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{} 1990-1994, 1996, 1998-2006, 2008-2012 Free
13 Software Foundation, Inc.
14 Contributed by Cygnus Solutions. Written by John Gilmore.
15 Second Edition by Stan Shebs.
17 Permission is granted to copy, distribute and/or modify this document
18 under the terms of the GNU Free Documentation License, Version 1.3 or
19 any later version published by the Free Software Foundation; with no
20 Invariant Sections, with no Front-Cover Texts, and with no Back-Cover
21 Texts. A copy of the license is included in the section entitled ``GNU
22 Free Documentation License''.
26 This file documents the internals of the GNU debugger @value{GDBN}.
36 @title @value{GDBN} Internals
37 @subtitle{A guide to the internals of the GNU debugger}
39 @author Cygnus Solutions
40 @author Second Edition:
42 @author Cygnus Solutions
45 \def\$#1${{#1}} % Kluge: collect RCS revision info without $...$
46 \xdef\manvers{\$Revision$} % For use in headers, footers too
48 \hfill Cygnus Solutions\par
50 \hfill \TeX{}info \texinfoversion\par
54 @vskip 0pt plus 1filll
61 @c Perhaps this should be the title of the document (but only for info,
62 @c not for TeX). Existing GNU manuals seem inconsistent on this point.
63 @top Scope of this Document
65 This document documents the internals of the GNU debugger, @value{GDBN}. It
66 includes description of @value{GDBN}'s key algorithms and operations, as well
67 as the mechanisms that adapt @value{GDBN} to specific hosts and targets.
80 * Target Architecture Definition::
81 * Target Descriptions::
82 * Target Vector Definition::
88 * Versions and Branches::
89 * Start of New Year Procedure::
94 * GDB Observers:: @value{GDBN} Currently available observers
95 * GNU Free Documentation License:: The license for this documentation
108 @section Requirements
109 @cindex requirements for @value{GDBN}
111 Before diving into the internals, you should understand the formal
112 requirements and other expectations for @value{GDBN}. Although some
113 of these may seem obvious, there have been proposals for @value{GDBN}
114 that have run counter to these requirements.
116 First of all, @value{GDBN} is a debugger. It's not designed to be a
117 front panel for embedded systems. It's not a text editor. It's not a
118 shell. It's not a programming environment.
120 @value{GDBN} is an interactive tool. Although a batch mode is
121 available, @value{GDBN}'s primary role is to interact with a human
124 @value{GDBN} should be responsive to the user. A programmer hot on
125 the trail of a nasty bug, and operating under a looming deadline, is
126 going to be very impatient of everything, including the response time
127 to debugger commands.
129 @value{GDBN} should be relatively permissive, such as for expressions.
130 While the compiler should be picky (or have the option to be made
131 picky), since source code lives for a long time usually, the
132 programmer doing debugging shouldn't be spending time figuring out to
133 mollify the debugger.
135 @value{GDBN} will be called upon to deal with really large programs.
136 Executable sizes of 50 to 100 megabytes occur regularly, and we've
137 heard reports of programs approaching 1 gigabyte in size.
139 @value{GDBN} should be able to run everywhere. No other debugger is
140 available for even half as many configurations as @value{GDBN}
144 @section Contributors
146 The first edition of this document was written by John Gilmore of
147 Cygnus Solutions. The current second edition was written by Stan Shebs
148 of Cygnus Solutions, who continues to update the manual.
150 Over the years, many others have made additions and changes to this
151 document. This section attempts to record the significant contributors
152 to that effort. One of the virtues of free software is that everyone
153 is free to contribute to it; with regret, we cannot actually
154 acknowledge everyone here.
157 @emph{Plea:} This section has only been added relatively recently (four
158 years after publication of the second edition). Additions to this
159 section are particularly welcome. If you or your friends (or enemies,
160 to be evenhanded) have been unfairly omitted from this list, we would
161 like to add your names!
164 A document such as this relies on being kept up to date by numerous
165 small updates by contributing engineers as they make changes to the
166 code base. The file @file{ChangeLog} in the @value{GDBN} distribution
167 approximates a blow-by-blow account. The most prolific contributors to
168 this important, but low profile task are Andrew Cagney (responsible
169 for over half the entries), Daniel Jacobowitz, Mark Kettenis, Jim
170 Blandy and Eli Zaretskii.
172 Eli Zaretskii and Daniel Jacobowitz wrote the sections documenting
175 Jeremy Bennett updated the sections on initializing a new architecture
176 and register representation, and added the section on Frame Interpretation.
179 @node Overall Structure
181 @chapter Overall Structure
183 @value{GDBN} consists of three major subsystems: user interface,
184 symbol handling (the @dfn{symbol side}), and target system handling (the
187 The user interface consists of several actual interfaces, plus
190 The symbol side consists of object file readers, debugging info
191 interpreters, symbol table management, source language expression
192 parsing, type and value printing.
194 The target side consists of execution control, stack frame analysis, and
195 physical target manipulation.
197 The target side/symbol side division is not formal, and there are a
198 number of exceptions. For instance, core file support involves symbolic
199 elements (the basic core file reader is in BFD) and target elements (it
200 supplies the contents of memory and the values of registers). Instead,
201 this division is useful for understanding how the minor subsystems
204 @section The Symbol Side
206 The symbolic side of @value{GDBN} can be thought of as ``everything
207 you can do in @value{GDBN} without having a live program running''.
208 For instance, you can look at the types of variables, and evaluate
209 many kinds of expressions.
211 @section The Target Side
213 The target side of @value{GDBN} is the ``bits and bytes manipulator''.
214 Although it may make reference to symbolic info here and there, most
215 of the target side will run with only a stripped executable
216 available---or even no executable at all, in remote debugging cases.
218 Operations such as disassembly, stack frame crawls, and register
219 display, are able to work with no symbolic info at all. In some cases,
220 such as disassembly, @value{GDBN} will use symbolic info to present addresses
221 relative to symbols rather than as raw numbers, but it will work either
224 @section Configurations
228 @dfn{Host} refers to attributes of the system where @value{GDBN} runs.
229 @dfn{Target} refers to the system where the program being debugged
230 executes. In most cases they are the same machine, in which case a
231 third type of @dfn{Native} attributes come into play.
233 Defines and include files needed to build on the host are host
234 support. Examples are tty support, system defined types, host byte
235 order, host float format. These are all calculated by @code{autoconf}
236 when the debugger is built.
238 Defines and information needed to handle the target format are target
239 dependent. Examples are the stack frame format, instruction set,
240 breakpoint instruction, registers, and how to set up and tear down the stack
243 Information that is only needed when the host and target are the same,
244 is native dependent. One example is Unix child process support; if the
245 host and target are not the same, calling @code{fork} to start the target
246 process is a bad idea. The various macros needed for finding the
247 registers in the @code{upage}, running @code{ptrace}, and such are all
248 in the native-dependent files.
250 Another example of native-dependent code is support for features that
251 are really part of the target environment, but which require
252 @code{#include} files that are only available on the host system. Core
253 file handling and @code{setjmp} handling are two common cases.
255 When you want to make @value{GDBN} work as the traditional native debugger
256 on a system, you will need to supply both target and native information.
258 @section Source Tree Structure
259 @cindex @value{GDBN} source tree structure
261 The @value{GDBN} source directory has a mostly flat structure---there
262 are only a few subdirectories. A file's name usually gives a hint as
263 to what it does; for example, @file{stabsread.c} reads stabs,
264 @file{dwarf2read.c} reads @sc{DWARF 2}, etc.
266 Files that are related to some common task have names that share
267 common substrings. For example, @file{*-thread.c} files deal with
268 debugging threads on various platforms; @file{*read.c} files deal with
269 reading various kinds of symbol and object files; @file{inf*.c} files
270 deal with direct control of the @dfn{inferior program} (@value{GDBN}
271 parlance for the program being debugged).
273 There are several dozens of files in the @file{*-tdep.c} family.
274 @samp{tdep} stands for @dfn{target-dependent code}---each of these
275 files implements debug support for a specific target architecture
276 (sparc, mips, etc). Usually, only one of these will be used in a
277 specific @value{GDBN} configuration (sometimes two, closely related).
279 Similarly, there are many @file{*-nat.c} files, each one for native
280 debugging on a specific system (e.g., @file{sparc-linux-nat.c} is for
281 native debugging of Sparc machines running the Linux kernel).
283 The few subdirectories of the source tree are:
287 Code that implements @dfn{CLI}, the @value{GDBN} Command-Line
288 Interpreter. @xref{User Interface, Command Interpreter}.
291 Code for the @value{GDBN} remote server.
294 Code for Insight, the @value{GDBN} TK-based GUI front-end.
297 The @dfn{GDB/MI}, the @value{GDBN} Machine Interface interpreter.
300 Target signal translation code.
303 Code for @dfn{TUI}, the @value{GDBN} Text-mode full-screen User
304 Interface. @xref{User Interface, TUI}.
312 @value{GDBN} uses a number of debugging-specific algorithms. They are
313 often not very complicated, but get lost in the thicket of special
314 cases and real-world issues. This chapter describes the basic
315 algorithms and mentions some of the specific target definitions that
318 @section Prologue Analysis
320 @cindex prologue analysis
321 @cindex call frame information
322 @cindex CFI (call frame information)
323 To produce a backtrace and allow the user to manipulate older frames'
324 variables and arguments, @value{GDBN} needs to find the base addresses
325 of older frames, and discover where those frames' registers have been
326 saved. Since a frame's ``callee-saves'' registers get saved by
327 younger frames if and when they're reused, a frame's registers may be
328 scattered unpredictably across younger frames. This means that
329 changing the value of a register-allocated variable in an older frame
330 may actually entail writing to a save slot in some younger frame.
332 Modern versions of GCC emit Dwarf call frame information (``CFI''),
333 which describes how to find frame base addresses and saved registers.
334 But CFI is not always available, so as a fallback @value{GDBN} uses a
335 technique called @dfn{prologue analysis} to find frame sizes and saved
336 registers. A prologue analyzer disassembles the function's machine
337 code starting from its entry point, and looks for instructions that
338 allocate frame space, save the stack pointer in a frame pointer
339 register, save registers, and so on. Obviously, this can't be done
340 accurately in general, but it's tractable to do well enough to be very
341 helpful. Prologue analysis predates the GNU toolchain's support for
342 CFI; at one time, prologue analysis was the only mechanism
343 @value{GDBN} used for stack unwinding at all, when the function
344 calling conventions didn't specify a fixed frame layout.
346 In the olden days, function prologues were generated by hand-written,
347 target-specific code in GCC, and treated as opaque and untouchable by
348 optimizers. Looking at this code, it was usually straightforward to
349 write a prologue analyzer for @value{GDBN} that would accurately
350 understand all the prologues GCC would generate. However, over time
351 GCC became more aggressive about instruction scheduling, and began to
352 understand more about the semantics of the prologue instructions
353 themselves; in response, @value{GDBN}'s analyzers became more complex
354 and fragile. Keeping the prologue analyzers working as GCC (and the
355 instruction sets themselves) evolved became a substantial task.
357 @cindex @file{prologue-value.c}
358 @cindex abstract interpretation of function prologues
359 @cindex pseudo-evaluation of function prologues
360 To try to address this problem, the code in @file{prologue-value.h}
361 and @file{prologue-value.c} provides a general framework for writing
362 prologue analyzers that are simpler and more robust than ad-hoc
363 analyzers. When we analyze a prologue using the prologue-value
364 framework, we're really doing ``abstract interpretation'' or
365 ``pseudo-evaluation'': running the function's code in simulation, but
366 using conservative approximations of the values registers and memory
367 would hold when the code actually runs. For example, if our function
368 starts with the instruction:
371 addi r1, 42 # add 42 to r1
374 we don't know exactly what value will be in @code{r1} after executing
375 this instruction, but we do know it'll be 42 greater than its original
378 If we then see an instruction like:
381 addi r1, 22 # add 22 to r1
384 we still don't know what @code{r1's} value is, but again, we can say
385 it is now 64 greater than its original value.
387 If the next instruction were:
390 mov r2, r1 # set r2 to r1's value
393 then we can say that @code{r2's} value is now the original value of
396 It's common for prologues to save registers on the stack, so we'll
397 need to track the values of stack frame slots, as well as the
398 registers. So after an instruction like this:
404 then we'd know that the stack slot four bytes above the frame pointer
405 holds the original value of @code{r1} plus 64.
409 Of course, this can only go so far before it gets unreasonable. If we
410 wanted to be able to say anything about the value of @code{r1} after
414 xor r1, r3 # exclusive-or r1 and r3, place result in r1
417 then things would get pretty complex. But remember, we're just doing
418 a conservative approximation; if exclusive-or instructions aren't
419 relevant to prologues, we can just say @code{r1}'s value is now
420 ``unknown''. We can ignore things that are too complex, if that loss of
421 information is acceptable for our application.
423 So when we say ``conservative approximation'' here, what we mean is an
424 approximation that is either accurate, or marked ``unknown'', but
427 Using this framework, a prologue analyzer is simply an interpreter for
428 machine code, but one that uses conservative approximations for the
429 contents of registers and memory instead of actual values. Starting
430 from the function's entry point, you simulate instructions up to the
431 current PC, or an instruction that you don't know how to simulate.
432 Now you can examine the state of the registers and stack slots you've
438 To see how large your stack frame is, just check the value of the
439 stack pointer register; if it's the original value of the SP
440 minus a constant, then that constant is the stack frame's size.
441 If the SP's value has been marked as ``unknown'', then that means
442 the prologue has done something too complex for us to track, and
443 we don't know the frame size.
446 To see where we've saved the previous frame's registers, we just
447 search the values we've tracked --- stack slots, usually, but
448 registers, too, if you want --- for something equal to the register's
449 original value. If the calling conventions suggest a standard place
450 to save a given register, then we can check there first, but really,
451 anything that will get us back the original value will probably work.
454 This does take some work. But prologue analyzers aren't
455 quick-and-simple pattern patching to recognize a few fixed prologue
456 forms any more; they're big, hairy functions. Along with inferior
457 function calls, prologue analysis accounts for a substantial portion
458 of the time needed to stabilize a @value{GDBN} port. So it's
459 worthwhile to look for an approach that will be easier to understand
460 and maintain. In the approach described above:
465 It's easier to see that the analyzer is correct: you just see
466 whether the analyzer properly (albeit conservatively) simulates
467 the effect of each instruction.
470 It's easier to extend the analyzer: you can add support for new
471 instructions, and know that you haven't broken anything that
472 wasn't already broken before.
475 It's orthogonal: to gather new information, you don't need to
476 complicate the code for each instruction. As long as your domain
477 of conservative values is already detailed enough to tell you
478 what you need, then all the existing instruction simulations are
479 already gathering the right data for you.
483 The file @file{prologue-value.h} contains detailed comments explaining
484 the framework and how to use it.
487 @section Breakpoint Handling
490 In general, a breakpoint is a user-designated location in the program
491 where the user wants to regain control if program execution ever reaches
494 There are two main ways to implement breakpoints; either as ``hardware''
495 breakpoints or as ``software'' breakpoints.
497 @cindex hardware breakpoints
498 @cindex program counter
499 Hardware breakpoints are sometimes available as a builtin debugging
500 features with some chips. Typically these work by having dedicated
501 register into which the breakpoint address may be stored. If the PC
502 (shorthand for @dfn{program counter})
503 ever matches a value in a breakpoint registers, the CPU raises an
504 exception and reports it to @value{GDBN}.
506 Another possibility is when an emulator is in use; many emulators
507 include circuitry that watches the address lines coming out from the
508 processor, and force it to stop if the address matches a breakpoint's
511 A third possibility is that the target already has the ability to do
512 breakpoints somehow; for instance, a ROM monitor may do its own
513 software breakpoints. So although these are not literally ``hardware
514 breakpoints'', from @value{GDBN}'s point of view they work the same;
515 @value{GDBN} need not do anything more than set the breakpoint and wait
516 for something to happen.
518 Since they depend on hardware resources, hardware breakpoints may be
519 limited in number; when the user asks for more, @value{GDBN} will
520 start trying to set software breakpoints. (On some architectures,
521 notably the 32-bit x86 platforms, @value{GDBN} cannot always know
522 whether there's enough hardware resources to insert all the hardware
523 breakpoints and watchpoints. On those platforms, @value{GDBN} prints
524 an error message only when the program being debugged is continued.)
526 @cindex software breakpoints
527 Software breakpoints require @value{GDBN} to do somewhat more work.
528 The basic theory is that @value{GDBN} will replace a program
529 instruction with a trap, illegal divide, or some other instruction
530 that will cause an exception, and then when it's encountered,
531 @value{GDBN} will take the exception and stop the program. When the
532 user says to continue, @value{GDBN} will restore the original
533 instruction, single-step, re-insert the trap, and continue on.
535 Since it literally overwrites the program being tested, the program area
536 must be writable, so this technique won't work on programs in ROM. It
537 can also distort the behavior of programs that examine themselves,
538 although such a situation would be highly unusual.
540 Also, the software breakpoint instruction should be the smallest size of
541 instruction, so it doesn't overwrite an instruction that might be a jump
542 target, and cause disaster when the program jumps into the middle of the
543 breakpoint instruction. (Strictly speaking, the breakpoint must be no
544 larger than the smallest interval between instructions that may be jump
545 targets; perhaps there is an architecture where only even-numbered
546 instructions may jumped to.) Note that it's possible for an instruction
547 set not to have any instructions usable for a software breakpoint,
548 although in practice only the ARC has failed to define such an
551 Basic breakpoint object handling is in @file{breakpoint.c}. However,
552 much of the interesting breakpoint action is in @file{infrun.c}.
555 @cindex insert or remove software breakpoint
556 @findex target_remove_breakpoint
557 @findex target_insert_breakpoint
558 @item target_remove_breakpoint (@var{bp_tgt})
559 @itemx target_insert_breakpoint (@var{bp_tgt})
560 Insert or remove a software breakpoint at address
561 @code{@var{bp_tgt}->placed_address}. Returns zero for success,
562 non-zero for failure. On input, @var{bp_tgt} contains the address of the
563 breakpoint, and is otherwise initialized to zero. The fields of the
564 @code{struct bp_target_info} pointed to by @var{bp_tgt} are updated
565 to contain other information about the breakpoint on output. The field
566 @code{placed_address} may be updated if the breakpoint was placed at a
567 related address; the field @code{shadow_contents} contains the real
568 contents of the bytes where the breakpoint has been inserted,
569 if reading memory would return the breakpoint instead of the
570 underlying memory; the field @code{shadow_len} is the length of
571 memory cached in @code{shadow_contents}, if any; and the field
572 @code{placed_size} is optionally set and used by the target, if
573 it could differ from @code{shadow_len}.
575 For example, the remote target @samp{Z0} packet does not require
576 shadowing memory, so @code{shadow_len} is left at zero. However,
577 the length reported by @code{gdbarch_breakpoint_from_pc} is cached in
578 @code{placed_size}, so that a matching @samp{z0} packet can be
579 used to remove the breakpoint.
581 @cindex insert or remove hardware breakpoint
582 @findex target_remove_hw_breakpoint
583 @findex target_insert_hw_breakpoint
584 @item target_remove_hw_breakpoint (@var{bp_tgt})
585 @itemx target_insert_hw_breakpoint (@var{bp_tgt})
586 Insert or remove a hardware-assisted breakpoint at address
587 @code{@var{bp_tgt}->placed_address}. Returns zero for success,
588 non-zero for failure. See @code{target_insert_breakpoint} for
589 a description of the @code{struct bp_target_info} pointed to by
590 @var{bp_tgt}; the @code{shadow_contents} and
591 @code{shadow_len} members are not used for hardware breakpoints,
592 but @code{placed_size} may be.
595 @section Single Stepping
597 @section Signal Handling
599 @section Thread Handling
601 @section Inferior Function Calls
603 @section Longjmp Support
605 @cindex @code{longjmp} debugging
606 @value{GDBN} has support for figuring out that the target is doing a
607 @code{longjmp} and for stopping at the target of the jump, if we are
608 stepping. This is done with a few specialized internal breakpoints,
609 which are visible in the output of the @samp{maint info breakpoint}
612 @findex gdbarch_get_longjmp_target
613 To make this work, you need to define a function called
614 @code{gdbarch_get_longjmp_target}, which will examine the
615 @code{jmp_buf} structure and extract the @code{longjmp} target address.
616 Since @code{jmp_buf} is target specific and typically defined in a
617 target header not available to @value{GDBN}, you will need to
618 determine the offset of the PC manually and return that; many targets
619 define a @code{jb_pc_offset} field in the tdep structure to save the
620 value once calculated.
625 Watchpoints are a special kind of breakpoints (@pxref{Algorithms,
626 breakpoints}) which break when data is accessed rather than when some
627 instruction is executed. When you have data which changes without
628 your knowing what code does that, watchpoints are the silver bullet to
629 hunt down and kill such bugs.
631 @cindex hardware watchpoints
632 @cindex software watchpoints
633 Watchpoints can be either hardware-assisted or not; the latter type is
634 known as ``software watchpoints.'' @value{GDBN} always uses
635 hardware-assisted watchpoints if they are available, and falls back on
636 software watchpoints otherwise. Typical situations where @value{GDBN}
637 will use software watchpoints are:
641 The watched memory region is too large for the underlying hardware
642 watchpoint support. For example, each x86 debug register can watch up
643 to 4 bytes of memory, so trying to watch data structures whose size is
644 more than 16 bytes will cause @value{GDBN} to use software
648 The value of the expression to be watched depends on data held in
649 registers (as opposed to memory).
652 Too many different watchpoints requested. (On some architectures,
653 this situation is impossible to detect until the debugged program is
654 resumed.) Note that x86 debug registers are used both for hardware
655 breakpoints and for watchpoints, so setting too many hardware
656 breakpoints might cause watchpoint insertion to fail.
659 No hardware-assisted watchpoints provided by the target
663 Software watchpoints are very slow, since @value{GDBN} needs to
664 single-step the program being debugged and test the value of the
665 watched expression(s) after each instruction. The rest of this
666 section is mostly irrelevant for software watchpoints.
668 When the inferior stops, @value{GDBN} tries to establish, among other
669 possible reasons, whether it stopped due to a watchpoint being hit.
670 It first uses @code{STOPPED_BY_WATCHPOINT} to see if any watchpoint
671 was hit. If not, all watchpoint checking is skipped.
673 Then @value{GDBN} calls @code{target_stopped_data_address} exactly
674 once. This method returns the address of the watchpoint which
675 triggered, if the target can determine it. If the triggered address
676 is available, @value{GDBN} compares the address returned by this
677 method with each watched memory address in each active watchpoint.
678 For data-read and data-access watchpoints, @value{GDBN} announces
679 every watchpoint that watches the triggered address as being hit.
680 For this reason, data-read and data-access watchpoints
681 @emph{require} that the triggered address be available; if not, read
682 and access watchpoints will never be considered hit. For data-write
683 watchpoints, if the triggered address is available, @value{GDBN}
684 considers only those watchpoints which match that address;
685 otherwise, @value{GDBN} considers all data-write watchpoints. For
686 each data-write watchpoint that @value{GDBN} considers, it evaluates
687 the expression whose value is being watched, and tests whether the
688 watched value has changed. Watchpoints whose watched values have
689 changed are announced as hit.
691 @c FIXME move these to the main lists of target/native defns
693 @value{GDBN} uses several macros and primitives to support hardware
697 @findex TARGET_CAN_USE_HARDWARE_WATCHPOINT
698 @item TARGET_CAN_USE_HARDWARE_WATCHPOINT (@var{type}, @var{count}, @var{other})
699 Return the number of hardware watchpoints of type @var{type} that are
700 possible to be set. The value is positive if @var{count} watchpoints
701 of this type can be set, zero if setting watchpoints of this type is
702 not supported, and negative if @var{count} is more than the maximum
703 number of watchpoints of type @var{type} that can be set. @var{other}
704 is non-zero if other types of watchpoints are currently enabled (there
705 are architectures which cannot set watchpoints of different types at
708 @findex TARGET_REGION_OK_FOR_HW_WATCHPOINT
709 @item TARGET_REGION_OK_FOR_HW_WATCHPOINT (@var{addr}, @var{len})
710 Return non-zero if hardware watchpoints can be used to watch a region
711 whose address is @var{addr} and whose length in bytes is @var{len}.
713 @cindex insert or remove hardware watchpoint
714 @findex target_insert_watchpoint
715 @findex target_remove_watchpoint
716 @item target_insert_watchpoint (@var{addr}, @var{len}, @var{type})
717 @itemx target_remove_watchpoint (@var{addr}, @var{len}, @var{type})
718 Insert or remove a hardware watchpoint starting at @var{addr}, for
719 @var{len} bytes. @var{type} is the watchpoint type, one of the
720 possible values of the enumerated data type @code{target_hw_bp_type},
721 defined by @file{breakpoint.h} as follows:
724 enum target_hw_bp_type
726 hw_write = 0, /* Common (write) HW watchpoint */
727 hw_read = 1, /* Read HW watchpoint */
728 hw_access = 2, /* Access (read or write) HW watchpoint */
729 hw_execute = 3 /* Execute HW breakpoint */
734 These two macros should return 0 for success, non-zero for failure.
736 @findex target_stopped_data_address
737 @item target_stopped_data_address (@var{addr_p})
738 If the inferior has some watchpoint that triggered, place the address
739 associated with the watchpoint at the location pointed to by
740 @var{addr_p} and return non-zero. Otherwise, return zero. This
741 is required for data-read and data-access watchpoints. It is
742 not required for data-write watchpoints, but @value{GDBN} uses
743 it to improve handling of those also.
745 @value{GDBN} will only call this method once per watchpoint stop,
746 immediately after calling @code{STOPPED_BY_WATCHPOINT}. If the
747 target's watchpoint indication is sticky, i.e., stays set after
748 resuming, this method should clear it. For instance, the x86 debug
749 control register has sticky triggered flags.
751 @findex target_watchpoint_addr_within_range
752 @item target_watchpoint_addr_within_range (@var{target}, @var{addr}, @var{start}, @var{length})
753 Check whether @var{addr} (as returned by @code{target_stopped_data_address})
754 lies within the hardware-defined watchpoint region described by
755 @var{start} and @var{length}. This only needs to be provided if the
756 granularity of a watchpoint is greater than one byte, i.e., if the
757 watchpoint can also trigger on nearby addresses outside of the watched
760 @findex HAVE_STEPPABLE_WATCHPOINT
761 @item HAVE_STEPPABLE_WATCHPOINT
762 If defined to a non-zero value, it is not necessary to disable a
763 watchpoint to step over it. Like @code{gdbarch_have_nonsteppable_watchpoint},
764 this is usually set when watchpoints trigger at the instruction
765 which will perform an interesting read or write. It should be
766 set if there is a temporary disable bit which allows the processor
767 to step over the interesting instruction without raising the
768 watchpoint exception again.
770 @findex gdbarch_have_nonsteppable_watchpoint
771 @item int gdbarch_have_nonsteppable_watchpoint (@var{gdbarch})
772 If it returns a non-zero value, @value{GDBN} should disable a
773 watchpoint to step the inferior over it. This is usually set when
774 watchpoints trigger at the instruction which will perform an
775 interesting read or write.
777 @findex HAVE_CONTINUABLE_WATCHPOINT
778 @item HAVE_CONTINUABLE_WATCHPOINT
779 If defined to a non-zero value, it is possible to continue the
780 inferior after a watchpoint has been hit. This is usually set
781 when watchpoints trigger at the instruction following an interesting
784 @findex STOPPED_BY_WATCHPOINT
785 @item STOPPED_BY_WATCHPOINT (@var{wait_status})
786 Return non-zero if stopped by a watchpoint. @var{wait_status} is of
787 the type @code{struct target_waitstatus}, defined by @file{target.h}.
788 Normally, this macro is defined to invoke the function pointed to by
789 the @code{to_stopped_by_watchpoint} member of the structure (of the
790 type @code{target_ops}, defined on @file{target.h}) that describes the
791 target-specific operations; @code{to_stopped_by_watchpoint} ignores
792 the @var{wait_status} argument.
794 @value{GDBN} does not require the non-zero value returned by
795 @code{STOPPED_BY_WATCHPOINT} to be 100% correct, so if a target cannot
796 determine for sure whether the inferior stopped due to a watchpoint,
797 it could return non-zero ``just in case''.
800 @subsection Watchpoints and Threads
801 @cindex watchpoints, with threads
803 @value{GDBN} only supports process-wide watchpoints, which trigger
804 in all threads. @value{GDBN} uses the thread ID to make watchpoints
805 act as if they were thread-specific, but it cannot set hardware
806 watchpoints that only trigger in a specific thread. Therefore, even
807 if the target supports threads, per-thread debug registers, and
808 watchpoints which only affect a single thread, it should set the
809 per-thread debug registers for all threads to the same value. On
810 @sc{gnu}/Linux native targets, this is accomplished by using
811 @code{ALL_LWPS} in @code{target_insert_watchpoint} and
812 @code{target_remove_watchpoint} and by using
813 @code{linux_set_new_thread} to register a handler for newly created
816 @value{GDBN}'s @sc{gnu}/Linux support only reports a single event
817 at a time, although multiple events can trigger simultaneously for
818 multi-threaded programs. When multiple events occur, @file{linux-nat.c}
819 queues subsequent events and returns them the next time the program
820 is resumed. This means that @code{STOPPED_BY_WATCHPOINT} and
821 @code{target_stopped_data_address} only need to consult the current
822 thread's state---the thread indicated by @code{inferior_ptid}. If
823 two threads have hit watchpoints simultaneously, those routines
824 will be called a second time for the second thread.
826 @subsection x86 Watchpoints
827 @cindex x86 debug registers
828 @cindex watchpoints, on x86
830 The 32-bit Intel x86 (a.k.a.@: ia32) processors feature special debug
831 registers designed to facilitate debugging. @value{GDBN} provides a
832 generic library of functions that x86-based ports can use to implement
833 support for watchpoints and hardware-assisted breakpoints. This
834 subsection documents the x86 watchpoint facilities in @value{GDBN}.
836 (At present, the library functions read and write debug registers directly, and are
837 thus only available for native configurations.)
839 To use the generic x86 watchpoint support, a port should do the
843 @findex I386_USE_GENERIC_WATCHPOINTS
845 Define the macro @code{I386_USE_GENERIC_WATCHPOINTS} somewhere in the
846 target-dependent headers.
849 Include the @file{config/i386/nm-i386.h} header file @emph{after}
850 defining @code{I386_USE_GENERIC_WATCHPOINTS}.
853 Add @file{i386-nat.o} to the value of the Make variable
854 @code{NATDEPFILES} (@pxref{Native Debugging, NATDEPFILES}).
857 Provide implementations for the @code{I386_DR_LOW_*} macros described
858 below. Typically, each macro should call a target-specific function
859 which does the real work.
862 The x86 watchpoint support works by maintaining mirror images of the
863 debug registers. Values are copied between the mirror images and the
864 real debug registers via a set of macros which each target needs to
868 @findex I386_DR_LOW_SET_CONTROL
869 @item I386_DR_LOW_SET_CONTROL (@var{val})
870 Set the Debug Control (DR7) register to the value @var{val}.
872 @findex I386_DR_LOW_SET_ADDR
873 @item I386_DR_LOW_SET_ADDR (@var{idx}, @var{addr})
874 Put the address @var{addr} into the debug register number @var{idx}.
876 @findex I386_DR_LOW_RESET_ADDR
877 @item I386_DR_LOW_RESET_ADDR (@var{idx})
878 Reset (i.e.@: zero out) the address stored in the debug register
881 @findex I386_DR_LOW_GET_STATUS
882 @item I386_DR_LOW_GET_STATUS
883 Return the value of the Debug Status (DR6) register. This value is
884 used immediately after it is returned by
885 @code{I386_DR_LOW_GET_STATUS}, so as to support per-thread status
889 For each one of the 4 debug registers (whose indices are from 0 to 3)
890 that store addresses, a reference count is maintained by @value{GDBN},
891 to allow sharing of debug registers by several watchpoints. This
892 allows users to define several watchpoints that watch the same
893 expression, but with different conditions and/or commands, without
894 wasting debug registers which are in short supply. @value{GDBN}
895 maintains the reference counts internally, targets don't have to do
896 anything to use this feature.
898 The x86 debug registers can each watch a region that is 1, 2, or 4
899 bytes long. The ia32 architecture requires that each watched region
900 be appropriately aligned: 2-byte region on 2-byte boundary, 4-byte
901 region on 4-byte boundary. However, the x86 watchpoint support in
902 @value{GDBN} can watch unaligned regions and regions larger than 4
903 bytes (up to 16 bytes) by allocating several debug registers to watch
904 a single region. This allocation of several registers per a watched
905 region is also done automatically without target code intervention.
907 The generic x86 watchpoint support provides the following API for the
908 @value{GDBN}'s application code:
911 @findex i386_region_ok_for_watchpoint
912 @item i386_region_ok_for_watchpoint (@var{addr}, @var{len})
913 The macro @code{TARGET_REGION_OK_FOR_HW_WATCHPOINT} is set to call
914 this function. It counts the number of debug registers required to
915 watch a given region, and returns a non-zero value if that number is
916 less than 4, the number of debug registers available to x86
919 @findex i386_stopped_data_address
920 @item i386_stopped_data_address (@var{addr_p})
922 @code{target_stopped_data_address} is set to call this function.
924 function examines the breakpoint condition bits in the DR6 Debug
925 Status register, as returned by the @code{I386_DR_LOW_GET_STATUS}
926 macro, and returns the address associated with the first bit that is
929 @findex i386_stopped_by_watchpoint
930 @item i386_stopped_by_watchpoint (void)
931 The macro @code{STOPPED_BY_WATCHPOINT}
932 is set to call this function. The
933 argument passed to @code{STOPPED_BY_WATCHPOINT} is ignored. This
934 function examines the breakpoint condition bits in the DR6 Debug
935 Status register, as returned by the @code{I386_DR_LOW_GET_STATUS}
936 macro, and returns true if any bit is set. Otherwise, false is
939 @findex i386_insert_watchpoint
940 @findex i386_remove_watchpoint
941 @item i386_insert_watchpoint (@var{addr}, @var{len}, @var{type})
942 @itemx i386_remove_watchpoint (@var{addr}, @var{len}, @var{type})
943 Insert or remove a watchpoint. The macros
944 @code{target_insert_watchpoint} and @code{target_remove_watchpoint}
945 are set to call these functions. @code{i386_insert_watchpoint} first
946 looks for a debug register which is already set to watch the same
947 region for the same access types; if found, it just increments the
948 reference count of that debug register, thus implementing debug
949 register sharing between watchpoints. If no such register is found,
950 the function looks for a vacant debug register, sets its mirrored
951 value to @var{addr}, sets the mirrored value of DR7 Debug Control
952 register as appropriate for the @var{len} and @var{type} parameters,
953 and then passes the new values of the debug register and DR7 to the
954 inferior by calling @code{I386_DR_LOW_SET_ADDR} and
955 @code{I386_DR_LOW_SET_CONTROL}. If more than one debug register is
956 required to cover the given region, the above process is repeated for
959 @code{i386_remove_watchpoint} does the opposite: it resets the address
960 in the mirrored value of the debug register and its read/write and
961 length bits in the mirrored value of DR7, then passes these new
962 values to the inferior via @code{I386_DR_LOW_RESET_ADDR} and
963 @code{I386_DR_LOW_SET_CONTROL}. If a register is shared by several
964 watchpoints, each time a @code{i386_remove_watchpoint} is called, it
965 decrements the reference count, and only calls
966 @code{I386_DR_LOW_RESET_ADDR} and @code{I386_DR_LOW_SET_CONTROL} when
967 the count goes to zero.
969 @findex i386_insert_hw_breakpoint
970 @findex i386_remove_hw_breakpoint
971 @item i386_insert_hw_breakpoint (@var{bp_tgt})
972 @itemx i386_remove_hw_breakpoint (@var{bp_tgt})
973 These functions insert and remove hardware-assisted breakpoints. The
974 macros @code{target_insert_hw_breakpoint} and
975 @code{target_remove_hw_breakpoint} are set to call these functions.
976 The argument is a @code{struct bp_target_info *}, as described in
977 the documentation for @code{target_insert_breakpoint}.
978 These functions work like @code{i386_insert_watchpoint} and
979 @code{i386_remove_watchpoint}, respectively, except that they set up
980 the debug registers to watch instruction execution, and each
981 hardware-assisted breakpoint always requires exactly one debug
984 @findex i386_cleanup_dregs
985 @item i386_cleanup_dregs (void)
986 This function clears all the reference counts, addresses, and control
987 bits in the mirror images of the debug registers. It doesn't affect
988 the actual debug registers in the inferior process.
