in changes since the last
6755 For further information on @sc{cvs}, see
6756 @uref{http://www.gnu.org/software/cvs/, Concurrent Versions System}.
6758 @node Start of New Year Procedure
6759 @chapter Start of New Year Procedure
6760 @cindex new year procedure
6762 At the start of each new year, the following actions should be performed:
6766 Rotate the ChangeLog file
6768 The current @file{ChangeLog} file should be renamed into
6769 @file{ChangeLog-YYYY} where YYYY is the year that has just passed.
6770 A new @file{ChangeLog} file should be created, and its contents should
6771 contain a reference to the previous ChangeLog. The following should
6772 also be preserved at the end of the new ChangeLog, in order to provide
6773 the appropriate settings when editing this file with Emacs:
6779 version-control: never
6785 Add an entry for the newly created ChangeLog file (@file{ChangeLog-YYYY})
6786 in @file{gdb/config/djgpp/fnchange.lst}.
6789 Update the copyright year in the startup message
6791 Update the copyright year in:
6794 file @file{top.c}, function @code{print_gdb_version}
6796 file @file{gdbserver/server.c}, function @code{gdbserver_version}
6798 file @file{gdbserver/gdbreplay.c}, function @code{gdbreplay_version}
6802 Run the @file{copyright.sh} script to add the new year in the copyright
6803 notices of most source files. This script requires Emacs 22 or later to
6807 The new year also needs to be added manually in all other files that
6808 are not already taken care of by the @file{copyright.sh} script:
6836 @chapter Releasing @value{GDBN}
6837 @cindex making a new release of gdb
6839 @section Branch Commit Policy
6841 The branch commit policy is pretty slack. @value{GDBN} releases 5.0,
6842 5.1 and 5.2 all used the below:
6846 The @file{gdb/MAINTAINERS} file still holds.
6848 Don't fix something on the branch unless/until it is also fixed in the
6849 trunk. If this isn't possible, mentioning it in the @file{gdb/PROBLEMS}
6850 file is better than committing a hack.
6852 When considering a patch for the branch, suggested criteria include:
6853 Does it fix a build? Does it fix the sequence @kbd{break main; run}
6854 when debugging a static binary?
6856 The further a change is from the core of @value{GDBN}, the less likely
6857 the change will worry anyone (e.g., target specific code).
6859 Only post a proposal to change the core of @value{GDBN} after you've
6860 sent individual bribes to all the people listed in the
6861 @file{MAINTAINERS} file @t{;-)}
6864 @emph{Pragmatics: Provided updates are restricted to non-core
6865 functionality there is little chance that a broken change will be fatal.
6866 This means that changes such as adding a new architectures or (within
6867 reason) support for a new host are considered acceptable.}
6870 @section Obsoleting code
6872 Before anything else, poke the other developers (and around the source
6873 code) to see if there is anything that can be removed from @value{GDBN}
6874 (an old target, an unused file).
6876 Obsolete code is identified by adding an @code{OBSOLETE} prefix to every
6877 line. Doing this means that it is easy to identify something that has
6878 been obsoleted when greping through the sources.
6880 The process is done in stages --- this is mainly to ensure that the
6881 wider @value{GDBN} community has a reasonable opportunity to respond.
6882 Remember, everything on the Internet takes a week.
6886 Post the proposal on @email{gdb@@sourceware.org, the GDB mailing
6887 list} Creating a bug report to track the task's state, is also highly
6892 Post the proposal on @email{gdb-announce@@sourceware.org, the GDB
6893 Announcement mailing list}.
6897 Go through and edit all relevant files and lines so that they are
6898 prefixed with the word @code{OBSOLETE}.
6900 Wait until the next GDB version, containing this obsolete code, has been
6903 Remove the obsolete code.
6907 @emph{Maintainer note: While removing old code is regrettable it is
6908 hopefully better for @value{GDBN}'s long term development. Firstly it
6909 helps the developers by removing code that is either no longer relevant
6910 or simply wrong. Secondly since it removes any history associated with
6911 the file (effectively clearing the slate) the developer has a much freer
6912 hand when it comes to fixing broken files.}
6916 @section Before the Branch
6918 The most important objective at this stage is to find and fix simple
6919 changes that become a pain to track once the branch is created. For
6920 instance, configuration problems that stop @value{GDBN} from even
6921 building. If you can't get the problem fixed, document it in the
6922 @file{gdb/PROBLEMS} file.
6924 @subheading Prompt for @file{gdb/NEWS}
6926 People always forget. Send a post reminding them but also if you know
6927 something interesting happened add it yourself. The @code{schedule}
6928 script will mention this in its e-mail.
6930 @subheading Review @file{gdb/README}
6932 Grab one of the nightly snapshots and then walk through the
6933 @file{gdb/README} looking for anything that can be improved. The
6934 @code{schedule} script will mention this in its e-mail.
6936 @subheading Refresh any imported files.
