1 \input texinfo @c -*- texinfo -*-
2 @setfilename gdbint.info
4 @dircategory Software development
6 * Gdb-Internals: (gdbint). The GNU debugger's internals.
10 This file documents the internals of the GNU debugger @value{GDBN}.
11 Copyright (C) 1990, 1991, 1992, 1993, 1994, 1996, 1998, 1999, 2000, 2001,
12 2002, 2003, 2004, 2005, 2006, 2008
13 Free 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 1.1 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''.
25 @setchapternewpage off
26 @settitle @value{GDBN} Internals
32 @title @value{GDBN} Internals
33 @subtitle{A guide to the internals of the GNU debugger}
35 @author Cygnus Solutions
36 @author Second Edition:
38 @author Cygnus Solutions
41 \def\$#1${{#1}} % Kluge: collect RCS revision info without $...$
42 \xdef\manvers{\$Revision$} % For use in headers, footers too
44 \hfill Cygnus Solutions\par
46 \hfill \TeX{}info \texinfoversion\par
50 @vskip 0pt plus 1filll
51 Copyright @copyright{} 1990,1991,1992,1993,1994,1996,1998,1999,2000,2001,
52 2002, 2003, 2004, 2005, 2006 Free Software Foundation, Inc.
54 Permission is granted to copy, distribute and/or modify this document
55 under the terms of the GNU Free Documentation License, Version 1.1 or
56 any later version published by the Free Software Foundation; with no
57 Invariant Sections, with no Front-Cover Texts, and with no Back-Cover
58 Texts. A copy of the license is included in the section entitled ``GNU
59 Free Documentation License''.
65 @c Perhaps this should be the title of the document (but only for info,
66 @c not for TeX). Existing GNU manuals seem inconsistent on this point.
67 @top Scope of this Document
69 This document documents the internals of the GNU debugger, @value{GDBN}. It
70 includes description of @value{GDBN}'s key algorithms and operations, as well
71 as the mechanisms that adapt @value{GDBN} to specific hosts and targets.
83 * Target Architecture Definition::
84 * Target Descriptions::
85 * Target Vector Definition::
90 * Versions and Branches::
91 * Start of New Year Procedure::
96 * GDB Observers:: @value{GDBN} Currently available observers
97 * GNU Free Documentation License:: The license for this documentation
103 @chapter Requirements
104 @cindex requirements for @value{GDBN}
106 Before diving into the internals, you should understand the formal
107 requirements and other expectations for @value{GDBN}. Although some
108 of these may seem obvious, there have been proposals for @value{GDBN}
109 that have run counter to these requirements.
111 First of all, @value{GDBN} is a debugger. It's not designed to be a
112 front panel for embedded systems. It's not a text editor. It's not a
113 shell. It's not a programming environment.
115 @value{GDBN} is an interactive tool. Although a batch mode is
116 available, @value{GDBN}'s primary role is to interact with a human
119 @value{GDBN} should be responsive to the user. A programmer hot on
120 the trail of a nasty bug, and operating under a looming deadline, is
121 going to be very impatient of everything, including the response time
122 to debugger commands.
124 @value{GDBN} should be relatively permissive, such as for expressions.
125 While the compiler should be picky (or have the option to be made
126 picky), since source code lives for a long time usually, the
127 programmer doing debugging shouldn't be spending time figuring out to
128 mollify the debugger.
130 @value{GDBN} will be called upon to deal with really large programs.
131 Executable sizes of 50 to 100 megabytes occur regularly, and we've
132 heard reports of programs approaching 1 gigabyte in size.
134 @value{GDBN} should be able to run everywhere. No other debugger is
135 available for even half as many configurations as @value{GDBN}
139 @node Overall Structure
141 @chapter Overall Structure
143 @value{GDBN} consists of three major subsystems: user interface,
144 symbol handling (the @dfn{symbol side}), and target system handling (the
147 The user interface consists of several actual interfaces, plus
150 The symbol side consists of object file readers, debugging info
151 interpreters, symbol table management, source language expression
152 parsing, type and value printing.
154 The target side consists of execution control, stack frame analysis, and
155 physical target manipulation.