995 x86 processors support setting watchpoints on I/O reads or writes.
996 However, since no target supports this (as of March 2001), and since
997 @code{enum target_hw_bp_type} doesn't even have an enumeration for I/O
998 watchpoints, this feature is not yet available to @value{GDBN} running
1002 x86 processors can enable watchpoints locally, for the current task
1003 only, or globally, for all the tasks. For each debug register,
1004 there's a bit in the DR7 Debug Control register that determines
1005 whether the associated address is watched locally or globally. The
1006 current implementation of x86 watchpoint support in @value{GDBN}
1007 always sets watchpoints to be locally enabled, since global
1008 watchpoints might interfere with the underlying OS and are probably
1009 unavailable in many platforms.
1012 @section Checkpoints
1015 In the abstract, a checkpoint is a point in the execution history of
1016 the program, which the user may wish to return to at some later time.
1018 Internally, a checkpoint is a saved copy of the program state, including
1019 whatever information is required in order to restore the program to that
1020 state at a later time. This can be expected to include the state of
1021 registers and memory, and may include external state such as the state
1022 of open files and devices.
1024 There are a number of ways in which checkpoints may be implemented
1025 in gdb, e.g.@: as corefiles, as forked processes, and as some opaque
1026 method implemented on the target side.
1028 A corefile can be used to save an image of target memory and register
1029 state, which can in principle be restored later --- but corefiles do
1030 not typically include information about external entities such as
1031 open files. Currently this method is not implemented in gdb.
1033 A forked process can save the state of user memory and registers,
1034 as well as some subset of external (kernel) state. This method
1035 is used to implement checkpoints on Linux, and in principle might
1036 be used on other systems.
1038 Some targets, e.g.@: simulators, might have their own built-in
1039 method for saving checkpoints, and gdb might be able to take
1040 advantage of that capability without necessarily knowing any
1041 details of how it is done.
1044 @section Observing changes in @value{GDBN} internals
1045 @cindex observer pattern interface
1046 @cindex notifications about changes in internals
1048 In order to function properly, several modules need to be notified when
1049 some changes occur in the @value{GDBN} internals. Traditionally, these
1050 modules have relied on several paradigms, the most common ones being
1051 hooks and gdb-events. Unfortunately, none of these paradigms was
1052 versatile enough to become the standard notification mechanism in
1053 @value{GDBN}. The fact that they only supported one ``client'' was also
1054 a strong limitation.
1056 A new paradigm, based on the Observer pattern of the @cite{Design
1057 Patterns} book, has therefore been implemented. The goal was to provide
1058 a new interface overcoming the issues with the notification mechanisms
1059 previously available. This new interface needed to be strongly typed,
1060 easy to extend, and versatile enough to be used as the standard
1061 interface when adding new notifications.
1063 See @ref{GDB Observers} for a brief description of the observers
1064 currently implemented in GDB. The rationale for the current
1065 implementation is also briefly discussed.
1067 @node User Interface
1069 @chapter User Interface
1071 @value{GDBN} has several user interfaces, of which the traditional
1072 command-line interface is perhaps the most familiar.
1074 @section Command Interpreter
1076 @cindex command interpreter
1078 The command interpreter in @value{GDBN} is fairly simple. It is designed to
1079 allow for the set of commands to be augmented dynamically, and also
1080 has a recursive subcommand capability, where the first argument to
1081 a command may itself direct a lookup on a different command list.
1083 For instance, the @samp{set} command just starts a lookup on the
1084 @code{setlist} command list, while @samp{set thread} recurses
1085 to the @code{set_thread_cmd_list}.
1089 To add commands in general, use @code{add_cmd}. @code{add_com} adds to
1090 the main command list, and should be used for those commands. The usual
1091 place to add commands is in the @code{_initialize_@var{xyz}} routines at
1092 the ends of most source files.
1094 @findex add_setshow_cmd
1095 @findex add_setshow_cmd_full
1096 To add paired @samp{set} and @samp{show} commands, use
1097 @code{add_setshow_cmd} or @code{add_setshow_cmd_full}. The former is
1098 a slightly simpler interface which is useful when you don't need to
1099 further modify the new command structures, while the latter returns
1100 the new command structures for manipulation.
1102 @cindex deprecating commands
1103 @findex deprecate_cmd
1104 Before removing commands from the command set it is a good idea to
1105 deprecate them for some time. Use @code{deprecate_cmd} on commands or
1106 aliases to set the deprecated flag. @code{deprecate_cmd} takes a
1107 @code{struct cmd_list_element} as it's first argument. You can use the
1108 return value from @code{add_com} or @code{add_cmd} to deprecate the
1109 command immediately after it is created.
1111 The first time a command is used the user will be warned and offered a
1112 replacement (if one exists). Note that the replacement string passed to
1113 @code{deprecate_cmd} should be the full name of the command, i.e., the
1114 entire string the user should type at the command line.
1116 @anchor{UI-Independent Output}
1117 @section UI-Independent Output---the @code{ui_out} Functions
1118 @c This section is based on the documentation written by Fernando
1119 @c Nasser <fnasser@redhat.com>.
1121 @cindex @code{ui_out} functions
1122 The @code{ui_out} functions present an abstraction level for the
1123 @value{GDBN} output code. They hide the specifics of different user
1124 interfaces supported by @value{GDBN}, and thus free the programmer
1125 from the need to write several versions of the same code, one each for
1126 every UI, to produce output.
1128 @subsection Overview and Terminology
1130 In general, execution of each @value{GDBN} command produces some sort
1131 of output, and can even generate an input request.
1133 Output can be generated for the following purposes:
1137 to display a @emph{result} of an operation;
1140 to convey @emph{info} or produce side-effects of a requested
1144 to provide a @emph{notification} of an asynchronous event (including
1145 progress indication of a prolonged asynchronous operation);
1148 to display @emph{error messages} (including warnings);
1151 to show @emph{debug data};
1154 to @emph{query} or prompt a user for input (a special case).
1158 This section mainly concentrates on how to build result output,
1159 although some of it also applies to other kinds of output.
1161 Generation of output that displays the results of an operation
1162 involves one or more of the following:
1166 output of the actual data
1169 formatting the output as appropriate for console output, to make it
1170 easily readable by humans
1173 machine oriented formatting--a more terse formatting to allow for easy
1174 parsing by programs which read @value{GDBN}'s output
1177 annotation, whose purpose is to help legacy GUIs to identify interesting
1181 The @code{ui_out} routines take care of the first three aspects.
1182 Annotations are provided by separate annotation routines. Note that use
1183 of annotations for an interface between a GUI and @value{GDBN} is
1186 Output can be in the form of a single item, which we call a @dfn{field};
1187 a @dfn{list} consisting of identical fields; a @dfn{tuple} consisting of
1188 non-identical fields; or a @dfn{table}, which is a tuple consisting of a
1189 header and a body. In a BNF-like form:
1192 @item <table> @expansion{}
1193 @code{<header> <body>}
1194 @item <header> @expansion{}
1195 @code{@{ <column> @}}
1196 @item <column> @expansion{}
1197 @code{<width> <alignment> <title>}
1198 @item <body> @expansion{}
1203 @subsection General Conventions
1205 Most @code{ui_out} routines are of type @code{void}, the exceptions are
1206 @code{ui_out_stream_new} (which returns a pointer to the newly created
1207 object) and the @code{make_cleanup} routines.
1209 The first parameter is always the @code{ui_out} vector object, a pointer
1210 to a @code{struct ui_out}.
1212 The @var{format} parameter is like in @code{printf} family of functions.
1213 When it is present, there must also be a variable list of arguments
1214 sufficient used to satisfy the @code{%} specifiers in the supplied
1217 When a character string argument is not used in a @code{ui_out} function
1218 call, a @code{NULL} pointer has to be supplied instead.
1221 @subsection Table, Tuple and List Functions
1223 @cindex list output functions
1224 @cindex table output functions
1225 @cindex tuple output functions
1226 This section introduces @code{ui_out} routines for building lists,
1227 tuples and tables. The routines to output the actual data items
1228 (fields) are presented in the next section.
1230 To recap: A @dfn{tuple} is a sequence of @dfn{fields}, each field
1231 containing information about an object; a @dfn{list} is a sequence of
1232 fields where each field describes an identical object.
1234 Use the @dfn{table} functions when your output consists of a list of
1235 rows (tuples) and the console output should include a heading. Use this
1236 even when you are listing just one object but you still want the header.
1238 @cindex nesting level in @code{ui_out} functions
1239 Tables can not be nested. Tuples and lists can be nested up to a
1240 maximum of five levels.
1242 The overall structure of the table output code is something like this:
1257 Here is the description of table-, tuple- and list-related @code{ui_out}
1260 @deftypefun void ui_out_table_begin (struct ui_out *@var{uiout}, int @var{nbrofcols}, int @var{nr_rows}, const char *@var{tblid})
1261 The function @code{ui_out_table_begin} marks the beginning of the output
1262 of a table. It should always be called before any other @code{ui_out}
1263 function for a given table. @var{nbrofcols} is the number of columns in
1264 the table. @var{nr_rows} is the number of rows in the table.
1265 @var{tblid} is an optional string identifying the table. The string
1266 pointed to by @var{tblid} is copied by the implementation of
1267 @code{ui_out_table_begin}, so the application can free the string if it
1268 was @code{malloc}ed.
1270 The companion function @code{ui_out_table_end}, described below, marks
1271 the end of the table's output.
1274 @deftypefun void ui_out_table_header (struct ui_out *@var{uiout}, int @var{width}, enum ui_align @var{alignment}, const char *@var{colhdr})
1275 @code{ui_out_table_header} provides the header information for a single
1276 table column. You call this function several times, one each for every
1277 column of the table, after @code{ui_out_table_begin}, but before
1278 @code{ui_out_table_body}.
1280 The value of @var{width} gives the column width in characters. The
1281 value of @var{alignment} is one of @code{left}, @code{center}, and
1282 @code{right}, and it specifies how to align the header: left-justify,
1283 center, or right-justify it. @var{colhdr} points to a string that
1284 specifies the column header; the implementation copies that string, so
1285 column header strings in @code{malloc}ed storage can be freed after the
1289 @deftypefun void ui_out_table_body (struct ui_out *@var{uiout})
1290 This function delimits the table header from the table body.
1293 @deftypefun void ui_out_table_end (struct ui_out *@var{uiout})
1294 This function signals the end of a table's output. It should be called
1295 after the table body has been produced by the list and field output
1298 There should be exactly one call to @code{ui_out_table_end} for each
1299 call to @code{ui_out_table_begin}, otherwise the @code{ui_out} functions
1300 will signal an internal error.
1303 The output of the tuples that represent the table rows must follow the
1304 call to @code{ui_out_table_body} and precede the call to
1305 @code{ui_out_table_end}. You build a tuple by calling
1306 @code{ui_out_tuple_begin} and @code{ui_out_tuple_end}, with suitable
1307 calls to functions which actually output fields between them.
1309 @deftypefun void ui_out_tuple_begin (struct ui_out *@var{uiout}, const char *@var{id})
1310 This function marks the beginning of a tuple output. @var{id} points
1311 to an optional string that identifies the tuple; it is copied by the
1312 implementation, and so strings in @code{malloc}ed storage can be freed
1316 @deftypefun void ui_out_tuple_end (struct ui_out *@var{uiout})
1317 This function signals an end of a tuple output. There should be exactly
1318 one call to @code{ui_out_tuple_end} for each call to
1319 @code{ui_out_tuple_begin}, otherwise an internal @value{GDBN} error will
1323 @deftypefun {struct cleanup *} make_cleanup_ui_out_tuple_begin_end (struct ui_out *@var{uiout}, const char *@var{id})
1324 This function first opens the tuple and then establishes a cleanup
1325 (@pxref{Misc Guidelines, Cleanups}) to close the tuple.
1326 It provides a convenient and correct implementation of the
1327 non-portable@footnote{The function cast is not portable ISO C.} code sequence:
1329 struct cleanup *old_cleanup;
1330 ui_out_tuple_begin (uiout, "...");
1331 old_cleanup = make_cleanup ((void(*)(void *)) ui_out_tuple_end,
1336 @deftypefun void ui_out_list_begin (struct ui_out *@var{uiout}, const char *@var{id})
1337 This function marks the beginning of a list output. @var{id} points to
1338 an optional string that identifies the list; it is copied by the
1339 implementation, and so strings in @code{malloc}ed storage can be freed
1343 @deftypefun void ui_out_list_end (struct ui_out *@var{uiout})
1344 This function signals an end of a list output. There should be exactly
1345 one call to @code{ui_out_list_end} for each call to
1346 @code{ui_out_list_begin}, otherwise an internal @value{GDBN} error will
1350 @deftypefun {struct cleanup *} make_cleanup_ui_out_list_begin_end (struct ui_out *@var{uiout}, const char *@var{id})
1351 Similar to @code{make_cleanup_ui_out_tuple_begin_end}, this function
1352 opens a list and then establishes cleanup
1353 (@pxref{Misc Guidelines, Cleanups})
1354 that will close the list.
1357 @subsection Item Output Functions
1359 @cindex item output functions
1360 @cindex field output functions
1362 The functions described below produce output for the actual data
1363 items, or fields, which contain information about the object.
1365 Choose the appropriate function accordingly to your particular needs.
1367 @deftypefun void ui_out_field_fmt (struct ui_out *@var{uiout}, char *@var{fldname}, char *@var{format}, ...)
1368 This is the most general output function. It produces the
1369 representation of the data in the variable-length argument list
1370 according to formatting specifications in @var{format}, a
1371 @code{printf}-like format string. The optional argument @var{fldname}
1372 supplies the name of the field. The data items themselves are
1373 supplied as additional arguments after @var{format}.
1375 This generic function should be used only when it is not possible to
1376 use one of the specialized versions (see below).
1379 @deftypefun void ui_out_field_int (struct ui_out *@var{uiout}, const char *@var{fldname}, int @var{value})
1380 This function outputs a value of an @code{int} variable. It uses the
1381 @code{"%d"} output conversion specification. @var{fldname} specifies
1382 the name of the field.
1385 @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})
1386 This function outputs a value of an @code{int} variable. It differs from
1387 @code{ui_out_field_int} in that the caller specifies the desired @var{width} and @var{alignment} of the output.
1388 @var{fldname} specifies
1389 the name of the field.
1392 @deftypefun void ui_out_field_core_addr (struct ui_out *@var{uiout}, const char *@var{fldname}, struct gdbarch *@var{gdbarch}, CORE_ADDR @var{address})
1393 This function outputs an address as appropriate for @var{gdbarch}.
1396 @deftypefun void ui_out_field_string (struct ui_out *@var{uiout}, const char *@var{fldname}, const char *@var{string})
1397 This function outputs a string using the @code{"%s"} conversion
1401 Sometimes, there's a need to compose your output piece by piece using
1402 functions that operate on a stream, such as @code{value_print} or
1403 @code{fprintf_symbol_filtered}. These functions accept an argument of
1404 the type @code{struct ui_file *}, a pointer to a @code{ui_file} object
1405 used to store the data stream used for the output. When you use one
1406 of these functions, you need a way to pass their results stored in a
1407 @code{ui_file} object to the @code{ui_out} functions. To this end,
1408 you first create a @code{ui_stream} object by calling
1409 @code{ui_out_stream_new}, pass the @code{stream} member of that
1410 @code{ui_stream} object to @code{value_print} and similar functions,
1411 and finally call @code{ui_out_field_stream} to output the field you
1412 constructed. When the @code{ui_stream} object is no longer needed,
1413 you should destroy it and free its memory by calling
1414 @code{ui_out_stream_delete}.
1416 @deftypefun {struct ui_stream *} ui_out_stream_new (struct ui_out *@var{uiout})
1417 This function creates a new @code{ui_stream} object which uses the
1418 same output methods as the @code{ui_out} object whose pointer is
1419 passed in @var{uiout}. It returns a pointer to the newly created
1420 @code{ui_stream} object.
1423 @deftypefun void ui_out_stream_delete (struct ui_stream *@var{streambuf})
1424 This functions destroys a @code{ui_stream} object specified by
1428 @deftypefun void ui_out_field_stream (struct ui_out *@var{uiout}, const char *@var{fieldname}, struct ui_stream *@var{streambuf})
1429 This function consumes all the data accumulated in
1430 @code{streambuf->stream} and outputs it like
1431 @code{ui_out_field_string} does. After a call to
1432 @code{ui_out_field_stream}, the accumulated data no longer exists, but
1433 the stream is still valid and may be used for producing more fields.
1436 @strong{Important:} If there is any chance that your code could bail
1437 out before completing output generation and reaching the point where
1438 @code{ui_out_stream_delete} is called, it is necessary to set up a
1439 cleanup, to avoid leaking memory and other resources. Here's a
1440 skeleton code to do that:
1443 struct ui_stream *mybuf = ui_out_stream_new (uiout);
1444 struct cleanup *old = make_cleanup (ui_out_stream_delete, mybuf);
1449 If the function already has the old cleanup chain set (for other kinds
1450 of cleanups), you just have to add your cleanup to it:
1453 mybuf = ui_out_stream_new (uiout);
1454 make_cleanup (ui_out_stream_delete, mybuf);
1457 Note that with cleanups in place, you should not call
1458 @code{ui_out_stream_delete} directly, or you would attempt to free the
1461 @subsection Utility Output Functions
1463 @deftypefun void ui_out_field_skip (struct ui_out *@var{uiout}, const char *@var{fldname})
1464 This function skips a field in a table. Use it if you have to leave
1465 an empty field without disrupting the table alignment. The argument
1466 @var{fldname} specifies a name for the (missing) filed.
1469 @deftypefun void ui_out_text (struct ui_out *@var{uiout}, const char *@var{string})
1470 This function outputs the text in @var{string} in a way that makes it
1471 easy to be read by humans. For example, the console implementation of
1472 this method filters the text through a built-in pager, to prevent it
1473 from scrolling off the visible portion of the screen.
1475 Use this function for printing relatively long chunks of text around
1476 the actual field data: the text it produces is not aligned according
1477 to the table's format. Use @code{ui_out_field_string} to output a
1478 string field, and use @code{ui_out_message}, described below, to
1479 output short messages.
1482 @deftypefun void ui_out_spaces (struct ui_out *@var{uiout}, int @var{nspaces})
1483 This function outputs @var{nspaces} spaces. It is handy to align the
1484 text produced by @code{ui_out_text} with the rest of the table or
1488 @deftypefun void ui_out_message (struct ui_out *@var{uiout}, int @var{verbosity}, const char *@var{format}, ...)
1489 This function produces a formatted message, provided that the current
1490 verbosity level is at least as large as given by @var{verbosity}. The
1491 current verbosity level is specified by the user with the @samp{set
1492 verbositylevel} command.@footnote{As of this writing (April 2001),
1493 setting verbosity level is not yet implemented, and is always returned
1494 as zero. So calling @code{ui_out_message} with a @var{verbosity}
1495 argument more than zero will cause the message to never be printed.}
1498 @deftypefun void ui_out_wrap_hint (struct ui_out *@var{uiout}, char *@var{indent})
1499 This function gives the console output filter (a paging filter) a hint
1500 of where to break lines which are too long. Ignored for all other
1501 output consumers. @var{indent}, if non-@code{NULL}, is the string to
1502 be printed to indent the wrapped text on the next line; it must remain
1503 accessible until the next call to @code{ui_out_wrap_hint}, or until an
1504 explicit newline is produced by one of the other functions. If
1505 @var{indent} is @code{NULL}, the wrapped text will not be indented.
1508 @deftypefun void ui_out_flush (struct ui_out *@var{uiout})
1509 This function flushes whatever output has been accumulated so far, if
1510 the UI buffers output.
1514 @subsection Examples of Use of @code{ui_out} functions
1516 @cindex using @code{ui_out} functions
1517 @cindex @code{ui_out} functions, usage examples
1518 This section gives some practical examples of using the @code{ui_out}
1519 functions to generalize the old console-oriented code in
1520 @value{GDBN}. The examples all come from functions defined on the
1521 @file{breakpoints.c} file.
1523 This example, from the @code{breakpoint_1} function, shows how to
1526 The original code was:
1529 if (!found_a_breakpoint++)
1531 annotate_breakpoints_headers ();
1534 printf_filtered ("Num ");
1536 printf_filtered ("Type ");
1538 printf_filtered ("Disp ");
1540 printf_filtered ("Enb ");
1544 printf_filtered ("Address ");
1547 printf_filtered ("What\n");
1549 annotate_breakpoints_table ();
1553 Here's the new version:
1556 nr_printable_breakpoints = @dots{};
1559 ui_out_table_begin (ui, 6, nr_printable_breakpoints, "BreakpointTable");
1561 ui_out_table_begin (ui, 5, nr_printable_breakpoints, "BreakpointTable");
1563 if (nr_printable_breakpoints > 0)
1564 annotate_breakpoints_headers ();
1565 if (nr_printable_breakpoints > 0)
1567 ui_out_table_header (uiout, 3, ui_left, "number", "Num"); /* 1 */
1568 if (nr_printable_breakpoints > 0)
1570 ui_out_table_header (uiout, 14, ui_left, "type", "Type"); /* 2 */
1571 if (nr_printable_breakpoints > 0)
1573 ui_out_table_header (uiout, 4, ui_left, "disp", "Disp"); /* 3 */
1574 if (nr_printable_breakpoints > 0)
1576 ui_out_table_header (uiout, 3, ui_left, "enabled", "Enb"); /* 4 */
1579 if (nr_printable_breakpoints > 0)
1581 if (print_address_bits <= 32)
1582 ui_out_table_header (uiout, 10, ui_left, "addr", "Address");/* 5 */
1584 ui_out_table_header (uiout, 18, ui_left, "addr", "Address");/* 5 */
1586 if (nr_printable_breakpoints > 0)
1588 ui_out_table_header (uiout, 40, ui_noalign, "what", "What"); /* 6 */
1589 ui_out_table_body (uiout);
1590 if (nr_printable_breakpoints > 0)
1591 annotate_breakpoints_table ();
1594 This example, from the @code{print_one_breakpoint} function, shows how
1595 to produce the actual data for the table whose structure was defined
1596 in the above example. The original code was:
1601 printf_filtered ("%-3d ", b->number);
1603 if ((int)b->type > (sizeof(bptypes)/sizeof(bptypes[0]))
1604 || ((int) b->type != bptypes[(int) b->type].type))
1605 internal_error ("bptypes table does not describe type #%d.",
1607 printf_filtered ("%-14s ", bptypes[(int)b->type].description);
1609 printf_filtered ("%-4s ", bpdisps[(int)b->disposition]);
1611 printf_filtered ("%-3c ", bpenables[(int)b->enable]);
1615 This is the new version:
1619 ui_out_tuple_begin (uiout, "bkpt");
1621 ui_out_field_int (uiout, "number", b->number);
1623 if (((int) b->type > (sizeof (bptypes) / sizeof (bptypes[0])))
1624 || ((int) b->type != bptypes[(int) b->type].type))
1625 internal_error ("bptypes table does not describe type #%d.",
1627 ui_out_field_string (uiout, "type", bptypes[(int)b->type].description);
1629 ui_out_field_string (uiout, "disp", bpdisps[(int)b->disposition]);
1631 ui_out_field_fmt (uiout, "enabled", "%c", bpenables[(int)b->enable]);
1635 This example, also from @code{print_one_breakpoint}, shows how to
1636 produce a complicated output field using the @code{print_expression}
1637 functions which requires a stream to be passed. It also shows how to
1638 automate stream destruction with cleanups. The original code was:
1642 print_expression (b->exp, gdb_stdout);
1648 struct ui_stream *stb = ui_out_stream_new (uiout);
1649 struct cleanup *old_chain = make_cleanup_ui_out_stream_delete (stb);
1652 print_expression (b->exp, stb->stream);
1653 ui_out_field_stream (uiout, "what", local_stream);
1656 This example, also from @code{print_one_breakpoint}, shows how to use
1657 @code{ui_out_text} and @code{ui_out_field_string}. The original code
1662 if (b->dll_pathname == NULL)
1663 printf_filtered ("<any library> ");
1665 printf_filtered ("library \"%s\" ", b->dll_pathname);
1672 if (b->dll_pathname == NULL)
1674 ui_out_field_string (uiout, "what", "<any library>");
1675 ui_out_spaces (uiout, 1);
1679 ui_out_text (uiout, "library \"");
1680 ui_out_field_string (uiout, "what", b->dll_pathname);
1681 ui_out_text (uiout, "\" ");
1685 The following example from @code{print_one_breakpoint} shows how to
1686 use @code{ui_out_field_int} and @code{ui_out_spaces}. The original
1691 if (b->forked_inferior_pid != 0)
1692 printf_filtered ("process %d ", b->forked_inferior_pid);
1699 if (b->forked_inferior_pid != 0)
1701 ui_out_text (uiout, "process ");
1702 ui_out_field_int (uiout, "what", b->forked_inferior_pid);
1703 ui_out_spaces (uiout, 1);
1707 Here's an example of using @code{ui_out_field_string}. The original
1712 if (b->exec_pathname != NULL)
1713 printf_filtered ("program \"%s\" ", b->exec_pathname);
1720 if (b->exec_pathname != NULL)
1722 ui_out_text (uiout, "program \"");
1723 ui_out_field_string (uiout, "what", b->exec_pathname);
1724 ui_out_text (uiout, "\" ");
1728 Finally, here's an example of printing an address. The original code:
1732 printf_filtered ("%s ",
1733 hex_string_custom ((unsigned long) b->address, 8));
1740 ui_out_field_core_addr (uiout, "Address", b->address);
1744 @section Console Printing
1753 @cindex @code{libgdb}
1754 @code{libgdb} 1.0 was an abortive project of years ago. The theory was
1755 to provide an API to @value{GDBN}'s functionality.
1758 @cindex @code{libgdb}
1759 @code{libgdb} 2.0 is an ongoing effort to update @value{GDBN} so that is
1760 better able to support graphical and other environments.
1762 Since @code{libgdb} development is on-going, its architecture is still
1763 evolving. The following components have so far been identified:
1767 Observer - @file{gdb-events.h}.
1769 Builder - @file{ui-out.h}
1771 Event Loop - @file{event-loop.h}
1773 Library - @file{gdb.h}
1776 The model that ties these components together is described below.
1778 @section The @code{libgdb} Model
1780 A client of @code{libgdb} interacts with the library in two ways.
1784 As an observer (using @file{gdb-events}) receiving notifications from
1785 @code{libgdb} of any internal state changes (break point changes, run
1788 As a client querying @code{libgdb} (using the @file{ui-out} builder) to
1789 obtain various status values from @value{GDBN}.
1792 Since @code{libgdb} could have multiple clients (e.g., a GUI supporting
1793 the existing @value{GDBN} CLI), those clients must co-operate when
1794 controlling @code{libgdb}. In particular, a client must ensure that
1795 @code{libgdb} is idle (i.e.@: no other client is using @code{libgdb})
1796 before responding to a @file{gdb-event} by making a query.
1798 @section CLI support
1800 At present @value{GDBN}'s CLI is very much entangled in with the core of
1801 @code{libgdb}. Consequently, a client wishing to include the CLI in
1802 their interface needs to carefully co-ordinate its own and the CLI's
1805 It is suggested that the client set @code{libgdb} up to be bi-modal
1806 (alternate between CLI and client query modes). The notes below sketch
1811 The client registers itself as an observer of @code{libgdb}.
1813 The client create and install @code{cli-out} builder using its own
1814 versions of the @code{ui-file} @code{gdb_stderr}, @code{gdb_stdtarg} and
1815 @code{gdb_stdout} streams.
1817 The client creates a separate custom @code{ui-out} builder that is only
1818 used while making direct queries to @code{libgdb}.
1821 When the client receives input intended for the CLI, it simply passes it
1822 along. Since the @code{cli-out} builder is installed by default, all
1823 the CLI output in response to that command is routed (pronounced rooted)
1824 through to the client controlled @code{gdb_stdout} et.@: al.@: streams.
1825 At the same time, the client is kept abreast of internal changes by
1826 virtue of being a @code{libgdb} observer.
1828 The only restriction on the client is that it must wait until
1829 @code{libgdb} becomes idle before initiating any queries (using the
1830 client's custom builder).
1832 @section @code{libgdb} components
1834 @subheading Observer - @file{gdb-events.h}
1835 @file{gdb-events} provides the client with a very raw mechanism that can
1836 be used to implement an observer. At present it only allows for one
1837 observer and that observer must, internally, handle the need to delay
1838 the processing of any event notifications until after @code{libgdb} has
1839 finished the current command.
1841 @subheading Builder - @file{ui-out.h}
1842 @file{ui-out} provides the infrastructure necessary for a client to
1843 create a builder. That builder is then passed down to @code{libgdb}
1844 when doing any queries.
1846 @subheading Event Loop - @file{event-loop.h}
1847 @c There could be an entire section on the event-loop
1848 @file{event-loop}, currently non-re-entrant, provides a simple event
1849 loop. A client would need to either plug its self into this loop or,
1850 implement a new event-loop that @value{GDBN} would use.
1852 The event-loop will eventually be made re-entrant. This is so that
1853 @value{GDBN} can better handle the problem of some commands blocking
1854 instead of returning.
1856 @subheading Library - @file{gdb.h}
1857 @file{libgdb} is the most obvious component of this system. It provides
1858 the query interface. Each function is parameterized by a @code{ui-out}
1859 builder. The result of the query is constructed using that builder
1860 before the query function returns.
1867 @cindex @code{value} structure
1868 @value{GDBN} uses @code{struct value}, or @dfn{values}, as an internal
1869 abstraction for the representation of a variety of inferior objects
1870 and @value{GDBN} convenience objects.
1872 Values have an associated @code{struct type}, that describes a virtual
1873 view of the raw data or object stored in or accessed through the
1876 A value is in addition discriminated by its lvalue-ness, given its
1877 @code{enum lval_type} enumeration type:
1879 @cindex @code{lval_type} enumeration, for values.
1881 @item @code{not_lval}
1882 This value is not an lval. It can't be assigned to.
1884 @item @code{lval_memory}
1885 This value represents an object in memory.
1887 @item @code{lval_register}
1888 This value represents an object that lives in a register.
1890 @item @code{lval_internalvar}
1891 Represents the value of an internal variable.
1893 @item @code{lval_internalvar_component}
1894 Represents part of a @value{GDBN} internal variable. E.g., a
1897 @cindex computed values
1898 @item @code{lval_computed}
1899 These are ``computed'' values. They allow creating specialized value
1900 objects for specific purposes, all abstracted away from the core value
1901 support code. The creator of such a value writes specialized
1902 functions to handle the reading and writing to/from the value's
1903 backend data, and optionally, a ``copy operator'' and a
1906 Pointers to these functions are stored in a @code{struct lval_funcs}
1907 instance (declared in @file{value.h}), and passed to the
1908 @code{allocate_computed_value} function, as in the example below.
1912 nil_value_read (struct value *v)
1914 /* This callback reads data from some backend, and stores it in V.
1915 In this case, we always read null data. You'll want to fill in
1916 something more interesting. */
1918 memset (value_contents_all_raw (v),
1920 TYPE_LENGTH (value_type (v)));
1924 nil_value_write (struct value *v, struct value *fromval)
1926 /* Takes the data from FROMVAL and stores it in the backend of V. */
1928 to_oblivion (value_contents_all_raw (fromval),
1930 TYPE_LENGTH (value_type (fromval)));
1933 static struct lval_funcs nil_value_funcs =
1940 make_nil_value (void)
1945 type = make_nils_type ();
1946 v = allocate_computed_value (type, &nil_value_funcs, NULL);
1952 See the implementation of the @code{$_siginfo} convenience variable in
1953 @file{infrun.c} as a real example use of lval_computed.
1958 @chapter Stack Frames
1961 @cindex call stack frame
1962 A frame is a construct that @value{GDBN} uses to keep track of calling
1963 and called functions.
1965 @cindex unwind frame
1966 @value{GDBN}'s frame model, a fresh design, was implemented with the
1967 need to support @sc{dwarf}'s Call Frame Information in mind. In fact,
1968 the term ``unwind'' is taken directly from that specification.
1969 Developers wishing to learn more about unwinders, are encouraged to
1970 read the @sc{dwarf} specification, available from
1971 @url{http://www.dwarfstd.org}.
1973 @findex frame_register_unwind
1974 @findex get_frame_register
1975 @value{GDBN}'s model is that you find a frame's registers by
1976 ``unwinding'' them from the next younger frame. That is,
1977 @samp{get_frame_register} which returns the value of a register in
1978 frame #1 (the next-to-youngest frame), is implemented by calling frame
1979 #0's @code{frame_register_unwind} (the youngest frame). But then the
1980 obvious question is: how do you access the registers of the youngest
1983 @cindex sentinel frame
1984 @findex get_frame_type
1985 @vindex SENTINEL_FRAME
1986 To answer this question, @value{GDBN} has the @dfn{sentinel} frame, the
1987 ``-1st'' frame. Unwinding registers from the sentinel frame gives you
1988 the current values of the youngest real frame's registers. If @var{f}
1989 is a sentinel frame, then @code{get_frame_type (@var{f}) @equiv{}
1992 @section Selecting an Unwinder
1994 @findex frame_unwind_prepend_unwinder
1995 @findex frame_unwind_append_unwinder
1996 The architecture registers a list of frame unwinders (@code{struct
1997 frame_unwind}), using the functions
1998 @code{frame_unwind_prepend_unwinder} and
1999 @code{frame_unwind_append_unwinder}. Each unwinder includes a
2000 sniffer. Whenever @value{GDBN} needs to unwind a frame (to fetch the
2001 previous frame's registers or the current frame's ID), it calls
2002 registered sniffers in order to find one which recognizes the frame.
2003 The first time a sniffer returns non-zero, the corresponding unwinder
2004 is assigned to the frame.
2006 @section Unwinding the Frame ID
2009 Every frame has an associated ID, of type @code{struct frame_id}.
2010 The ID includes the stack base and function start address for
2011 the frame. The ID persists through the entire life of the frame,
2012 including while other called frames are running; it is used to
2013 locate an appropriate @code{struct frame_info} from the cache.
2015 Every time the inferior stops, and at various other times, the frame
2016 cache is flushed. Because of this, parts of @value{GDBN} which need
2017 to keep track of individual frames cannot use pointers to @code{struct
2018 frame_info}. A frame ID provides a stable reference to a frame, even
2019 when the unwinder must be run again to generate a new @code{struct
2020 frame_info} for the same frame.
2022 The frame's unwinder's @code{this_id} method is called to find the ID.
2023 Note that this is different from register unwinding, where the next
2024 frame's @code{prev_register} is called to unwind this frame's
2027 Both stack base and function address are required to identify the
2028 frame, because a recursive function has the same function address for
2029 two consecutive frames and a leaf function may have the same stack
2030 address as its caller. On some platforms, a third address is part of
2031 the ID to further disambiguate frames---for instance, on IA-64
2032 the separate register stack address is included in the ID.
2034 An invalid frame ID (@code{outer_frame_id}) returned from the
2035 @code{this_id} method means to stop unwinding after this frame.
2037 @code{null_frame_id} is another invalid frame ID which should be used
2038 when there is no frame. For instance, certain breakpoints are attached
2039 to a specific frame, and that frame is identified through its frame ID
2040 (we use this to implement the "finish" command). Using
2041 @code{null_frame_id} as the frame ID for a given breakpoint means
2042 that the breakpoint is not specific to any frame. The @code{this_id}
2043 method should never return @code{null_frame_id}.
2045 @section Unwinding Registers
2047 Each unwinder includes a @code{prev_register} method. This method
2048 takes a frame, an associated cache pointer, and a register number.
2049 It returns a @code{struct value *} describing the requested register,
2050 as saved by this frame. This is the value of the register that is
2051 current in this frame's caller.
2053 The returned value must have the same type as the register. It may
2054 have any lvalue type. In most circumstances one of these routines
2055 will generate the appropriate value:
2058 @item frame_unwind_got_optimized
2059 @findex frame_unwind_got_optimized
2060 This register was not saved.
2062 @item frame_unwind_got_register
2063 @findex frame_unwind_got_register
2064 This register was copied into another register in this frame. This
2065 is also used for unchanged registers; they are ``copied'' into the
2068 @item frame_unwind_got_memory
2069 @findex frame_unwind_got_memory
2070 This register was saved in memory.
2072 @item frame_unwind_got_constant
2073 @findex frame_unwind_got_constant
2074 This register was not saved, but the unwinder can compute the previous
2075 value some other way.