6938 A number of files are taken from external repositories. They include:
6942 @file{texinfo/texinfo.tex}
6944 @file{config.guess} et.@: al.@: (see the top-level @file{MAINTAINERS}
6947 @file{etc/standards.texi}, @file{etc/make-stds.texi}
6950 @subheading Check the ARI
6952 @uref{http://sourceware.org/gdb/ari,,A.R.I.} is an @code{awk} script
6953 (Awk Regression Index ;-) that checks for a number of errors and coding
6954 conventions. The checks include things like using @code{malloc} instead
6955 of @code{xmalloc} and file naming problems. There shouldn't be any
6958 @subsection Review the bug data base
6960 Close anything obviously fixed.
6962 @subsection Check all cross targets build
6964 The targets are listed in @file{gdb/MAINTAINERS}.
6967 @section Cut the Branch
6969 @subheading Create the branch
6974 $ V=`echo $v | sed 's/\./_/g'`
6975 $ D=`date -u +%Y-%m-%d`
6978 $ echo cvs -f -d :ext:sourceware.org:/cvs/src rtag \
6979 -D $D-gmt gdb_$V-$D-branchpoint insight
6980 cvs -f -d :ext:sourceware.org:/cvs/src rtag
6981 -D 2002-03-03-gmt gdb_5_2-2002-03-03-branchpoint insight
6984 $ echo cvs -f -d :ext:sourceware.org:/cvs/src rtag \
6985 -b -r gdb_$V-$D-branchpoint gdb_$V-branch insight
6986 cvs -f -d :ext:sourceware.org:/cvs/src rtag \
6987 -b -r gdb_5_2-2002-03-03-branchpoint gdb_5_2-branch insight
6995 By using @kbd{-D YYYY-MM-DD-gmt}, the branch is forced to an exact
6998 The trunk is first tagged so that the branch point can easily be found.
7000 Insight, which includes @value{GDBN}, is tagged at the same time.
7002 @file{version.in} gets bumped to avoid version number conflicts.
7004 The reading of @file{.cvsrc} is disabled using @file{-f}.
7007 @subheading Update @file{version.in}
7012 $ V=`echo $v | sed 's/\./_/g'`
7016 $ echo cvs -f -d :ext:sourceware.org:/cvs/src co \
7017 -r gdb_$V-branch src/gdb/version.in
7018 cvs -f -d :ext:sourceware.org:/cvs/src co
7019 -r gdb_5_2-branch src/gdb/version.in
7021 U src/gdb/version.in
7023 $ echo $u.90-0000-00-00-cvs > version.in
7025 5.1.90-0000-00-00-cvs
7026 $ cvs -f commit version.in
7031 @file{0000-00-00} is used as a date to pump prime the version.in update
7034 @file{.90} and the previous branch version are used as fairly arbitrary
7035 initial branch version number.
7039 @subheading Update the web and news pages
7043 @subheading Tweak cron to track the new branch
7045 The file @file{gdbadmin/cron/crontab} contains gdbadmin's cron table.
7046 This file needs to be updated so that:
7050 A daily timestamp is added to the file @file{version.in}.
7052 The new branch is included in the snapshot process.
7056 See the file @file{gdbadmin/cron/README} for how to install the updated
7059 The file @file{gdbadmin/ss/README} should also be reviewed to reflect
7060 any changes. That file is copied to both the branch/ and current/
7061 snapshot directories.
7064 @subheading Update the NEWS and README files
7066 The @file{NEWS} file needs to be updated so that on the branch it refers
7067 to @emph{changes in the current release} while on the trunk it also
7068 refers to @emph{changes since the current release}.
7070 The @file{README} file needs to be updated so that it refers to the
7073 @subheading Post the branch info
7075 Send an announcement to the mailing lists:
7079 @email{gdb-announce@@sourceware.org, GDB Announcement mailing list}
7081 @email{gdb@@sourceware.org, GDB Discussion mailing list} and
7082 @email{gdb-testers@@sourceware.org, GDB Testers mailing list}
7085 @emph{Pragmatics: The branch creation is sent to the announce list to
7086 ensure that people people not subscribed to the higher volume discussion
7089 The announcement should include:
7095 How to check out the branch using CVS.
7097 The date/number of weeks until the release.
7099 The branch commit policy still holds.
7102 @section Stabilize the branch
7104 Something goes here.
7106 @section Create a Release
7108 The process of creating and then making available a release is broken
7109 down into a number of stages. The first part addresses the technical
7110 process of creating a releasable tar ball. The later stages address the
7111 process of releasing that tar ball.
7113 When making a release candidate just the first section is needed.
7115 @subsection Create a release candidate
7117 The objective at this stage is to create a set of tar balls that can be
7118 made available as a formal release (or as a less formal release
7121 @subsubheading Freeze the branch
7123 Send out an e-mail notifying everyone that the branch is frozen to
7124 @email{gdb-patches@@sourceware.org}.
7126 @subsubheading Establish a few defaults.
7131 $ t=/sourceware/snapshot-tmp/gdbadmin-tmp
7133 /sourceware/snapshot-tmp/gdbadmin-tmp/gdb_5_2-branch/5.2
7137 /sourceware/snapshot-tmp/gdbadmin-tmp/gdb_5_2-branch/5.2
7139 /home/gdbadmin/bin/autoconf
7148 Check the @code{autoconf} version carefully. You want to be using the
7149 version documented in the toplevel @file{README-maintainer-mode} file.