157 The target side/symbol side division is not formal, and there are a
158 number of exceptions. For instance, core file support involves symbolic
159 elements (the basic core file reader is in BFD) and target elements (it
160 supplies the contents of memory and the values of registers). Instead,
161 this division is useful for understanding how the minor subsystems
164 @section The Symbol Side
166 The symbolic side of @value{GDBN} can be thought of as ``everything
167 you can do in @value{GDBN} without having a live program running''.
168 For instance, you can look at the types of variables, and evaluate
169 many kinds of expressions.
171 @section The Target Side
173 The target side of @value{GDBN} is the ``bits and bytes manipulator''.
174 Although it may make reference to symbolic info here and there, most
175 of the target side will run with only a stripped executable
176 available---or even no executable at all, in remote debugging cases.
178 Operations such as disassembly, stack frame crawls, and register
179 display, are able to work with no symbolic info at all. In some cases,
180 such as disassembly, @value{GDBN} will use symbolic info to present addresses
181 relative to symbols rather than as raw numbers, but it will work either
184 @section Configurations
188 @dfn{Host} refers to attributes of the system where @value{GDBN} runs.
189 @dfn{Target} refers to the system where the program being debugged
190 executes. In most cases they are the same machine, in which case a
191 third type of @dfn{Native} attributes come into play.
193 Defines and include files needed to build on the host are host
194 support. Examples are tty support, system defined types, host byte
195 order, host float format. These are all calculated by @code{autoconf}
196 when the debugger is built.
198 Defines and information needed to handle the target format are target
199 dependent. Examples are the stack frame format, instruction set,
200 breakpoint instruction, registers, and how to set up and tear down the stack
203 Information that is only needed when the host and target are the same,
204 is native dependent. One example is Unix child process support; if the
205 host and target are not the same, calling @code{fork} to start the target
206 process is a bad idea. The various macros needed for finding the
207 registers in the @code{upage}, running @code{ptrace}, and such are all
208 in the native-dependent files.
210 Another example of native-dependent code is support for features that
211 are really part of the target environment, but which require
212 @code{#include} files that are only available on the host system. Core
213 file handling and @code{setjmp} handling are two common cases.
215 When you want to make @value{GDBN} work as the traditional native debugger
216 on a system, you will need to supply both target and native information.
218 @section Source Tree Structure
219 @cindex @value{GDBN} source tree structure
221 The @value{GDBN} source directory has a mostly flat structure---there
222 are only a few subdirectories. A file's name usually gives a hint as
223 to what it does; for example, @file{stabsread.c} reads stabs,
224 @file{dwarf2read.c} reads @sc{DWARF 2}, etc.
226 Files that are related to some common task have names that share
227 common substrings. For example, @file{*-thread.c} files deal with
228 debugging threads on various platforms; @file{*read.c} files deal with
229 reading various kinds of symbol and object files; @file{inf*.c} files
230 deal with direct control of the @dfn{inferior program} (@value{GDBN}
231 parlance for the program being debugged).
233 There are several dozens of files in the @file{*-tdep.c} family.
234 @samp{tdep} stands for @dfn{target-dependent code}---each of these
235 files implements debug support for a specific target architecture
236 (sparc, mips, etc). Usually, only one of these will be used in a
237 specific @value{GDBN} configuration (sometimes two, closely related).
239 Similarly, there are many @file{*-nat.c} files, each one for native
240 debugging on a specific system (e.g., @file{sparc-linux-nat.c} is for
241 native debugging of Sparc machines running the Linux kernel).
243 The few subdirectories of the source tree are:
247 Code that implements @dfn{CLI}, the @value{GDBN} Command-Line
248 Interpreter. @xref{User Interface, Command Interpreter}.
251 Code for the @value{GDBN} remote server.
254 Code for Insight, the @value{GDBN} TK-based GUI front-end.
257 The @dfn{GDB/MI}, the @value{GDBN} Machine Interface interpreter.
260 Target signal translation code.
263 Code for @dfn{TUI}, the @value{GDBN} Text-mode full-screen User
264 Interface. @xref{User Interface, TUI}.
272 @value{GDBN} uses a number of debugging-specific algorithms. They are
273 often not very complicated, but get lost in the thicket of special
274 cases and real-world issues. This chapter describes the basic
275 algorithms and mentions some of the specific target definitions that
278 @section Prologue Analysis
280 @cindex prologue analysis
281 @cindex call frame information
282 @cindex CFI (call frame information)
283 To produce a backtrace and allow the user to manipulate older frames'
284 variables and arguments, @value{GDBN} needs to find the base addresses
285 of older frames, and discover where those frames' registers have been
286 saved. Since a frame's ``callee-saves'' registers get saved by
287 younger frames if and when they're reused, a frame's registers may be
288 scattered unpredictably across younger frames. This means that
289 changing the value of a register-allocated variable in an older frame
290 may actually entail writing to a save slot in some younger frame.