2077 @item frame_unwind_got_address
2078 @findex frame_unwind_got_address
2079 Same as @code{frame_unwind_got_constant}, except that the value is a target
2080 address. This is frequently used for the stack pointer, which is not
2081 explicitly saved but has a known offset from this frame's stack
2082 pointer. For architectures with a flat unified address space, this is
2083 generally the same as @code{frame_unwind_got_constant}.
2086 @node Symbol Handling
2088 @chapter Symbol Handling
2090 Symbols are a key part of @value{GDBN}'s operation. Symbols include
2091 variables, functions, and types.
2093 Symbol information for a large program can be truly massive, and
2094 reading of symbol information is one of the major performance
2095 bottlenecks in @value{GDBN}; it can take many minutes to process it
2096 all. Studies have shown that nearly all the time spent is
2097 computational, rather than file reading.
2099 One of the ways for @value{GDBN} to provide a good user experience is
2100 to start up quickly, taking no more than a few seconds. It is simply
2101 not possible to process all of a program's debugging info in that
2102 time, and so we attempt to handle symbols incrementally. For instance,
2103 we create @dfn{partial symbol tables} consisting of only selected
2104 symbols, and only expand them to full symbol tables when necessary.
2106 @section Symbol Reading
2108 @cindex symbol reading
2109 @cindex reading of symbols
2110 @cindex symbol files
2111 @value{GDBN} reads symbols from @dfn{symbol files}. The usual symbol
2112 file is the file containing the program which @value{GDBN} is
2113 debugging. @value{GDBN} can be directed to use a different file for
2114 symbols (with the @samp{symbol-file} command), and it can also read
2115 more symbols via the @samp{add-file} and @samp{load} commands. In
2116 addition, it may bring in more symbols while loading shared
2119 @findex find_sym_fns
2120 Symbol files are initially opened by code in @file{symfile.c} using
2121 the BFD library (@pxref{Support Libraries}). BFD identifies the type
2122 of the file by examining its header. @code{find_sym_fns} then uses
2123 this identification to locate a set of symbol-reading functions.
2125 @findex add_symtab_fns
2126 @cindex @code{sym_fns} structure
2127 @cindex adding a symbol-reading module
2128 Symbol-reading modules identify themselves to @value{GDBN} by calling
2129 @code{add_symtab_fns} during their module initialization. The argument
2130 to @code{add_symtab_fns} is a @code{struct sym_fns} which contains the
2131 name (or name prefix) of the symbol format, the length of the prefix,
2132 and pointers to four functions. These functions are called at various
2133 times to process symbol files whose identification matches the specified
2136 The functions supplied by each module are:
2139 @item @var{xyz}_symfile_init(struct sym_fns *sf)
2141 @cindex secondary symbol file
2142 Called from @code{symbol_file_add} when we are about to read a new
2143 symbol file. This function should clean up any internal state (possibly
2144 resulting from half-read previous files, for example) and prepare to
2145 read a new symbol file. Note that the symbol file which we are reading
2146 might be a new ``main'' symbol file, or might be a secondary symbol file
2147 whose symbols are being added to the existing symbol table.
2149 The argument to @code{@var{xyz}_symfile_init} is a newly allocated
2150 @code{struct sym_fns} whose @code{bfd} field contains the BFD for the
2151 new symbol file being read. Its @code{private} field has been zeroed,
2152 and can be modified as desired. Typically, a struct of private
2153 information will be @code{malloc}'d, and a pointer to it will be placed
2154 in the @code{private} field.
2156 There is no result from @code{@var{xyz}_symfile_init}, but it can call
2157 @code{error} if it detects an unavoidable problem.
2159 @item @var{xyz}_new_init()
2161 Called from @code{symbol_file_add} when discarding existing symbols.
2162 This function needs only handle the symbol-reading module's internal
2163 state; the symbol table data structures visible to the rest of
2164 @value{GDBN} will be discarded by @code{symbol_file_add}. It has no
2165 arguments and no result. It may be called after
2166 @code{@var{xyz}_symfile_init}, if a new symbol table is being read, or
2167 may be called alone if all symbols are simply being discarded.
2169 @item @var{xyz}_symfile_read(struct sym_fns *sf, CORE_ADDR addr, int mainline)
2171 Called from @code{symbol_file_add} to actually read the symbols from a
2172 symbol-file into a set of psymtabs or symtabs.
2174 @code{sf} points to the @code{struct sym_fns} originally passed to
2175 @code{@var{xyz}_sym_init} for possible initialization. @code{addr} is
2176 the offset between the file's specified start address and its true
2177 address in memory. @code{mainline} is 1 if this is the main symbol
2178 table being read, and 0 if a secondary symbol file (e.g., shared library
2179 or dynamically loaded file) is being read.@refill
2182 In addition, if a symbol-reading module creates psymtabs when
2183 @var{xyz}_symfile_read is called, these psymtabs will contain a pointer
2184 to a function @code{@var{xyz}_psymtab_to_symtab}, which can be called
2185 from any point in the @value{GDBN} symbol-handling code.
2188 @item @var{xyz}_psymtab_to_symtab (struct partial_symtab *pst)
2190 Called from @code{psymtab_to_symtab} (or the @code{PSYMTAB_TO_SYMTAB} macro) if
2191 the psymtab has not already been read in and had its @code{pst->symtab}
2192 pointer set. The argument is the psymtab to be fleshed-out into a
2193 symtab. Upon return, @code{pst->readin} should have been set to 1, and
2194 @code{pst->symtab} should contain a pointer to the new corresponding symtab, or
2195 zero if there were no symbols in that part of the symbol file.
2198 @section Partial Symbol Tables
2200 @value{GDBN} has three types of symbol tables:
2203 @cindex full symbol table
2206 Full symbol tables (@dfn{symtabs}). These contain the main
2207 information about symbols and addresses.
2211 Partial symbol tables (@dfn{psymtabs}). These contain enough
2212 information to know when to read the corresponding part of the full
2215 @cindex minimal symbol table
2218 Minimal symbol tables (@dfn{msymtabs}). These contain information
2219 gleaned from non-debugging symbols.
2222 @cindex partial symbol table
2223 This section describes partial symbol tables.
2225 A psymtab is constructed by doing a very quick pass over an executable
2226 file's debugging information. Small amounts of information are
2227 extracted---enough to identify which parts of the symbol table will
2228 need to be re-read and fully digested later, when the user needs the
2229 information. The speed of this pass causes @value{GDBN} to start up very
2230 quickly. Later, as the detailed rereading occurs, it occurs in small
2231 pieces, at various times, and the delay therefrom is mostly invisible to
2233 @c (@xref{Symbol Reading}.)
2235 The symbols that show up in a file's psymtab should be, roughly, those
2236 visible to the debugger's user when the program is not running code from
2237 that file. These include external symbols and types, static symbols and
2238 types, and @code{enum} values declared at file scope.
2240 The psymtab also contains the range of instruction addresses that the
2241 full symbol table would represent.
2243 @cindex finding a symbol
2244 @cindex symbol lookup
2245 The idea is that there are only two ways for the user (or much of the
2246 code in the debugger) to reference a symbol:
2249 @findex find_pc_function
2250 @findex find_pc_line
2252 By its address (e.g., execution stops at some address which is inside a
2253 function in this file). The address will be noticed to be in the
2254 range of this psymtab, and the full symtab will be read in.
2255 @code{find_pc_function}, @code{find_pc_line}, and other
2256 @code{find_pc_@dots{}} functions handle this.
2258 @cindex lookup_symbol
2261 (e.g., the user asks to print a variable, or set a breakpoint on a
2262 function). Global names and file-scope names will be found in the
2263 psymtab, which will cause the symtab to be pulled in. Local names will
2264 have to be qualified by a global name, or a file-scope name, in which
2265 case we will have already read in the symtab as we evaluated the
2266 qualifier. Or, a local symbol can be referenced when we are ``in'' a
2267 local scope, in which case the first case applies. @code{lookup_symbol}
2268 does most of the work here.
2271 The only reason that psymtabs exist is to cause a symtab to be read in
2272 at the right moment. Any symbol that can be elided from a psymtab,
2273 while still causing that to happen, should not appear in it. Since
2274 psymtabs don't have the idea of scope, you can't put local symbols in
2275 them anyway. Psymtabs don't have the idea of the type of a symbol,
2276 either, so types need not appear, unless they will be referenced by
2279 It is a bug for @value{GDBN} to behave one way when only a psymtab has
2280 been read, and another way if the corresponding symtab has been read
2281 in. Such bugs are typically caused by a psymtab that does not contain
2282 all the visible symbols, or which has the wrong instruction address
2285 The psymtab for a particular section of a symbol file (objfile) could be
2286 thrown away after the symtab has been read in. The symtab should always
2287 be searched before the psymtab, so the psymtab will never be used (in a
2288 bug-free environment). Currently, psymtabs are allocated on an obstack,
2289 and all the psymbols themselves are allocated in a pair of large arrays
2290 on an obstack, so there is little to be gained by trying to free them
2291 unless you want to do a lot more work.
2293 Whether or not psymtabs are created depends on the objfile's symbol
2294 reader. The core of @value{GDBN} hides the details of partial symbols
2295 and partial symbol tables behind a set of function pointers known as
2296 the @dfn{quick symbol functions}. These are documented in
2301 @unnumberedsubsec Fundamental Types (e.g., @code{FT_VOID}, @code{FT_BOOLEAN}).
2303 @cindex fundamental types
2304 These are the fundamental types that @value{GDBN} uses internally. Fundamental
2305 types from the various debugging formats (stabs, ELF, etc) are mapped
2306 into one of these. They are basically a union of all fundamental types
2307 that @value{GDBN} knows about for all the languages that @value{GDBN}
2310 @unnumberedsubsec Type Codes (e.g., @code{TYPE_CODE_PTR}, @code{TYPE_CODE_ARRAY}).
2313 Each time @value{GDBN} builds an internal type, it marks it with one
2314 of these types. The type may be a fundamental type, such as
2315 @code{TYPE_CODE_INT}, or a derived type, such as @code{TYPE_CODE_PTR}
2316 which is a pointer to another type. Typically, several @code{FT_*}
2317 types map to one @code{TYPE_CODE_*} type, and are distinguished by
2318 other members of the type struct, such as whether the type is signed
2319 or unsigned, and how many bits it uses.
2321 @unnumberedsubsec Builtin Types (e.g., @code{builtin_type_void}, @code{builtin_type_char}).
2323 These are instances of type structs that roughly correspond to
2324 fundamental types and are created as global types for @value{GDBN} to
2325 use for various ugly historical reasons. We eventually want to
2326 eliminate these. Note for example that @code{builtin_type_int}
2327 initialized in @file{gdbtypes.c} is basically the same as a
2328 @code{TYPE_CODE_INT} type that is initialized in @file{c-lang.c} for
2329 an @code{FT_INTEGER} fundamental type. The difference is that the
2330 @code{builtin_type} is not associated with any particular objfile, and
2331 only one instance exists, while @file{c-lang.c} builds as many
2332 @code{TYPE_CODE_INT} types as needed, with each one associated with
2333 some particular objfile.
2335 @section Object File Formats
2336 @cindex object file formats
2340 @cindex @code{a.out} format
2341 The @code{a.out} format is the original file format for Unix. It
2342 consists of three sections: @code{text}, @code{data}, and @code{bss},
2343 which are for program code, initialized data, and uninitialized data,
2346 The @code{a.out} format is so simple that it doesn't have any reserved
2347 place for debugging information. (Hey, the original Unix hackers used
2348 @samp{adb}, which is a machine-language debugger!) The only debugging
2349 format for @code{a.out} is stabs, which is encoded as a set of normal
2350 symbols with distinctive attributes.
2352 The basic @code{a.out} reader is in @file{dbxread.c}.
2357 The COFF format was introduced with System V Release 3 (SVR3) Unix.
2358 COFF files may have multiple sections, each prefixed by a header. The
2359 number of sections is limited.
2361 The COFF specification includes support for debugging. Although this
2362 was a step forward, the debugging information was woefully limited.
2363 For instance, it was not possible to represent code that came from an
2364 included file. GNU's COFF-using configs often use stabs-type info,
2365 encapsulated in special sections.
2367 The COFF reader is in @file{coffread.c}.
2371 @cindex ECOFF format
2372 ECOFF is an extended COFF originally introduced for Mips and Alpha
2375 The basic ECOFF reader is in @file{mipsread.c}.
2379 @cindex XCOFF format
2380 The IBM RS/6000 running AIX uses an object file format called XCOFF.
2381 The COFF sections, symbols, and line numbers are used, but debugging
2382 symbols are @code{dbx}-style stabs whose strings are located in the
2383 @code{.debug} section (rather than the string table). For more
2384 information, see @ref{Top,,,stabs,The Stabs Debugging Format}.
2386 The shared library scheme has a clean interface for figuring out what
2387 shared libraries are in use, but the catch is that everything which
2388 refers to addresses (symbol tables and breakpoints at least) needs to be
2389 relocated for both shared libraries and the main executable. At least
2390 using the standard mechanism this can only be done once the program has
2391 been run (or the core file has been read).
2395 @cindex PE-COFF format
2396 Windows 95 and NT use the PE (@dfn{Portable Executable}) format for their
2397 executables. PE is basically COFF with additional headers.
2399 While BFD includes special PE support, @value{GDBN} needs only the basic
2405 The ELF format came with System V Release 4 (SVR4) Unix. ELF is
2406 similar to COFF in being organized into a number of sections, but it
2407 removes many of COFF's limitations. Debugging info may be either stabs
2408 encapsulated in ELF sections, or more commonly these days, DWARF.
2410 The basic ELF reader is in @file{elfread.c}.
2415 SOM is HP's object file and debug format (not to be confused with IBM's
2416 SOM, which is a cross-language ABI).
2418 The SOM reader is in @file{somread.c}.
2420 @section Debugging File Formats
2422 This section describes characteristics of debugging information that
2423 are independent of the object file format.
2427 @cindex stabs debugging info
2428 @code{stabs} started out as special symbols within the @code{a.out}
2429 format. Since then, it has been encapsulated into other file
2430 formats, such as COFF and ELF.
2432 While @file{dbxread.c} does some of the basic stab processing,
2433 including for encapsulated versions, @file{stabsread.c} does
2438 @cindex COFF debugging info
2439 The basic COFF definition includes debugging information. The level
2440 of support is minimal and non-extensible, and is not often used.
2442 @subsection Mips debug (Third Eye)
2444 @cindex ECOFF debugging info
2445 ECOFF includes a definition of a special debug format.
2447 The file @file{mdebugread.c} implements reading for this format.
2449 @c mention DWARF 1 as a formerly-supported format
2453 @cindex DWARF 2 debugging info
2454 DWARF 2 is an improved but incompatible version of DWARF 1.
2456 The DWARF 2 reader is in @file{dwarf2read.c}.
2458 @subsection Compressed DWARF 2
2460 @cindex Compressed DWARF 2 debugging info
2461 Compressed DWARF 2 is not technically a separate debugging format, but
2462 merely DWARF 2 debug information that has been compressed. In this
2463 format, every object-file section holding DWARF 2 debugging
2464 information is compressed and prepended with a header. (The section
2465 is also typically renamed, so a section called @code{.debug_info} in a
2466 DWARF 2 binary would be called @code{.zdebug_info} in a compressed
2467 DWARF 2 binary.) The header is 12 bytes long:
2471 4 bytes: the literal string ``ZLIB''
2473 8 bytes: the uncompressed size of the section, in big-endian byte
2477 The same reader is used for both compressed an normal DWARF 2 info.
2478 Section decompression is done in @code{zlib_decompress_section} in
2479 @file{dwarf2read.c}.
2483 @cindex DWARF 3 debugging info
2484 DWARF 3 is an improved version of DWARF 2.
2488 @cindex SOM debugging info
2489 Like COFF, the SOM definition includes debugging information.
2491 @section Adding a New Symbol Reader to @value{GDBN}
2493 @cindex adding debugging info reader
2494 If you are using an existing object file format (@code{a.out}, COFF, ELF, etc),
2495 there is probably little to be done.
2497 If you need to add a new object file format, you must first add it to
2498 BFD. This is beyond the scope of this document.
2500 You must then arrange for the BFD code to provide access to the
2501 debugging symbols. Generally @value{GDBN} will have to call swapping
2502 routines from BFD and a few other BFD internal routines to locate the
2503 debugging information. As much as possible, @value{GDBN} should not
2504 depend on the BFD internal data structures.
2506 For some targets (e.g., COFF), there is a special transfer vector used
2507 to call swapping routines, since the external data structures on various
2508 platforms have different sizes and layouts. Specialized routines that
2509 will only ever be implemented by one object file format may be called
2510 directly. This interface should be described in a file
2511 @file{bfd/lib@var{xyz}.h}, which is included by @value{GDBN}.
2513 @section Memory Management for Symbol Files
2515 Most memory associated with a loaded symbol file is stored on
2516 its @code{objfile_obstack}. This includes symbols, types,
2517 namespace data, and other information produced by the symbol readers.
2519 Because this data lives on the objfile's obstack, it is automatically
2520 released when the objfile is unloaded or reloaded. Therefore one
2521 objfile must not reference symbol or type data from another objfile;
2522 they could be unloaded at different times.
2524 User convenience variables, et cetera, have associated types. Normally
2525 these types live in the associated objfile. However, when the objfile
2526 is unloaded, those types are deep copied to global memory, so that
2527 the values of the user variables and history items are not lost.
2530 @node Language Support
2532 @chapter Language Support
2534 @cindex language support
2535 @value{GDBN}'s language support is mainly driven by the symbol reader,
2536 although it is possible for the user to set the source language
2539 @value{GDBN} chooses the source language by looking at the extension
2540 of the file recorded in the debug info; @file{.c} means C, @file{.f}
2541 means Fortran, etc. It may also use a special-purpose language
2542 identifier if the debug format supports it, like with DWARF.
2544 @section Adding a Source Language to @value{GDBN}
2546 @cindex adding source language
2547 To add other languages to @value{GDBN}'s expression parser, follow the
2551 @item Create the expression parser.
2553 @cindex expression parser
2554 This should reside in a file @file{@var{lang}-exp.y}. Routines for
2555 building parsed expressions into a @code{union exp_element} list are in
2558 @cindex language parser
2559 Since we can't depend upon everyone having Bison, and YACC produces
2560 parsers that define a bunch of global names, the following lines
2561 @strong{must} be included at the top of the YACC parser, to prevent the
2562 various parsers from defining the same global names:
2565 #define yyparse @var{lang}_parse
2566 #define yylex @var{lang}_lex
2567 #define yyerror @var{lang}_error
2568 #define yylval @var{lang}_lval
2569 #define yychar @var{lang}_char
2570 #define yydebug @var{lang}_debug
2571 #define yypact @var{lang}_pact
2572 #define yyr1 @var{lang}_r1
2573 #define yyr2 @var{lang}_r2
2574 #define yydef @var{lang}_def
2575 #define yychk @var{lang}_chk
2576 #define yypgo @var{lang}_pgo
2577 #define yyact @var{lang}_act
2578 #define yyexca @var{lang}_exca
2579 #define yyerrflag @var{lang}_errflag
2580 #define yynerrs @var{lang}_nerrs
2583 At the bottom of your parser, define a @code{struct language_defn} and
2584 initialize it with the right values for your language. Define an
2585 @code{initialize_@var{lang}} routine and have it call
2586 @samp{add_language(@var{lang}_language_defn)} to tell the rest of @value{GDBN}
2587 that your language exists. You'll need some other supporting variables
2588 and functions, which will be used via pointers from your
2589 @code{@var{lang}_language_defn}. See the declaration of @code{struct
2590 language_defn} in @file{language.h}, and the other @file{*-exp.y} files,
2591 for more information.
2593 @item Add any evaluation routines, if necessary
2595 @cindex expression evaluation routines
2596 @findex evaluate_subexp
2597 @findex prefixify_subexp
2598 @findex length_of_subexp
2599 If you need new opcodes (that represent the operations of the language),
2600 add them to the enumerated type in @file{expression.h}. Add support
2601 code for these operations in the @code{evaluate_subexp} function
2602 defined in the file @file{eval.c}. Add cases
2603 for new opcodes in two functions from @file{parse.c}:
2604 @code{prefixify_subexp} and @code{length_of_subexp}. These compute
2605 the number of @code{exp_element}s that a given operation takes up.
2607 @item Update some existing code
2609 Add an enumerated identifier for your language to the enumerated type
2610 @code{enum language} in @file{defs.h}.
2612 Update the routines in @file{language.c} so your language is included.
2613 These routines include type predicates and such, which (in some cases)
2614 are language dependent. If your language does not appear in the switch
2615 statement, an error is reported.
2617 @vindex current_language
2618 Also included in @file{language.c} is the code that updates the variable
2619 @code{current_language}, and the routines that translate the
2620 @code{language_@var{lang}} enumerated identifier into a printable
2623 @findex _initialize_language
2624 Update the function @code{_initialize_language} to include your
2625 language. This function picks the default language upon startup, so is
2626 dependent upon which languages that @value{GDBN} is built for.
2628 @findex allocate_symtab
2629 Update @code{allocate_symtab} in @file{symfile.c} and/or symbol-reading
2630 code so that the language of each symtab (source file) is set properly.
2631 This is used to determine the language to use at each stack frame level.
2632 Currently, the language is set based upon the extension of the source
2633 file. If the language can be better inferred from the symbol
2634 information, please set the language of the symtab in the symbol-reading
2637 @findex print_subexp
2638 @findex op_print_tab
2639 Add helper code to @code{print_subexp} (in @file{expprint.c}) to handle any new
2640 expression opcodes you have added to @file{expression.h}. Also, add the
2641 printed representations of your operators to @code{op_print_tab}.
2643 @item Add a place of call
2646 Add a call to @code{@var{lang}_parse()} and @code{@var{lang}_error} in
2647 @code{parse_exp_1} (defined in @file{parse.c}).
2649 @item Edit @file{Makefile.in}
2651 Add dependencies in @file{Makefile.in}. Make sure you update the macro
2652 variables such as @code{HFILES} and @code{OBJS}, otherwise your code may
2653 not get linked in, or, worse yet, it may not get @code{tar}red into the
2658 @node Host Definition
2660 @chapter Host Definition
2662 With the advent of Autoconf, it's rarely necessary to have host
2663 definition machinery anymore. The following information is provided,
2664 mainly, as an historical reference.
2666 @section Adding a New Host
2668 @cindex adding a new host
2669 @cindex host, adding
2670 @value{GDBN}'s host configuration support normally happens via Autoconf.
2671 New host-specific definitions should not be needed. Older hosts
2672 @value{GDBN} still use the host-specific definitions and files listed
2673 below, but these mostly exist for historical reasons, and will
2674 eventually disappear.
2677 @item gdb/config/@var{arch}/@var{xyz}.mh
2678 This file is a Makefile fragment that once contained both host and
2679 native configuration information (@pxref{Native Debugging}) for the
2680 machine @var{xyz}. The host configuration information is now handled
2683 Host configuration information included definitions for @code{CC},
2684 @code{SYSV_DEFINE}, @code{XM_CFLAGS}, @code{XM_ADD_FILES},
2685 @code{XM_CLIBS}, @code{XM_CDEPS}, etc.; see @file{Makefile.in}.
2687 New host-only configurations do not need this file.
2691 (Files named @file{gdb/config/@var{arch}/xm-@var{xyz}.h} were once
2692 used to define host-specific macros, but were no longer needed and
2693 have all been removed.)
2695 @subheading Generic Host Support Files
2697 @cindex generic host support
2698 There are some ``generic'' versions of routines that can be used by
2702 @cindex remote debugging support
2703 @cindex serial line support
2705 This contains serial line support for Unix systems. It is included by
2706 default on all Unix-like hosts.
2709 This contains serial pipe support for Unix systems. It is included by
2710 default on all Unix-like hosts.
2713 This contains serial line support for 32-bit programs running under
2714 Windows using MinGW.
2717 This contains serial line support for 32-bit programs running under DOS,
2718 using the DJGPP (a.k.a.@: GO32) execution environment.
2720 @cindex TCP remote support
2722 This contains generic TCP support using sockets. It is included by
2723 default on all Unix-like hosts and with MinGW.
2726 @section Host Conditionals
2728 When @value{GDBN} is configured and compiled, various macros are
2729 defined or left undefined, to control compilation based on the
2730 attributes of the host system. While formerly they could be set in
2731 host-specific header files, at present they can be changed only by
2732 setting @code{CFLAGS} when building, or by editing the source code.
2734 These macros and their meanings (or if the meaning is not documented
2735 here, then one of the source files where they are used is indicated)
2739 @item @value{GDBN}INIT_FILENAME
2740 The default name of @value{GDBN}'s initialization file (normally
2743 @item SIGWINCH_HANDLER
2744 If your host defines @code{SIGWINCH}, you can define this to be the name
2745 of a function to be called if @code{SIGWINCH} is received.
2747 @item SIGWINCH_HANDLER_BODY
2748 Define this to expand into code that will define the function named by
2749 the expansion of @code{SIGWINCH_HANDLER}.
2751 @item CRLF_SOURCE_FILES
2752 @cindex DOS text files
2753 Define this if host files use @code{\r\n} rather than @code{\n} as a
2754 line terminator. This will cause source file listings to omit @code{\r}
2755 characters when printing and it will allow @code{\r\n} line endings of files
2756 which are ``sourced'' by gdb. It must be possible to open files in binary
2757 mode using @code{O_BINARY} or, for fopen, @code{"rb"}.
2759 @item DEFAULT_PROMPT
2761 The default value of the prompt string (normally @code{"(gdb) "}).
2764 @cindex terminal device
2765 The name of the generic TTY device, defaults to @code{"/dev/tty"}.
2768 Substitute for isatty, if not available.
2771 Define this if binary files are opened the same way as text files.
2773 @item CC_HAS_LONG_LONG
2774 @cindex @code{long long} data type
2775 Define this if the host C compiler supports @code{long long}. This is set
2776 by the @code{configure} script.
2778 @item PRINTF_HAS_LONG_LONG
2779 Define this if the host can handle printing of long long integers via
2780 the printf format conversion specifier @code{ll}. This is set by the
2781 @code{configure} script.
2783 @item LSEEK_NOT_LINEAR
2784 Define this if @code{lseek (n)} does not necessarily move to byte number
2785 @code{n} in the file. This is only used when reading source files. It
2786 is normally faster to define @code{CRLF_SOURCE_FILES} when possible.
2789 Define this to help placate @code{lint} in some situations.
2792 Define this to override the defaults of @code{__volatile__} or
2797 @node Target Architecture Definition
2799 @chapter Target Architecture Definition
2801 @cindex target architecture definition
2802 @value{GDBN}'s target architecture defines what sort of
2803 machine-language programs @value{GDBN} can work with, and how it works
2806 The target architecture object is implemented as the C structure
2807 @code{struct gdbarch *}. The structure, and its methods, are generated
2808 using the Bourne shell script @file{gdbarch.sh}.
2811 * OS ABI Variant Handling::
2812 * Initialize New Architecture::
2813 * Registers and Memory::
2814 * Pointers and Addresses::
2816 * Register Representation::
2817 * Frame Interpretation::
2818 * Inferior Call Setup::
2819 * Adding support for debugging core files::
2820 * Defining Other Architecture Features::
2821 * Adding a New Target::
2824 @node OS ABI Variant Handling
2825 @section Operating System ABI Variant Handling
2826 @cindex OS ABI variants
2828 @value{GDBN} provides a mechanism for handling variations in OS
2829 ABIs. An OS ABI variant may have influence over any number of
2830 variables in the target architecture definition. There are two major
2831 components in the OS ABI mechanism: sniffers and handlers.
2833 A @dfn{sniffer} examines a file matching a BFD architecture/flavour pair
2834 (the architecture may be wildcarded) in an attempt to determine the
2835 OS ABI of that file. Sniffers with a wildcarded architecture are considered
2836 to be @dfn{generic}, while sniffers for a specific architecture are
2837 considered to be @dfn{specific}. A match from a specific sniffer
2838 overrides a match from a generic sniffer. Multiple sniffers for an
2839 architecture/flavour may exist, in order to differentiate between two
2840 different operating systems which use the same basic file format. The
2841 OS ABI framework provides a generic sniffer for ELF-format files which
2842 examines the @code{EI_OSABI} field of the ELF header, as well as note
2843 sections known to be used by several operating systems.
2845 @cindex fine-tuning @code{gdbarch} structure
2846 A @dfn{handler} is used to fine-tune the @code{gdbarch} structure for the
2847 selected OS ABI. There may be only one handler for a given OS ABI
2848 for each BFD architecture.
2850 The following OS ABI variants are defined in @file{defs.h}:
2854 @findex GDB_OSABI_UNINITIALIZED
2855 @item GDB_OSABI_UNINITIALIZED
2856 Used for struct gdbarch_info if ABI is still uninitialized.
2858 @findex GDB_OSABI_UNKNOWN
2859 @item GDB_OSABI_UNKNOWN
2860 The ABI of the inferior is unknown. The default @code{gdbarch}
2861 settings for the architecture will be used.
2863 @findex GDB_OSABI_SVR4
2864 @item GDB_OSABI_SVR4
2865 UNIX System V Release 4.
2867 @findex GDB_OSABI_HURD
2868 @item GDB_OSABI_HURD
2869 GNU using the Hurd kernel.
2871 @findex GDB_OSABI_SOLARIS
2872 @item GDB_OSABI_SOLARIS
2875 @findex GDB_OSABI_OSF1
2876 @item GDB_OSABI_OSF1
2877 OSF/1, including Digital UNIX and Compaq Tru64 UNIX.
2879 @findex GDB_OSABI_LINUX
2880 @item GDB_OSABI_LINUX
2881 GNU using the Linux kernel.
2883 @findex GDB_OSABI_FREEBSD_AOUT
2884 @item GDB_OSABI_FREEBSD_AOUT
2885 FreeBSD using the @code{a.out} executable format.
2887 @findex GDB_OSABI_FREEBSD_ELF
2888 @item GDB_OSABI_FREEBSD_ELF
2889 FreeBSD using the ELF executable format.
2891 @findex GDB_OSABI_NETBSD_AOUT
2892 @item GDB_OSABI_NETBSD_AOUT
2893 NetBSD using the @code{a.out} executable format.
2895 @findex GDB_OSABI_NETBSD_ELF
2896 @item GDB_OSABI_NETBSD_ELF
2897 NetBSD using the ELF executable format.
2899 @findex GDB_OSABI_OPENBSD_ELF
2900 @item GDB_OSABI_OPENBSD_ELF
2901 OpenBSD using the ELF executable format.
2903 @findex GDB_OSABI_WINCE
2904 @item GDB_OSABI_WINCE
2907 @findex GDB_OSABI_GO32
2908 @item GDB_OSABI_GO32
2911 @findex GDB_OSABI_IRIX
2912 @item GDB_OSABI_IRIX
2915 @findex GDB_OSABI_INTERIX
2916 @item GDB_OSABI_INTERIX
2917 Interix (Posix layer for MS-Windows systems).
2919 @findex GDB_OSABI_HPUX_ELF
2920 @item GDB_OSABI_HPUX_ELF
2921 HP/UX using the ELF executable format.
2923 @findex GDB_OSABI_HPUX_SOM
2924 @item GDB_OSABI_HPUX_SOM
2925 HP/UX using the SOM executable format.
2927 @findex GDB_OSABI_QNXNTO
2928 @item GDB_OSABI_QNXNTO
2931 @findex GDB_OSABI_CYGWIN
2932 @item GDB_OSABI_CYGWIN
2935 @findex GDB_OSABI_AIX
2941 Here are the functions that make up the OS ABI framework:
2943 @deftypefun {const char *} gdbarch_osabi_name (enum gdb_osabi @var{osabi})
2944 Return the name of the OS ABI corresponding to @var{osabi}.
2947 @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}))
2948 Register the OS ABI handler specified by @var{init_osabi} for the
2949 architecture, machine type and OS ABI specified by @var{arch},
2950 @var{machine} and @var{osabi}. In most cases, a value of zero for the
2951 machine type, which implies the architecture's default machine type,
2955 @deftypefun void gdbarch_register_osabi_sniffer (enum bfd_architecture @var{arch}, enum bfd_flavour @var{flavour}, enum gdb_osabi (*@var{sniffer})(bfd *@var{abfd}))
2956 Register the OS ABI file sniffer specified by @var{sniffer} for the
2957 BFD architecture/flavour pair specified by @var{arch} and @var{flavour}.
2958 If @var{arch} is @code{bfd_arch_unknown}, the sniffer is considered to
2959 be generic, and is allowed to examine @var{flavour}-flavoured files for
2963 @deftypefun {enum gdb_osabi} gdbarch_lookup_osabi (bfd *@var{abfd})
2964 Examine the file described by @var{abfd} to determine its OS ABI.
2965 The value @code{GDB_OSABI_UNKNOWN} is returned if the OS ABI cannot
2969 @deftypefun void gdbarch_init_osabi (struct gdbarch info @var{info}, struct gdbarch *@var{gdbarch}, enum gdb_osabi @var{osabi})
2970 Invoke the OS ABI handler corresponding to @var{osabi} to fine-tune the
2971 @code{gdbarch} structure specified by @var{gdbarch}. If a handler
2972 corresponding to @var{osabi} has not been registered for @var{gdbarch}'s
2973 architecture, a warning will be issued and the debugging session will continue
2974 with the defaults already established for @var{gdbarch}.
2977 @deftypefun void generic_elf_osabi_sniff_abi_tag_sections (bfd *@var{abfd}, asection *@var{sect}, void *@var{obj})
2978 Helper routine for ELF file sniffers. Examine the file described by
2979 @var{abfd} and look at ABI tag note sections to determine the OS ABI
2980 from the note. This function should be called via
2981 @code{bfd_map_over_sections}.
2984 @node Initialize New Architecture
2985 @section Initializing a New Architecture
2988 * How an Architecture is Represented::
2989 * Looking Up an Existing Architecture::
2990 * Creating a New Architecture::
2993 @node How an Architecture is Represented
2994 @subsection How an Architecture is Represented
2995 @cindex architecture representation
2996 @cindex representation of architecture
2998 Each @code{gdbarch} is associated with a single @sc{bfd} architecture,
2999 via a @code{bfd_arch_@var{arch}} in the @code{bfd_architecture}
3000 enumeration. The @code{gdbarch} is registered by a call to
3001 @code{register_gdbarch_init}, usually from the file's
3002 @code{_initialize_@var{filename}} routine, which will be automatically
3003 called during @value{GDBN} startup. The arguments are a @sc{bfd}
3004 architecture constant and an initialization function.
3006 @findex _initialize_@var{arch}_tdep
3007 @cindex @file{@var{arch}-tdep.c}
3008 A @value{GDBN} description for a new architecture, @var{arch} is created by
3009 defining a global function @code{_initialize_@var{arch}_tdep}, by
3010 convention in the source file @file{@var{arch}-tdep.c}. For example,
3011 in the case of the OpenRISC 1000, this function is called
3012 @code{_initialize_or1k_tdep} and is found in the file
3015 @cindex @file{configure.tgt}
3016 @cindex @code{gdbarch}
3017 @findex gdbarch_register
3018 The resulting object files containing the implementation of the
3019 @code{_initialize_@var{arch}_tdep} function are specified in the @value{GDBN}
3020 @file{configure.tgt} file, which includes a large case statement
3021 pattern matching against the @code{--target} option of the
3022 @code{configure} script. The new @code{struct gdbarch} is created
3023 within the @code{_initialize_@var{arch}_tdep} function by calling
3024 @code{gdbarch_register}:
3027 void gdbarch_register (enum bfd_architecture @var{architecture},
3028 gdbarch_init_ftype *@var{init_func},
3029 gdbarch_dump_tdep_ftype *@var{tdep_dump_func});
3032 The @var{architecture} will identify the unique @sc{bfd} to be
3033 associated with this @code{gdbarch}. The @var{init_func} funciton is
3034 called to create and return the new @code{struct gdbarch}. The
3035 @var{tdep_dump_func} function will dump the target specific details
3036 associated with this architecture.
3038 For example the function @code{_initialize_or1k_tdep} creates its
3039 architecture for 32-bit OpenRISC 1000 architectures by calling:
3042 gdbarch_register (bfd_arch_or32, or1k_gdbarch_init, or1k_dump_tdep);
3045 @node Looking Up an Existing Architecture
3046 @subsection Looking Up an Existing Architecture
3047 @cindex @code{gdbarch} lookup
3049 The initialization function has this prototype:
3052 static struct gdbarch *
3053 @var{arch}_gdbarch_init (struct gdbarch_info @var{info},
3054 struct gdbarch_list *@var{arches})
3057 The @var{info} argument contains parameters used to select the correct
3058 architecture, and @var{arches} is a list of architectures which
3059 have already been created with the same @code{bfd_arch_@var{arch}}
3062 The initialization function should first make sure that @var{info}
3063 is acceptable, and return @code{NULL} if it is not. Then, it should
3064 search through @var{arches} for an exact match to @var{info}, and
3065 return one if found. Lastly, if no exact match was found, it should
3066 create a new architecture based on @var{info} and return it.