7150 It is very unlikely that the version of @code{autoconf} installed in
7151 system directories (e.g., @file{/usr/bin/autoconf}) is correct.
7154 @subsubheading Check out the relevant modules:
7157 $ for m in gdb insight
7159 ( mkdir -p $m && cd $m && cvs -q -f -d /cvs/src co -P -r $b $m )
7169 The reading of @file{.cvsrc} is disabled (@file{-f}) so that there isn't
7170 any confusion between what is written here and what your local
7171 @code{cvs} really does.
7174 @subsubheading Update relevant files.
7180 Major releases get their comments added as part of the mainline. Minor
7181 releases should probably mention any significant bugs that were fixed.
7183 Don't forget to include the @file{ChangeLog} entry.
7186 $ emacs gdb/src/gdb/NEWS
7191 $ cp gdb/src/gdb/NEWS insight/src/gdb/NEWS
7192 $ cp gdb/src/gdb/ChangeLog insight/src/gdb/ChangeLog
7197 You'll need to update:
7209 $ emacs gdb/src/gdb/README
7214 $ cp gdb/src/gdb/README insight/src/gdb/README
7215 $ cp gdb/src/gdb/ChangeLog insight/src/gdb/ChangeLog
7218 @emph{Maintainer note: Hopefully the @file{README} file was reviewed
7219 before the initial branch was cut so just a simple substitute is needed
7222 @emph{Maintainer note: Other projects generate @file{README} and
7223 @file{INSTALL} from the core documentation. This might be worth
7226 @item gdb/version.in
7229 $ echo $v > gdb/src/gdb/version.in
7230 $ cat gdb/src/gdb/version.in
7232 $ emacs gdb/src/gdb/version.in
7235 ... Bump to version ...
7237 $ cp gdb/src/gdb/version.in insight/src/gdb/version.in
7238 $ cp gdb/src/gdb/ChangeLog insight/src/gdb/ChangeLog
7243 @subsubheading Do the dirty work
7245 This is identical to the process used to create the daily snapshot.
7248 $ for m in gdb insight
7250 ( cd $m/src && gmake -f src-release $m.tar )
7254 If the top level source directory does not have @file{src-release}
7255 (@value{GDBN} version 5.3.1 or earlier), try these commands instead:
7258 $ for m in gdb insight
7260 ( cd $m/src && gmake -f Makefile.in $m.tar )
7264 @subsubheading Check the source files
7266 You're looking for files that have mysteriously disappeared.
7267 @kbd{distclean} has the habit of deleting files it shouldn't. Watch out
7268 for the @file{version.in} update @kbd{cronjob}.
7271 $ ( cd gdb/src && cvs -f -q -n update )
7275 @dots{} lots of generated files @dots{}
7280 @dots{} lots of generated files @dots{}
7285 @emph{Don't worry about the @file{gdb.info-??} or
7286 @file{gdb/p-exp.tab.c}. They were generated (and yes @file{gdb.info-1}
7287 was also generated only something strange with CVS means that they
7288 didn't get suppressed). Fixing it would be nice though.}
7290 @subsubheading Create compressed versions of the release
7296 gdb/ gdb-5.2.tar insight/ insight-5.2.tar
7297 $ for m in gdb insight
7299 bzip2 -v -9 -c $m-$v.tar > $m-$v.tar.bz2
7300 gzip -v -9 -c $m-$v.tar > $m-$v.tar.gz
7310 A pipe such as @kbd{bunzip2 < xxx.bz2 | gzip -9 > xxx.gz} is not since,
7311 in that mode, @code{gzip} does not know the name of the file and, hence,
7312 can not include it in the compressed file. This is also why the release
7313 process runs @code{tar} and @code{bzip2} as separate passes.
7316 @subsection Sanity check the tar ball
7318 Pick a popular machine (Solaris/PPC?) and try the build on that.
7321 $ bunzip2 < gdb-5.2.tar.bz2 | tar xpf -
7326 $ ./gdb/gdb ./gdb/gdb
7330 Breakpoint 1 at 0x80732bc: file main.c, line 734.
7332 Starting program: /tmp/gdb-5.2/gdb/gdb
7334 Breakpoint 1, main (argc=1, argv=0xbffff8b4) at main.c:734
7335 734 catch_errors (captured_main, &args, "", RETURN_MASK_ALL);
7337 $1 = @{argc = 136426532, argv = 0x821b7f0@}
7341 @subsection Make a release candidate available
7343 If this is a release candidate then the only remaining steps are:
7347 Commit @file{version.in} and @file{ChangeLog}
7349 Tweak @file{version.in} (and @file{ChangeLog} to read
7350 @var{L}.@var{M}.@var{N}-0000-00-00-cvs so that the version update
7351 process can restart.
7353 Make the release candidate available in
7354 @uref{ftp://sourceware.org/pub/gdb/snapshots/branch}
7356 Notify the relevant mailing lists ( @email{gdb@@sourceware.org} and
7357 @email{gdb-testers@@sourceware.org} that the candidate is available.
7360 @subsection Make a formal release available
7362 (And you thought all that was required was to post an e-mail.)