292 Modern versions of GCC emit Dwarf call frame information (``CFI''),
293 which describes how to find frame base addresses and saved registers.
294 But CFI is not always available, so as a fallback @value{GDBN} uses a
295 technique called @dfn{prologue analysis} to find frame sizes and saved
296 registers. A prologue analyzer disassembles the function's machine
297 code starting from its entry point, and looks for instructions that
298 allocate frame space, save the stack pointer in a frame pointer
299 register, save registers, and so on. Obviously, this can't be done
300 accurately in general, but it's tractable to do well enough to be very
301 helpful. Prologue analysis predates the GNU toolchain's support for
302 CFI; at one time, prologue analysis was the only mechanism
303 @value{GDBN} used for stack unwinding at all, when the function
304 calling conventions didn't specify a fixed frame layout.
306 In the olden days, function prologues were generated by hand-written,
307 target-specific code in GCC, and treated as opaque and untouchable by
308 optimizers. Looking at this code, it was usually straightforward to
309 write a prologue analyzer for @value{GDBN} that would accurately
310 understand all the prologues GCC would generate. However, over time
311 GCC became more aggressive about instruction scheduling, and began to
312 understand more about the semantics of the prologue instructions
313 themselves; in response, @value{GDBN}'s analyzers became more complex
314 and fragile. Keeping the prologue analyzers working as GCC (and the
315 instruction sets themselves) evolved became a substantial task.
317 @cindex @file{prologue-value.c}
318 @cindex abstract interpretation of function prologues
319 @cindex pseudo-evaluation of function prologues
320 To try to address this problem, the code in @file{prologue-value.h}
321 and @file{prologue-value.c} provides a general framework for writing
322 prologue analyzers that are simpler and more robust than ad-hoc
323 analyzers. When we analyze a prologue using the prologue-value
324 framework, we're really doing ``abstract interpretation'' or
325 ``pseudo-evaluation'': running the function's code in simulation, but
326 using conservative approximations of the values registers and memory
327 would hold when the code actually runs. For example, if our function
328 starts with the instruction:
331 addi r1, 42 # add 42 to r1
334 we don't know exactly what value will be in @code{r1} after executing
335 this instruction, but we do know it'll be 42 greater than its original
338 If we then see an instruction like:
341 addi r1, 22 # add 22 to r1
344 we still don't know what @code{r1's} value is, but again, we can say
345 it is now 64 greater than its original value.
347 If the next instruction were:
350 mov r2, r1 # set r2 to r1's value
353 then we can say that @code{r2's} value is now the original value of
356 It's common for prologues to save registers on the stack, so we'll
357 need to track the values of stack frame slots, as well as the
358 registers. So after an instruction like this:
364 then we'd know that the stack slot four bytes above the frame pointer
365 holds the original value of @code{r1} plus 64.
369 Of course, this can only go so far before it gets unreasonable. If we
370 wanted to be able to say anything about the value of @code{r1} after
374 xor r1, r3 # exclusive-or r1 and r3, place result in r1
377 then things would get pretty complex. But remember, we're just doing
378 a conservative approximation; if exclusive-or instructions aren't
379 relevant to prologues, we can just say @code{r1}'s value is now
380 ``unknown''. We can ignore things that are too complex, if that loss of
381 information is acceptable for our application.
383 So when we say ``conservative approximation'' here, what we mean is an
384 approximation that is either accurate, or marked ``unknown'', but
387 Using this framework, a prologue analyzer is simply an interpreter for
388 machine code, but one that uses conservative approximations for the
389 contents of registers and memory instead of actual values. Starting
390 from the function's entry point, you simulate instructions up to the
391 current PC, or an instruction that you don't know how to simulate.