3068 @findex gdbarch_list_lookup_by_info
3069 @cindex @code{gdbarch_info}
3070 The lookup is done using @code{gdbarch_list_lookup_by_info}. It is
3071 passed the list of existing architectures, @var{arches}, and the
3072 @code{struct gdbarch_info}, @var{info}, and returns the first matching
3073 architecture it finds, or @code{NULL} if none are found. If an
3074 architecture is found it can be returned as the result from the
3075 initialization function, otherwise a new @code{struct gdbach} will need
3078 The struct gdbarch_info has the following components:
3083 const struct bfd_arch_info *bfd_arch_info;
3086 struct gdbarch_tdep_info *tdep_info;
3087 enum gdb_osabi osabi;
3088 const struct target_desc *target_desc;
3092 @vindex bfd_arch_info
3093 The @code{bfd_arch_info} member holds the key details about the
3094 architecture. The @code{byte_order} member is a value in an
3095 enumeration indicating the endianism. The @code{abfd} member is a
3096 pointer to the full @sc{bfd}, the @code{tdep_info} member is
3097 additional custom target specific information, @code{osabi} identifies
3098 which (if any) of a number of operating specific ABIs are used by this
3099 architecture and the @code{target_desc} member is a set of name-value
3100 pairs with information about register usage in this target.
3102 When the @code{struct gdbarch} initialization function is called, not
3103 all the fields are provided---only those which can be deduced from the
3104 @sc{bfd}. The @code{struct gdbarch_info}, @var{info} is used as a
3105 look-up key with the list of existing architectures, @var{arches} to
3106 see if a suitable architecture already exists. The @var{tdep_info},
3107 @var{osabi} and @var{target_desc} fields may be added before this
3108 lookup to refine the search.
3110 Only information in @var{info} should be used to choose the new
3111 architecture. Historically, @var{info} could be sparse, and
3112 defaults would be collected from the first element on @var{arches}.
3113 However, @value{GDBN} now fills in @var{info} more thoroughly,
3114 so new @code{gdbarch} initialization functions should not take
3115 defaults from @var{arches}.
3117 @node Creating a New Architecture
3118 @subsection Creating a New Architecture
3119 @cindex @code{struct gdbarch} creation
3121 @findex gdbarch_alloc
3122 @cindex @code{gdbarch_tdep} when allocating new @code{gdbarch}
3123 If no architecture is found, then a new architecture must be created,
3124 by calling @code{gdbarch_alloc} using the supplied @code{@w{struct
3125 gdbarch_info}} and any additional custom target specific
3126 information in a @code{struct gdbarch_tdep}. The prototype for
3127 @code{gdbarch_alloc} is:
3130 struct gdbarch *gdbarch_alloc (const struct gdbarch_info *@var{info},
3131 struct gdbarch_tdep *@var{tdep});
3134 @cindex @code{set_gdbarch} functions
3135 @cindex @code{gdbarch} accessor functions
3136 The newly created struct gdbarch must then be populated. Although
3137 there are default values, in most cases they are not what is
3140 For each element, @var{X}, there is are a pair of corresponding accessor
3141 functions, one to set the value of that element,
3142 @code{set_gdbarch_@var{X}}, the second to either get the value of an
3143 element (if it is a variable) or to apply the element (if it is a
3144 function), @code{gdbarch_@var{X}}. Note that both accessor functions
3145 take a pointer to the @code{@w{struct gdbarch}} as first
3146 argument. Populating the new @code{gdbarch} should use the
3147 @code{set_gdbarch} functions.
3149 The following sections identify the main elements that should be set
3150 in this way. This is not the complete list, but represents the
3151 functions and elements that must commonly be specified for a new
3152 architecture. Many of the functions and variables are described in the
3153 header file @file{gdbarch.h}.
3155 This is the main work in defining a new architecture. Implementing the
3156 set of functions to populate the @code{struct gdbarch}.
3158 @cindex @code{gdbarch_tdep} definition
3159 @code{struct gdbarch_tdep} is not defined within @value{GDBN}---it is up
3160 to the user to define this struct if it is needed to hold custom target
3161 information that is not covered by the standard @code{@w{struct
3162 gdbarch}}. For example with the OpenRISC 1000 architecture it is used to
3163 hold the number of matchpoints available in the target (along with other
3166 If there is no additional target specific information, it can be set to
3169 @node Registers and Memory
3170 @section Registers and Memory
3172 @value{GDBN}'s model of the target machine is rather simple.
3173 @value{GDBN} assumes the machine includes a bank of registers and a
3174 block of memory. Each register may have a different size.
3176 @value{GDBN} does not have a magical way to match up with the
3177 compiler's idea of which registers are which; however, it is critical
3178 that they do match up accurately. The only way to make this work is
3179 to get accurate information about the order that the compiler uses,
3180 and to reflect that in the @code{gdbarch_register_name} and related functions.
3182 @value{GDBN} can handle big-endian, little-endian, and bi-endian architectures.
3184 @node Pointers and Addresses
3185 @section Pointers Are Not Always Addresses
3186 @cindex pointer representation
3187 @cindex address representation
3188 @cindex word-addressed machines
3189 @cindex separate data and code address spaces
3190 @cindex spaces, separate data and code address
3191 @cindex address spaces, separate data and code
3192 @cindex code pointers, word-addressed
3193 @cindex converting between pointers and addresses
3194 @cindex D10V addresses
3196 On almost all 32-bit architectures, the representation of a pointer is
3197 indistinguishable from the representation of some fixed-length number
3198 whose value is the byte address of the object pointed to. On such
3199 machines, the words ``pointer'' and ``address'' can be used interchangeably.
3200 However, architectures with smaller word sizes are often cramped for
3201 address space, so they may choose a pointer representation that breaks this
3202 identity, and allows a larger code address space.
3204 @c D10V is gone from sources - more current example?
3206 For example, the Renesas D10V is a 16-bit VLIW processor whose
3207 instructions are 32 bits long@footnote{Some D10V instructions are
3208 actually pairs of 16-bit sub-instructions. However, since you can't
3209 jump into the middle of such a pair, code addresses can only refer to
3210 full 32 bit instructions, which is what matters in this explanation.}.
3211 If the D10V used ordinary byte addresses to refer to code locations,
3212 then the processor would only be able to address 64kb of instructions.
3213 However, since instructions must be aligned on four-byte boundaries, the
3214 low two bits of any valid instruction's byte address are always
3215 zero---byte addresses waste two bits. So instead of byte addresses,
3216 the D10V uses word addresses---byte addresses shifted right two bits---to
3217 refer to code. Thus, the D10V can use 16-bit words to address 256kb of
3220 However, this means that code pointers and data pointers have different
3221 forms on the D10V. The 16-bit word @code{0xC020} refers to byte address
3222 @code{0xC020} when used as a data address, but refers to byte address
3223 @code{0x30080} when used as a code address.
3225 (The D10V also uses separate code and data address spaces, which also
3226 affects the correspondence between pointers and addresses, but we're
3227 going to ignore that here; this example is already too long.)
3229 To cope with architectures like this---the D10V is not the only
3230 one!---@value{GDBN} tries to distinguish between @dfn{addresses}, which are
3231 byte numbers, and @dfn{pointers}, which are the target's representation
3232 of an address of a particular type of data. In the example above,
3233 @code{0xC020} is the pointer, which refers to one of the addresses
3234 @code{0xC020} or @code{0x30080}, depending on the type imposed upon it.
3235 @value{GDBN} provides functions for turning a pointer into an address
3236 and vice versa, in the appropriate way for the current architecture.
3238 Unfortunately, since addresses and pointers are identical on almost all
3239 processors, this distinction tends to bit-rot pretty quickly. Thus,
3240 each time you port @value{GDBN} to an architecture which does
3241 distinguish between pointers and addresses, you'll probably need to
3242 clean up some architecture-independent code.
3244 Here are functions which convert between pointers and addresses:
3246 @deftypefun CORE_ADDR extract_typed_address (void *@var{buf}, struct type *@var{type})
3247 Treat the bytes at @var{buf} as a pointer or reference of type
3248 @var{type}, and return the address it represents, in a manner
3249 appropriate for the current architecture. This yields an address
3250 @value{GDBN} can use to read target memory, disassemble, etc. Note that
3251 @var{buf} refers to a buffer in @value{GDBN}'s memory, not the
3254 For example, if the current architecture is the Intel x86, this function
3255 extracts a little-endian integer of the appropriate length from
3256 @var{buf} and returns it. However, if the current architecture is the
3257 D10V, this function will return a 16-bit integer extracted from
3258 @var{buf}, multiplied by four if @var{type} is a pointer to a function.
3260 If @var{type} is not a pointer or reference type, then this function
3261 will signal an internal error.
3264 @deftypefun CORE_ADDR store_typed_address (void *@var{buf}, struct type *@var{type}, CORE_ADDR @var{addr})
3265 Store the address @var{addr} in @var{buf}, in the proper format for a
3266 pointer of type @var{type} in the current architecture. Note that
3267 @var{buf} refers to a buffer in @value{GDBN}'s memory, not the
3270 For example, if the current architecture is the Intel x86, this function
3271 stores @var{addr} unmodified as a little-endian integer of the
3272 appropriate length in @var{buf}. However, if the current architecture
3273 is the D10V, this function divides @var{addr} by four if @var{type} is
3274 a pointer to a function, and then stores it in @var{buf}.
3276 If @var{type} is not a pointer or reference type, then this function
3277 will signal an internal error.
3280 @deftypefun CORE_ADDR value_as_address (struct value *@var{val})
3281 Assuming that @var{val} is a pointer, return the address it represents,
3282 as appropriate for the current architecture.
3284 This function actually works on integral values, as well as pointers.
3285 For pointers, it performs architecture-specific conversions as
3286 described above for @code{extract_typed_address}.
3289 @deftypefun CORE_ADDR value_from_pointer (struct type *@var{type}, CORE_ADDR @var{addr})
3290 Create and return a value representing a pointer of type @var{type} to
3291 the address @var{addr}, as appropriate for the current architecture.
3292 This function performs architecture-specific conversions as described
3293 above for @code{store_typed_address}.
3296 Here are two functions which architectures can define to indicate the
3297 relationship between pointers and addresses. These have default
3298 definitions, appropriate for architectures on which all pointers are
3299 simple unsigned byte addresses.
3301 @deftypefun CORE_ADDR gdbarch_pointer_to_address (struct gdbarch *@var{gdbarch}, struct type *@var{type}, char *@var{buf})
3302 Assume that @var{buf} holds a pointer of type @var{type}, in the
3303 appropriate format for the current architecture. Return the byte
3304 address the pointer refers to.
3306 This function may safely assume that @var{type} is either a pointer or a
3307 C@t{++} reference type.
3310 @deftypefun void gdbarch_address_to_pointer (struct gdbarch *@var{gdbarch}, struct type *@var{type}, char *@var{buf}, CORE_ADDR @var{addr})
3311 Store in @var{buf} a pointer of type @var{type} representing the address
3312 @var{addr}, in the appropriate format for the current architecture.
3314 This function may safely assume that @var{type} is either a pointer or a
3315 C@t{++} reference type.
3318 @node Address Classes
3319 @section Address Classes
3320 @cindex address classes
3321 @cindex DW_AT_byte_size
3322 @cindex DW_AT_address_class
3324 Sometimes information about different kinds of addresses is available
3325 via the debug information. For example, some programming environments
3326 define addresses of several different sizes. If the debug information
3327 distinguishes these kinds of address classes through either the size
3328 info (e.g, @code{DW_AT_byte_size} in @w{DWARF 2}) or through an explicit
3329 address class attribute (e.g, @code{DW_AT_address_class} in @w{DWARF 2}), the
3330 following macros should be defined in order to disambiguate these
3331 types within @value{GDBN} as well as provide the added information to
3332 a @value{GDBN} user when printing type expressions.
3334 @deftypefun int gdbarch_address_class_type_flags (struct gdbarch *@var{gdbarch}, int @var{byte_size}, int @var{dwarf2_addr_class})
3335 Returns the type flags needed to construct a pointer type whose size
3336 is @var{byte_size} and whose address class is @var{dwarf2_addr_class}.
3337 This function is normally called from within a symbol reader. See
3338 @file{dwarf2read.c}.
3341 @deftypefun {char *} gdbarch_address_class_type_flags_to_name (struct gdbarch *@var{gdbarch}, int @var{type_flags})
3342 Given the type flags representing an address class qualifier, return
3345 @deftypefun int gdbarch_address_class_name_to_type_flags (struct gdbarch *@var{gdbarch}, int @var{name}, int *@var{type_flags_ptr})
3346 Given an address qualifier name, set the @code{int} referenced by @var{type_flags_ptr} to the type flags
3347 for that address class qualifier.
3350 Since the need for address classes is rather rare, none of
3351 the address class functions are defined by default. Predicate
3352 functions are provided to detect when they are defined.
3354 Consider a hypothetical architecture in which addresses are normally
3355 32-bits wide, but 16-bit addresses are also supported. Furthermore,
3356 suppose that the @w{DWARF 2} information for this architecture simply
3357 uses a @code{DW_AT_byte_size} value of 2 to indicate the use of one
3358 of these "short" pointers. The following functions could be defined
3359 to implement the address class functions:
3362 somearch_address_class_type_flags (int byte_size,
3363 int dwarf2_addr_class)
3366 return TYPE_FLAG_ADDRESS_CLASS_1;
3372 somearch_address_class_type_flags_to_name (int type_flags)
3374 if (type_flags & TYPE_FLAG_ADDRESS_CLASS_1)
3381 somearch_address_class_name_to_type_flags (char *name,
3382 int *type_flags_ptr)
3384 if (strcmp (name, "short") == 0)
3386 *type_flags_ptr = TYPE_FLAG_ADDRESS_CLASS_1;
3394 The qualifier @code{@@short} is used in @value{GDBN}'s type expressions
3395 to indicate the presence of one of these ``short'' pointers. For
3396 example if the debug information indicates that @code{short_ptr_var} is
3397 one of these short pointers, @value{GDBN} might show the following
3401 (gdb) ptype short_ptr_var
3402 type = int * @@short
3406 @node Register Representation
3407 @section Register Representation
3410 * Raw and Cooked Registers::
3411 * Register Architecture Functions & Variables::
3412 * Register Information Functions::
3413 * Register and Memory Data::
3414 * Register Caching::
3417 @node Raw and Cooked Registers
3418 @subsection Raw and Cooked Registers
3419 @cindex raw register representation
3420 @cindex cooked register representation
3421 @cindex representations, raw and cooked registers
3423 @value{GDBN} considers registers to be a set with members numbered
3424 linearly from 0 upwards. The first part of that set corresponds to real
3425 physical registers, the second part to any @dfn{pseudo-registers}.
3426 Pseudo-registers have no independent physical existence, but are useful
3427 representations of information within the architecture. For example the
3428 OpenRISC 1000 architecture has up to 32 general purpose registers, which
3429 are typically represented as 32-bit (or 64-bit) integers. However the
3430 GPRs are also used as operands to the floating point operations, and it
3431 could be convenient to define a set of pseudo-registers, to show the
3432 GPRs represented as floating point values.
3434 For any architecture, the implementer will decide on a mapping from
3435 hardware to @value{GDBN} register numbers. The registers corresponding to real
3436 hardware are referred to as @dfn{raw} registers, the remaining registers are
3437 @dfn{pseudo-registers}. The total register set (raw and pseudo) is called
3438 the @dfn{cooked} register set.
3441 @node Register Architecture Functions & Variables
3442 @subsection Functions and Variables Specifying the Register Architecture
3443 @cindex @code{gdbarch} register architecture functions
3445 These @code{struct gdbarch} functions and variables specify the number
3446 and type of registers in the architecture.
3448 @deftypefn {Architecture Function} CORE_ADDR read_pc (struct regcache *@var{regcache})
3450 @deftypefn {Architecture Function} void write_pc (struct regcache *@var{regcache}, CORE_ADDR @var{val})
3452 Read or write the program counter. The default value of both
3453 functions is @code{NULL} (no function available). If the program
3454 counter is just an ordinary register, it can be specified in
3455 @code{struct gdbarch} instead (see @code{pc_regnum} below) and it will
3456 be read or written using the standard routines to access registers. This
3457 function need only be specified if the program counter is not an
3460 Any register information can be obtained using the supplied register
3461 cache, @var{regcache}. @xref{Register Caching, , Register Caching}.
3465 @deftypefn {Architecture Function} void pseudo_register_read (struct gdbarch *@var{gdbarch}, struct regcache *@var{regcache}, int @var{regnum}, const gdb_byte *@var{buf})
3467 @deftypefn {Architecture Function} void pseudo_register_write (struct gdbarch *@var{gdbarch}, struct regcache *@var{regcache}, int @var{regnum}, const gdb_byte *@var{buf})
3469 These functions should be defined if there are any pseudo-registers.
3470 The default value is @code{NULL}. @var{regnum} is the number of the
3471 register to read or write (which will be a @dfn{cooked} register
3472 number) and @var{buf} is the buffer where the value read will be
3473 placed, or from which the value to be written will be taken. The
3474 value in the buffer may be converted to or from a signed or unsigned
3475 integral value using one of the utility functions (@pxref{Register and
3476 Memory Data, , Using Different Register and Memory Data
3479 The access should be for the specified architecture,
3480 @var{gdbarch}. Any register information can be obtained using the
3481 supplied register cache, @var{regcache}. @xref{Register Caching, ,
3486 @deftypevr {Architecture Variable} int sp_regnum
3488 @cindex stack pointer
3491 This specifies the register holding the stack pointer, which may be a
3492 raw or pseudo-register. It defaults to -1 (not defined), but it is an
3493 error for it not to be defined.
3495 The value of the stack pointer register can be accessed withing
3496 @value{GDBN} as the variable @kbd{$sp}.
3500 @deftypevr {Architecture Variable} int pc_regnum
3502 @cindex program counter
3505 This specifies the register holding the program counter, which may be a
3506 raw or pseudo-register. It defaults to -1 (not defined). If
3507 @code{pc_regnum} is not defined, then the functions @code{read_pc} and
3508 @code{write_pc} (see above) must be defined.
3510 The value of the program counter (whether defined as a register, or
3511 through @code{read_pc} and @code{write_pc}) can be accessed withing
3512 @value{GDBN} as the variable @kbd{$pc}.
3516 @deftypevr {Architecture Variable} int ps_regnum
3518 @cindex processor status register
3519 @cindex status register
3522 This specifies the register holding the processor status (often called
3523 the status register), which may be a raw or pseudo-register. It
3524 defaults to -1 (not defined).
3526 If defined, the value of this register can be accessed withing
3527 @value{GDBN} as the variable @kbd{$ps}.
3531 @deftypevr {Architecture Variable} int fp0_regnum
3533 @cindex first floating point register
3535 This specifies the first floating point register. It defaults to
3536 0. @code{fp0_regnum} is not needed unless the target offers support
3541 @node Register Information Functions
3542 @subsection Functions Giving Register Information
3543 @cindex @code{gdbarch} register information functions
3545 These functions return information about registers.
3547 @deftypefn {Architecture Function} {const char *} register_name (struct gdbarch *@var{gdbarch}, int @var{regnum})
3549 This function should convert a register number (raw or pseudo) to a
3550 register name (as a C @code{const char *}). This is used both to
3551 determine the name of a register for output and to work out the meaning
3552 of any register names used as input. The function may also return
3553 @code{NULL}, to indicate that @var{regnum} is not a valid register.
3555 For example with the OpenRISC 1000, @value{GDBN} registers 0-31 are the
3556 General Purpose Registers, register 32 is the program counter and
3557 register 33 is the supervision register (i.e.@: the processor status
3558 register), which map to the strings @code{"gpr00"} through
3559 @code{"gpr31"}, @code{"pc"} and @code{"sr"} respectively. This means
3560 that the @value{GDBN} command @kbd{print $gpr5} should print the value of
3561 the OR1K general purpose register 5@footnote{
3562 @cindex frame pointer
3564 Historically, @value{GDBN} always had a concept of a frame pointer
3565 register, which could be accessed via the @value{GDBN} variable,
3566 @kbd{$fp}. That concept is now deprecated, recognizing that not all
3567 architectures have a frame pointer. However if an architecture does
3568 have a frame pointer register, and defines a register or
3569 pseudo-register with the name @code{"fp"}, then that register will be
3570 used as the value of the @kbd{$fp} variable.}.
3572 The default value for this function is @code{NULL}, meaning
3573 undefined. It should always be defined.
3575 The access should be for the specified architecture, @var{gdbarch}.
3579 @deftypefn {Architecture Function} {struct type *} register_type (struct gdbarch *@var{gdbarch}, int @var{regnum})
3581 Given a register number, this function identifies the type of data it
3582 may be holding, specified as a @code{struct type}. @value{GDBN} allows
3583 creation of arbitrary types, but a number of built in types are
3584 provided (@code{builtin_type_void}, @code{builtin_type_int32} etc),
3585 together with functions to derive types from these.
3587 Typically the program counter will have a type of ``pointer to
3588 function'' (it points to code), the frame pointer and stack pointer
3589 will have types of ``pointer to void'' (they point to data on the stack)
3590 and all other integer registers will have a type of 32-bit integer or
3593 This information guides the formatting when displaying register
3594 information. The default value is @code{NULL} meaning no information is
3595 available to guide formatting when displaying registers.
3599 @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})
3601 Define this function to print out one or all of the registers for the
3602 @value{GDBN} @kbd{info registers} command. The default value is the
3603 function @code{default_print_registers_info}, which uses the register
3604 type information (see @code{register_type} above) to determine how each
3605 register should be printed. Define a custom version of this function
3606 for fuller control over how the registers are displayed.
3608 The access should be for the specified architecture, @var{gdbarch},
3609 with output to the file specified by the User Interface
3610 Independent Output file handle, @var{file} (@pxref{UI-Independent
3611 Output, , UI-Independent Output---the @code{ui_out}
3614 The registers should show their values in the frame specified by
3615 @var{frame}. If @var{regnum} is -1 and @var{all} is zero, then all
3616 the ``significant'' registers should be shown (the implementer should
3617 decide which registers are ``significant''). Otherwise only the value of
3618 the register specified by @var{regnum} should be output. If
3619 @var{regnum} is -1 and @var{all} is non-zero (true), then the value of
3620 all registers should be shown.
3622 By default @code{default_print_registers_info} prints one register per
3623 line, and if @var{all} is zero omits floating-point registers.
3627 @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})
3629 Define this function to provide output about the floating point unit and
3630 registers for the @value{GDBN} @kbd{info float} command respectively.
3631 The default value is @code{NULL} (not defined), meaning no information
3634 The @var{gdbarch} and @var{file} and @var{frame} arguments have the same
3635 meaning as in the @code{print_registers_info} function above. The string
3636 @var{args} contains any supplementary arguments to the @kbd{info float}
3639 Define this function if the target supports floating point operations.
3643 @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})
3645 Define this function to provide output about the vector unit and
3646 registers for the @value{GDBN} @kbd{info vector} command respectively.
3647 The default value is @code{NULL} (not defined), meaning no information
3650 The @var{gdbarch}, @var{file} and @var{frame} arguments have the
3651 same meaning as in the @code{print_registers_info} function above. The
3652 string @var{args} contains any supplementary arguments to the @kbd{info
3655 Define this function if the target supports vector operations.
3659 @deftypefn {Architecture Function} int register_reggroup_p (struct gdbarch *@var{gdbarch}, int @var{regnum}, struct reggroup *@var{group})
3661 @value{GDBN} groups registers into different categories (general,
3662 vector, floating point etc). This function, given a register,
3663 @var{regnum}, and group, @var{group}, returns 1 (true) if the register
3664 is in the group and 0 (false) otherwise.
3666 The information should be for the specified architecture,
3669 The default value is the function @code{default_register_reggroup_p}
3670 which will do a reasonable job based on the type of the register (see
3671 the function @code{register_type} above), with groups for general
3672 purpose registers, floating point registers, vector registers and raw
3673 (i.e not pseudo) registers.
3677 @node Register and Memory Data
3678 @subsection Using Different Register and Memory Data Representations
3679 @cindex register representation
3680 @cindex memory representation
3681 @cindex representations, register and memory
3682 @cindex register data formats, converting
3683 @cindex @code{struct value}, converting register contents to
3685 Some architectures have different representations of data objects,
3686 depending whether the object is held in a register or memory. For
3692 The Alpha architecture can represent 32 bit integer values in
3693 floating-point registers.
3696 The x86 architecture supports 80-bit floating-point registers. The
3697 @code{long double} data type occupies 96 bits in memory but only 80
3698 bits when stored in a register.
3702 In general, the register representation of a data type is determined by
3703 the architecture, or @value{GDBN}'s interface to the architecture, while
3704 the memory representation is determined by the Application Binary
3707 For almost all data types on almost all architectures, the two
3708 representations are identical, and no special handling is needed.
3709 However, they do occasionally differ. An architecture may define the
3710 following @code{struct gdbarch} functions to request conversions
3711 between the register and memory representations of a data type:
3713 @deftypefn {Architecture Function} int gdbarch_convert_register_p (struct gdbarch *@var{gdbarch}, int @var{reg})
3715 Return non-zero (true) if the representation of a data value stored in
3716 this register may be different to the representation of that same data
3717 value when stored in memory. The default value is @code{NULL}
3720 If this function is defined and returns non-zero, the @code{struct
3721 gdbarch} functions @code{gdbarch_register_to_value} and
3722 @code{gdbarch_value_to_register} (see below) should be used to perform
3723 any necessary conversion.
3725 If defined, this function should return zero for the register's native
3726 type, when no conversion is necessary.
3729 @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})
3731 Convert the value of register number @var{reg} to a data object of
3732 type @var{type}. The buffer at @var{from} holds the register's value
3733 in raw format; the converted value should be placed in the buffer at
3737 @emph{Note:} @code{gdbarch_register_to_value} and
3738 @code{gdbarch_value_to_register} take their @var{reg} and @var{type}
3739 arguments in different orders.
3742 @code{gdbarch_register_to_value} should only be used with registers
3743 for which the @code{gdbarch_convert_register_p} function returns a
3748 @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})
3750 Convert a data value of type @var{type} to register number @var{reg}'
3754 @emph{Note:} @code{gdbarch_register_to_value} and
3755 @code{gdbarch_value_to_register} take their @var{reg} and @var{type}
3756 arguments in different orders.
3759 @code{gdbarch_value_to_register} should only be used with registers
3760 for which the @code{gdbarch_convert_register_p} function returns a
3765 @node Register Caching
3766 @subsection Register Caching
3767 @cindex register caching
3769 Caching of registers is used, so that the target does not need to be
3770 accessed and reanalyzed multiple times for each register in
3771 circumstances where the register value cannot have changed.
3773 @cindex @code{struct regcache}
3774 @value{GDBN} provides @code{struct regcache}, associated with a
3775 particular @code{struct gdbarch} to hold the cached values of the raw
3776 registers. A set of functions is provided to access both the raw
3777 registers (with @code{raw} in their name) and the full set of cooked
3778 registers (with @code{cooked} in their name). Functions are provided
3779 to ensure the register cache is kept synchronized with the values of
3780 the actual registers in the target.
3782 Accessing registers through the @code{struct regcache} routines will
3783 ensure that the appropriate @code{struct gdbarch} functions are called
3784 when necessary to access the underlying target architecture. In general
3785 users should use the @dfn{cooked} functions, since these will map to the
3786 @dfn{raw} functions automatically as appropriate.
3788 @findex regcache_cooked_read
3789 @findex regcache_cooked_write
3790 @cindex @code{gdb_byte}
3791 @findex regcache_cooked_read_signed
3792 @findex regcache_cooked_read_unsigned
3793 @findex regcache_cooked_write_signed
3794 @findex regcache_cooked_write_unsigned
3795 The two key functions are @code{regcache_cooked_read} and
3796 @code{regcache_cooked_write} which read or write a register from or to
3797 a byte buffer (type @code{gdb_byte *}). For convenience the wrapper
3798 functions @code{regcache_cooked_read_signed},
3799 @code{regcache_cooked_read_unsigned},
3800 @code{regcache_cooked_write_signed} and
3801 @code{regcache_cooked_write_unsigned} are provided, which read or
3802 write the value using the buffer and convert to or from an integral
3803 value as appropriate.
3805 @node Frame Interpretation
3806 @section Frame Interpretation
3809 * All About Stack Frames::
3810 * Frame Handling Terminology::
3812 * Functions and Variable to Analyze Frames::
3813 * Functions to Access Frame Data::
3814 * Analyzing Stacks---Frame Sniffers::
3817 @node All About Stack Frames
3818 @subsection All About Stack Frames
3820 @value{GDBN} needs to understand the stack on which local (automatic)
3821 variables are stored. The area of the stack containing all the local
3822 variables for a function invocation is known as the @dfn{stack frame}
3823 for that function (or colloquially just as the @dfn{frame}). In turn the
3824 function that called the function will have its stack frame, and so on
3825 back through the chain of functions that have been called.
3827 Almost all architectures have one register dedicated to point to the
3828 end of the stack (the @dfn{stack pointer}). Many have a second register
3829 which points to the start of the currently active stack frame (the
3830 @dfn{frame pointer}). The specific arrangements for an architecture are
3831 a key part of the ABI.
3833 A diagram helps to explain this. Here is a simple program to compute
3846 return n * fact (n - 1);
3854 for (i = 0; i < 10; i++)
3857 printf ("%d! = %d\n", i, f);
3862 Consider the state of the stack when the code reaches line 6 after the
3863 main program has called @code{fact@w{ }(3)}. The chain of function
3864 calls will be @code{main ()}, @code{fact@w{ }(3)}, @code{fact@w{
3865 }(2)}, @code{@w{fact (1)}} and @code{fact@w{ }(0)}.
3867 In this illustration the stack is falling (as used for example by the
3868 OpenRISC 1000 ABI). The stack pointer (SP) is at the end of the stack
3869 (lowest address) and the frame pointer (FP) is at the highest address
3870 in the current stack frame. The following diagram shows how the stack
3873 @center @image{stack_frame,14cm}
3875 In each stack frame, offset 0 from the stack pointer is the frame
3876 pointer of the previous frame and offset 4 (this is illustrating a
3877 32-bit architecture) from the stack pointer is the return address.
3878 Local variables are indexed from the frame pointer, with negative
3879 indexes. In the function @code{fact}, offset -4 from the frame
3880 pointer is the argument @var{n}. In the @code{main} function, offset
3881 -4 from the frame pointer is the local variable @var{i} and offset -8
3882 from the frame pointer is the local variable @var{f}@footnote{This is
3883 a simplified example for illustrative purposes only. Good optimizing
3884 compilers would not put anything on the stack for such simple
3885 functions. Indeed they might eliminate the recursion and use of the
3888 It is very easy to get confused when examining stacks. @value{GDBN}
3889 has terminology it uses rigorously throughout. The stack frame of the
3890 function currently executing, or where execution stopped is numbered
3891 zero. In this example frame #0 is the stack frame of the call to
3892 @code{fact@w{ }(0)}. The stack frame of its calling function
3893 (@code{fact@w{ }(1)} in this case) is numbered #1 and so on back
3894 through the chain of calls.
3896 The main @value{GDBN} data structure describing frames is
3897 @code{@w{struct frame_info}}. It is not used directly, but only via
3898 its accessor functions. @code{frame_info} includes information about
3899 the registers in the frame and a pointer to the code of the function
3900 with which the frame is associated. The entire stack is represented as
3901 a linked list of @code{frame_info} structs.
3903 @node Frame Handling Terminology
3904 @subsection Frame Handling Terminology
3906 It is easy to get confused when referencing stack frames. @value{GDBN}
3907 uses some precise terminology.
3913 @cindex stack frame, definition of THIS frame
3914 @cindex frame, definition of THIS frame
3915 @dfn{THIS} frame is the frame currently under consideration.
3919 @cindex stack frame, definition of NEXT frame
3920 @cindex frame, definition of NEXT frame
3921 The @dfn{NEXT} frame, also sometimes called the inner or newer frame is the
3922 frame of the function called by the function of THIS frame.
3925 @cindex PREVIOUS frame
3926 @cindex stack frame, definition of PREVIOUS frame
3927 @cindex frame, definition of PREVIOUS frame
3928 The @dfn{PREVIOUS} frame, also sometimes called the outer or older frame is
3929 the frame of the function which called the function of THIS frame.
3933 So in the example in the previous section (@pxref{All About Stack
3934 Frames, , All About Stack Frames}), if THIS frame is #3 (the call to
3935 @code{fact@w{ }(3)}), the NEXT frame is frame #2 (the call to
3936 @code{fact@w{ }(2)}) and the PREVIOUS frame is frame #4 (the call to
3937 @code{main@w{ }()}).
3939 @cindex innermost frame
3940 @cindex stack frame, definition of innermost frame
3941 @cindex frame, definition of innermost frame
3942 The @dfn{innermost} frame is the frame of the current executing
3943 function, or where the program stopped, in this example, in the middle
3944 of the call to @code{@w{fact (0))}}. It is always numbered frame #0.
3946 @cindex base of a frame
3947 @cindex stack frame, definition of base of a frame
3948 @cindex frame, definition of base of a frame
3949 The @dfn{base} of a frame is the address immediately before the start
3950 of the NEXT frame. For a stack which grows down in memory (a
3951 @dfn{falling} stack) this will be the lowest address and for a stack
3952 which grows up in memory (a @dfn{rising} stack) this will be the
3953 highest address in the frame.
3955 @value{GDBN} functions to analyze the stack are typically given a
3956 pointer to the NEXT frame to determine information about THIS
3957 frame. Information about THIS frame includes data on where the
3958 registers of the PREVIOUS frame are stored in this stack frame. In
3959 this example the frame pointer of the PREVIOUS frame is stored at
3960 offset 0 from the stack pointer of THIS frame.
3963 @cindex stack frame, definition of unwinding
3964 @cindex frame, definition of unwinding
3965 The process whereby a function is given a pointer to the NEXT
3966 frame to work out information about THIS frame is referred to as
3967 @dfn{unwinding}. The @value{GDBN} functions involved in this typically
3968 include unwind in their name.
3971 @cindex stack frame, definition of sniffing
3972 @cindex frame, definition of sniffing
3973 The process of analyzing a target to determine the information that
3974 should go in struct frame_info is called @dfn{sniffing}. The functions
3975 that carry this out are called sniffers and typically include sniffer
3976 in their name. More than one sniffer may be required to extract all
3977 the information for a particular frame.
3979 @cindex sentinel frame
3980 @cindex stack frame, definition of sentinel frame
3981 @cindex frame, definition of sentinel frame
3982 Because so many functions work using the NEXT frame, there is an issue
3983 about addressing the innermost frame---it has no NEXT frame. To solve
3984 this @value{GDBN} creates a dummy frame #-1, known as the
3985 @dfn{sentinel} frame.
3987 @node Prologue Caches
3988 @subsection Prologue Caches
3990 @cindex function prologue
3991 @cindex prologue of a function
3992 All the frame sniffing functions typically examine the code at the
3993 start of the corresponding function, to determine the state of
3994 registers. The ABI will save old values and set new values of key
3995 registers at the start of each function in what is known as the
3996 function @dfn{prologue}.
3998 @cindex prologue cache
3999 For any particular stack frame this data does not change, so all the
4000 standard unwinding functions, in addition to receiving a pointer to
4001 the NEXT frame as their first argument, receive a pointer to a
4002 @dfn{prologue cache} as their second argument. This can be used to store
4003 values associated with a particular frame, for reuse on subsequent
4004 calls involving the same frame.
4006 It is up to the user to define the structure used (it is a
4007 @code{void@w{ }*} pointer) and arrange allocation and deallocation of
4008 storage. However for general use, @value{GDBN} provides
4009 @code{@w{struct trad_frame_cache}}, with a set of accessor
4010 routines. This structure holds the stack and code address of
4011 THIS frame, the base address of the frame, a pointer to the
4012 struct @code{frame_info} for the NEXT frame and details of
4013 where the registers of the PREVIOUS frame may be found in THIS
4016 Typically the first time any sniffer function is called with NEXT
4017 frame, the prologue sniffer for THIS frame will be @code{NULL}. The
4018 sniffer will analyze the frame, allocate a prologue cache structure
4019 and populate it. Subsequent calls using the same NEXT frame will
4020 pass in this prologue cache, so the data can be returned with no
4021 additional analysis.