7364 @subsubheading Install on sware
7366 Copy the new files to both the release and the old release directory:
7369 $ cp *.bz2 *.gz ~ftp/pub/gdb/old-releases/
7370 $ cp *.bz2 *.gz ~ftp/pub/gdb/releases
7374 Clean up the releases directory so that only the most recent releases
7375 are available (e.g.@: keep 5.2 and 5.2.1 but remove 5.1):
7378 $ cd ~ftp/pub/gdb/releases
7383 Update the file @file{README} and @file{.message} in the releases
7390 $ ln README .message
7393 @subsubheading Update the web pages.
7397 @item htdocs/download/ANNOUNCEMENT
7398 This file, which is posted as the official announcement, includes:
7401 General announcement.
7403 News. If making an @var{M}.@var{N}.1 release, retain the news from
7404 earlier @var{M}.@var{N} release.
7409 @item htdocs/index.html
7410 @itemx htdocs/news/index.html
7411 @itemx htdocs/download/index.html
7412 These files include:
7415 Announcement of the most recent release.
7417 News entry (remember to update both the top level and the news directory).
7419 These pages also need to be regenerate using @code{index.sh}.
7421 @item download/onlinedocs/
7422 You need to find the magic command that is used to generate the online
7423 docs from the @file{.tar.bz2}. The best way is to look in the output
7424 from one of the nightly @code{cron} jobs and then just edit accordingly.
7428 $ ~/ss/update-web-docs \
7429 ~ftp/pub/gdb/releases/gdb-5.2.tar.bz2 \
7431 /www/sourceware/htdocs/gdb/download/onlinedocs \
7436 Just like the online documentation. Something like:
7439 $ /bin/sh ~/ss/update-web-ari \
7440 ~ftp/pub/gdb/releases/gdb-5.2.tar.bz2 \
7442 /www/sourceware/htdocs/gdb/download/ari \
7448 @subsubheading Shadow the pages onto gnu
7450 Something goes here.
7453 @subsubheading Install the @value{GDBN} tar ball on GNU
7455 At the time of writing, the GNU machine was @kbd{gnudist.gnu.org} in
7456 @file{~ftp/gnu/gdb}.
7458 @subsubheading Make the @file{ANNOUNCEMENT}
7460 Post the @file{ANNOUNCEMENT} file you created above to:
7464 @email{gdb-announce@@sourceware.org, GDB Announcement mailing list}
7466 @email{info-gnu@@gnu.org, General GNU Announcement list} (but delay it a
7467 day or so to let things get out)
7469 @email{bug-gdb@@gnu.org, GDB Bug Report mailing list}
7474 The release is out but you're still not finished.
7476 @subsubheading Commit outstanding changes
7478 In particular you'll need to commit any changes to:
7482 @file{gdb/ChangeLog}
7484 @file{gdb/version.in}
7491 @subsubheading Tag the release
7496 $ d=`date -u +%Y-%m-%d`
7499 $ ( cd insight/src/gdb && cvs -f -q update )
7500 $ ( cd insight/src && cvs -f -q tag gdb_5_2-$d-release )
7503 Insight is used since that contains more of the release than
7506 @subsubheading Mention the release on the trunk
7508 Just put something in the @file{ChangeLog} so that the trunk also
7509 indicates when the release was made.
7511 @subsubheading Restart @file{gdb/version.in}
7513 If @file{gdb/version.in} does not contain an ISO date such as
7514 @kbd{2002-01-24} then the daily @code{cronjob} won't update it. Having
7515 committed all the release changes it can be set to
7516 @file{5.2.0_0000-00-00-cvs} which will restart things (yes the @kbd{_}
7517 is important - it affects the snapshot process).
7519 Don't forget the @file{ChangeLog}.
7521 @subsubheading Merge into trunk
7523 The files committed to the branch may also need changes merged into the
7526 @subsubheading Revise the release schedule
7528 Post a revised release schedule to @email{gdb@@sourceware.org, GDB
7529 Discussion List} with an updated announcement. The schedule can be
7530 generated by running:
7533 $ ~/ss/schedule `date +%s` schedule
7537 The first parameter is approximate date/time in seconds (from the epoch)
7538 of the most recent release.
7540 Also update the schedule @code{cronjob}.
7542 @section Post release
7544 Remove any @code{OBSOLETE} code.
7551 The testsuite is an important component of the @value{GDBN} package.
7552 While it is always worthwhile to encourage user testing, in practice
7553 this is rarely sufficient; users typically use only a small subset of
7554 the available commands, and it has proven all too common for a change
7555 to cause a significant regression that went unnoticed for some time.
7557 The @value{GDBN} testsuite uses the DejaGNU testing framework. The
7558 tests themselves are calls to various @code{Tcl} procs; the framework
7559 runs all the procs and summarizes the passes and fails.