392 Now you can examine the state of the registers and stack slots you've
398 To see how large your stack frame is, just check the value of the
399 stack pointer register; if it's the original value of the SP
400 minus a constant, then that constant is the stack frame's size.
401 If the SP's value has been marked as ``unknown'', then that means
402 the prologue has done something too complex for us to track, and
403 we don't know the frame size.
406 To see where we've saved the previous frame's registers, we just
407 search the values we've tracked --- stack slots, usually, but
408 registers, too, if you want --- for something equal to the register's
409 original value. If the calling conventions suggest a standard place
410 to save a given register, then we can check there first, but really,
411 anything that will get us back the original value will probably work.
414 This does take some work. But prologue analyzers aren't
415 quick-and-simple pattern patching to recognize a few fixed prologue
416 forms any more; they're big, hairy functions. Along with inferior
417 function calls, prologue analysis accounts for a substantial portion
418 of the time needed to stabilize a @value{GDBN} port. So it's
419 worthwhile to look for an approach that will be easier to understand
420 and maintain. In the approach described above:
425 It's easier to see that the analyzer is correct: you just see
426 whether the analyzer properly (albeit conservatively) simulates
427 the effect of each instruction.
430 It's easier to extend the analyzer: you can add support for new
431 instructions, and know that you haven't broken anything that
432 wasn't already broken before.
435 It's orthogonal: to gather new information, you don't need to
436 complicate the code for each instruction. As long as your domain
437 of conservative values is already detailed enough to tell you
438 what you need, then all the existing instruction simulations are
439 already gathering the right data for you.
443 The file @file{prologue-value.h} contains detailed comments explaining
444 the framework and how to use it.
447 @section Breakpoint Handling
450 In general, a breakpoint is a user-designated location in the program
451 where the user wants to regain control if program execution ever reaches
454 There are two main ways to implement breakpoints; either as ``hardware''
455 breakpoints or as ``software'' breakpoints.
457 @cindex hardware breakpoints
458 @cindex program counter
459 Hardware breakpoints are sometimes available as a builtin debugging
460 features with some chips. Typically these work by having dedicated
461 register into which the breakpoint address may be stored. If the PC
462 (shorthand for @dfn{program counter})
463 ever matches a value in a breakpoint registers, the CPU raises an
464 exception and reports it to @value{GDBN}.
466 Another possibility is when an emulator is in use; many emulators
467 include circuitry that watches the address lines coming out from the
468 processor, and force it to stop if the address matches a breakpoint's
471 A third possibility is that the target already has the ability to do
472 breakpoints somehow; for instance, a ROM monitor may do its own
473 software breakpoints. So although these are not literally ``hardware
474 breakpoints'', from @value{GDBN}'s point of view they work the same;
475 @value{GDBN} need not do anything more than set the breakpoint and wait
476 for something to happen.
478 Since they depend on hardware resources, hardware breakpoints may be
479 limited in number; when the user asks for more, @value{GDBN} will
480 start trying to set software breakpoints. (On some architectures,
481 notably the 32-bit x86 platforms, @value{GDBN} cannot always know
482 whether there's enough hardware resources to insert all the hardware
483 breakpoints and watchpoints. On those platforms, @value{GDBN} prints
484 an error message only when the program being debugged is continued.)
486 @cindex software breakpoints
487 Software breakpoints require @value{GDBN} to do somewhat more work.
488 The basic theory is that @value{GDBN} will replace a program
489 instruction with a trap, illegal divide, or some other instruction
490 that will cause an exception, and then when it's encountered,
491 @value{GDBN} will take the exception and stop the program. When the
492 user says to continue, @value{GDBN} will restore the original
493 instruction, single-step, re-insert the trap, and continue on.
495 Since it literally overwrites the program being tested, the program area
496 must be writable, so this technique won't work on programs in ROM. It
497 can also distort the behavior of programs that examine themselves,
498 although such a situation would be highly unusual.