4023 @node Functions and Variable to Analyze Frames
4024 @subsection Functions and Variable to Analyze Frames
4026 These struct @code{gdbarch} functions and variable should be defined
4027 to provide analysis of the stack frame and allow it to be adjusted as
4030 @deftypefn {Architecture Function} CORE_ADDR skip_prologue (struct gdbarch *@var{gdbarch}, CORE_ADDR @var{pc})
4032 The prologue of a function is the code at the beginning of the
4033 function which sets up the stack frame, saves the return address
4034 etc. The code representing the behavior of the function starts after
4037 This function skips past the prologue of a function if the program
4038 counter, @var{pc}, is within the prologue of a function. The result is
4039 the program counter immediately after the prologue. With modern
4040 optimizing compilers, this may be a far from trivial exercise. However
4041 the required information may be within the binary as DWARF2 debugging
4042 information, making the job much easier.
4044 The default value is @code{NULL} (not defined). This function should always
4045 be provided, but can take advantage of DWARF2 debugging information,
4046 if that is available.
4050 @deftypefn {Architecture Function} int inner_than (CORE_ADDR @var{lhs}, CORE_ADDR @var{rhs})
4051 @findex core_addr_lessthan
4052 @findex core_addr_greaterthan
4054 Given two frame or stack pointers, return non-zero (true) if the first
4055 represents the @dfn{inner} stack frame and 0 (false) otherwise. This
4056 is used to determine whether the target has a stack which grows up in
4057 memory (rising stack) or grows down in memory (falling stack).
4058 @xref{All About Stack Frames, , All About Stack Frames}, for an
4059 explanation of @dfn{inner} frames.
4061 The default value of this function is @code{NULL} and it should always
4062 be defined. However for almost all architectures one of the built-in
4063 functions can be used: @code{core_addr_lessthan} (for stacks growing
4064 down in memory) or @code{core_addr_greaterthan} (for stacks growing up
4069 @anchor{frame_align}
4070 @deftypefn {Architecture Function} CORE_ADDR frame_align (struct gdbarch *@var{gdbarch}, CORE_ADDR @var{address})
4074 The architecture may have constraints on how its frames are
4075 aligned. For example the OpenRISC 1000 ABI requires stack frames to be
4076 double-word aligned, but 32-bit versions of the architecture allocate
4077 single-word values to the stack. Thus extra padding may be needed at
4078 the end of a stack frame.
4080 Given a proposed address for the stack pointer, this function
4081 returns a suitably aligned address (by expanding the stack frame).
4083 The default value is @code{NULL} (undefined). This function should be defined
4084 for any architecture where it is possible the stack could become
4085 misaligned. The utility functions @code{align_down} (for falling
4086 stacks) and @code{align_up} (for rising stacks) will facilitate the
4087 implementation of this function.
4091 @deftypevr {Architecture Variable} int frame_red_zone_size
4093 Some ABIs reserve space beyond the end of the stack for use by leaf
4094 functions without prologue or epilogue or by exception handlers (for
4095 example the OpenRISC 1000).
4097 This is known as a @dfn{red zone} (AMD terminology). The @sc{amd64}
4098 (nee x86-64) ABI documentation refers to the @dfn{red zone} when
4099 describing this scratch area.
4101 The default value is 0. Set this field if the architecture has such a
4102 red zone. The value must be aligned as required by the ABI (see
4103 @code{frame_align} above for an explanation of stack frame alignment).
4107 @node Functions to Access Frame Data
4108 @subsection Functions to Access Frame Data
4110 These functions provide access to key registers and arguments in the
4113 @deftypefn {Architecture Function} CORE_ADDR unwind_pc (struct gdbarch *@var{gdbarch}, struct frame_info *@var{next_frame})
4115 This function is given a pointer to the NEXT stack frame (@pxref{All
4116 About Stack Frames, , All About Stack Frames}, for how frames are
4117 represented) and returns the value of the program counter in the
4118 PREVIOUS frame (i.e.@: the frame of the function that called THIS
4119 one). This is commonly referred to as the @dfn{return address}.
4121 The implementation, which must be frame agnostic (work with any frame),
4122 is typically no more than:
4126 pc = frame_unwind_register_unsigned (next_frame, @var{ARCH}_PC_REGNUM);
4127 return gdbarch_addr_bits_remove (gdbarch, pc);
4132 @deftypefn {Architecture Function} CORE_ADDR unwind_sp (struct gdbarch *@var{gdbarch}, struct frame_info *@var{next_frame})
4134 This function is given a pointer to the NEXT stack frame
4135 (@pxref{All About Stack Frames, , All About Stack Frames} for how
4136 frames are represented) and returns the value of the stack pointer in
4137 the PREVIOUS frame (i.e.@: the frame of the function that called
4140 The implementation, which must be frame agnostic (work with any frame),
4141 is typically no more than:
4145 sp = frame_unwind_register_unsigned (next_frame, @var{ARCH}_SP_REGNUM);
4146 return gdbarch_addr_bits_remove (gdbarch, sp);
4151 @deftypefn {Architecture Function} int frame_num_args (struct gdbarch *@var{gdbarch}, struct frame_info *@var{this_frame})
4153 This function is given a pointer to THIS stack frame (@pxref{All
4154 About Stack Frames, , All About Stack Frames} for how frames are
4155 represented), and returns the number of arguments that are being
4156 passed, or -1 if not known.
4158 The default value is @code{NULL} (undefined), in which case the number of
4159 arguments passed on any stack frame is always unknown. For many
4160 architectures this will be a suitable default.
4164 @node Analyzing Stacks---Frame Sniffers
4165 @subsection Analyzing Stacks---Frame Sniffers
4167 When a program stops, @value{GDBN} needs to construct the chain of
4168 struct @code{frame_info} representing the state of the stack using
4169 appropriate @dfn{sniffers}.
4171 Each architecture requires appropriate sniffers, but they do not form
4172 entries in @code{@w{struct gdbarch}}, since more than one sniffer may
4173 be required and a sniffer may be suitable for more than one
4174 @code{@w{struct gdbarch}}. Instead sniffers are associated with
4175 architectures using the following functions.
4180 @findex frame_unwind_append_sniffer
4181 @code{frame_unwind_append_sniffer} is used to add a new sniffer to
4182 analyze THIS frame when given a pointer to the NEXT frame.
4185 @findex frame_base_append_sniffer
4186 @code{frame_base_append_sniffer} is used to add a new sniffer
4187 which can determine information about the base of a stack frame.
4190 @findex frame_base_set_default
4191 @code{frame_base_set_default} is used to specify the default base
4196 These functions all take a reference to @code{@w{struct gdbarch}}, so
4197 they are associated with a specific architecture. They are usually
4198 called in the @code{gdbarch} initialization function, after the
4199 @code{gdbarch} struct has been set up. Unless a default has been set, the
4200 most recently appended sniffer will be tried first.
4202 The main frame unwinding sniffer (as set by
4203 @code{frame_unwind_append_sniffer)} returns a structure specifying
4204 a set of sniffing functions:
4206 @cindex @code{frame_unwind}
4210 enum frame_type type;
4211 frame_this_id_ftype *this_id;
4212 frame_prev_register_ftype *prev_register;
4213 const struct frame_data *unwind_data;
4214 frame_sniffer_ftype *sniffer;
4215 frame_prev_pc_ftype *prev_pc;
4216 frame_dealloc_cache_ftype *dealloc_cache;
4220 The @code{type} field indicates the type of frame this sniffer can
4221 handle: normal, dummy (@pxref{Functions Creating Dummy Frames, ,
4222 Functions Creating Dummy Frames}), signal handler or sentinel. Signal
4223 handlers sometimes have their own simplified stack structure for
4224 efficiency, so may need their own handlers.
4226 The @code{unwind_data} field holds additional information which may be
4227 relevant to particular types of frame. For example it may hold
4228 additional information for signal handler frames.
4230 The remaining fields define functions that yield different types of
4231 information when given a pointer to the NEXT stack frame. Not all
4232 functions need be provided. If an entry is @code{NULL}, the next sniffer will
4238 @code{this_id} determines the stack pointer and function (code
4239 entry point) for THIS stack frame.
4242 @code{prev_register} determines where the values of registers for
4243 the PREVIOUS stack frame are stored in THIS stack frame.
4246 @code{sniffer} takes a look at THIS frame's registers to
4247 determine if this is the appropriate unwinder.
4250 @code{prev_pc} determines the program counter for THIS
4251 frame. Only needed if the program counter is not an ordinary register
4252 (@pxref{Register Architecture Functions & Variables,
4253 , Functions and Variables Specifying the Register Architecture}).
4256 @code{dealloc_cache} frees any additional memory associated with
4257 the prologue cache for this frame (@pxref{Prologue Caches, , Prologue
4262 In general it is only the @code{this_id} and @code{prev_register}
4263 fields that need be defined for custom sniffers.
4265 The frame base sniffer is much simpler. It is a @code{@w{struct
4266 frame_base}}, which refers to the corresponding @code{frame_unwind}
4267 struct and whose fields refer to functions yielding various addresses
4270 @cindex @code{frame_base}
4274 const struct frame_unwind *unwind;
4275 frame_this_base_ftype *this_base;
4276 frame_this_locals_ftype *this_locals;
4277 frame_this_args_ftype *this_args;
4281 All the functions referred to take a pointer to the NEXT frame as
4282 argument. The function referred to by @code{this_base} returns the
4283 base address of THIS frame, the function referred to by
4284 @code{this_locals} returns the base address of local variables in THIS
4285 frame and the function referred to by @code{this_args} returns the
4286 base address of the function arguments in this frame.
4288 As described above, the base address of a frame is the address
4289 immediately before the start of the NEXT frame. For a falling
4290 stack, this is the lowest address in the frame and for a rising stack
4291 it is the highest address in the frame. For most architectures the
4292 same address is also the base address for local variables and
4293 arguments, in which case the same function can be used for all three
4294 entries@footnote{It is worth noting that if it cannot be determined in any
4295 other way (for example by there being a register with the name
4296 @code{"fp"}), then the result of the @code{this_base} function will be
4297 used as the value of the frame pointer variable @kbd{$fp} in
4298 @value{GDBN}. This is very often not correct (for example with the
4299 OpenRISC 1000, this value is the stack pointer, @kbd{$sp}). In this
4300 case a register (raw or pseudo) with the name @code{"fp"} should be
4301 defined. It will be used in preference as the value of @kbd{$fp}.}.
4303 @node Inferior Call Setup
4304 @section Inferior Call Setup
4305 @cindex calls to the inferior
4308 * About Dummy Frames::
4309 * Functions Creating Dummy Frames::
4312 @node About Dummy Frames
4313 @subsection About Dummy Frames
4314 @cindex dummy frames
4316 @value{GDBN} can call functions in the target code (for example by
4317 using the @kbd{call} or @kbd{print} commands). These functions may be
4318 breakpointed, and it is essential that if a function does hit a
4319 breakpoint, commands like @kbd{backtrace} work correctly.
4321 This is achieved by making the stack look as though the function had
4322 been called from the point where @value{GDBN} had previously stopped.
4323 This requires that @value{GDBN} can set up stack frames appropriate for
4324 such function calls.
4326 @node Functions Creating Dummy Frames
4327 @subsection Functions Creating Dummy Frames
4329 The following functions provide the functionality to set up such
4330 @dfn{dummy} stack frames.
4332 @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})
4334 This function sets up a dummy stack frame for the function about to be
4335 called. @code{push_dummy_call} is given the arguments to be passed
4336 and must copy them into registers or push them on to the stack as
4337 appropriate for the ABI.
4339 @var{function} is a pointer to the function
4340 that will be called and @var{regcache} the register cache from which
4341 values should be obtained. @var{bp_addr} is the address to which the
4342 function should return (which is breakpointed, so @value{GDBN} can
4343 regain control, hence the name). @var{nargs} is the number of
4344 arguments to pass and @var{args} an array containing the argument
4345 values. @var{struct_return} is non-zero (true) if the function returns
4346 a structure, and if so @var{struct_addr} is the address in which the
4347 structure should be returned.
4349 After calling this function, @value{GDBN} will pass control to the
4350 target at the address of the function, which will find the stack and
4351 registers set up just as expected.
4353 The default value of this function is @code{NULL} (undefined). If the
4354 function is not defined, then @value{GDBN} will not allow the user to
4355 call functions within the target being debugged.
4359 @deftypefn {Architecture Function} {struct frame_id} unwind_dummy_id (struct gdbarch *@var{gdbarch}, struct frame_info *@var{next_frame})
4361 This is the inverse of @code{push_dummy_call} which restores the stack
4362 pointer and program counter after a call to evaluate a function using
4363 a dummy stack frame. The result is a @code{@w{struct frame_id}}, which
4364 contains the value of the stack pointer and program counter to be
4367 The NEXT frame pointer is provided as argument,
4368 @var{next_frame}. THIS frame is the frame of the dummy function,
4369 which can be unwound, to yield the required stack pointer and program
4370 counter from the PREVIOUS frame.
4372 The default value is @code{NULL} (undefined). If @code{push_dummy_call} is
4373 defined, then this function should also be defined.
4377 @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})
4379 If this function is not defined (its default value is @code{NULL}), a dummy
4380 call will use the entry point of the currently loaded code on the
4381 target as its return address. A temporary breakpoint will be set
4382 there, so the location must be writable and have room for a
4385 It is possible that this default is not suitable. It might not be
4386 writable (in ROM possibly), or the ABI might require code to be
4387 executed on return from a call to unwind the stack before the
4388 breakpoint is encountered.
4390 If either of these is the case, then push_dummy_code should be defined
4391 to push an instruction sequence onto the end of the stack to which the
4392 dummy call should return.
4394 The arguments are essentially the same as those to
4395 @code{push_dummy_call}. However the function is provided with the
4396 type of the function result, @var{value_type}, @var{bp_addr} is used
4397 to return a value (the address at which the breakpoint instruction
4398 should be inserted) and @var{real pc} is used to specify the resume
4399 address when starting the call sequence. The function should return
4400 the updated innermost stack address.
4403 @emph{Note:} This does require that code in the stack can be executed.
4404 Some Harvard architectures may not allow this.
4409 @node Adding support for debugging core files
4410 @section Adding support for debugging core files
4413 The prerequisite for adding core file support in @value{GDBN} is to have
4414 core file support in BFD.
4416 Once BFD support is available, writing the apropriate
4417 @code{regset_from_core_section} architecture function should be all
4418 that is needed in order to add support for core files in @value{GDBN}.
4420 @node Defining Other Architecture Features
4421 @section Defining Other Architecture Features
4423 This section describes other functions and values in @code{gdbarch},
4424 together with some useful macros, that you can use to define the
4425 target architecture.
4429 @item CORE_ADDR gdbarch_addr_bits_remove (@var{gdbarch}, @var{addr})
4430 @findex gdbarch_addr_bits_remove
4431 If a raw machine instruction address includes any bits that are not
4432 really part of the address, then this function is used to zero those bits in
4433 @var{addr}. This is only used for addresses of instructions, and even then not
4436 For example, the two low-order bits of the PC on the Hewlett-Packard PA
4437 2.0 architecture contain the privilege level of the corresponding
4438 instruction. Since instructions must always be aligned on four-byte
4439 boundaries, the processor masks out these bits to generate the actual
4440 address of the instruction. @code{gdbarch_addr_bits_remove} would then for
4441 example look like that:
4443 arch_addr_bits_remove (CORE_ADDR addr)
4445 return (addr &= ~0x3);
4449 @item int address_class_name_to_type_flags (@var{gdbarch}, @var{name}, @var{type_flags_ptr})
4450 @findex address_class_name_to_type_flags
4451 If @var{name} is a valid address class qualifier name, set the @code{int}
4452 referenced by @var{type_flags_ptr} to the mask representing the qualifier
4453 and return 1. If @var{name} is not a valid address class qualifier name,
4456 The value for @var{type_flags_ptr} should be one of
4457 @code{TYPE_FLAG_ADDRESS_CLASS_1}, @code{TYPE_FLAG_ADDRESS_CLASS_2}, or
4458 possibly some combination of these values or'd together.
4459 @xref{Target Architecture Definition, , Address Classes}.
4461 @item int address_class_name_to_type_flags_p (@var{gdbarch})
4462 @findex address_class_name_to_type_flags_p
4463 Predicate which indicates whether @code{address_class_name_to_type_flags}
4466 @item int gdbarch_address_class_type_flags (@var{gdbarch}, @var{byte_size}, @var{dwarf2_addr_class})
4467 @findex gdbarch_address_class_type_flags
4468 Given a pointers byte size (as described by the debug information) and
4469 the possible @code{DW_AT_address_class} value, return the type flags
4470 used by @value{GDBN} to represent this address class. The value
4471 returned should be one of @code{TYPE_FLAG_ADDRESS_CLASS_1},
4472 @code{TYPE_FLAG_ADDRESS_CLASS_2}, or possibly some combination of these
4473 values or'd together.
4474 @xref{Target Architecture Definition, , Address Classes}.
4476 @item int gdbarch_address_class_type_flags_p (@var{gdbarch})
4477 @findex gdbarch_address_class_type_flags_p
4478 Predicate which indicates whether @code{gdbarch_address_class_type_flags_p} has
4481 @item const char *gdbarch_address_class_type_flags_to_name (@var{gdbarch}, @var{type_flags})
4482 @findex gdbarch_address_class_type_flags_to_name
4483 Return the name of the address class qualifier associated with the type
4484 flags given by @var{type_flags}.
4486 @item int gdbarch_address_class_type_flags_to_name_p (@var{gdbarch})
4487 @findex gdbarch_address_class_type_flags_to_name_p
4488 Predicate which indicates whether @code{gdbarch_address_class_type_flags_to_name} has been defined.
4489 @xref{Target Architecture Definition, , Address Classes}.
4491 @item void gdbarch_address_to_pointer (@var{gdbarch}, @var{type}, @var{buf}, @var{addr})
4492 @findex gdbarch_address_to_pointer
4493 Store in @var{buf} a pointer of type @var{type} representing the address
4494 @var{addr}, in the appropriate format for the current architecture.
4495 This function may safely assume that @var{type} is either a pointer or a
4496 C@t{++} reference type.
4497 @xref{Target Architecture Definition, , Pointers Are Not Always Addresses}.
4499 @item int gdbarch_believe_pcc_promotion (@var{gdbarch})
4500 @findex gdbarch_believe_pcc_promotion
4501 Used to notify if the compiler promotes a @code{short} or @code{char}
4502 parameter to an @code{int}, but still reports the parameter as its
4503 original type, rather than the promoted type.
4505 @item gdbarch_bits_big_endian (@var{gdbarch})
4506 @findex gdbarch_bits_big_endian
4507 This is used if the numbering of bits in the targets does @strong{not} match
4508 the endianism of the target byte order. A value of 1 means that the bits
4509 are numbered in a big-endian bit order, 0 means little-endian.
4511 @item set_gdbarch_bits_big_endian (@var{gdbarch}, @var{bits_big_endian})
4512 @findex set_gdbarch_bits_big_endian
4513 Calling set_gdbarch_bits_big_endian with a value of 1 indicates that the
4514 bits in the target are numbered in a big-endian bit order, 0 indicates
4519 This is the character array initializer for the bit pattern to put into
4520 memory where a breakpoint is set. Although it's common to use a trap
4521 instruction for a breakpoint, it's not required; for instance, the bit
4522 pattern could be an invalid instruction. The breakpoint must be no
4523 longer than the shortest instruction of the architecture.
4525 @code{BREAKPOINT} has been deprecated in favor of
4526 @code{gdbarch_breakpoint_from_pc}.
4528 @item BIG_BREAKPOINT
4529 @itemx LITTLE_BREAKPOINT
4530 @findex LITTLE_BREAKPOINT
4531 @findex BIG_BREAKPOINT
4532 Similar to BREAKPOINT, but used for bi-endian targets.
4534 @code{BIG_BREAKPOINT} and @code{LITTLE_BREAKPOINT} have been deprecated in
4535 favor of @code{gdbarch_breakpoint_from_pc}.
4537 @item const gdb_byte *gdbarch_breakpoint_from_pc (@var{gdbarch}, @var{pcptr}, @var{lenptr})
4538 @findex gdbarch_breakpoint_from_pc
4539 @anchor{gdbarch_breakpoint_from_pc} Use the program counter to determine the
4540 contents and size of a breakpoint instruction. It returns a pointer to
4541 a static string of bytes that encode a breakpoint instruction, stores the
4542 length of the string to @code{*@var{lenptr}}, and adjusts the program
4543 counter (if necessary) to point to the actual memory location where the
4544 breakpoint should be inserted. May return @code{NULL} to indicate that
4545 software breakpoints are not supported.
4547 Although it is common to use a trap instruction for a breakpoint, it's
4548 not required; for instance, the bit pattern could be an invalid
4549 instruction. The breakpoint must be no longer than the shortest
4550 instruction of the architecture.
4552 Provided breakpoint bytes can be also used by @code{bp_loc_is_permanent} to
4553 detect permanent breakpoints. @code{gdbarch_breakpoint_from_pc} should return
4554 an unchanged memory copy if it was called for a location with permanent
4555 breakpoint as some architectures use breakpoint instructions containing
4556 arbitrary parameter value.
4558 Replaces all the other @var{BREAKPOINT} macros.
4560 @item int gdbarch_memory_insert_breakpoint (@var{gdbarch}, @var{bp_tgt})
4561 @itemx gdbarch_memory_remove_breakpoint (@var{gdbarch}, @var{bp_tgt})
4562 @findex gdbarch_memory_remove_breakpoint
4563 @findex gdbarch_memory_insert_breakpoint
4564 Insert or remove memory based breakpoints. Reasonable defaults
4565 (@code{default_memory_insert_breakpoint} and
4566 @code{default_memory_remove_breakpoint} respectively) have been
4567 provided so that it is not necessary to set these for most
4568 architectures. Architectures which may want to set
4569 @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
4570 conventional manner.
4572 It may also be desirable (from an efficiency standpoint) to define
4573 custom breakpoint insertion and removal routines if
4574 @code{gdbarch_breakpoint_from_pc} needs to read the target's memory for some
4577 @item CORE_ADDR gdbarch_adjust_breakpoint_address (@var{gdbarch}, @var{bpaddr})
4578 @findex gdbarch_adjust_breakpoint_address
4579 @cindex breakpoint address adjusted
4580 Given an address at which a breakpoint is desired, return a breakpoint
4581 address adjusted to account for architectural constraints on
4582 breakpoint placement. This method is not needed by most targets.
4584 The FR-V target (see @file{frv-tdep.c}) requires this method.
4585 The FR-V is a VLIW architecture in which a number of RISC-like
4586 instructions are grouped (packed) together into an aggregate
4587 instruction or instruction bundle. When the processor executes
4588 one of these bundles, the component instructions are executed
4591 In the course of optimization, the compiler may group instructions
4592 from distinct source statements into the same bundle. The line number
4593 information associated with one of the latter statements will likely
4594 refer to some instruction other than the first one in the bundle. So,
4595 if the user attempts to place a breakpoint on one of these latter
4596 statements, @value{GDBN} must be careful to @emph{not} place the break
4597 instruction on any instruction other than the first one in the bundle.
4598 (Remember though that the instructions within a bundle execute
4599 in parallel, so the @emph{first} instruction is the instruction
4600 at the lowest address and has nothing to do with execution order.)
4602 The FR-V's @code{gdbarch_adjust_breakpoint_address} method will adjust a
4603 breakpoint's address by scanning backwards for the beginning of
4604 the bundle, returning the address of the bundle.
4606 Since the adjustment of a breakpoint may significantly alter a user's
4607 expectation, @value{GDBN} prints a warning when an adjusted breakpoint
4608 is initially set and each time that that breakpoint is hit.
4610 @item int gdbarch_call_dummy_location (@var{gdbarch})
4611 @findex gdbarch_call_dummy_location
4612 See the file @file{inferior.h}.
4614 This method has been replaced by @code{gdbarch_push_dummy_code}
4615 (@pxref{gdbarch_push_dummy_code}).
4617 @item int gdbarch_cannot_fetch_register (@var{gdbarch}, @var{regum})
4618 @findex gdbarch_cannot_fetch_register
4619 This function should return nonzero if @var{regno} cannot be fetched
4620 from an inferior process.
4622 @item int gdbarch_cannot_store_register (@var{gdbarch}, @var{regnum})
4623 @findex gdbarch_cannot_store_register
4624 This function should return nonzero if @var{regno} should not be
4625 written to the target. This is often the case for program counters,
4626 status words, and other special registers. This function returns 0 as
4627 default so that @value{GDBN} will assume that all registers may be written.
4629 @item int gdbarch_convert_register_p (@var{gdbarch}, @var{regnum}, struct type *@var{type})
4630 @findex gdbarch_convert_register_p
4631 Return non-zero if register @var{regnum} represents data values of type
4632 @var{type} in a non-standard form.
4633 @xref{Target Architecture Definition, , Using Different Register and Memory Data Representations}.
4635 @item int gdbarch_fp0_regnum (@var{gdbarch})
4636 @findex gdbarch_fp0_regnum
4637 This function returns the number of the first floating point register,
4638 if the machine has such registers. Otherwise, it returns -1.
4640 @item CORE_ADDR gdbarch_decr_pc_after_break (@var{gdbarch})
4641 @findex gdbarch_decr_pc_after_break
4642 This function shall return the amount by which to decrement the PC after the
4643 program encounters a breakpoint. This is often the number of bytes in
4644 @code{BREAKPOINT}, though not always. For most targets this value will be 0.
4646 @item DISABLE_UNSETTABLE_BREAK (@var{addr})
4647 @findex DISABLE_UNSETTABLE_BREAK
4648 If defined, this should evaluate to 1 if @var{addr} is in a shared
4649 library in which breakpoints cannot be set and so should be disabled.
4651 @item int gdbarch_dwarf2_reg_to_regnum (@var{gdbarch}, @var{dwarf2_regnr})
4652 @findex gdbarch_dwarf2_reg_to_regnum
4653 Convert DWARF2 register number @var{dwarf2_regnr} into @value{GDBN} regnum.
4654 If not defined, no conversion will be performed.
4656 @item int gdbarch_ecoff_reg_to_regnum (@var{gdbarch}, @var{ecoff_regnr})
4657 @findex gdbarch_ecoff_reg_to_regnum
4658 Convert ECOFF register number @var{ecoff_regnr} into @value{GDBN} regnum. If
4659 not defined, no conversion will be performed.
4661 @item GCC_COMPILED_FLAG_SYMBOL
4662 @itemx GCC2_COMPILED_FLAG_SYMBOL
4663 @findex GCC2_COMPILED_FLAG_SYMBOL
4664 @findex GCC_COMPILED_FLAG_SYMBOL
4665 If defined, these are the names of the symbols that @value{GDBN} will
4666 look for to detect that GCC compiled the file. The default symbols
4667 are @code{gcc_compiled.} and @code{gcc2_compiled.},
4668 respectively. (Currently only defined for the Delta 68.)
4670 @item gdbarch_get_longjmp_target
4671 @findex gdbarch_get_longjmp_target
4672 This function determines the target PC address that @code{longjmp}
4673 will jump to, assuming that we have just stopped at a @code{longjmp}
4674 breakpoint. It takes a @code{CORE_ADDR *} as argument, and stores the
4675 target PC value through this pointer. It examines the current state
4676 of the machine as needed, typically by using a manually-determined
4677 offset into the @code{jmp_buf}. (While we might like to get the offset
4678 from the target's @file{jmpbuf.h}, that header file cannot be assumed
4679 to be available when building a cross-debugger.)
4681 @item DEPRECATED_IBM6000_TARGET
4682 @findex DEPRECATED_IBM6000_TARGET
4683 Shows that we are configured for an IBM RS/6000 system. This
4684 conditional should be eliminated (FIXME) and replaced by
4685 feature-specific macros. It was introduced in haste and we are
4686 repenting at leisure.
4688 @item I386_USE_GENERIC_WATCHPOINTS
4689 An x86-based target can define this to use the generic x86 watchpoint
4690 support; see @ref{Algorithms, I386_USE_GENERIC_WATCHPOINTS}.
4692 @item gdbarch_in_function_epilogue_p (@var{gdbarch}, @var{addr})
4693 @findex gdbarch_in_function_epilogue_p
4694 Returns non-zero if the given @var{addr} is in the epilogue of a function.
4695 The epilogue of a function is defined as the part of a function where
4696 the stack frame of the function already has been destroyed up to the
4697 final `return from function call' instruction.
4699 @item int gdbarch_in_solib_return_trampoline (@var{gdbarch}, @var{pc}, @var{name})
4700 @findex gdbarch_in_solib_return_trampoline
4701 Define this function to return nonzero if the program is stopped in the
4702 trampoline that returns from a shared library.
4704 @item target_so_ops.in_dynsym_resolve_code (@var{pc})
4705 @findex in_dynsym_resolve_code
4706 Define this to return nonzero if the program is stopped in the
4709 @item SKIP_SOLIB_RESOLVER (@var{pc})
4710 @findex SKIP_SOLIB_RESOLVER
4711 Define this to evaluate to the (nonzero) address at which execution
4712 should continue to get past the dynamic linker's symbol resolution
4713 function. A zero value indicates that it is not important or necessary
4714 to set a breakpoint to get through the dynamic linker and that single
4715 stepping will suffice.
4717 @item CORE_ADDR gdbarch_integer_to_address (@var{gdbarch}, @var{type}, @var{buf})
4718 @findex gdbarch_integer_to_address
4719 @cindex converting integers to addresses
4720 Define this when the architecture needs to handle non-pointer to address
4721 conversions specially. Converts that value to an address according to
4722 the current architectures conventions.
4724 @emph{Pragmatics: When the user copies a well defined expression from
4725 their source code and passes it, as a parameter, to @value{GDBN}'s
4726 @code{print} command, they should get the same value as would have been
4727 computed by the target program. Any deviation from this rule can cause
4728 major confusion and annoyance, and needs to be justified carefully. In
4729 other words, @value{GDBN} doesn't really have the freedom to do these
4730 conversions in clever and useful ways. It has, however, been pointed
4731 out that users aren't complaining about how @value{GDBN} casts integers
4732 to pointers; they are complaining that they can't take an address from a
4733 disassembly listing and give it to @code{x/i}. Adding an architecture
4734 method like @code{gdbarch_integer_to_address} certainly makes it possible for
4735 @value{GDBN} to ``get it right'' in all circumstances.}
4737 @xref{Target Architecture Definition, , Pointers Are Not Always
4740 @item CORE_ADDR gdbarch_pointer_to_address (@var{gdbarch}, @var{type}, @var{buf})
4741 @findex gdbarch_pointer_to_address
4742 Assume that @var{buf} holds a pointer of type @var{type}, in the
4743 appropriate format for the current architecture. Return the byte
4744 address the pointer refers to.
4745 @xref{Target Architecture Definition, , Pointers Are Not Always Addresses}.
4747 @item void gdbarch_register_to_value(@var{gdbarch}, @var{frame}, @var{regnum}, @var{type}, @var{fur})
4748 @findex gdbarch_register_to_value
4749 Convert the raw contents of register @var{regnum} into a value of type
4751 @xref{Target Architecture Definition, , Using Different Register and Memory Data Representations}.
4753 @item REGISTER_CONVERT_TO_VIRTUAL(@var{reg}, @var{type}, @var{from}, @var{to})
4754 @findex REGISTER_CONVERT_TO_VIRTUAL
4755 Convert the value of register @var{reg} from its raw form to its virtual
4757 @xref{Target Architecture Definition, , Raw and Virtual Register Representations}.
4759 @item REGISTER_CONVERT_TO_RAW(@var{type}, @var{reg}, @var{from}, @var{to})
4760 @findex REGISTER_CONVERT_TO_RAW
4761 Convert the value of register @var{reg} from its virtual form to its raw
4763 @xref{Target Architecture Definition, , Raw and Virtual Register Representations}.
4765 @item const struct regset *regset_from_core_section (struct gdbarch * @var{gdbarch}, const char * @var{sect_name}, size_t @var{sect_size})
4766 @findex regset_from_core_section
4767 Return the appropriate register set for a core file section with name
4768 @var{sect_name} and size @var{sect_size}.
4770 @item SOFTWARE_SINGLE_STEP_P()
4771 @findex SOFTWARE_SINGLE_STEP_P
4772 Define this as 1 if the target does not have a hardware single-step
4773 mechanism. The macro @code{SOFTWARE_SINGLE_STEP} must also be defined.
4775 @item SOFTWARE_SINGLE_STEP(@var{signal}, @var{insert_breakpoints_p})
4776 @findex SOFTWARE_SINGLE_STEP
4777 A function that inserts or removes (depending on
4778 @var{insert_breakpoints_p}) breakpoints at each possible destinations of
4779 the next instruction. See @file{sparc-tdep.c} and @file{rs6000-tdep.c}
4782 @item set_gdbarch_sofun_address_maybe_missing (@var{gdbarch}, @var{set})
4783 @findex set_gdbarch_sofun_address_maybe_missing
4784 Somebody clever observed that, the more actual addresses you have in the
4785 debug information, the more time the linker has to spend relocating
4786 them. So whenever there's some other way the debugger could find the
4787 address it needs, you should omit it from the debug info, to make
4790 Calling @code{set_gdbarch_sofun_address_maybe_missing} with a non-zero
4791 argument @var{set} indicates that a particular set of hacks of this sort
4792 are in use, affecting @code{N_SO} and @code{N_FUN} entries in stabs-format
4793 debugging information. @code{N_SO} stabs mark the beginning and ending
4794 addresses of compilation units in the text segment. @code{N_FUN} stabs
4795 mark the starts and ends of functions.
4797 In this case, @value{GDBN} assumes two things:
4801 @code{N_FUN} stabs have an address of zero. Instead of using those
4802 addresses, you should find the address where the function starts by
4803 taking the function name from the stab, and then looking that up in the
4804 minsyms (the linker/assembler symbol table). In other words, the stab
4805 has the name, and the linker/assembler symbol table is the only place
4806 that carries the address.
4809 @code{N_SO} stabs have an address of zero, too. You just look at the
4810 @code{N_FUN} stabs that appear before and after the @code{N_SO} stab, and
4811 guess the starting and ending addresses of the compilation unit from them.
4814 @item int gdbarch_stabs_argument_has_addr (@var{gdbarch}, @var{type})
4815 @findex gdbarch_stabs_argument_has_addr
4816 @anchor{gdbarch_stabs_argument_has_addr} Define this function to return
4817 nonzero if a function argument of type @var{type} is passed by reference
4820 @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})
4821 @findex gdbarch_push_dummy_call
4822 @anchor{gdbarch_push_dummy_call} Define this to push the dummy frame's call to
4823 the inferior function onto the stack. In addition to pushing @var{nargs}, the
4824 code should push @var{struct_addr} (when @var{struct_return} is non-zero), and
4825 the return address (@var{bp_addr}).
4827 @var{function} is a pointer to a @code{struct value}; on architectures that use
4828 function descriptors, this contains the function descriptor value.
4830 Returns the updated top-of-stack pointer.
4832 @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})
4833 @findex gdbarch_push_dummy_code
4834 @anchor{gdbarch_push_dummy_code} Given a stack based call dummy, push the
4835 instruction sequence (including space for a breakpoint) to which the
4836 called function should return.
4838 Set @var{bp_addr} to the address at which the breakpoint instruction
4839 should be inserted, @var{real_pc} to the resume address when starting
4840 the call sequence, and return the updated inner-most stack address.
4842 By default, the stack is grown sufficient to hold a frame-aligned
4843 (@pxref{frame_align}) breakpoint, @var{bp_addr} is set to the address
4844 reserved for that breakpoint, and @var{real_pc} set to @var{funaddr}.
4846 This method replaces @w{@code{gdbarch_call_dummy_location (@var{gdbarch})}}.
4848 @item int gdbarch_sdb_reg_to_regnum (@var{gdbarch}, @var{sdb_regnr})
4849 @findex gdbarch_sdb_reg_to_regnum
4850 Use this function to convert sdb register @var{sdb_regnr} into @value{GDBN}
4851 regnum. If not defined, no conversion will be done.
4853 @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})
4854 @findex gdbarch_return_value
4855 @anchor{gdbarch_return_value} Given a function with a return-value of
4856 type @var{rettype}, return which return-value convention that function
4859 @value{GDBN} currently recognizes two function return-value conventions:
4860 @code{RETURN_VALUE_REGISTER_CONVENTION} where the return value is found
4861 in registers; and @code{RETURN_VALUE_STRUCT_CONVENTION} where the return
4862 value is found in memory and the address of that memory location is
4863 passed in as the function's first parameter.