7561 @section Using the Testsuite
7563 @cindex running the test suite
7564 To run the testsuite, simply go to the @value{GDBN} object directory (or to the
7565 testsuite's objdir) and type @code{make check}. This just sets up some
7566 environment variables and invokes DejaGNU's @code{runtest} script. While
7567 the testsuite is running, you'll get mentions of which test file is in use,
7568 and a mention of any unexpected passes or fails. When the testsuite is
7569 finished, you'll get a summary that looks like this:
7574 # of expected passes 6016
7575 # of unexpected failures 58
7576 # of unexpected successes 5
7577 # of expected failures 183
7578 # of unresolved testcases 3
7579 # of untested testcases 5
7582 To run a specific test script, type:
7584 make check RUNTESTFLAGS='@var{tests}'
7586 where @var{tests} is a list of test script file names, separated by
7589 If you use GNU make, you can use its @option{-j} option to run the
7590 testsuite in parallel. This can greatly reduce the amount of time it
7591 takes for the testsuite to run. In this case, if you set
7592 @code{RUNTESTFLAGS} then, by default, the tests will be run serially
7593 even under @option{-j}. You can override this and force a parallel run
7594 by setting the @code{make} variable @code{FORCE_PARALLEL} to any
7595 non-empty value. Note that the parallel @kbd{make check} assumes
7596 that you want to run the entire testsuite, so it is not compatible
7597 with some dejagnu options, like @option{--directory}.
7599 The ideal test run consists of expected passes only; however, reality
7600 conspires to keep us from this ideal. Unexpected failures indicate
7601 real problems, whether in @value{GDBN} or in the testsuite. Expected
7602 failures are still failures, but ones which have been decided are too
7603 hard to deal with at the time; for instance, a test case might work
7604 everywhere except on AIX, and there is no prospect of the AIX case
7605 being fixed in the near future. Expected failures should not be added
7606 lightly, since you may be masking serious bugs in @value{GDBN}.
7607 Unexpected successes are expected fails that are passing for some
7608 reason, while unresolved and untested cases often indicate some minor
7609 catastrophe, such as the compiler being unable to deal with a test
7612 When making any significant change to @value{GDBN}, you should run the
7613 testsuite before and after the change, to confirm that there are no
7614 regressions. Note that truly complete testing would require that you
7615 run the testsuite with all supported configurations and a variety of
7616 compilers; however this is more than really necessary. In many cases
7617 testing with a single configuration is sufficient. Other useful
7618 options are to test one big-endian (Sparc) and one little-endian (x86)
7619 host, a cross config with a builtin simulator (powerpc-eabi,
7620 mips-elf), or a 64-bit host (Alpha).
7622 If you add new functionality to @value{GDBN}, please consider adding
7623 tests for it as well; this way future @value{GDBN} hackers can detect
7624 and fix their changes that break the functionality you added.
7625 Similarly, if you fix a bug that was not previously reported as a test
7626 failure, please add a test case for it. Some cases are extremely
7627 difficult to test, such as code that handles host OS failures or bugs
7628 in particular versions of compilers, and it's OK not to try to write
7629 tests for all of those.
7631 DejaGNU supports separate build, host, and target machines. However,
7632 some @value{GDBN} test scripts do not work if the build machine and
7633 the host machine are not the same. In such an environment, these scripts
7634 will give a result of ``UNRESOLVED'', like this:
7637 UNRESOLVED: gdb.base/example.exp: This test script does not work on a remote host.
7640 @section Testsuite Parameters
7642 Several variables exist to modify the behavior of the testsuite.
7646 @item @code{TRANSCRIPT}
7648 Sometimes it is convenient to get a transcript of the commands which
7649 the testsuite sends to @value{GDBN}. For example, if @value{GDBN}
7650 crashes during testing, a transcript can be used to more easily
7651 reconstruct the failure when running @value{GDBN} under @value{GDBN}.
7653 You can instruct the @value{GDBN} testsuite to write transcripts by
7654 setting the DejaGNU variable @code{TRANSCRIPT} (to any value)
7655 before invoking @code{runtest} or @kbd{make check}. The transcripts
7656 will be written into DejaGNU's output directory. One transcript will
7657 be made for each invocation of @value{GDBN}; they will be named
7658 @file{transcript.@var{n}}, where @var{n} is an integer. The first
7659 line of the transcript file will show how @value{GDBN} was invoked;
7660 each subsequent line is a command sent as input to @value{GDBN}.
7663 make check RUNTESTFLAGS=TRANSCRIPT=y
7666 Note that the transcript is not always complete. In particular, tests
7667 of completion can yield partial command lines.
7671 Sometimes one wishes to test a different @value{GDBN} than the one in the build
7672 directory. For example, one may wish to run the testsuite on
7673 @file{/usr/bin/gdb}.
7676 make check RUNTESTFLAGS=GDB=/usr/bin/gdb
7679 @item @code{GDBSERVER}
7681 When testing a different @value{GDBN}, it is often useful to also test a
7682 different gdbserver.
7685 make check RUNTESTFLAGS="GDB=/usr/bin/gdb GDBSERVER=/usr/bin/gdbserver"
7688 @item @code{INTERNAL_GDBFLAGS}
7690 When running the testsuite normally one doesn't want whatever is in
7691 @file{~/.gdbinit} to interfere with the tests, therefore the test harness
7692 passes @option{-nx} to @value{GDBN}. One also doesn't want any windowed
7693 version of @value{GDBN}, e.g., @command{gdbtui}, to run.