500 Also, the software breakpoint instruction should be the smallest size of
501 instruction, so it doesn't overwrite an instruction that might be a jump
502 target, and cause disaster when the program jumps into the middle of the
503 breakpoint instruction. (Strictly speaking, the breakpoint must be no
504 larger than the smallest interval between instructions that may be jump
505 targets; perhaps there is an architecture where only even-numbered
506 instructions may jumped to.) Note that it's possible for an instruction
507 set not to have any instructions usable for a software breakpoint,
508 although in practice only the ARC has failed to define such an
511 Basic breakpoint object handling is in @file{breakpoint.c}. However,
512 much of the interesting breakpoint action is in @file{infrun.c}.
515 @cindex insert or remove software breakpoint
516 @findex target_remove_breakpoint
517 @findex target_insert_breakpoint
518 @item target_remove_breakpoint (@var{bp_tgt})
519 @itemx target_insert_breakpoint (@var{bp_tgt})
520 Insert or remove a software breakpoint at address
521 @code{@var{bp_tgt}->placed_address}. Returns zero for success,
522 non-zero for failure. On input, @var{bp_tgt} contains the address of the
523 breakpoint, and is otherwise initialized to zero. The fields of the
524 @code{struct bp_target_info} pointed to by @var{bp_tgt} are updated
525 to contain other information about the breakpoint on output. The field
526 @code{placed_address} may be updated if the breakpoint was placed at a
527 related address; the field @code{shadow_contents} contains the real
528 contents of the bytes where the breakpoint has been inserted,
529 if reading memory would return the breakpoint instead of the
530 underlying memory; the field @code{shadow_len} is the length of
531 memory cached in @code{shadow_contents}, if any; and the field
532 @code{placed_size} is optionally set and used by the target, if
533 it could differ from @code{shadow_len}.
535 For example, the remote target @samp{Z0} packet does not require
536 shadowing memory, so @code{shadow_len} is left at zero. However,
537 the length reported by @code{gdbarch_breakpoint_from_pc} is cached in
538 @code{placed_size}, so that a matching @samp{z0} packet can be
539 used to remove the breakpoint.
541 @cindex insert or remove hardware breakpoint
542 @findex target_remove_hw_breakpoint
543 @findex target_insert_hw_breakpoint
544 @item target_remove_hw_breakpoint (@var{bp_tgt})
545 @itemx target_insert_hw_breakpoint (@var{bp_tgt})
546 Insert or remove a hardware-assisted breakpoint at address
547 @code{@var{bp_tgt}->placed_address}. Returns zero for success,
548 non-zero for failure. See @code{target_insert_breakpoint} for
549 a description of the @code{struct bp_target_info} pointed to by
550 @var{bp_tgt}; the @code{shadow_contents} and
551 @code{shadow_len} members are not used for hardware breakpoints,
552 but @code{placed_size} may be.
555 @section Single Stepping
557 @section Signal Handling
559 @section Thread Handling
561 @section Inferior Function Calls
563 @section Longjmp Support
565 @cindex @code{longjmp} debugging
566 @value{GDBN} has support for figuring out that the target is doing a
567 @code{longjmp} and for stopping at the target of the jump, if we are
568 stepping. This is done with a few specialized internal breakpoints,
569 which are visible in the output of the @samp{maint info breakpoint}
572 @findex gdbarch_get_longjmp_target
573 To make this work, you need to define a function called
574 @code{gdbarch_get_longjmp_target}, which will examine the
575 @code{jmp_buf} structure and extract the @code{longjmp} target address.
576 Since @code{jmp_buf} is target specific and typically defined in a
577 target header not available to @value{GDBN}, you will need to
578 determine the offset of the PC manually and return that; many targets
579 define a @code{jb_pc_offset} field in the tdep structure to save the
580 value once calculated.
585 Watchpoints are a special kind of breakpoints (@pxref{Algorithms,
586 breakpoints}) which break when data is accessed rather than when some
587 instruction is executed. When you have data which changes without
588 your knowing what code does that, watchpoints are the silver bullet to
589 hunt down and kill such bugs.
591 @cindex hardware watchpoints
592 @cindex software watchpoints
593 Watchpoints can be either hardware-assisted or not; the latter type is
594 known as ``software watchpoints.'' @value{GDBN} always uses
595 hardware-assisted watchpoints if they are available, and falls back on
596 software watchpoints otherwise. Typical situations where @value{GDBN}
597 will use software watchpoints are:
601 The watched memory region is too large for the underlying hardware
602 watchpoint support. For example, each x86 debug register can watch up
603 to 4 bytes of memory, so trying to watch data structures whose size is
604 more than 16 bytes will cause @value{GDBN} to use software
608 The value of the expression to be watched depends on data held in
609 registers (as opposed to memory).