4865 If the register convention is being used, and @var{writebuf} is
4866 non-@code{NULL}, also copy the return-value in @var{writebuf} into
4869 If the register convention is being used, and @var{readbuf} is
4870 non-@code{NULL}, also copy the return value from @var{regcache} into
4871 @var{readbuf} (@var{regcache} contains a copy of the registers from the
4872 just returned function).
4874 @emph{Maintainer note: This method replaces separate predicate, extract,
4875 store methods. By having only one method, the logic needed to determine
4876 the return-value convention need only be implemented in one place. If
4877 @value{GDBN} were written in an @sc{oo} language, this method would
4878 instead return an object that knew how to perform the register
4879 return-value extract and store.}
4881 @emph{Maintainer note: This method does not take a @var{gcc_p}
4882 parameter, and such a parameter should not be added. If an architecture
4883 that requires per-compiler or per-function information be identified,
4884 then the replacement of @var{rettype} with @code{struct value}
4885 @var{function} should be pursued.}
4887 @emph{Maintainer note: The @var{regcache} parameter limits this methods
4888 to the inner most frame. While replacing @var{regcache} with a
4889 @code{struct frame_info} @var{frame} parameter would remove that
4890 limitation there has yet to be a demonstrated need for such a change.}
4892 @item void gdbarch_skip_permanent_breakpoint (@var{gdbarch}, @var{regcache})
4893 @findex gdbarch_skip_permanent_breakpoint
4894 Advance the inferior's PC past a permanent breakpoint. @value{GDBN} normally
4895 steps over a breakpoint by removing it, stepping one instruction, and
4896 re-inserting the breakpoint. However, permanent breakpoints are
4897 hardwired into the inferior, and can't be removed, so this strategy
4898 doesn't work. Calling @code{gdbarch_skip_permanent_breakpoint} adjusts the
4899 processor's state so that execution will resume just after the breakpoint.
4900 This function does the right thing even when the breakpoint is in the delay slot
4901 of a branch or jump.
4903 @item CORE_ADDR gdbarch_skip_trampoline_code (@var{gdbarch}, @var{frame}, @var{pc})
4904 @findex gdbarch_skip_trampoline_code
4905 If the target machine has trampoline code that sits between callers and
4906 the functions being called, then define this function to return a new PC
4907 that is at the start of the real function.
4909 @item int gdbarch_deprecated_fp_regnum (@var{gdbarch})
4910 @findex gdbarch_deprecated_fp_regnum
4911 If the frame pointer is in a register, use this function to return the
4912 number of that register.
4914 @item int gdbarch_stab_reg_to_regnum (@var{gdbarch}, @var{stab_regnr})
4915 @findex gdbarch_stab_reg_to_regnum
4916 Use this function to convert stab register @var{stab_regnr} into @value{GDBN}
4917 regnum. If not defined, no conversion will be done.
4919 @item SYMBOL_RELOADING_DEFAULT
4920 @findex SYMBOL_RELOADING_DEFAULT
4921 The default value of the ``symbol-reloading'' variable. (Never defined in
4924 @item TARGET_CHAR_BIT
4925 @findex TARGET_CHAR_BIT
4926 Number of bits in a char; defaults to 8.
4928 @item int gdbarch_char_signed (@var{gdbarch})
4929 @findex gdbarch_char_signed
4930 Non-zero if @code{char} is normally signed on this architecture; zero if
4931 it should be unsigned.
4933 The ISO C standard requires the compiler to treat @code{char} as
4934 equivalent to either @code{signed char} or @code{unsigned char}; any
4935 character in the standard execution set is supposed to be positive.
4936 Most compilers treat @code{char} as signed, but @code{char} is unsigned
4937 on the IBM S/390, RS6000, and PowerPC targets.
4939 @item int gdbarch_double_bit (@var{gdbarch})
4940 @findex gdbarch_double_bit
4941 Number of bits in a double float; defaults to @w{@code{8 * TARGET_CHAR_BIT}}.
4943 @item int gdbarch_float_bit (@var{gdbarch})
4944 @findex gdbarch_float_bit
4945 Number of bits in a float; defaults to @w{@code{4 * TARGET_CHAR_BIT}}.
4947 @item int gdbarch_int_bit (@var{gdbarch})
4948 @findex gdbarch_int_bit
4949 Number of bits in an integer; defaults to @w{@code{4 * TARGET_CHAR_BIT}}.
4951 @item int gdbarch_long_bit (@var{gdbarch})
4952 @findex gdbarch_long_bit
4953 Number of bits in a long integer; defaults to @w{@code{4 * TARGET_CHAR_BIT}}.
4955 @item int gdbarch_long_double_bit (@var{gdbarch})
4956 @findex gdbarch_long_double_bit
4957 Number of bits in a long double float;
4958 defaults to @w{@code{2 * gdbarch_double_bit (@var{gdbarch})}}.
4960 @item int gdbarch_long_long_bit (@var{gdbarch})
4961 @findex gdbarch_long_long_bit
4962 Number of bits in a long long integer; defaults to
4963 @w{@code{2 * gdbarch_long_bit (@var{gdbarch})}}.
4965 @item int gdbarch_ptr_bit (@var{gdbarch})
4966 @findex gdbarch_ptr_bit
4967 Number of bits in a pointer; defaults to
4968 @w{@code{gdbarch_int_bit (@var{gdbarch})}}.
4970 @item int gdbarch_short_bit (@var{gdbarch})
4971 @findex gdbarch_short_bit
4972 Number of bits in a short integer; defaults to @w{@code{2 * TARGET_CHAR_BIT}}.
4974 @item void gdbarch_virtual_frame_pointer (@var{gdbarch}, @var{pc}, @var{frame_regnum}, @var{frame_offset})
4975 @findex gdbarch_virtual_frame_pointer
4976 Returns a @code{(@var{register}, @var{offset})} pair representing the virtual
4977 frame pointer in use at the code address @var{pc}. If virtual frame
4978 pointers are not used, a default definition simply returns
4979 @code{gdbarch_deprecated_fp_regnum} (or @code{gdbarch_sp_regnum}, if
4980 no frame pointer is defined), with an offset of zero.
4982 @c need to explain virtual frame pointers, they are recorded in agent
4983 @c expressions for tracepoints
4985 @item TARGET_HAS_HARDWARE_WATCHPOINTS
4986 If non-zero, the target has support for hardware-assisted
4987 watchpoints. @xref{Algorithms, watchpoints}, for more details and
4988 other related macros.
4990 @item int gdbarch_print_insn (@var{gdbarch}, @var{vma}, @var{info})
4991 @findex gdbarch_print_insn
4992 This is the function used by @value{GDBN} to print an assembly
4993 instruction. It prints the instruction at address @var{vma} in
4994 debugged memory and returns the length of the instruction, in bytes.
4995 This usually points to a function in the @code{opcodes} library
4996 (@pxref{Support Libraries, ,Opcodes}). @var{info} is a structure (of
4997 type @code{disassemble_info}) defined in the header file
4998 @file{include/dis-asm.h}, and used to pass information to the
4999 instruction decoding routine.
5001 @item frame_id gdbarch_dummy_id (@var{gdbarch}, @var{frame})
5002 @findex gdbarch_dummy_id
5003 @anchor{gdbarch_dummy_id} Given @var{frame} return a @w{@code{struct
5004 frame_id}} that uniquely identifies an inferior function call's dummy
5005 frame. The value returned must match the dummy frame stack value
5006 previously saved by @code{call_function_by_hand}.
5008 @item void gdbarch_value_to_register (@var{gdbarch}, @var{frame}, @var{type}, @var{buf})
5009 @findex gdbarch_value_to_register
5010 Convert a value of type @var{type} into the raw contents of a register.
5011 @xref{Target Architecture Definition, , Using Different Register and Memory Data Representations}.
5015 Motorola M68K target conditionals.
5019 Define this to be the 4-bit location of the breakpoint trap vector. If
5020 not defined, it will default to @code{0xf}.
5022 @item REMOTE_BPT_VECTOR
5023 Defaults to @code{1}.
5027 @node Adding a New Target
5028 @section Adding a New Target
5030 @cindex adding a target
5031 The following files add a target to @value{GDBN}:
5034 @cindex target dependent files
5036 @item gdb/@var{ttt}-tdep.c
5037 Contains any miscellaneous code required for this target machine. On
5038 some machines it doesn't exist at all.
5040 @item gdb/@var{arch}-tdep.c
5041 @itemx gdb/@var{arch}-tdep.h
5042 This is required to describe the basic layout of the target machine's
5043 processor chip (registers, stack, etc.). It can be shared among many
5044 targets that use the same processor architecture.
5048 (Target header files such as
5049 @file{gdb/config/@var{arch}/tm-@var{ttt}.h},
5050 @file{gdb/config/@var{arch}/tm-@var{arch}.h}, and
5051 @file{config/tm-@var{os}.h} are no longer used.)
5053 @findex _initialize_@var{arch}_tdep
5054 A @value{GDBN} description for a new architecture, arch is created by
5055 defining a global function @code{_initialize_@var{arch}_tdep}, by
5056 convention in the source file @file{@var{arch}-tdep.c}. For
5057 example, in the case of the OpenRISC 1000, this function is called
5058 @code{_initialize_or1k_tdep} and is found in the file
5061 The object file resulting from compiling this source file, which will
5062 contain the implementation of the
5063 @code{_initialize_@var{arch}_tdep} function is specified in the
5064 @value{GDBN} @file{configure.tgt} file, which includes a large case
5065 statement pattern matching against the @code{--target} option of the
5066 @kbd{configure} script.
5069 @emph{Note:} If the architecture requires multiple source files, the
5070 corresponding binaries should be included in
5071 @file{configure.tgt}. However if there are header files, the
5072 dependencies on these will not be picked up from the entries in
5073 @file{configure.tgt}. The @file{Makefile.in} file will need extending to
5074 show these dependencies.
5077 @findex gdbarch_register
5078 A new struct gdbarch, defining the new architecture, is created within
5079 the @code{_initialize_@var{arch}_tdep} function by calling
5080 @code{gdbarch_register}:
5083 void gdbarch_register (enum bfd_architecture architecture,
5084 gdbarch_init_ftype *init_func,
5085 gdbarch_dump_tdep_ftype *tdep_dump_func);
5088 This function has been described fully in an earlier
5089 section. @xref{How an Architecture is Represented, , How an
5090 Architecture is Represented}.
5092 The new @code{@w{struct gdbarch}} should contain implementations of
5093 the necessary functions (described in the previous sections) to
5094 describe the basic layout of the target machine's processor chip
5095 (registers, stack, etc.). It can be shared among many targets that use
5096 the same processor architecture.
5098 @node Target Descriptions
5099 @chapter Target Descriptions
5100 @cindex target descriptions
5102 The target architecture definition (@pxref{Target Architecture Definition})
5103 contains @value{GDBN}'s hard-coded knowledge about an architecture. For
5104 some platforms, it is handy to have more flexible knowledge about a specific
5105 instance of the architecture---for instance, a processor or development board.
5106 @dfn{Target descriptions} provide a mechanism for the user to tell @value{GDBN}
5107 more about what their target supports, or for the target to tell @value{GDBN}
5110 For details on writing, automatically supplying, and manually selecting
5111 target descriptions, see @ref{Target Descriptions, , , gdb,
5112 Debugging with @value{GDBN}}. This section will cover some related
5113 topics about the @value{GDBN} internals.
5116 * Target Descriptions Implementation::
5117 * Adding Target Described Register Support::
5120 @node Target Descriptions Implementation
5121 @section Target Descriptions Implementation
5122 @cindex target descriptions, implementation
5124 Before @value{GDBN} connects to a new target, or runs a new program on
5125 an existing target, it discards any existing target description and
5126 reverts to a default gdbarch. Then, after connecting, it looks for a
5127 new target description by calling @code{target_find_description}.
5129 A description may come from a user specified file (XML), the remote
5130 @samp{qXfer:features:read} packet (also XML), or from any custom
5131 @code{to_read_description} routine in the target vector. For instance,
5132 the remote target supports guessing whether a MIPS target is 32-bit or
5133 64-bit based on the size of the @samp{g} packet.
5135 If any target description is found, @value{GDBN} creates a new gdbarch
5136 incorporating the description by calling @code{gdbarch_update_p}. Any
5137 @samp{<architecture>} element is handled first, to determine which
5138 architecture's gdbarch initialization routine is called to create the
5139 new architecture. Then the initialization routine is called, and has
5140 a chance to adjust the constructed architecture based on the contents
5141 of the target description. For instance, it can recognize any
5142 properties set by a @code{to_read_description} routine. Also
5143 see @ref{Adding Target Described Register Support}.
5145 @node Adding Target Described Register Support
5146 @section Adding Target Described Register Support
5147 @cindex target descriptions, adding register support
5149 Target descriptions can report additional registers specific to an
5150 instance of the target. But it takes a little work in the architecture
5151 specific routines to support this.
5153 A target description must either have no registers or a complete
5154 set---this avoids complexity in trying to merge standard registers
5155 with the target defined registers. It is the architecture's
5156 responsibility to validate that a description with registers has
5157 everything it needs. To keep architecture code simple, the same
5158 mechanism is used to assign fixed internal register numbers to
5161 If @code{tdesc_has_registers} returns 1, the description contains
5162 registers. The architecture's @code{gdbarch_init} routine should:
5167 Call @code{tdesc_data_alloc} to allocate storage, early, before
5168 searching for a matching gdbarch or allocating a new one.
5171 Use @code{tdesc_find_feature} to locate standard features by name.
5174 Use @code{tdesc_numbered_register} and @code{tdesc_numbered_register_choices}
5175 to locate the expected registers in the standard features.
5178 Return @code{NULL} if a required feature is missing, or if any standard
5179 feature is missing expected registers. This will produce a warning that
5180 the description was incomplete.
5183 Free the allocated data before returning, unless @code{tdesc_use_registers}
5187 Call @code{set_gdbarch_num_regs} as usual, with a number higher than any
5188 fixed number passed to @code{tdesc_numbered_register}.
5191 Call @code{tdesc_use_registers} after creating a new gdbarch, before
5196 After @code{tdesc_use_registers} has been called, the architecture's
5197 @code{register_name}, @code{register_type}, and @code{register_reggroup_p}
5198 routines will not be called; that information will be taken from
5199 the target description. @code{num_regs} may be increased to account
5200 for any additional registers in the description.
5202 Pseudo-registers require some extra care:
5207 Using @code{tdesc_numbered_register} allows the architecture to give
5208 constant register numbers to standard architectural registers, e.g.@:
5209 as an @code{enum} in @file{@var{arch}-tdep.h}. But because
5210 pseudo-registers are always numbered above @code{num_regs},
5211 which may be increased by the description, constant numbers
5212 can not be used for pseudos. They must be numbered relative to
5213 @code{num_regs} instead.
5216 The description will not describe pseudo-registers, so the
5217 architecture must call @code{set_tdesc_pseudo_register_name},
5218 @code{set_tdesc_pseudo_register_type}, and
5219 @code{set_tdesc_pseudo_register_reggroup_p} to supply routines
5220 describing pseudo registers. These routines will be passed
5221 internal register numbers, so the same routines used for the
5222 gdbarch equivalents are usually suitable.
5227 @node Target Vector Definition
5229 @chapter Target Vector Definition
5230 @cindex target vector
5232 The target vector defines the interface between @value{GDBN}'s
5233 abstract handling of target systems, and the nitty-gritty code that
5234 actually exercises control over a process or a serial port.
5235 @value{GDBN} includes some 30-40 different target vectors; however,
5236 each configuration of @value{GDBN} includes only a few of them.
5239 * Managing Execution State::
5240 * Existing Targets::
5243 @node Managing Execution State
5244 @section Managing Execution State
5245 @cindex execution state
5247 A target vector can be completely inactive (not pushed on the target
5248 stack), active but not running (pushed, but not connected to a fully
5249 manifested inferior), or completely active (pushed, with an accessible
5250 inferior). Most targets are only completely inactive or completely
5251 active, but some support persistent connections to a target even
5252 when the target has exited or not yet started.
5254 For example, connecting to the simulator using @code{target sim} does
5255 not create a running program. Neither registers nor memory are
5256 accessible until @code{run}. Similarly, after @code{kill}, the
5257 program can not continue executing. But in both cases @value{GDBN}
5258 remains connected to the simulator, and target-specific commands
5259 are directed to the simulator.
5261 A target which only supports complete activation should push itself
5262 onto the stack in its @code{to_open} routine (by calling
5263 @code{push_target}), and unpush itself from the stack in its
5264 @code{to_mourn_inferior} routine (by calling @code{unpush_target}).
5266 A target which supports both partial and complete activation should
5267 still call @code{push_target} in @code{to_open}, but not call
5268 @code{unpush_target} in @code{to_mourn_inferior}. Instead, it should
5269 call either @code{target_mark_running} or @code{target_mark_exited}
5270 in its @code{to_open}, depending on whether the target is fully active
5271 after connection. It should also call @code{target_mark_running} any
5272 time the inferior becomes fully active (e.g.@: in
5273 @code{to_create_inferior} and @code{to_attach}), and
5274 @code{target_mark_exited} when the inferior becomes inactive (in
5275 @code{to_mourn_inferior}). The target should also make sure to call
5276 @code{target_mourn_inferior} from its @code{to_kill}, to return the
5277 target to inactive state.
5279 @node Existing Targets
5280 @section Existing Targets
5283 @subsection File Targets
5285 Both executables and core files have target vectors.
5287 @subsection Standard Protocol and Remote Stubs
5289 @value{GDBN}'s file @file{remote.c} talks a serial protocol to code that
5290 runs in the target system. @value{GDBN} provides several sample
5291 @dfn{stubs} that can be integrated into target programs or operating
5292 systems for this purpose; they are named @file{@var{cpu}-stub.c}. Many
5293 operating systems, embedded targets, emulators, and simulators already
5294 have a @value{GDBN} stub built into them, and maintenance of the remote
5295 protocol must be careful to preserve compatibility.
5297 The @value{GDBN} user's manual describes how to put such a stub into
5298 your target code. What follows is a discussion of integrating the
5299 SPARC stub into a complicated operating system (rather than a simple
5300 program), by Stu Grossman, the author of this stub.
5302 The trap handling code in the stub assumes the following upon entry to
5307 %l1 and %l2 contain pc and npc respectively at the time of the trap;
5313 you are in the correct trap window.
5316 As long as your trap handler can guarantee those conditions, then there
5317 is no reason why you shouldn't be able to ``share'' traps with the stub.
5318 The stub has no requirement that it be jumped to directly from the
5319 hardware trap vector. That is why it calls @code{exceptionHandler()},
5320 which is provided by the external environment. For instance, this could
5321 set up the hardware traps to actually execute code which calls the stub
5322 first, and then transfers to its own trap handler.
5324 For the most point, there probably won't be much of an issue with
5325 ``sharing'' traps, as the traps we use are usually not used by the kernel,
5326 and often indicate unrecoverable error conditions. Anyway, this is all
5327 controlled by a table, and is trivial to modify. The most important
5328 trap for us is for @code{ta 1}. Without that, we can't single step or
5329 do breakpoints. Everything else is unnecessary for the proper operation
5330 of the debugger/stub.
5332 From reading the stub, it's probably not obvious how breakpoints work.
5333 They are simply done by deposit/examine operations from @value{GDBN}.
5335 @subsection ROM Monitor Interface
5337 @subsection Custom Protocols
5339 @subsection Transport Layer
5341 @subsection Builtin Simulator
5344 @node Native Debugging
5346 @chapter Native Debugging
5347 @cindex native debugging
5349 Several files control @value{GDBN}'s configuration for native support:
5353 @item gdb/config/@var{arch}/@var{xyz}.mh
5354 Specifies Makefile fragments needed by a @emph{native} configuration on
5355 machine @var{xyz}. In particular, this lists the required
5356 native-dependent object files, by defining @samp{NATDEPFILES=@dots{}}.
5357 Also specifies the header file which describes native support on
5358 @var{xyz}, by defining @samp{NAT_FILE= nm-@var{xyz}.h}. You can also
5359 define @samp{NAT_CFLAGS}, @samp{NAT_ADD_FILES}, @samp{NAT_CLIBS},
5360 @samp{NAT_CDEPS}, @samp{NAT_GENERATED_FILES}, etc.; see @file{Makefile.in}.
5362 @emph{Maintainer's note: The @file{.mh} suffix is because this file
5363 originally contained @file{Makefile} fragments for hosting @value{GDBN}
5364 on machine @var{xyz}. While the file is no longer used for this
5365 purpose, the @file{.mh} suffix remains. Perhaps someone will
5366 eventually rename these fragments so that they have a @file{.mn}
5369 @item gdb/config/@var{arch}/nm-@var{xyz}.h
5370 (@file{nm.h} is a link to this file, created by @code{configure}). Contains C
5371 macro definitions describing the native system environment, such as
5372 child process control and core file support.
5374 @item gdb/@var{xyz}-nat.c
5375 Contains any miscellaneous C code required for this native support of
5376 this machine. On some machines it doesn't exist at all.
5379 There are some ``generic'' versions of routines that can be used by
5380 various systems. These can be customized in various ways by macros
5381 defined in your @file{nm-@var{xyz}.h} file. If these routines work for
5382 the @var{xyz} host, you can just include the generic file's name (with
5383 @samp{.o}, not @samp{.c}) in @code{NATDEPFILES}.
5385 Otherwise, if your machine needs custom support routines, you will need
5386 to write routines that perform the same functions as the generic file.
5387 Put them into @file{@var{xyz}-nat.c}, and put @file{@var{xyz}-nat.o}
5388 into @code{NATDEPFILES}.
5392 This contains the @emph{target_ops vector} that supports Unix child
5393 processes on systems which use ptrace and wait to control the child.
5396 This contains the @emph{target_ops vector} that supports Unix child
5397 processes on systems which use /proc to control the child.
5400 This does the low-level grunge that uses Unix system calls to do a ``fork
5401 and exec'' to start up a child process.
5404 This is the low level interface to inferior processes for systems using
5405 the Unix @code{ptrace} call in a vanilla way.
5414 @section shared libraries
5416 @section Native Conditionals
5417 @cindex native conditionals
5419 When @value{GDBN} is configured and compiled, various macros are
5420 defined or left undefined, to control compilation when the host and
5421 target systems are the same. These macros should be defined (or left
5422 undefined) in @file{nm-@var{system}.h}.
5426 @item I386_USE_GENERIC_WATCHPOINTS
5427 An x86-based machine can define this to use the generic x86 watchpoint
5428 support; see @ref{Algorithms, I386_USE_GENERIC_WATCHPOINTS}.
5430 @item SOLIB_ADD (@var{filename}, @var{from_tty}, @var{targ}, @var{readsyms})
5432 Define this to expand into an expression that will cause the symbols in
5433 @var{filename} to be added to @value{GDBN}'s symbol table. If
5434 @var{readsyms} is zero symbols are not read but any necessary low level
5435 processing for @var{filename} is still done.
5437 @item SOLIB_CREATE_INFERIOR_HOOK
5438 @findex SOLIB_CREATE_INFERIOR_HOOK
5439 Define this to expand into any shared-library-relocation code that you
5440 want to be run just after the child process has been forked.
5442 @item START_INFERIOR_TRAPS_EXPECTED
5443 @findex START_INFERIOR_TRAPS_EXPECTED
5444 When starting an inferior, @value{GDBN} normally expects to trap
5446 the shell execs, and once when the program itself execs. If the actual
5447 number of traps is something other than 2, then define this macro to
5448 expand into the number expected.
5452 @node Support Libraries
5454 @chapter Support Libraries
5459 BFD provides support for @value{GDBN} in several ways:
5462 @item identifying executable and core files
5463 BFD will identify a variety of file types, including a.out, coff, and
5464 several variants thereof, as well as several kinds of core files.
5466 @item access to sections of files
5467 BFD parses the file headers to determine the names, virtual addresses,
5468 sizes, and file locations of all the various named sections in files
5469 (such as the text section or the data section). @value{GDBN} simply
5470 calls BFD to read or write section @var{x} at byte offset @var{y} for
5473 @item specialized core file support
5474 BFD provides routines to determine the failing command name stored in a
5475 core file, the signal with which the program failed, and whether a core
5476 file matches (i.e.@: could be a core dump of) a particular executable
5479 @item locating the symbol information
5480 @value{GDBN} uses an internal interface of BFD to determine where to find the
5481 symbol information in an executable file or symbol-file. @value{GDBN} itself
5482 handles the reading of symbols, since BFD does not ``understand'' debug
5483 symbols, but @value{GDBN} uses BFD's cached information to find the symbols,
5488 @cindex opcodes library
5490 The opcodes library provides @value{GDBN}'s disassembler. (It's a separate
5491 library because it's also used in binutils, for @file{objdump}).
5494 @cindex readline library
5495 The @code{readline} library provides a set of functions for use by applications
5496 that allow users to edit command lines as they are typed in.
5499 @cindex @code{libiberty} library
5501 The @code{libiberty} library provides a set of functions and features
5502 that integrate and improve on functionality found in modern operating
5503 systems. Broadly speaking, such features can be divided into three
5504 groups: supplemental functions (functions that may be missing in some
5505 environments and operating systems), replacement functions (providing
5506 a uniform and easier to use interface for commonly used standard
5507 functions), and extensions (which provide additional functionality
5508 beyond standard functions).
5510 @value{GDBN} uses various features provided by the @code{libiberty}
5511 library, for instance the C@t{++} demangler, the @acronym{IEEE}
5512 floating format support functions, the input options parser
5513 @samp{getopt}, the @samp{obstack} extension, and other functions.
5515 @subsection @code{obstacks} in @value{GDBN}
5516 @cindex @code{obstacks}
5518 The obstack mechanism provides a convenient way to allocate and free
5519 chunks of memory. Each obstack is a pool of memory that is managed
5520 like a stack. Objects (of any nature, size and alignment) are
5521 allocated and freed in a @acronym{LIFO} fashion on an obstack (see
5522 @code{libiberty}'s documentation for a more detailed explanation of
5525 The most noticeable use of the @code{obstacks} in @value{GDBN} is in
5526 object files. There is an obstack associated with each internal
5527 representation of an object file. Lots of things get allocated on
5528 these @code{obstacks}: dictionary entries, blocks, blockvectors,
5529 symbols, minimal symbols, types, vectors of fundamental types, class
5530 fields of types, object files section lists, object files section
5531 offset lists, line tables, symbol tables, partial symbol tables,
5532 string tables, symbol table private data, macros tables, debug
5533 information sections and entries, import and export lists (som),
5534 unwind information (hppa), dwarf2 location expressions data. Plus
5535 various strings such as directory names strings, debug format strings,
5538 An essential and convenient property of all data on @code{obstacks} is
5539 that memory for it gets allocated (with @code{obstack_alloc}) at
5540 various times during a debugging session, but it is released all at
5541 once using the @code{obstack_free} function. The @code{obstack_free}
5542 function takes a pointer to where in the stack it must start the
5543 deletion from (much like the cleanup chains have a pointer to where to
5544 start the cleanups). Because of the stack like structure of the
5545 @code{obstacks}, this allows to free only a top portion of the
5546 obstack. There are a few instances in @value{GDBN} where such thing
5547 happens. Calls to @code{obstack_free} are done after some local data
5548 is allocated to the obstack. Only the local data is deleted from the
5549 obstack. Of course this assumes that nothing between the
5550 @code{obstack_alloc} and the @code{obstack_free} allocates anything
5551 else on the same obstack. For this reason it is best and safest to
5552 use temporary @code{obstacks}.
5554 Releasing the whole obstack is also not safe per se. It is safe only
5555 under the condition that we know the @code{obstacks} memory is no
5556 longer needed. In @value{GDBN} we get rid of the @code{obstacks} only
5557 when we get rid of the whole objfile(s), for instance upon reading a
5561 @cindex regular expressions library
5572 @item SIGN_EXTEND_CHAR
5574 @item SWITCH_ENUM_BUG
5583 @section Array Containers
5584 @cindex Array Containers
5587 Often it is necessary to manipulate a dynamic array of a set of
5588 objects. C forces some bookkeeping on this, which can get cumbersome
5589 and repetitive. The @file{vec.h} file contains macros for defining
5590 and using a typesafe vector type. The functions defined will be
5591 inlined when compiling, and so the abstraction cost should be zero.
5592 Domain checks are added to detect programming errors.
5594 An example use would be an array of symbols or section information.
5595 The array can be grown as symbols are read in (or preallocated), and
5596 the accessor macros provided keep care of all the necessary
5597 bookkeeping. Because the arrays are type safe, there is no danger of
5598 accidentally mixing up the contents. Think of these as C++ templates,
5599 but implemented in C.
5601 Because of the different behavior of structure objects, scalar objects
5602 and of pointers, there are three flavors of vector, one for each of
5603 these variants. Both the structure object and pointer variants pass
5604 pointers to objects around --- in the former case the pointers are
5605 stored into the vector and in the latter case the pointers are
5606 dereferenced and the objects copied into the vector. The scalar
5607 object variant is suitable for @code{int}-like objects, and the vector
5608 elements are returned by value.
5610 There are both @code{index} and @code{iterate} accessors. The iterator
5611 returns a boolean iteration condition and updates the iteration
5612 variable passed by reference. Because the iterator will be inlined,
5613 the address-of can be optimized away.
5615 The vectors are implemented using the trailing array idiom, thus they
5616 are not resizeable without changing the address of the vector object
5617 itself. This means you cannot have variables or fields of vector type
5618 --- always use a pointer to a vector. The one exception is the final
5619 field of a structure, which could be a vector type. You will have to
5620 use the @code{embedded_size} & @code{embedded_init} calls to create
5621 such objects, and they will probably not be resizeable (so don't use
5622 the @dfn{safe} allocation variants). The trailing array idiom is used
5623 (rather than a pointer to an array of data), because, if we allow
5624 @code{NULL} to also represent an empty vector, empty vectors occupy
5625 minimal space in the structure containing them.
5627 Each operation that increases the number of active elements is
5628 available in @dfn{quick} and @dfn{safe} variants. The former presumes
5629 that there is sufficient allocated space for the operation to succeed
5630 (it dies if there is not). The latter will reallocate the vector, if
5631 needed. Reallocation causes an exponential increase in vector size.
5632 If you know you will be adding N elements, it would be more efficient
5633 to use the reserve operation before adding the elements with the
5634 @dfn{quick} operation. This will ensure there are at least as many
5635 elements as you ask for, it will exponentially increase if there are
5636 too few spare slots. If you want reserve a specific number of slots,
5637 but do not want the exponential increase (for instance, you know this
5638 is the last allocation), use a negative number for reservation. You
5639 can also create a vector of a specific size from the get go.
5641 You should prefer the push and pop operations, as they append and
5642 remove from the end of the vector. If you need to remove several items
5643 in one go, use the truncate operation. The insert and remove
5644 operations allow you to change elements in the middle of the vector.
5645 There are two remove operations, one which preserves the element
5646 ordering @code{ordered_remove}, and one which does not
5647 @code{unordered_remove}. The latter function copies the end element
5648 into the removed slot, rather than invoke a memmove operation. The
5649 @code{lower_bound} function will determine where to place an item in
5650 the array using insert that will maintain sorted order.
5652 If you need to directly manipulate a vector, then the @code{address}
5653 accessor will return the address of the start of the vector. Also the
5654 @code{space} predicate will tell you whether there is spare capacity in the
5655 vector. You will not normally need to use these two functions.
5657 Vector types are defined using a
5658 @code{DEF_VEC_@{O,P,I@}(@var{typename})} macro. Variables of vector
5659 type are declared using a @code{VEC(@var{typename})} macro. The
5660 characters @code{O}, @code{P} and @code{I} indicate whether
5661 @var{typename} is an object (@code{O}), pointer (@code{P}) or integral
5662 (@code{I}) type. Be careful to pick the correct one, as you'll get an
5663 awkward and inefficient API if you use the wrong one. There is a
5664 check, which results in a compile-time warning, for the @code{P} and
5665 @code{I} versions, but there is no check for the @code{O} versions, as
5666 that is not possible in plain C.
5668 An example of their use would be,
5671 DEF_VEC_P(tree); // non-managed tree vector.
5674 VEC(tree) *v; // A (pointer to) a vector of tree pointers.
5677 struct my_struct *s;
5679 if (VEC_length(tree, s->v)) @{ we have some contents @}
5680 VEC_safe_push(tree, s->v, decl); // append some decl onto the end
5681 for (ix = 0; VEC_iterate(tree, s->v, ix, elt); ix++)
5682 @{ do something with elt @}
5686 The @file{vec.h} file provides details on how to invoke the various
5687 accessors provided. They are enumerated here:
5691 Return the number of items in the array,
5694 Return true if the array has no elements.
5698 Return the last or arbitrary item in the array.
5701 Access an array element and indicate whether the array has been
5706 Create and destroy an array.
5708 @item VEC_embedded_size
5709 @itemx VEC_embedded_init
5710 Helpers for embedding an array as the final element of another struct.
5716 Return the amount of free space in an array.
5719 Ensure a certain amount of free space.
5721 @item VEC_quick_push
5722 @itemx VEC_safe_push
5723 Append to an array, either assuming the space is available, or making
5727 Remove the last item from an array.
5730 Remove several items from the end of an array.
5733 Add several items to the end of an array.
5736 Overwrite an item in the array.
5738 @item VEC_quick_insert
5739 @itemx VEC_safe_insert
5740 Insert an item into the middle of the array. Either the space must
5741 already exist, or the space is created.
5743 @item VEC_ordered_remove
5744 @itemx VEC_unordered_remove
5745 Remove an item from the array, preserving order or not.
5747 @item VEC_block_remove
5748 Remove a set of items from the array.
5751 Provide the address of the first element.
5753 @item VEC_lower_bound
5754 Binary search the array.
5760 @node Coding Standards
5762 @chapter Coding Standards
5763 @cindex coding standards
5765 @section @value{GDBN} C Coding Standards
5767 @value{GDBN} follows the GNU coding standards, as described in
5768 @file{etc/standards.texi}. This file is also available for anonymous
5769 FTP from GNU archive sites. @value{GDBN} takes a strict interpretation
5770 of the standard; in general, when the GNU standard recommends a practice
5771 but does not require it, @value{GDBN} requires it.
5773 @value{GDBN} follows an additional set of coding standards specific to
5774 @value{GDBN}, as described in the following sections.
5778 @value{GDBN} assumes an ISO/IEC 9899:1990 (a.k.a.@: ISO C90) compliant
5781 @value{GDBN} does not assume an ISO C or POSIX compliant C library.
5783 @subsection Formatting
5785 @cindex source code formatting
5786 The standard GNU recommendations for formatting must be followed
5787 strictly. Any @value{GDBN}-specific deviation from GNU
5788 recomendations is described below.
5790 A function declaration should not have its name in column zero. A
5791 function definition should have its name in column zero.
5795 static void foo (void);
5803 @emph{Pragmatics: This simplifies scripting. Function definitions can
5804 be found using @samp{^function-name}.}
5806 There must be a space between a function or macro name and the opening
5807 parenthesis of its argument list (except for macro definitions, as
5808 required by C). There must not be a space after an open paren/bracket
5809 or before a close paren/bracket.
5811 While additional whitespace is generally helpful for reading, do not use
5812 more than one blank line to separate blocks, and avoid adding whitespace
5813 after the end of a program line (as of 1/99, some 600 lines had
5814 whitespace after the semicolon). Excess whitespace causes difficulties
5815 for @code{diff} and @code{patch} utilities.
5817 Pointers are declared using the traditional K&R C style:
5831 In addition, whitespace around casts and unary operators should follow
5832 the following guidelines:
5834 @multitable @columnfractions .2 .2 .8
5835 @item Use... @tab ...instead of @tab
5844 @item @code{(foo) x}
5849 @tab (pointer dereference)
5852 @subsection Comments
5854 @cindex comment formatting
5855 The standard GNU requirements on comments must be followed strictly.
5857 Block comments must appear in the following form, with no @code{/*}- or
5858 @code{*/}-only lines, and no leading @code{*}:
5861 /* Wait for control to return from inferior to debugger. If inferior
5862 gets a signal, we may decide to start it up again instead of
5863 returning. That is why there is a loop in this function. When
5864 this function actually returns it means the inferior should be left
5865 stopped and @value{GDBN} should read more commands. */
5868 (Note that this format is encouraged by Emacs; tabbing for a multi-line
5869 comment works correctly, and @kbd{M-q} fills the block consistently.)