7694 This is achieved via @code{INTERNAL_GDBFLAGS}.
7697 set INTERNAL_GDBFLAGS "-nw -nx"
7700 This is all well and good, except when testing an installed @value{GDBN}
7701 that has been configured with @option{--with-system-gdbinit}. Here one
7702 does not want @file{~/.gdbinit} loaded but one may want the system
7703 @file{.gdbinit} file loaded. This can be achieved by pointing @code{$HOME}
7704 at a directory without a @file{.gdbinit} and by overriding
7705 @code{INTERNAL_GDBFLAGS} and removing @option{-nx}.
7709 HOME=`pwd` runtest \
7711 GDBSERVER=/usr/bin/gdbserver \
7712 INTERNAL_GDBFLAGS=-nw
7717 There are two ways to run the testsuite and pass additional parameters
7718 to DejaGnu. The first is with @kbd{make check} and specifying the
7719 makefile variable @samp{RUNTESTFLAGS}.
7722 make check RUNTESTFLAGS=TRANSCRIPT=y
7725 The second is to cd to the @file{testsuite} directory and invoke the DejaGnu
7726 @command{runtest} command directly.
7731 runtest TRANSCRIPT=y
7734 @section Testsuite Configuration
7735 @cindex Testsuite Configuration
7737 It is possible to adjust the behavior of the testsuite by defining
7738 the global variables listed below, either in a @file{site.exp} file,
7743 @item @code{gdb_test_timeout}
7745 Defining this variable changes the default timeout duration used during
7746 communication with @value{GDBN}. More specifically, the global variable
7747 used during testing is @code{timeout}, but this variable gets reset to
7748 @code{gdb_test_timeout} at the beginning of each testcase, making sure
7749 that any local change to @code{timeout} in a testcase does not affect
7750 subsequent testcases.
7752 This global variable comes in handy when the debugger is slower than
7753 normal due to the testing environment, triggering unexpected @code{TIMEOUT}
7754 test failures. Examples include when testing on a remote machine, or
7755 against a system where communications are slow.
7757 If not specifically defined, this variable gets automatically defined
7758 to the same value as @code{timeout} during the testsuite initialization.
7759 The default value of the timeout is defined in the file
7760 @file{gdb/testsuite/config/unix.exp} that is part of the @value{GDBN}
7761 test suite@footnote{If you are using a board file, it could override
7762 the test-suite default; search the board file for "timeout".}.
7766 @section Testsuite Organization
7768 @cindex test suite organization
7769 The testsuite is entirely contained in @file{gdb/testsuite}. While the
7770 testsuite includes some makefiles and configury, these are very minimal,
7771 and used for little besides cleaning up, since the tests themselves
7772 handle the compilation of the programs that @value{GDBN} will run. The file
7773 @file{testsuite/lib/gdb.exp} contains common utility procs useful for
7774 all @value{GDBN} tests, while the directory @file{testsuite/config} contains
7775 configuration-specific files, typically used for special-purpose
7776 definitions of procs like @code{gdb_load} and @code{gdb_start}.
7778 The tests themselves are to be found in @file{testsuite/gdb.*} and
7779 subdirectories of those. The names of the test files must always end
7780 with @file{.exp}. DejaGNU collects the test files by wildcarding
7781 in the test directories, so both subdirectories and individual files
7782 get chosen and run in alphabetical order.
7784 The following table lists the main types of subdirectories and what they
7785 are for. Since DejaGNU finds test files no matter where they are
7786 located, and since each test file sets up its own compilation and
7787 execution environment, this organization is simply for convenience and
7792 This is the base testsuite. The tests in it should apply to all
7793 configurations of @value{GDBN} (but generic native-only tests may live here).
7794 The test programs should be in the subset of C that is valid K&R,
7795 ANSI/ISO, and C@t{++} (@code{#ifdef}s are allowed if necessary, for instance
7798 @item gdb.@var{lang}
7799 Language-specific tests for any language @var{lang} besides C. Examples are
7800 @file{gdb.cp} and @file{gdb.java}.
7802 @item gdb.@var{platform}
7803 Non-portable tests. The tests are specific to a specific configuration
7804 (host or target), such as HP-UX or eCos. Example is @file{gdb.hp}, for
7807 @item gdb.@var{compiler}
7808 Tests specific to a particular compiler. As of this writing (June
7809 1999), there aren't currently any groups of tests in this category that
7810 couldn't just as sensibly be made platform-specific, but one could
7811 imagine a @file{gdb.gcc}, for tests of @value{GDBN}'s handling of GCC
7814 @item gdb.@var{subsystem}
7815 Tests that exercise a specific @value{GDBN} subsystem in more depth. For
7816 instance, @file{gdb.disasm} exercises various disassemblers, while
7817 @file{gdb.stabs} tests pathways through the stabs symbol reader.
7820 @section Writing Tests
7821 @cindex writing tests
7823 In many areas, the @value{GDBN} tests are already quite comprehensive; you
7824 should be able to copy existing tests to handle new cases.
7826 You should try to use @code{gdb_test} whenever possible, since it
7827 includes cases to handle all the unexpected errors that might happen.