5871 Put a blank line between the block comments preceding function or
5872 variable definitions, and the definition itself.
5874 In general, put function-body comments on lines by themselves, rather
5875 than trying to fit them into the 20 characters left at the end of a
5876 line, since either the comment or the code will inevitably get longer
5877 than will fit, and then somebody will have to move it anyhow.
5881 @cindex C data types
5882 Code must not depend on the sizes of C data types, the format of the
5883 host's floating point numbers, the alignment of anything, or the order
5884 of evaluation of expressions.
5886 @cindex function usage
5887 Use functions freely. There are only a handful of compute-bound areas
5888 in @value{GDBN} that might be affected by the overhead of a function
5889 call, mainly in symbol reading. Most of @value{GDBN}'s performance is
5890 limited by the target interface (whether serial line or system call).
5892 However, use functions with moderation. A thousand one-line functions
5893 are just as hard to understand as a single thousand-line function.
5895 @emph{Macros are bad, M'kay.}
5896 (But if you have to use a macro, make sure that the macro arguments are
5897 protected with parentheses.)
5901 Declarations like @samp{struct foo *} should be used in preference to
5902 declarations like @samp{typedef struct foo @{ @dots{} @} *foo_ptr}.
5904 @subsection Function Prototypes
5905 @cindex function prototypes
5907 Prototypes must be used when both @emph{declaring} and @emph{defining}
5908 a function. Prototypes for @value{GDBN} functions must include both the
5909 argument type and name, with the name matching that used in the actual
5910 function definition.
5912 All external functions should have a declaration in a header file that
5913 callers include, except for @code{_initialize_*} functions, which must
5914 be external so that @file{init.c} construction works, but shouldn't be
5915 visible to random source files.
5917 Where a source file needs a forward declaration of a static function,
5918 that declaration must appear in a block near the top of the source file.
5920 @subsection File Names
5922 Any file used when building the core of @value{GDBN} must be in lower
5923 case. Any file used when building the core of @value{GDBN} must be 8.3
5924 unique. These requirements apply to both source and generated files.
5926 @emph{Pragmatics: The core of @value{GDBN} must be buildable on many
5927 platforms including DJGPP and MacOS/HFS. Every time an unfriendly file
5928 is introduced to the build process both @file{Makefile.in} and
5929 @file{configure.in} need to be modified accordingly. Compare the
5930 convoluted conversion process needed to transform @file{COPYING} into
5931 @file{copying.c} with the conversion needed to transform
5932 @file{version.in} into @file{version.c}.}
5934 Any file non 8.3 compliant file (that is not used when building the core
5935 of @value{GDBN}) must be added to @file{gdb/config/djgpp/fnchange.lst}.
5937 @emph{Pragmatics: This is clearly a compromise.}
5939 When @value{GDBN} has a local version of a system header file (ex
5940 @file{string.h}) the file name based on the POSIX header prefixed with
5941 @file{gdb_} (@file{gdb_string.h}). These headers should be relatively
5942 independent: they should use only macros defined by @file{configure},
5943 the compiler, or the host; they should include only system headers; they
5944 should refer only to system types. They may be shared between multiple
5945 programs, e.g.@: @value{GDBN} and @sc{gdbserver}.
5947 For other files @samp{-} is used as the separator.
5949 @subsection Include Files
5951 A @file{.c} file should include @file{defs.h} first.
5953 A @file{.c} file should directly include the @code{.h} file of every
5954 declaration and/or definition it directly refers to. It cannot rely on
5957 A @file{.h} file should directly include the @code{.h} file of every
5958 declaration and/or definition it directly refers to. It cannot rely on
5959 indirect inclusion. Exception: The file @file{defs.h} does not need to
5960 be directly included.
5962 An external declaration should only appear in one include file.
5964 An external declaration should never appear in a @code{.c} file.
5965 Exception: a declaration for the @code{_initialize} function that
5966 pacifies @option{-Wmissing-declaration}.
5968 A @code{typedef} definition should only appear in one include file.
5970 An opaque @code{struct} declaration can appear in multiple @file{.h}
5971 files. Where possible, a @file{.h} file should use an opaque
5972 @code{struct} declaration instead of an include.
5974 All @file{.h} files should be wrapped in:
5977 #ifndef INCLUDE_FILE_NAME_H
5978 #define INCLUDE_FILE_NAME_H
5983 @section @value{GDBN} Python Coding Standards
5985 @value{GDBN} follows the published @code{Python} coding standards in
5986 @uref{http://www.python.org/dev/peps/pep-0008/, @code{PEP008}}.
5988 In addition, the guidelines in the
5989 @uref{http://google-styleguide.googlecode.com/svn/trunk/pyguide.html,
5990 Google Python Style Guide} are also followed where they do not
5991 conflict with @code{PEP008}.
5993 @subsection @value{GDBN}-specific exceptions
5995 There are a few exceptions to the published standards.
5996 They exist mainly for consistency with the @code{C} standards.
5998 @c It is expected that there are a few more exceptions,
5999 @c so we use itemize here.
6004 Use @code{FIXME} instead of @code{TODO}.
6008 @node Misc Guidelines
6010 @chapter Misc Guidelines
6012 This chapter covers topics that are lower-level than the major
6013 algorithms of @value{GDBN}.
6018 Cleanups are a structured way to deal with things that need to be done
6021 When your code does something (e.g., @code{xmalloc} some memory, or
6022 @code{open} a file) that needs to be undone later (e.g., @code{xfree}
6023 the memory or @code{close} the file), it can make a cleanup. The
6024 cleanup will be done at some future point: when the command is finished
6025 and control returns to the top level; when an error occurs and the stack
6026 is unwound; or when your code decides it's time to explicitly perform
6027 cleanups. Alternatively you can elect to discard the cleanups you
6033 @item struct cleanup *@var{old_chain};
6034 Declare a variable which will hold a cleanup chain handle.
6036 @findex make_cleanup
6037 @item @var{old_chain} = make_cleanup (@var{function}, @var{arg});
6038 Make a cleanup which will cause @var{function} to be called with
6039 @var{arg} (a @code{char *}) later. The result, @var{old_chain}, is a
6040 handle that can later be passed to @code{do_cleanups} or
6041 @code{discard_cleanups}. Unless you are going to call
6042 @code{do_cleanups} or @code{discard_cleanups}, you can ignore the result
6043 from @code{make_cleanup}.
6046 @item do_cleanups (@var{old_chain});
6047 Do all cleanups added to the chain since the corresponding
6048 @code{make_cleanup} call was made.
6050 @findex discard_cleanups
6051 @item discard_cleanups (@var{old_chain});
6052 Same as @code{do_cleanups} except that it just removes the cleanups from
6053 the chain and does not call the specified functions.
6056 Cleanups are implemented as a chain. The handle returned by
6057 @code{make_cleanups} includes the cleanup passed to the call and any
6058 later cleanups appended to the chain (but not yet discarded or
6062 make_cleanup (a, 0);
6064 struct cleanup *old = make_cleanup (b, 0);
6072 will call @code{c()} and @code{b()} but will not call @code{a()}. The
6073 cleanup that calls @code{a()} will remain in the cleanup chain, and will
6074 be done later unless otherwise discarded.@refill
6076 Your function should explicitly do or discard the cleanups it creates.
6077 Failing to do this leads to non-deterministic behavior since the caller
6078 will arbitrarily do or discard your functions cleanups. This need leads
6079 to two common cleanup styles.
6081 The first style is try/finally. Before it exits, your code-block calls
6082 @code{do_cleanups} with the old cleanup chain and thus ensures that your
6083 code-block's cleanups are always performed. For instance, the following
6084 code-segment avoids a memory leak problem (even when @code{error} is
6085 called and a forced stack unwind occurs) by ensuring that the
6086 @code{xfree} will always be called:
6089 struct cleanup *old = make_cleanup (null_cleanup, 0);
6090 data = xmalloc (sizeof blah);
6091 make_cleanup (xfree, data);
6096 The second style is try/except. Before it exits, your code-block calls
6097 @code{discard_cleanups} with the old cleanup chain and thus ensures that
6098 any created cleanups are not performed. For instance, the following
6099 code segment, ensures that the file will be closed but only if there is
6103 FILE *file = fopen ("afile", "r");
6104 struct cleanup *old = make_cleanup (close_file, file);
6106 discard_cleanups (old);
6110 Some functions, e.g., @code{fputs_filtered()} or @code{error()}, specify
6111 that they ``should not be called when cleanups are not in place''. This
6112 means that any actions you need to reverse in the case of an error or
6113 interruption must be on the cleanup chain before you call these
6114 functions, since they might never return to your code (they
6115 @samp{longjmp} instead).
6117 @section Per-architecture module data
6118 @cindex per-architecture module data
6119 @cindex multi-arch data
6120 @cindex data-pointer, per-architecture/per-module
6122 The multi-arch framework includes a mechanism for adding module
6123 specific per-architecture data-pointers to the @code{struct gdbarch}
6124 architecture object.
6126 A module registers one or more per-architecture data-pointers using:
6128 @deftypefn {Architecture Function} {struct gdbarch_data *} gdbarch_data_register_pre_init (gdbarch_data_pre_init_ftype *@var{pre_init})
6129 @var{pre_init} is used to, on-demand, allocate an initial value for a
6130 per-architecture data-pointer using the architecture's obstack (passed
6131 in as a parameter). Since @var{pre_init} can be called during
6132 architecture creation, it is not parameterized with the architecture.
6133 and must not call modules that use per-architecture data.
6136 @deftypefn {Architecture Function} {struct gdbarch_data *} gdbarch_data_register_post_init (gdbarch_data_post_init_ftype *@var{post_init})
6137 @var{post_init} is used to obtain an initial value for a
6138 per-architecture data-pointer @emph{after}. Since @var{post_init} is
6139 always called after architecture creation, it both receives the fully
6140 initialized architecture and is free to call modules that use
6141 per-architecture data (care needs to be taken to ensure that those
6142 other modules do not try to call back to this module as that will
6143 create in cycles in the initialization call graph).
6146 These functions return a @code{struct gdbarch_data} that is used to
6147 identify the per-architecture data-pointer added for that module.
6149 The per-architecture data-pointer is accessed using the function:
6151 @deftypefn {Architecture Function} {void *} gdbarch_data (struct gdbarch *@var{gdbarch}, struct gdbarch_data *@var{data_handle})
6152 Given the architecture @var{arch} and module data handle
6153 @var{data_handle} (returned by @code{gdbarch_data_register_pre_init}
6154 or @code{gdbarch_data_register_post_init}), this function returns the
6155 current value of the per-architecture data-pointer. If the data
6156 pointer is @code{NULL}, it is first initialized by calling the
6157 corresponding @var{pre_init} or @var{post_init} method.
6160 The examples below assume the following definitions:
6163 struct nozel @{ int total; @};
6164 static struct gdbarch_data *nozel_handle;
6167 A module can extend the architecture vector, adding additional
6168 per-architecture data, using the @var{pre_init} method. The module's
6169 per-architecture data is then initialized during architecture
6172 In the below, the module's per-architecture @emph{nozel} is added. An
6173 architecture can specify its nozel by calling @code{set_gdbarch_nozel}
6174 from @code{gdbarch_init}.
6178 nozel_pre_init (struct obstack *obstack)
6180 struct nozel *data = OBSTACK_ZALLOC (obstack, struct nozel);
6187 set_gdbarch_nozel (struct gdbarch *gdbarch, int total)
6189 struct nozel *data = gdbarch_data (gdbarch, nozel_handle);
6190 data->total = nozel;
6194 A module can on-demand create architecture dependent data structures
6195 using @code{post_init}.
6197 In the below, the nozel's total is computed on-demand by
6198 @code{nozel_post_init} using information obtained from the
6203 nozel_post_init (struct gdbarch *gdbarch)
6205 struct nozel *data = GDBARCH_OBSTACK_ZALLOC (gdbarch, struct nozel);
6206 nozel->total = gdbarch@dots{} (gdbarch);
6213 nozel_total (struct gdbarch *gdbarch)
6215 struct nozel *data = gdbarch_data (gdbarch, nozel_handle);
6220 @section Wrapping Output Lines
6221 @cindex line wrap in output
6224 Output that goes through @code{printf_filtered} or @code{fputs_filtered}
6225 or @code{fputs_demangled} needs only to have calls to @code{wrap_here}
6226 added in places that would be good breaking points. The utility
6227 routines will take care of actually wrapping if the line width is
6230 The argument to @code{wrap_here} is an indentation string which is
6231 printed @emph{only} if the line breaks there. This argument is saved
6232 away and used later. It must remain valid until the next call to
6233 @code{wrap_here} or until a newline has been printed through the
6234 @code{*_filtered} functions. Don't pass in a local variable and then
6237 It is usually best to call @code{wrap_here} after printing a comma or
6238 space. If you call it before printing a space, make sure that your
6239 indentation properly accounts for the leading space that will print if
6240 the line wraps there.
6242 Any function or set of functions that produce filtered output must
6243 finish by printing a newline, to flush the wrap buffer, before switching
6244 to unfiltered (@code{printf}) output. Symbol reading routines that
6245 print warnings are a good example.
6247 @section Memory Management
6249 @value{GDBN} does not use the functions @code{malloc}, @code{realloc},
6250 @code{calloc}, @code{free} and @code{asprintf}.
6252 @value{GDBN} uses the functions @code{xmalloc}, @code{xrealloc} and
6253 @code{xcalloc} when allocating memory. Unlike @code{malloc} et.al.@:
6254 these functions do not return when the memory pool is empty. Instead,
6255 they unwind the stack using cleanups. These functions return
6256 @code{NULL} when requested to allocate a chunk of memory of size zero.
6258 @emph{Pragmatics: By using these functions, the need to check every
6259 memory allocation is removed. These functions provide portable
6262 @value{GDBN} does not use the function @code{free}.
6264 @value{GDBN} uses the function @code{xfree} to return memory to the
6265 memory pool. Consistent with ISO-C, this function ignores a request to
6266 free a @code{NULL} pointer.
6268 @emph{Pragmatics: On some systems @code{free} fails when passed a
6269 @code{NULL} pointer.}
6271 @value{GDBN} can use the non-portable function @code{alloca} for the
6272 allocation of small temporary values (such as strings).
6274 @emph{Pragmatics: This function is very non-portable. Some systems
6275 restrict the memory being allocated to no more than a few kilobytes.}
6277 @value{GDBN} uses the string function @code{xstrdup} and the print
6278 function @code{xstrprintf}.
6280 @emph{Pragmatics: @code{asprintf} and @code{strdup} can fail. Print
6281 functions such as @code{sprintf} are very prone to buffer overflow
6285 @section Compiler Warnings
6286 @cindex compiler warnings
6288 With few exceptions, developers should avoid the configuration option
6289 @samp{--disable-werror} when building @value{GDBN}. The exceptions
6290 are listed in the file @file{gdb/MAINTAINERS}. The default, when
6291 building with @sc{gcc}, is @samp{--enable-werror}.
6293 This option causes @value{GDBN} (when built using GCC) to be compiled
6294 with a carefully selected list of compiler warning flags. Any warnings
6295 from those flags are treated as errors.
6297 The current list of warning flags includes:
6301 Recommended @sc{gcc} warnings.
6303 @item -Wdeclaration-after-statement
6305 @sc{gcc} 3.x (and later) and @sc{c99} allow declarations mixed with
6306 code, but @sc{gcc} 2.x and @sc{c89} do not.
6308 @item -Wpointer-arith
6310 @item -Wformat-nonliteral
6311 Non-literal format strings, with a few exceptions, are bugs - they
6312 might contain unintended user-supplied format specifiers.
6313 Since @value{GDBN} uses the @code{format printf} attribute on all
6314 @code{printf} like functions this checks not just @code{printf} calls
6315 but also calls to functions such as @code{fprintf_unfiltered}.
6317 @item -Wno-pointer-sign
6318 In version 4.0, GCC began warning about pointer argument passing or
6319 assignment even when the source and destination differed only in
6320 signedness. However, most @value{GDBN} code doesn't distinguish
6321 carefully between @code{char} and @code{unsigned char}. In early 2006
6322 the @value{GDBN} developers decided correcting these warnings wasn't
6323 worth the time it would take.
6325 @item -Wno-unused-parameter
6326 Due to the way that @value{GDBN} is implemented many functions have
6327 unused parameters. Consequently this warning is avoided. The macro
6328 @code{ATTRIBUTE_UNUSED} is not used as it leads to false negatives ---
6329 it is not an error to have @code{ATTRIBUTE_UNUSED} on a parameter that
6334 @itemx -Wno-char-subscripts
6335 These are warnings which might be useful for @value{GDBN}, but are
6336 currently too noisy to enable with @samp{-Werror}.
6340 @section Internal Error Recovery
6342 During its execution, @value{GDBN} can encounter two types of errors.
6343 User errors and internal errors. User errors include not only a user
6344 entering an incorrect command but also problems arising from corrupt
6345 object files and system errors when interacting with the target.
6346 Internal errors include situations where @value{GDBN} has detected, at
6347 run time, a corrupt or erroneous situation.
6349 When reporting an internal error, @value{GDBN} uses
6350 @code{internal_error} and @code{gdb_assert}.
6352 @value{GDBN} must not call @code{abort} or @code{assert}.
6354 @emph{Pragmatics: There is no @code{internal_warning} function. Either
6355 the code detected a user error, recovered from it and issued a
6356 @code{warning} or the code failed to correctly recover from the user
6357 error and issued an @code{internal_error}.}
6359 @section Command Names
6361 GDB U/I commands are written @samp{foo-bar}, not @samp{foo_bar}.
6363 @section Clean Design and Portable Implementation
6366 In addition to getting the syntax right, there's the little question of
6367 semantics. Some things are done in certain ways in @value{GDBN} because long
6368 experience has shown that the more obvious ways caused various kinds of
6371 @cindex assumptions about targets
6372 You can't assume the byte order of anything that comes from a target
6373 (including @var{value}s, object files, and instructions). Such things
6374 must be byte-swapped using @code{SWAP_TARGET_AND_HOST} in
6375 @value{GDBN}, or one of the swap routines defined in @file{bfd.h},
6376 such as @code{bfd_get_32}.
6378 You can't assume that you know what interface is being used to talk to
6379 the target system. All references to the target must go through the
6380 current @code{target_ops} vector.
6382 You can't assume that the host and target machines are the same machine
6383 (except in the ``native'' support modules). In particular, you can't
6384 assume that the target machine's header files will be available on the
6385 host machine. Target code must bring along its own header files --
6386 written from scratch or explicitly donated by their owner, to avoid
6390 Insertion of new @code{#ifdef}'s will be frowned upon. It's much better
6391 to write the code portably than to conditionalize it for various
6394 @cindex system dependencies
6395 New @code{#ifdef}'s which test for specific compilers or manufacturers
6396 or operating systems are unacceptable. All @code{#ifdef}'s should test
6397 for features. The information about which configurations contain which
6398 features should be segregated into the configuration files. Experience
6399 has proven far too often that a feature unique to one particular system
6400 often creeps into other systems; and that a conditional based on some
6401 predefined macro for your current system will become worthless over
6402 time, as new versions of your system come out that behave differently
6403 with regard to this feature.
6405 Adding code that handles specific architectures, operating systems,
6406 target interfaces, or hosts, is not acceptable in generic code.
6408 @cindex portable file name handling
6409 @cindex file names, portability
6410 One particularly notorious area where system dependencies tend to
6411 creep in is handling of file names. The mainline @value{GDBN} code
6412 assumes Posix semantics of file names: absolute file names begin with
6413 a forward slash @file{/}, slashes are used to separate leading
6414 directories, case-sensitive file names. These assumptions are not
6415 necessarily true on non-Posix systems such as MS-Windows. To avoid
6416 system-dependent code where you need to take apart or construct a file
6417 name, use the following portable macros:
6420 @findex HAVE_DOS_BASED_FILE_SYSTEM
6421 @item HAVE_DOS_BASED_FILE_SYSTEM
6422 This preprocessing symbol is defined to a non-zero value on hosts
6423 whose filesystems belong to the MS-DOS/MS-Windows family. Use this
6424 symbol to write conditional code which should only be compiled for
6427 @findex IS_DIR_SEPARATOR
6428 @item IS_DIR_SEPARATOR (@var{c})
6429 Evaluates to a non-zero value if @var{c} is a directory separator
6430 character. On Unix and GNU/Linux systems, only a slash @file{/} is
6431 such a character, but on Windows, both @file{/} and @file{\} will
6434 @findex IS_ABSOLUTE_PATH
6435 @item IS_ABSOLUTE_PATH (@var{file})
6436 Evaluates to a non-zero value if @var{file} is an absolute file name.
6437 For Unix and GNU/Linux hosts, a name which begins with a slash
6438 @file{/} is absolute. On DOS and Windows, @file{d:/foo} and
6439 @file{x:\bar} are also absolute file names.
6441 @findex FILENAME_CMP
6442 @item FILENAME_CMP (@var{f1}, @var{f2})
6443 Calls a function which compares file names @var{f1} and @var{f2} as
6444 appropriate for the underlying host filesystem. For Posix systems,
6445 this simply calls @code{strcmp}; on case-insensitive filesystems it
6446 will call @code{strcasecmp} instead.
6448 @findex DIRNAME_SEPARATOR
6449 @item DIRNAME_SEPARATOR
6450 Evaluates to a character which separates directories in
6451 @code{PATH}-style lists, typically held in environment variables.
6452 This character is @samp{:} on Unix, @samp{;} on DOS and Windows.
6454 @findex SLASH_STRING
6456 This evaluates to a constant string you should use to produce an
6457 absolute filename from leading directories and the file's basename.
6458 @code{SLASH_STRING} is @code{"/"} on most systems, but might be
6459 @code{"\\"} for some Windows-based ports.
6462 In addition to using these macros, be sure to use portable library
6463 functions whenever possible. For example, to extract a directory or a
6464 basename part from a file name, use the @code{dirname} and
6465 @code{basename} library functions (available in @code{libiberty} for
6466 platforms which don't provide them), instead of searching for a slash
6467 with @code{strrchr}.
6469 Another way to generalize @value{GDBN} along a particular interface is with an
6470 attribute struct. For example, @value{GDBN} has been generalized to handle
6471 multiple kinds of remote interfaces---not by @code{#ifdef}s everywhere, but
6472 by defining the @code{target_ops} structure and having a current target (as
6473 well as a stack of targets below it, for memory references). Whenever
6474 something needs to be done that depends on which remote interface we are
6475 using, a flag in the current target_ops structure is tested (e.g.,
6476 @code{target_has_stack}), or a function is called through a pointer in the
6477 current target_ops structure. In this way, when a new remote interface
6478 is added, only one module needs to be touched---the one that actually
6479 implements the new remote interface. Other examples of
6480 attribute-structs are BFD access to multiple kinds of object file
6481 formats, or @value{GDBN}'s access to multiple source languages.
6483 Please avoid duplicating code. For example, in @value{GDBN} 3.x all
6484 the code interfacing between @code{ptrace} and the rest of
6485 @value{GDBN} was duplicated in @file{*-dep.c}, and so changing
6486 something was very painful. In @value{GDBN} 4.x, these have all been
6487 consolidated into @file{infptrace.c}. @file{infptrace.c} can deal
6488 with variations between systems the same way any system-independent
6489 file would (hooks, @code{#if defined}, etc.), and machines which are
6490 radically different don't need to use @file{infptrace.c} at all.
6492 All debugging code must be controllable using the @samp{set debug
6493 @var{module}} command. Do not use @code{printf} to print trace
6494 messages. Use @code{fprintf_unfiltered(gdb_stdlog, ...}. Do not use
6495 @code{#ifdef DEBUG}.
6499 @chapter Porting @value{GDBN}
6500 @cindex porting to new machines
6502 Most of the work in making @value{GDBN} compile on a new machine is in
6503 specifying the configuration of the machine. Porting a new
6504 architecture to @value{GDBN} can be broken into a number of steps.
6509 Ensure a @sc{bfd} exists for executables of the target architecture in
6510 the @file{bfd} directory. If one does not exist, create one by
6511 modifying an existing similar one.
6514 Implement a disassembler for the target architecture in the @file{opcodes}
6518 Define the target architecture in the @file{gdb} directory
6519 (@pxref{Adding a New Target, , Adding a New Target}). Add the pattern
6520 for the new target to @file{configure.tgt} with the names of the files
6521 that contain the code. By convention the target architecture
6522 definition for an architecture @var{arch} is placed in
6523 @file{@var{arch}-tdep.c}.
6525 Within @file{@var{arch}-tdep.c} define the function
6526 @code{_initialize_@var{arch}_tdep} which calls
6527 @code{gdbarch_register} to create the new @code{@w{struct
6528 gdbarch}} for the architecture.
6531 If a new remote target is needed, consider adding a new remote target
6532 by defining a function
6533 @code{_initialize_remote_@var{arch}}. However if at all possible
6534 use the @value{GDBN} @emph{Remote Serial Protocol} for this and implement
6535 the server side protocol independently with the target.
6538 If desired implement a simulator in the @file{sim} directory. This
6539 should create the library @file{libsim.a} implementing the interface
6540 in @file{remote-sim.h} (found in the @file{include} directory).
6543 Build and test. If desired, lobby the @sc{gdb} steering group to
6544 have the new port included in the main distribution!
6547 Add a description of the new architecture to the main @value{GDBN} user
6548 guide (@pxref{Configuration Specific Information, , Configuration
6549 Specific Information, gdb, Debugging with @value{GDBN}}).
6553 @node Versions and Branches
6554 @chapter Versions and Branches
6558 @value{GDBN}'s version is determined by the file
6559 @file{gdb/version.in} and takes one of the following forms:
6562 @item @var{major}.@var{minor}
6563 @itemx @var{major}.@var{minor}.@var{patchlevel}
6564 an official release (e.g., 6.2 or 6.2.1)
6565 @item @var{major}.@var{minor}.@var{patchlevel}.@var{YYYY}@var{MM}@var{DD}
6566 a snapshot taken at @var{YYYY}-@var{MM}-@var{DD}-gmt (e.g.,
6567 6.1.50.20020302, 6.1.90.20020304, or 6.1.0.20020308)
6568 @item @var{major}.@var{minor}.@var{patchlevel}.@var{YYYY}@var{MM}@var{DD}-cvs
6569 a @sc{cvs} check out drawn on @var{YYYY}-@var{MM}-@var{DD} (e.g.,
6570 6.1.50.20020302-cvs, 6.1.90.20020304-cvs, or 6.1.0.20020308-cvs)
6571 @item @var{major}.@var{minor}.@var{patchlevel}.@var{YYYY}@var{MM}@var{DD} (@var{vendor})
6572 a vendor specific release of @value{GDBN}, that while based on@*
6573 @var{major}.@var{minor}.@var{patchlevel}.@var{YYYY}@var{MM}@var{DD},
6574 may include additional changes
6577 @value{GDBN}'s mainline uses the @var{major} and @var{minor} version
6578 numbers from the most recent release branch, with a @var{patchlevel}
6579 of 50. At the time each new release branch is created, the mainline's
6580 @var{major} and @var{minor} version numbers are updated.
6582 @value{GDBN}'s release branch is similar. When the branch is cut, the
6583 @var{patchlevel} is changed from 50 to 90. As draft releases are
6584 drawn from the branch, the @var{patchlevel} is incremented. Once the
6585 first release (@var{major}.@var{minor}) has been made, the
6586 @var{patchlevel} is set to 0 and updates have an incremented
6589 For snapshots, and @sc{cvs} check outs, it is also possible to
6590 identify the @sc{cvs} origin:
6593 @item @var{major}.@var{minor}.50.@var{YYYY}@var{MM}@var{DD}
6594 drawn from the @sc{head} of mainline @sc{cvs} (e.g., 6.1.50.20020302)
6595 @item @var{major}.@var{minor}.90.@var{YYYY}@var{MM}@var{DD}
6596 @itemx @var{major}.@var{minor}.91.@var{YYYY}@var{MM}@var{DD} @dots{}
6597 drawn from a release branch prior to the release (e.g.,
6599 @item @var{major}.@var{minor}.0.@var{YYYY}@var{MM}@var{DD}
6600 @itemx @var{major}.@var{minor}.1.@var{YYYY}@var{MM}@var{DD} @dots{}
6601 drawn from a release branch after the release (e.g., 6.2.0.20020308)
6604 If the previous @value{GDBN} version is 6.1 and the current version is
6605 6.2, then, substituting 6 for @var{major} and 1 or 2 for @var{minor},
6606 here's an illustration of a typical sequence:
6613 +--------------------------.
6616 6.2.50.20020303-cvs 6.1.90 (draft #1)
6618 6.2.50.20020304-cvs 6.1.90.20020304-cvs
6620 6.2.50.20020305-cvs 6.1.91 (draft #2)
6622 6.2.50.20020306-cvs 6.1.91.20020306-cvs
6624 6.2.50.20020307-cvs 6.2 (release)
6626 6.2.50.20020308-cvs 6.2.0.20020308-cvs
6628 6.2.50.20020309-cvs 6.2.1 (update)
6630 6.2.50.20020310-cvs <branch closed>
6634 +--------------------------.
6637 6.3.50.20020312-cvs 6.2.90 (draft #1)
6641 @section Release Branches
6642 @cindex Release Branches
6644 @value{GDBN} draws a release series (6.2, 6.2.1, @dots{}) from a
6645 single release branch, and identifies that branch using the @sc{cvs}
6649 gdb_@var{major}_@var{minor}-@var{YYYY}@var{MM}@var{DD}-branchpoint
6650 gdb_@var{major}_@var{minor}-branch
6651 gdb_@var{major}_@var{minor}-@var{YYYY}@var{MM}@var{DD}-release
6654 @emph{Pragmatics: To help identify the date at which a branch or
6655 release is made, both the branchpoint and release tags include the
6656 date that they are cut (@var{YYYY}@var{MM}@var{DD}) in the tag. The
6657 branch tag, denoting the head of the branch, does not need this.}
6659 @section Vendor Branches
6660 @cindex vendor branches
6662 To avoid version conflicts, vendors are expected to modify the file
6663 @file{gdb/version.in} to include a vendor unique alphabetic identifier
6664 (an official @value{GDBN} release never uses alphabetic characters in
6665 its version identifier). E.g., @samp{6.2widgit2}, or @samp{6.2 (Widgit
6668 @section Experimental Branches
6669 @cindex experimental branches
6671 @subsection Guidelines
6673 @value{GDBN} permits the creation of branches, cut from the @sc{cvs}
6674 repository, for experimental development. Branches make it possible
6675 for developers to share preliminary work, and maintainers to examine
6676 significant new developments.
6678 The following are a set of guidelines for creating such branches:
6682 @item a branch has an owner
6683 The owner can set further policy for a branch, but may not change the
6684 ground rules. In particular, they can set a policy for commits (be it
6685 adding more reviewers or deciding who can commit).
6687 @item all commits are posted
6688 All changes committed to a branch shall also be posted to
6689 @email{gdb-patches@@sourceware.org, the @value{GDBN} patches
6690 mailing list}. While commentary on such changes are encouraged, people
6691 should remember that the changes only apply to a branch.
6693 @item all commits are covered by an assignment
6694 This ensures that all changes belong to the Free Software Foundation,
6695 and avoids the possibility that the branch may become contaminated.
6697 @item a branch is focused
6698 A focused branch has a single objective or goal, and does not contain
6699 unnecessary or irrelevant changes. Cleanups, where identified, being
6700 be pushed into the mainline as soon as possible.
6702 @item a branch tracks mainline
6703 This keeps the level of divergence under control. It also keeps the
6704 pressure on developers to push cleanups and other stuff into the
6707 @item a branch shall contain the entire @value{GDBN} module
6708 The @value{GDBN} module @code{gdb} should be specified when creating a
6709 branch (branches of individual files should be avoided). @xref{Tags}.
6711 @item a branch shall be branded using @file{version.in}
6712 The file @file{gdb/version.in} shall be modified so that it identifies
6713 the branch @var{owner} and branch @var{name}, e.g.,
6714 @samp{6.2.50.20030303_owner_name} or @samp{6.2 (Owner Name)}.
6721 To simplify the identification of @value{GDBN} branches, the following
6722 branch tagging convention is strongly recommended:
6726 @item @var{owner}_@var{name}-@var{YYYYMMDD}-branchpoint
6727 @itemx @var{owner}_@var{name}-@var{YYYYMMDD}-branch
6728 The branch point and corresponding branch tag. @var{YYYYMMDD} is the
6729 date that the branch was created. A branch is created using the
6730 sequence: @anchor{experimental branch tags}
6732 cvs rtag @var{owner}_@var{name}-@var{YYYYMMDD}-branchpoint gdb
6733 cvs rtag -b -r @var{owner}_@var{name}-@var{YYYYMMDD}-branchpoint \
6734 @var{owner}_@var{name}-@var{YYYYMMDD}-branch gdb
6737 @item @var{owner}_@var{name}-@var{yyyymmdd}-mergepoint
6738 The tagged point, on the mainline, that was used when merging the branch
6739 on @var{yyyymmdd}. To merge in all changes since the branch was cut,
6740 use a command sequence like:
6742 cvs rtag @var{owner}_@var{name}-@var{yyyymmdd}-mergepoint gdb
6744 -j@var{owner}_@var{name}-@var{YYYYMMDD}-branchpoint
6745 -j@var{owner}_@var{name}-@var{yyyymmdd}-mergepoint
6748 Similar sequences can be used to just merge in changes since the last
6754 For further information on @sc{cvs}, see
6755 @uref{http://www.gnu.org/software/cvs/, Concurrent Versions System}.
6757 @node Start of New Year Procedure
6758 @chapter Start of New Year Procedure
6759 @cindex new year procedure
6761 At the start of each new year, the following actions should be performed:
6765 Rotate the ChangeLog file
6767 The current @file{ChangeLog} file should be renamed into
6768 @file{ChangeLog-YYYY} where YYYY is the year that has just passed.
6769 A new @file{ChangeLog} file should be created, and its contents should
6770 contain a reference to the previous ChangeLog. The following should
6771 also be preserved at the end of the new ChangeLog, in order to provide
6772 the appropriate settings when editing this file with Emacs:
6778 version-control: never
6784 Add an entry for the newly created ChangeLog file (@file{ChangeLog-YYYY})
6785 in @file{gdb/config/djgpp/fnchange.lst}.
6788 Update the copyright year in the startup message
6790 Update the copyright year in:
6793 file @file{top.c}, function @code{print_gdb_version}
6795 file @file{gdbserver/server.c}, function @code{gdbserver_version}
6797 file @file{gdbserver/gdbreplay.c}, function @code{gdbreplay_version}
6801 Run the @file{copyright.sh} script to add the new year in the copyright
6802 notices of most source files. This script requires Emacs 22 or later to
6806 The new year also needs to be added manually in all other files that
6807 are not already taken care of by the @file{copyright.sh} script:
6835 @chapter Releasing @value{GDBN}
6836 @cindex making a new release of gdb
6838 @section Branch Commit Policy
6840 The branch commit policy is pretty slack. @value{GDBN} releases 5.0,
6841 5.1 and 5.2 all used the below:
6845 The @file{gdb/MAINTAINERS} file still holds.
6847 Don't fix something on the branch unless/until it is also fixed in the
6848 trunk. If this isn't possible, mentioning it in the @file{gdb/PROBLEMS}
6849 file is better than committing a hack.
6851 When considering a patch for the branch, suggested criteria include:
6852 Does it fix a build? Does it fix the sequence @kbd{break main; run}
6853 when debugging a static binary?
6855 The further a change is from the core of @value{GDBN}, the less likely
6856 the change will worry anyone (e.g., target specific code).
6858 Only post a proposal to change the core of @value{GDBN} after you've
6859 sent individual bribes to all the people listed in the
6860 @file{MAINTAINERS} file @t{;-)}
6863 @emph{Pragmatics: Provided updates are restricted to non-core
6864 functionality there is little chance that a broken change will be fatal.
6865 This means that changes such as adding a new architectures or (within
6866 reason) support for a new host are considered acceptable.}
6869 @section Obsoleting code
6871 Before anything else, poke the other developers (and around the source
6872 code) to see if there is anything that can be removed from @value{GDBN}
6873 (an old target, an unused file).
6875 Obsolete code is identified by adding an @code{OBSOLETE} prefix to every
6876 line. Doing this means that it is easy to identify something that has
6877 been obsoleted when greping through the sources.
6879 The process is done in stages --- this is mainly to ensure that the
6880 wider @value{GDBN} community has a reasonable opportunity to respond.