7828 However, it doesn't cost anything to add new test procedures; for
7829 instance, @file{gdb.base/exprs.exp} defines a @code{test_expr} that
7830 calls @code{gdb_test} multiple times.
7832 Only use @code{send_gdb} and @code{gdb_expect} when absolutely
7833 necessary. Even if @value{GDBN} has several valid responses to
7834 a command, you can use @code{gdb_test_multiple}. Like @code{gdb_test},
7835 @code{gdb_test_multiple} recognizes internal errors and unexpected
7838 Do not write tests which expect a literal tab character from @value{GDBN}.
7839 On some operating systems (e.g.@: OpenBSD) the TTY layer expands tabs to
7840 spaces, so by the time @value{GDBN}'s output reaches expect the tab is gone.
7842 The source language programs do @emph{not} need to be in a consistent
7843 style. Since @value{GDBN} is used to debug programs written in many different
7844 styles, it's worth having a mix of styles in the testsuite; for
7845 instance, some @value{GDBN} bugs involving the display of source lines would
7846 never manifest themselves if the programs used GNU coding style
7853 Check the @file{README} file, it often has useful information that does not
7854 appear anywhere else in the directory.
7857 * Getting Started:: Getting started working on @value{GDBN}
7858 * Debugging GDB:: Debugging @value{GDBN} with itself
7861 @node Getting Started
7863 @section Getting Started
7865 @value{GDBN} is a large and complicated program, and if you first starting to
7866 work on it, it can be hard to know where to start. Fortunately, if you
7867 know how to go about it, there are ways to figure out what is going on.
7869 This manual, the @value{GDBN} Internals manual, has information which applies
7870 generally to many parts of @value{GDBN}.
7872 Information about particular functions or data structures are located in
7873 comments with those functions or data structures. If you run across a
7874 function or a global variable which does not have a comment correctly
7875 explaining what is does, this can be thought of as a bug in @value{GDBN}; feel
7876 free to submit a bug report, with a suggested comment if you can figure
7877 out what the comment should say. If you find a comment which is
7878 actually wrong, be especially sure to report that.
7880 Comments explaining the function of macros defined in host, target, or
7881 native dependent files can be in several places. Sometimes they are
7882 repeated every place the macro is defined. Sometimes they are where the
7883 macro is used. Sometimes there is a header file which supplies a
7884 default definition of the macro, and the comment is there. This manual
7885 also documents all the available macros.
7886 @c (@pxref{Host Conditionals}, @pxref{Target
7887 @c Conditionals}, @pxref{Native Conditionals}, and @pxref{Obsolete
7890 Start with the header files. Once you have some idea of how
7891 @value{GDBN}'s internal symbol tables are stored (see @file{symtab.h},
7892 @file{gdbtypes.h}), you will find it much easier to understand the
7893 code which uses and creates those symbol tables.
7895 You may wish to process the information you are getting somehow, to
7896 enhance your understanding of it. Summarize it, translate it to another
7897 language, add some (perhaps trivial or non-useful) feature to @value{GDBN}, use
7898 the code to predict what a test case would do and write the test case
7899 and verify your prediction, etc. If you are reading code and your eyes
7900 are starting to glaze over, this is a sign you need to use a more active
7903 Once you have a part of @value{GDBN} to start with, you can find more
7904 specifically the part you are looking for by stepping through each
7905 function with the @code{next} command. Do not use @code{step} or you
7906 will quickly get distracted; when the function you are stepping through
7907 calls another function try only to get a big-picture understanding
7908 (perhaps using the comment at the beginning of the function being
7909 called) of what it does. This way you can identify which of the
7910 functions being called by the function you are stepping through is the
7911 one which you are interested in. You may need to examine the data
7912 structures generated at each stage, with reference to the comments in
7913 the header files explaining what the data structures are supposed to
7916 Of course, this same technique can be used if you are just reading the
7917 code, rather than actually stepping through it. The same general
7918 principle applies---when the code you are looking at calls something
7919 else, just try to understand generally what the code being called does,
7920 rather than worrying about all its details.
7922 @cindex command implementation
7923 A good place to start when tracking down some particular area is with
7924 a command which invokes that feature. Suppose you want to know how
7925 single-stepping works. As a @value{GDBN} user, you know that the
7926 @code{step} command invokes single-stepping. The command is invoked
7927 via command tables (see @file{command.h}); by convention the function
7928 which actually performs the command is formed by taking the name of
7929 the command and adding @samp{_command}, or in the case of an
7930 @code{info} subcommand, @samp{_info}. For example, the @code{step}
7931 command invokes the @code{step_command} function and the @code{info
7932 display} command invokes @code{display_info}. When this convention is
7933 not followed, you might have to use @code{grep} or @kbd{M-x
7934 tags-search} in emacs, or run @value{GDBN} on itself and set a
7935 breakpoint in @code{execute_command}.
7937 @cindex @code{bug-gdb} mailing list
7938 If all of the above fail, it may be appropriate to ask for information
7939 on @code{bug-gdb}. But @emph{never} post a generic question like ``I was
7940 wondering if anyone could give me some tips about understanding
7941 @value{GDBN}''---if we had some magic secret we would put it in this manual.
7942 Suggestions for improving the manual are always welcome, of course.
7946 @section Debugging @value{GDBN} with itself
7947 @cindex debugging @value{GDBN}
7949 If @value{GDBN} is limping on your machine, this is the preferred way to get it
7950 fully functional. Be warned that in some ancient Unix systems, like
7951 Ultrix 4.2, a program can't be running in one process while it is being
7952 debugged in another. Rather than typing the command @kbd{@w{./gdb
7953 ./gdb}}, which works on Suns and such, you can copy @file{gdb} to
7954 @file{gdb2} and then type @kbd{@w{./gdb ./gdb2}}.
7956 When you run @value{GDBN} in the @value{GDBN} source directory, it will read a
7957 @file{.gdbinit} file that sets up some simple things to make debugging
7958 gdb easier. The @code{info} command, when executed without a subcommand
7959 in a @value{GDBN} being debugged by gdb, will pop you back up to the top level
7960 gdb. See @file{.gdbinit} for details.
7962 If you use emacs, you will probably want to do a @code{make TAGS} after
7963 you configure your distribution; this will put the machine dependent
7964 routines for your local machine where they will be accessed first by
7967 Also, make sure that you've either compiled @value{GDBN} with your local cc, or
7968 have run @code{fixincludes} if you are compiling with gcc.
7970 @section Submitting Patches
7972 @cindex submitting patches
7973 Thanks for thinking of offering your changes back to the community of
7974 @value{GDBN} users. In general we like to get well designed enhancements.
7975 Thanks also for checking in advance about the best way to transfer the
7978 The @value{GDBN} maintainers will only install ``cleanly designed'' patches.
7979 This manual summarizes what we believe to be clean design for @value{GDBN}.
7981 If the maintainers don't have time to put the patch in when it arrives,
7982 or if there is any question about a patch, it goes into a large queue
7983 with everyone else's patches and bug reports.
7985 @cindex legal papers for code contributions
7986 The legal issue is that to incorporate substantial changes requires a
7987 copyright assignment from you and/or your employer, granting ownership
7988 of the changes to the Free Software Foundation. You can get the
7989 standard documents for doing this by sending mail to @code{gnu@@gnu.org}
7990 and asking for it. We recommend that people write in "All programs
7991 owned by the Free Software Foundation" as "NAME OF PROGRAM", so that
7992 changes in many programs (not just @value{GDBN}, but GAS, Emacs, GCC,
7994 contributed with only one piece of legalese pushed through the
7995 bureaucracy and filed with the FSF. We can't start merging changes until
7996 this paperwork is received by the FSF (their rules, which we follow
7997 since we maintain it for them).
7999 Technically, the easiest way to receive changes is to receive each
8000 feature as a small context diff or unidiff, suitable for @code{patch}.
8001 Each message sent to me should include the changes to C code and
8002 header files for a single feature, plus @file{ChangeLog} entries for
8003 each directory where files were modified, and diffs for any changes
8004 needed to the manuals (@file{gdb/doc/gdb.texinfo} or
8005 @file{gdb/doc/gdbint.texinfo}). If there are a lot of changes for a
8006 single feature, they can be split down into multiple messages.
8008 In this way, if we read and like the feature, we can add it to the
8009 sources with a single patch command, do some testing, and check it in.
8010 If you leave out the @file{ChangeLog}, we have to write one. If you leave
8011 out the doc, we have to puzzle out what needs documenting. Etc., etc.
8013 The reason to send each change in a separate message is that we will not
8014 install some of the changes. They'll be returned to you with questions
8015 or comments. If we're doing our job correctly, the message back to you
8016 will say what you have to fix in order to make the change acceptable.
8017 The reason to have separate messages for separate features is so that
8018 the acceptable changes can be installed while one or more changes are
8019 being reworked. If multiple features are sent in a single message, we
8020 tend to not put in the effort to sort out the acceptable changes from
8021 the unacceptable, so none of the features get installed until all are
8024 If this sounds painful or authoritarian, well, it is. But we get a lot
8025 of bug reports and a lot of patches, and many of them don't get
8026 installed because we don't have the time to finish the job that the bug
8027 reporter or the contributor could have done. Patches that arrive
8028 complete, working, and well designed, tend to get installed on the day
8029 they arrive. The others go into a queue and get installed as time
8030 permits, which, since the maintainers have many demands to meet, may not
8031 be for quite some time.
8033 Please send patches directly to
8034 @email{gdb-patches@@sourceware.org, the @value{GDBN} maintainers}.
8036 @section Build Script
8038 @cindex build script
8040 The script @file{gdb_buildall.sh} builds @value{GDBN} with flag
8041 @option{--enable-targets=all} set. This builds @value{GDBN} with all supported
8042 targets activated. This helps testing @value{GDBN} when doing changes that
8043 affect more than one architecture and is much faster than using
8044 @file{gdb_mbuild.sh}.
8046 After building @value{GDBN} the script checks which architectures are
8047 supported and then switches the current architecture to each of those to get
8048 information about the architecture. The test results are stored in log files
8049 in the directory the script was called from.
8051 @include observer.texi
8053 @node GNU Free Documentation License
8054 @appendix GNU Free Documentation License