6881 Remember, everything on the Internet takes a week.
6885 Post the proposal on @email{gdb@@sourceware.org, the GDB mailing
6886 list} Creating a bug report to track the task's state, is also highly
6891 Post the proposal on @email{gdb-announce@@sourceware.org, the GDB
6892 Announcement mailing list}.
6896 Go through and edit all relevant files and lines so that they are
6897 prefixed with the word @code{OBSOLETE}.
6899 Wait until the next GDB version, containing this obsolete code, has been
6902 Remove the obsolete code.
6906 @emph{Maintainer note: While removing old code is regrettable it is
6907 hopefully better for @value{GDBN}'s long term development. Firstly it
6908 helps the developers by removing code that is either no longer relevant
6909 or simply wrong. Secondly since it removes any history associated with
6910 the file (effectively clearing the slate) the developer has a much freer
6911 hand when it comes to fixing broken files.}
6915 @section Before the Branch
6917 The most important objective at this stage is to find and fix simple
6918 changes that become a pain to track once the branch is created. For
6919 instance, configuration problems that stop @value{GDBN} from even
6920 building. If you can't get the problem fixed, document it in the
6921 @file{gdb/PROBLEMS} file.
6923 @subheading Prompt for @file{gdb/NEWS}
6925 People always forget. Send a post reminding them but also if you know
6926 something interesting happened add it yourself. The @code{schedule}
6927 script will mention this in its e-mail.
6929 @subheading Review @file{gdb/README}
6931 Grab one of the nightly snapshots and then walk through the
6932 @file{gdb/README} looking for anything that can be improved. The
6933 @code{schedule} script will mention this in its e-mail.
6935 @subheading Refresh any imported files.
6937 A number of files are taken from external repositories. They include:
6941 @file{texinfo/texinfo.tex}
6943 @file{config.guess} et.@: al.@: (see the top-level @file{MAINTAINERS}
6946 @file{etc/standards.texi}, @file{etc/make-stds.texi}
6949 @subheading Check the ARI
6951 @uref{http://sourceware.org/gdb/ari,,A.R.I.} is an @code{awk} script
6952 (Awk Regression Index ;-) that checks for a number of errors and coding
6953 conventions. The checks include things like using @code{malloc} instead
6954 of @code{xmalloc} and file naming problems. There shouldn't be any
6957 @subsection Review the bug data base
6959 Close anything obviously fixed.
6961 @subsection Check all cross targets build
6963 The targets are listed in @file{gdb/MAINTAINERS}.
6966 @section Cut the Branch
6968 @subheading Create the branch
6973 $ V=`echo $v | sed 's/\./_/g'`
6974 $ D=`date -u +%Y-%m-%d`
6977 $ echo cvs -f -d :ext:sourceware.org:/cvs/src rtag \
6978 -D $D-gmt gdb_$V-$D-branchpoint insight
6979 cvs -f -d :ext:sourceware.org:/cvs/src rtag
6980 -D 2002-03-03-gmt gdb_5_2-2002-03-03-branchpoint insight
6983 $ echo cvs -f -d :ext:sourceware.org:/cvs/src rtag \
6984 -b -r gdb_$V-$D-branchpoint gdb_$V-branch insight
6985 cvs -f -d :ext:sourceware.org:/cvs/src rtag \
6986 -b -r gdb_5_2-2002-03-03-branchpoint gdb_5_2-branch insight
6994 By using @kbd{-D YYYY-MM-DD-gmt}, the branch is forced to an exact
6997 The trunk is first tagged so that the branch point can easily be found.
6999 Insight, which includes @value{GDBN}, is tagged at the same time.
7001 @file{version.in} gets bumped to avoid version number conflicts.
7003 The reading of @file{.cvsrc} is disabled using @file{-f}.
7006 @subheading Update @file{version.in}
7011 $ V=`echo $v | sed 's/\./_/g'`
7015 $ echo cvs -f -d :ext:sourceware.org:/cvs/src co \
7016 -r gdb_$V-branch src/gdb/version.in
7017 cvs -f -d :ext:sourceware.org:/cvs/src co
7018 -r gdb_5_2-branch src/gdb/version.in
7020 U src/gdb/version.in
7022 $ echo $u.90-0000-00-00-cvs > version.in
7024 5.1.90-0000-00-00-cvs
7025 $ cvs -f commit version.in
7030 @file{0000-00-00} is used as a date to pump prime the version.in update
7033 @file{.90} and the previous branch version are used as fairly arbitrary
7034 initial branch version number.
7038 @subheading Update the web and news pages
7042 @subheading Tweak cron to track the new branch
7044 The file @file{gdbadmin/cron/crontab} contains gdbadmin's cron table.
7045 This file needs to be updated so that:
7049 A daily timestamp is added to the file @file{version.in}.
7051 The new branch is included in the snapshot process.
7055 See the file @file{gdbadmin/cron/README} for how to install the updated
7058 The file @file{gdbadmin/ss/README} should also be reviewed to reflect
7059 any changes. That file is copied to both the branch/ and current/
7060 snapshot directories.
7063 @subheading Update the NEWS and README files
7065 The @file{NEWS} file needs to be updated so that on the branch it refers
7066 to @emph{changes in the current release} while on the trunk it also
7067 refers to @emph{changes since the current release}.
7069 The @file{README} file needs to be updated so that it refers to the
7072 @subheading Post the branch info
7074 Send an announcement to the mailing lists:
7078 @email{gdb-announce@@sourceware.org, GDB Announcement mailing list}
7080 @email{gdb@@sourceware.org, GDB Discussion mailing list} and
7081 @email{gdb-testers@@sourceware.org, GDB Testers mailing list}
7084 @emph{Pragmatics: The branch creation is sent to the announce list to
7085 ensure that people people not subscribed to the higher volume discussion
7088 The announcement should include:
7094 How to check out the branch using CVS.
7096 The date/number of weeks until the release.
7098 The branch commit policy still holds.
7101 @section Stabilize the branch
7103 Something goes here.
7105 @section Create a Release
7107 The process of creating and then making available a release is broken
7108 down into a number of stages. The first part addresses the technical
7109 process of creating a releasable tar ball. The later stages address the
7110 process of releasing that tar ball.
7112 When making a release candidate just the first section is needed.
7114 @subsection Create a release candidate
7116 The objective at this stage is to create a set of tar balls that can be
7117 made available as a formal release (or as a less formal release
7120 @subsubheading Freeze the branch
7122 Send out an e-mail notifying everyone that the branch is frozen to
7123 @email{gdb-patches@@sourceware.org}.
7125 @subsubheading Establish a few defaults.
7130 $ t=/sourceware/snapshot-tmp/gdbadmin-tmp
7132 /sourceware/snapshot-tmp/gdbadmin-tmp/gdb_5_2-branch/5.2
7136 /sourceware/snapshot-tmp/gdbadmin-tmp/gdb_5_2-branch/5.2
7138 /home/gdbadmin/bin/autoconf
7147 Check the @code{autoconf} version carefully. You want to be using the
7148 version documented in the toplevel @file{README-maintainer-mode} file.
7149 It is very unlikely that the version of @code{autoconf} installed in
7150 system directories (e.g., @file{/usr/bin/autoconf}) is correct.
7153 @subsubheading Check out the relevant modules:
7156 $ for m in gdb insight
7158 ( mkdir -p $m && cd $m && cvs -q -f -d /cvs/src co -P -r $b $m )
7168 The reading of @file{.cvsrc} is disabled (@file{-f}) so that there isn't
7169 any confusion between what is written here and what your local
7170 @code{cvs} really does.
7173 @subsubheading Update relevant files.
7179 Major releases get their comments added as part of the mainline. Minor
7180 releases should probably mention any significant bugs that were fixed.
7182 Don't forget to include the @file{ChangeLog} entry.
7185 $ emacs gdb/src/gdb/NEWS
7190 $ cp gdb/src/gdb/NEWS insight/src/gdb/NEWS
7191 $ cp gdb/src/gdb/ChangeLog insight/src/gdb/ChangeLog
7196 You'll need to update:
7208 $ emacs gdb/src/gdb/README
7213 $ cp gdb/src/gdb/README insight/src/gdb/README
7214 $ cp gdb/src/gdb/ChangeLog insight/src/gdb/ChangeLog
7217 @emph{Maintainer note: Hopefully the @file{README} file was reviewed
7218 before the initial branch was cut so just a simple substitute is needed
7221 @emph{Maintainer note: Other projects generate @file{README} and
7222 @file{INSTALL} from the core documentation. This might be worth
7225 @item gdb/version.in
7228 $ echo $v > gdb/src/gdb/version.in
7229 $ cat gdb/src/gdb/version.in
7231 $ emacs gdb/src/gdb/version.in
7234 ... Bump to version ...
7236 $ cp gdb/src/gdb/version.in insight/src/gdb/version.in
7237 $ cp gdb/src/gdb/ChangeLog insight/src/gdb/ChangeLog
7242 @subsubheading Do the dirty work
7244 This is identical to the process used to create the daily snapshot.
7247 $ for m in gdb insight
7249 ( cd $m/src && gmake -f src-release $m.tar )
7253 If the top level source directory does not have @file{src-release}
7254 (@value{GDBN} version 5.3.1 or earlier), try these commands instead:
7257 $ for m in gdb insight
7259 ( cd $m/src && gmake -f Makefile.in $m.tar )
7263 @subsubheading Check the source files
7265 You're looking for files that have mysteriously disappeared.
7266 @kbd{distclean} has the habit of deleting files it shouldn't. Watch out
7267 for the @file{version.in} update @kbd{cronjob}.
7270 $ ( cd gdb/src && cvs -f -q -n update )
7274 @dots{} lots of generated files @dots{}
7279 @dots{} lots of generated files @dots{}
7284 @emph{Don't worry about the @file{gdb.info-??} or
7285 @file{gdb/p-exp.tab.c}. They were generated (and yes @file{gdb.info-1}
7286 was also generated only something strange with CVS means that they
7287 didn't get suppressed). Fixing it would be nice though.}
7289 @subsubheading Create compressed versions of the release
7295 gdb/ gdb-5.2.tar insight/ insight-5.2.tar
7296 $ for m in gdb insight
7298 bzip2 -v -9 -c $m-$v.tar > $m-$v.tar.bz2
7299 gzip -v -9 -c $m-$v.tar > $m-$v.tar.gz
7309 A pipe such as @kbd{bunzip2 < xxx.bz2 | gzip -9 > xxx.gz} is not since,
7310 in that mode, @code{gzip} does not know the name of the file and, hence,
7311 can not include it in the compressed file. This is also why the release
7312 process runs @code{tar} and @code{bzip2} as separate passes.
7315 @subsection Sanity check the tar ball
7317 Pick a popular machine (Solaris/PPC?) and try the build on that.
7320 $ bunzip2 < gdb-5.2.tar.bz2 | tar xpf -
7325 $ ./gdb/gdb ./gdb/gdb
7329 Breakpoint 1 at 0x80732bc: file main.c, line 734.
7331 Starting program: /tmp/gdb-5.2/gdb/gdb
7333 Breakpoint 1, main (argc=1, argv=0xbffff8b4) at main.c:734
7334 734 catch_errors (captured_main, &args, "", RETURN_MASK_ALL);
7336 $1 = @{argc = 136426532, argv = 0x821b7f0@}
7340 @subsection Make a release candidate available
7342 If this is a release candidate then the only remaining steps are:
7346 Commit @file{version.in} and @file{ChangeLog}
7348 Tweak @file{version.in} (and @file{ChangeLog} to read
7349 @var{L}.@var{M}.@var{N}-0000-00-00-cvs so that the version update
7350 process can restart.
7352 Make the release candidate available in
7353 @uref{ftp://sourceware.org/pub/gdb/snapshots/branch}
7355 Notify the relevant mailing lists ( @email{gdb@@sourceware.org} and
7356 @email{gdb-testers@@sourceware.org} that the candidate is available.
7359 @subsection Make a formal release available
7361 (And you thought all that was required was to post an e-mail.)
7363 @subsubheading Install on sware
7365 Copy the new files to both the release and the old release directory:
7368 $ cp *.bz2 *.gz ~ftp/pub/gdb/old-releases/
7369 $ cp *.bz2 *.gz ~ftp/pub/gdb/releases
7373 Clean up the releases directory so that only the most recent releases
7374 are available (e.g.@: keep 5.2 and 5.2.1 but remove 5.1):
7377 $ cd ~ftp/pub/gdb/releases
7382 Update the file @file{README} and @file{.message} in the releases
7389 $ ln README .message
7392 @subsubheading Update the web pages.
7396 @item htdocs/download/ANNOUNCEMENT
7397 This file, which is posted as the official announcement, includes:
7400 General announcement.
7402 News. If making an @var{M}.@var{N}.1 release, retain the news from
7403 earlier @var{M}.@var{N} release.
7408 @item htdocs/index.html
7409 @itemx htdocs/news/index.html
7410 @itemx htdocs/download/index.html
7411 These files include:
7414 Announcement of the most recent release.
7416 News entry (remember to update both the top level and the news directory).
7418 These pages also need to be regenerate using @code{index.sh}.
7420 @item download/onlinedocs/
7421 You need to find the magic command that is used to generate the online
7422 docs from the @file{.tar.bz2}. The best way is to look in the output
7423 from one of the nightly @code{cron} jobs and then just edit accordingly.
7427 $ ~/ss/update-web-docs \
7428 ~ftp/pub/gdb/releases/gdb-5.2.tar.bz2 \
7430 /www/sourceware/htdocs/gdb/download/onlinedocs \
7435 Just like the online documentation. Something like:
7438 $ /bin/sh ~/ss/update-web-ari \
7439 ~ftp/pub/gdb/releases/gdb-5.2.tar.bz2 \
7441 /www/sourceware/htdocs/gdb/download/ari \
7447 @subsubheading Shadow the pages onto gnu
7449 Something goes here.
7452 @subsubheading Install the @value{GDBN} tar ball on GNU
7454 At the time of writing, the GNU machine was @kbd{gnudist.gnu.org} in
7455 @file{~ftp/gnu/gdb}.
7457 @subsubheading Make the @file{ANNOUNCEMENT}
7459 Post the @file{ANNOUNCEMENT} file you created above to:
7463 @email{gdb-announce@@sourceware.org, GDB Announcement mailing list}
7465 @email{info-gnu@@gnu.org, General GNU Announcement list} (but delay it a
7466 day or so to let things get out)
7468 @email{bug-gdb@@gnu.org, GDB Bug Report mailing list}
7473 The release is out but you're still not finished.
7475 @subsubheading Commit outstanding changes
7477 In particular you'll need to commit any changes to:
7481 @file{gdb/ChangeLog}
7483 @file{gdb/version.in}
7490 @subsubheading Tag the release
7495 $ d=`date -u +%Y-%m-%d`
7498 $ ( cd insight/src/gdb && cvs -f -q update )
7499 $ ( cd insight/src && cvs -f -q tag gdb_5_2-$d-release )
7502 Insight is used since that contains more of the release than
7505 @subsubheading Mention the release on the trunk
7507 Just put something in the @file{ChangeLog} so that the trunk also
7508 indicates when the release was made.
7510 @subsubheading Restart @file{gdb/version.in}
7512 If @file{gdb/version.in} does not contain an ISO date such as
7513 @kbd{2002-01-24} then the daily @code{cronjob} won't update it. Having
7514 committed all the release changes it can be set to
7515 @file{5.2.0_0000-00-00-cvs} which will restart things (yes the @kbd{_}
7516 is important - it affects the snapshot process).
7518 Don't forget the @file{ChangeLog}.
7520 @subsubheading Merge into trunk
7522 The files committed to the branch may also need changes merged into the
7525 @subsubheading Revise the release schedule
7527 Post a revised release schedule to @email{gdb@@sourceware.org, GDB
7528 Discussion List} with an updated announcement. The schedule can be
7529 generated by running:
7532 $ ~/ss/schedule `date +%s` schedule
7536 The first parameter is approximate date/time in seconds (from the epoch)
7537 of the most recent release.
7539 Also update the schedule @code{cronjob}.
7541 @section Post release
7543 Remove any @code{OBSOLETE} code.
7550 The testsuite is an important component of the @value{GDBN} package.
7551 While it is always worthwhile to encourage user testing, in practice
7552 this is rarely sufficient; users typically use only a small subset of
7553 the available commands, and it has proven all too common for a change
7554 to cause a significant regression that went unnoticed for some time.
7556 The @value{GDBN} testsuite uses the DejaGNU testing framework. The
7557 tests themselves are calls to various @code{Tcl} procs; the framework
7558 runs all the procs and summarizes the passes and fails.
7560 @section Using the Testsuite
7562 @cindex running the test suite
7563 To run the testsuite, simply go to the @value{GDBN} object directory (or to the
7564 testsuite's objdir) and type @code{make check}. This just sets up some
7565 environment variables and invokes DejaGNU's @code{runtest} script. While
7566 the testsuite is running, you'll get mentions of which test file is in use,
7567 and a mention of any unexpected passes or fails. When the testsuite is
7568 finished, you'll get a summary that looks like this:
7573 # of expected passes 6016
7574 # of unexpected failures 58
7575 # of unexpected successes 5
7576 # of expected failures 183
7577 # of unresolved testcases 3
7578 # of untested testcases 5
7581 To run a specific test script, type:
7583 make check RUNTESTFLAGS='@var{tests}'
7585 where @var{tests} is a list of test script file names, separated by
7588 If you use GNU make, you can use its @option{-j} option to run the
7589 testsuite in parallel. This can greatly reduce the amount of time it
7590 takes for the testsuite to run. In this case, if you set
7591 @code{RUNTESTFLAGS} then, by default, the tests will be run serially
7592 even under @option{-j}. You can override this and force a parallel run
7593 by setting the @code{make} variable @code{FORCE_PARALLEL} to any
7594 non-empty value. Note that the parallel @kbd{make check} assumes
7595 that you want to run the entire testsuite, so it is not compatible
7596 with some dejagnu options, like @option{--directory}.
7598 The ideal test run consists of expected passes only; however, reality
7599 conspires to keep us from this ideal. Unexpected failures indicate
7600 real problems, whether in @value{GDBN} or in the testsuite. Expected
7601 failures are still failures, but ones which have been decided are too
7602 hard to deal with at the time; for instance, a test case might work
7603 everywhere except on AIX, and there is no prospect of the AIX case
7604 being fixed in the near future. Expected failures should not be added
7605 lightly, since you may be masking serious bugs in @value{GDBN}.
7606 Unexpected successes are expected fails that are passing for some
7607 reason, while unresolved and untested cases often indicate some minor
7608 catastrophe, such as the compiler being unable to deal with a test
7611 When making any significant change to @value{GDBN}, you should run the
7612 testsuite before and after the change, to confirm that there are no
7613 regressions. Note that truly complete testing would require that you
7614 run the testsuite with all supported configurations and a variety of
7615 compilers; however this is more than really necessary. In many cases
7616 testing with a single configuration is sufficient. Other useful
7617 options are to test one big-endian (Sparc) and one little-endian (x86)
7618 host, a cross config with a builtin simulator (powerpc-eabi,
7619 mips-elf), or a 64-bit host (Alpha).
7621 If you add new functionality to @value{GDBN}, please consider adding
7622 tests for it as well; this way future @value{GDBN} hackers can detect
7623 and fix their changes that break the functionality you added.
7624 Similarly, if you fix a bug that was not previously reported as a test
7625 failure, please add a test case for it. Some cases are extremely
7626 difficult to test, such as code that handles host OS failures or bugs
7627 in particular versions of compilers, and it's OK not to try to write
7628 tests for all of those.
7630 DejaGNU supports separate build, host, and target machines. However,
7631 some @value{GDBN} test scripts do not work if the build machine and
7632 the host machine are not the same. In such an environment, these scripts
7633 will give a result of ``UNRESOLVED'', like this:
7636 UNRESOLVED: gdb.base/example.exp: This test script does not work on a remote host.
7639 @section Testsuite Parameters
7641 Several variables exist to modify the behavior of the testsuite.
7645 @item @code{TRANSCRIPT}
7647 Sometimes it is convenient to get a transcript of the commands which
7648 the testsuite sends to @value{GDBN}. For example, if @value{GDBN}
7649 crashes during testing, a transcript can be used to more easily
7650 reconstruct the failure when running @value{GDBN} under @value{GDBN}.
7652 You can instruct the @value{GDBN} testsuite to write transcripts by
7653 setting the DejaGNU variable @code{TRANSCRIPT} (to any value)
7654 before invoking @code{runtest} or @kbd{make check}. The transcripts
7655 will be written into DejaGNU's output directory. One transcript will
7656 be made for each invocation of @value{GDBN}; they will be named
7657 @file{transcript.@var{n}}, where @var{n} is an integer. The first
7658 line of the transcript file will show how @value{GDBN} was invoked;
7659 each subsequent line is a command sent as input to @value{GDBN}.
7662 make check RUNTESTFLAGS=TRANSCRIPT=y
7665 Note that the transcript is not always complete. In particular, tests
7666 of completion can yield partial command lines.
7670 Sometimes one wishes to test a different @value{GDBN} than the one in the build
7671 directory. For example, one may wish to run the testsuite on
7672 @file{/usr/bin/gdb}.
7675 make check RUNTESTFLAGS=GDB=/usr/bin/gdb
7678 @item @code{GDBSERVER}
7680 When testing a different @value{GDBN}, it is often useful to also test a
7681 different gdbserver.
7684 make check RUNTESTFLAGS="GDB=/usr/bin/gdb GDBSERVER=/usr/bin/gdbserver"
7687 @item @code{INTERNAL_GDBFLAGS}
7689 When running the testsuite normally one doesn't want whatever is in
7690 @file{~/.gdbinit} to interfere with the tests, therefore the test harness
7691 passes @option{-nx} to @value{GDBN}. One also doesn't want any windowed
7692 version of @value{GDBN}, e.g., @command{gdbtui}, to run.
7693 This is achieved via @code{INTERNAL_GDBFLAGS}.
7696 set INTERNAL_GDBFLAGS "-nw -nx"
7699 This is all well and good, except when testing an installed @value{GDBN}
7700 that has been configured with @option{--with-system-gdbinit}. Here one
7701 does not want @file{~/.gdbinit} loaded but one may want the system
7702 @file{.gdbinit} file loaded. This can be achieved by pointing @code{$HOME}
7703 at a directory without a @file{.gdbinit} and by overriding
7704 @code{INTERNAL_GDBFLAGS} and removing @option{-nx}.
7708 HOME=`pwd` runtest \
7710 GDBSERVER=/usr/bin/gdbserver \
7711 INTERNAL_GDBFLAGS=-nw
7716 There are two ways to run the testsuite and pass additional parameters
7717 to DejaGnu. The first is with @kbd{make check} and specifying the
7718 makefile variable @samp{RUNTESTFLAGS}.
7721 make check RUNTESTFLAGS=TRANSCRIPT=y
7724 The second is to cd to the @file{testsuite} directory and invoke the DejaGnu
7725 @command{runtest} command directly.
7730 runtest TRANSCRIPT=y
7733 @section Testsuite Configuration
7734 @cindex Testsuite Configuration
7736 It is possible to adjust the behavior of the testsuite by defining
7737 the global variables listed below, either in a @file{site.exp} file,
7742 @item @code{gdb_test_timeout}
7744 Defining this variable changes the default timeout duration used during
7745 communication with @value{GDBN}. More specifically, the global variable
7746 used during testing is @code{timeout}, but this variable gets reset to
7747 @code{gdb_test_timeout} at the beginning of each testcase, making sure
7748 that any local change to @code{timeout} in a testcase does not affect
7749 subsequent testcases.
7751 This global variable comes in handy when the debugger is slower than
7752 normal due to the testing environment, triggering unexpected @code{TIMEOUT}
7753 test failures. Examples include when testing on a remote machine, or
7754 against a system where communications are slow.
7756 If not specifically defined, this variable gets automatically defined
7757 to the same value as @code{timeout} during the testsuite initialization.
7758 The default value of the timeout is defined in the file
7759 @file{gdb/testsuite/config/unix.exp} that is part of the @value{GDBN}
7760 test suite@footnote{If you are using a board file, it could override
7761 the test-suite default; search the board file for "timeout".}.
7765 @section Testsuite Organization
7767 @cindex test suite organization
7768 The testsuite is entirely contained in @file{gdb/testsuite}. While the
7769 testsuite includes some makefiles and configury, these are very minimal,
7770 and used for little besides cleaning up, since the tests themselves
7771 handle the compilation of the programs that @value{GDBN} will run. The file
7772 @file{testsuite/lib/gdb.exp} contains common utility procs useful for
7773 all @value{GDBN} tests, while the directory @file{testsuite/config} contains
7774 configuration-specific files, typically used for special-purpose
7775 definitions of procs like @code{gdb_load} and @code{gdb_start}.
7777 The tests themselves are to be found in @file{testsuite/gdb.*} and
7778 subdirectories of those. The names of the test files must always end
7779 with @file{.exp}. DejaGNU collects the test files by wildcarding
7780 in the test directories, so both subdirectories and individual files
7781 get chosen and run in alphabetical order.
7783 The following table lists the main types of subdirectories and what they
7784 are for. Since DejaGNU finds test files no matter where they are
7785 located, and since each test file sets up its own compilation and
7786 execution environment, this organization is simply for convenience and
7791 This is the base testsuite. The tests in it should apply to all
7792 configurations of @value{GDBN} (but generic native-only tests may live here).
7793 The test programs should be in the subset of C that is valid K&R,
7794 ANSI/ISO, and C@t{++} (@code{#ifdef}s are allowed if necessary, for instance
7797 @item gdb.@var{lang}
7798 Language-specific tests for any language @var{lang} besides C. Examples are
7799 @file{gdb.cp} and @file{gdb.java}.
7801 @item gdb.@var{platform}
7802 Non-portable tests. The tests are specific to a specific configuration
7803 (host or target), such as HP-UX or eCos. Example is @file{gdb.hp}, for
7806 @item gdb.@var{compiler}
7807 Tests specific to a particular compiler. As of this writing (June
7808 1999), there aren't currently any groups of tests in this category that
7809 couldn't just as sensibly be made platform-specific, but one could
7810 imagine a @file{gdb.gcc}, for tests of @value{GDBN}'s handling of GCC
7813 @item gdb.@var{subsystem}
7814 Tests that exercise a specific @value{GDBN} subsystem in more depth. For
7815 instance, @file{gdb.disasm} exercises various disassemblers, while
7816 @file{gdb.stabs} tests pathways through the stabs symbol reader.
7819 @section Writing Tests
7820 @cindex writing tests
7822 In many areas, the @value{GDBN} tests are already quite comprehensive; you
7823 should be able to copy existing tests to handle new cases.
7825 You should try to use @code{gdb_test} whenever possible, since it
7826 includes cases to handle all the unexpected errors that might happen.
7827 However, it doesn't cost anything to add new test procedures; for
7828 instance, @file{gdb.base/exprs.exp} defines a @code{test_expr} that
7829 calls @code{gdb_test} multiple times.
7831 Only use @code{send_gdb} and @code{gdb_expect} when absolutely
7832 necessary. Even if @value{GDBN} has several valid responses to
7833 a command, you can use @code{gdb_test_multiple}. Like @code{gdb_test},
7834 @code{gdb_test_multiple} recognizes internal errors and unexpected
7837 Do not write tests which expect a literal tab character from @value{GDBN}.
7838 On some operating systems (e.g.@: OpenBSD) the TTY layer expands tabs to
7839 spaces, so by the time @value{GDBN}'s output reaches expect the tab is gone.
7841 The source language programs do @emph{not} need to be in a consistent
7842 style. Since @value{GDBN} is used to debug programs written in many different
7843 styles, it's worth having a mix of styles in the testsuite; for
7844 instance, some @value{GDBN} bugs involving the display of source lines would
7845 never manifest themselves if the programs used GNU coding style
7852 Check the @file{README} file, it often has useful information that does not
7853 appear anywhere else in the directory.
7856 * Getting Started:: Getting started working on @value{GDBN}
7857 * Debugging GDB:: Debugging @value{GDBN} with itself
7860 @node Getting Started
7862 @section Getting Started
7864 @value{GDBN} is a large and complicated program, and if you first starting to
7865 work on it, it can be hard to know where to start. Fortunately, if you
7866 know how to go about it, there are ways to figure out what is going on.
7868 This manual, the @value{GDBN} Internals manual, has information which applies
7869 generally to many parts of @value{GDBN}.
7871 Information about particular functions or data structures are located in
7872 comments with those functions or data structures. If you run across a
7873 function or a global variable which does not have a comment correctly
7874 explaining what is does, this can be thought of as a bug in @value{GDBN}; feel
7875 free to submit a bug report, with a suggested comment if you can figure
7876 out what the comment should say. If you find a comment which is
7877 actually wrong, be especially sure to report that.
7879 Comments explaining the function of macros defined in host, target, or
7880 native dependent files can be in several places. Sometimes they are
7881 repeated every place the macro is defined. Sometimes they are where the
7882 macro is used. Sometimes there is a header file which supplies a
7883 default definition of the macro, and the comment is there. This manual
7884 also documents all the available macros.
7885 @c (@pxref{Host Conditionals}, @pxref{Target
7886 @c Conditionals}, @pxref{Native Conditionals}, and @pxref{Obsolete
7889 Start with the header files. Once you have some idea of how
7890 @value{GDBN}'s internal symbol tables are stored (see @file{symtab.h},
7891 @file{gdbtypes.h}), you will find it much easier to understand the
7892 code which uses and creates those symbol tables.
7894 You may wish to process the information you are getting somehow, to
7895 enhance your understanding of it. Summarize it, translate it to another
7896 language, add some (perhaps trivial or non-useful) feature to @value{GDBN}, use
7897 the code to predict what a test case would do and write the test case
7898 and verify your prediction, etc. If you are reading code and your eyes
7899 are starting to glaze over, this is a sign you need to use a more active
7902 Once you have a part of @value{GDBN} to start with, you can find more
7903 specifically the part you are looking for by stepping through each
7904 function with the @code{next} command. Do not use @code{step} or you
7905 will quickly get distracted; when the function you are stepping through
7906 calls another function try only to get a big-picture understanding
7907 (perhaps using the comment at the beginning of the function being
7908 called) of what it does. This way you can identify which of the
7909 functions being called by the function you are stepping through is the
7910 one which you are interested in. You may need to examine the data
7911 structures generated at each stage, with reference to the comments in
7912 the header files explaining what the data structures are supposed to
7915 Of course, this same technique can be used if you are just reading the
7916 code, rather than actually stepping through it. The same general
7917 principle applies---when the code you are looking at calls something
7918 else, just try to understand generally what the code being called does,
7919 rather than worrying about all its details.
7921 @cindex command implementation
7922 A good place to start when tracking down some particular area is with
7923 a command which invokes that feature. Suppose you want to know how
7924 single-stepping works. As a @value{GDBN} user, you know that the
7925 @code{step} command invokes single-stepping. The command is invoked
7926 via command tables (see @file{command.h}); by convention the function
7927 which actually performs the command is formed by taking the name of
7928 the command and adding @samp{_command}, or in the case of an
7929 @code{info} subcommand, @samp{_info}. For example, the @code{step}
7930 command invokes the @code{step_command} function and the @code{info
7931 display} command invokes @code{display_info}. When this convention is
7932 not followed, you might have to use @code{grep} or @kbd{M-x
7933 tags-search} in emacs, or run @value{GDBN} on itself and set a
7934 breakpoint in @code{execute_command}.
7936 @cindex @code{bug-gdb} mailing list
7937 If all of the above fail, it may be appropriate to ask for information
7938 on @code{bug-gdb}. But @emph{never} post a generic question like ``I was
7939 wondering if anyone could give me some tips about understanding
7940 @value{GDBN}''---if we had some magic secret we would put it in this manual.
7941 Suggestions for improving the manual are always welcome, of course.
7945 @section Debugging @value{GDBN} with itself
7946 @cindex debugging @value{GDBN}
7948 If @value{GDBN} is limping on your machine, this is the preferred way to get it
7949 fully functional. Be warned that in some ancient Unix systems, like
7950 Ultrix 4.2, a program can't be running in one process while it is being
7951 debugged in another. Rather than typing the command @kbd{@w{./gdb
7952 ./gdb}}, which works on Suns and such, you can copy @file{gdb} to
7953 @file{gdb2} and then type @kbd{@w{./gdb ./gdb2}}.
7955 When you run @value{GDBN} in the @value{GDBN} source directory, it will read a
7956 @file{.gdbinit} file that sets up some simple things to make debugging
7957 gdb easier. The @code{info} command, when executed without a subcommand
7958 in a @value{GDBN} being debugged by gdb, will pop you back up to the top level
7959 gdb. See @file{.gdbinit} for details.
7961 If you use emacs, you will probably want to do a @code{make TAGS} after
7962 you configure your distribution; this will put the machine dependent
7963 routines for your local machine where they will be accessed first by
7966 Also, make sure that you've either compiled @value{GDBN} with your local cc, or
7967 have run @code{fixincludes} if you are compiling with gcc.
7969 @section Submitting Patches
7971 @cindex submitting patches
7972 Thanks for thinking of offering your changes back to the community of
7973 @value{GDBN} users. In general we like to get well designed enhancements.
7974 Thanks also for checking in advance about the best way to transfer the
7977 The @value{GDBN} maintainers will only install ``cleanly designed'' patches.
7978 This manual summarizes what we believe to be clean design for @value{GDBN}.
7980 If the maintainers don't have time to put the patch in when it arrives,
7981 or if there is any question about a patch, it goes into a large queue
7982 with everyone else's patches and bug reports.
7984 @cindex legal papers for code contributions
7985 The legal issue is that to incorporate substantial changes requires a
7986 copyright assignment from you and/or your employer, granting ownership
7987 of the changes to the Free Software Foundation. You can get the
7988 standard documents for doing this by sending mail to @code{gnu@@gnu.org}
7989 and asking for it. We recommend that people write in "All programs
7990 owned by the Free Software Foundation" as "NAME OF PROGRAM", so that
7991 changes in many programs (not just @value{GDBN}, but GAS, Emacs, GCC,
7993 contributed with only one piece of legalese pushed through the
7994 bureaucracy and filed with the FSF. We can't start merging changes until
7995 this paperwork is received by the FSF (their rules, which we follow
7996 since we maintain it for them).
7998 Technically, the easiest way to receive changes is to receive each
7999 feature as a small context diff or unidiff, suitable for @code{patch}.
8000 Each message sent to me should include the changes to C code and
8001 header files for a single feature, plus @file{ChangeLog} entries for
8002 each directory where files were modified, and diffs for any changes
8003 needed to the manuals (@file{gdb/doc/gdb.texinfo} or
8004 @file{gdb/doc/gdbint.texinfo}). If there are a lot of changes for a
8005 single feature, they can be split down into multiple messages.
8007 In this way, if we read and like the feature, we can add it to the
8008 sources with a single patch command, do some testing, and check it in.
8009 If you leave out the @file{ChangeLog}, we have to write one. If you leave
8010 out the doc, we have to puzzle out what needs documenting. Etc., etc.
8012 The reason to send each change in a separate message is that we will not
8013 install some of the changes. They'll be returned to you with questions
8014 or comments. If we're doing our job correctly, the message back to you
8015 will say what you have to fix in order to make the change acceptable.
8016 The reason to have separate messages for separate features is so that
8017 the acceptable changes can be installed while one or more changes are
8018 being reworked. If multiple features are sent in a single message, we
8019 tend to not put in the effort to sort out the acceptable changes from
8020 the unacceptable, so none of the features get installed until all are
8023 If this sounds painful or authoritarian, well, it is. But we get a lot
8024 of bug reports and a lot of patches, and many of them don't get
8025 installed because we don't have the time to finish the job that the bug
8026 reporter or the contributor could have done. Patches that arrive
8027 complete, working, and well designed, tend to get installed on the day
8028 they arrive. The others go into a queue and get installed as time
8029 permits, which, since the maintainers have many demands to meet, may not
8030 be for quite some time.
8032 Please send patches directly to
8033 @email{gdb-patches@@sourceware.org, the @value{GDBN} maintainers}.
8035 @section Build Script
8037 @cindex build script
8039 The script @file{gdb_buildall.sh} builds @value{GDBN} with flag
8040 @option{--enable-targets=all} set. This builds @value{GDBN} with all supported
8041 targets activated. This helps testing @value{GDBN} when doing changes that
8042 affect more than one architecture and is much faster than using
8043 @file{gdb_mbuild.sh}.
8045 After building @value{GDBN} the script checks which architectures are
8046 supported and then switches the current architecture to each of those to get
8047 information about the architecture. The test results are stored in log files
8048 in the directory the script was called from.
8050 @include observer.texi
8052 @node GNU Free Documentation License
8053 @appendix GNU Free Documentation License