612 Too many different watchpoints requested. (On some architectures,
613 this situation is impossible to detect until the debugged program is
614 resumed.) Note that x86 debug registers are used both for hardware
615 breakpoints and for watchpoints, so setting too many hardware
616 breakpoints might cause watchpoint insertion to fail.
619 No hardware-assisted watchpoints provided by the target
623 Software watchpoints are very slow, since @value{GDBN} needs to
624 single-step the program being debugged and test the value of the
625 watched expression(s) after each instruction. The rest of this
626 section is mostly irrelevant for software watchpoints.
628 When the inferior stops, @value{GDBN} tries to establish, among other
629 possible reasons, whether it stopped due to a watchpoint being hit.
630 It first uses @code{STOPPED_BY_WATCHPOINT} to see if any watchpoint
631 was hit. If not, all watchpoint checking is skipped.
633 Then @value{GDBN} calls @code{target_stopped_data_address} exactly
634 once. This method returns the address of the watchpoint which
635 triggered, if the target can determine it. If the triggered address
636 is available, @value{GDBN} compares the address returned by this
637 method with each watched memory address in each active watchpoint.
638 For data-read and data-access watchpoints, @value{GDBN} announces
639 every watchpoint that watches the triggered address as being hit.
640 For this reason, data-read and data-access watchpoints
641 @emph{require} that the triggered address be available; if not, read
642 and access watchpoints will never be considered hit. For data-write
643 watchpoints, if the triggered address is available, @value{GDBN}
644 considers only those watchpoints which match that address;
645 otherwise, @value{GDBN} considers all data-write watchpoints. For
646 each data-write watchpoint that @value{GDBN} considers, it evaluates
647 the expression whose value is being watched, and tests whether the
648 watched value has changed. Watchpoints whose watched values have
649 changed are announced as hit.
651 @c FIXME move these to the main lists of target/native defns
653 @value{GDBN} uses several macros and primitives to support hardware
657 @findex TARGET_HAS_HARDWARE_WATCHPOINTS
658 @item TARGET_HAS_HARDWARE_WATCHPOINTS
659 If defined, the target supports hardware watchpoints.
660 (Currently only used for several native configs.)
662 @findex TARGET_CAN_USE_HARDWARE_WATCHPOINT
663 @item TARGET_CAN_USE_HARDWARE_WATCHPOINT (@var{type}, @var{count}, @var{other})
664 Return the number of hardware watchpoints of type @var{type} that are
665 possible to be set. The value is positive if @var{count} watchpoints
666 of this type can be set, zero if setting watchpoints of this type is
667 not supported, and negative if @var{count} is more than the maximum
668 number of watchpoints of type @var{type} that can be set. @var{other}
669 is non-zero if other types of watchpoints are currently enabled (there
670 are architectures which cannot set watchpoints of different types at
673 @findex TARGET_REGION_OK_FOR_HW_WATCHPOINT
674 @item TARGET_REGION_OK_FOR_HW_WATCHPOINT (@var{addr}, @var{len})
675 Return non-zero if hardware watchpoints can be used to watch a region
676 whose address is @var{addr} and whose length in bytes is @var{len}.
678 @cindex insert or remove hardware watchpoint
679 @findex target_insert_watchpoint
680 @findex target_remove_watchpoint
681 @item target_insert_watchpoint (@var{addr}, @var{len}, @var{type})
682 @itemx target_remove_watchpoint (@var{addr}, @var{len}, @var{type})
683 Insert or remove a hardware watchpoint starting at @var{addr}, for
684 @var{len} bytes. @var{type} is the watchpoint type, one of the
685 possible values of the enumerated data type @code{target_hw_bp_type},
686 defined by @file{breakpoint.h} as follows:
689 enum target_hw_bp_type
691 hw_write = 0, /* Common (write) HW watchpoint */
692 hw_read = 1, /* Read HW watchpoint */
693 hw_access = 2, /* Access (read or write) HW watchpoint */
694 hw_execute = 3 /* Execute HW breakpoint */
699 These two macros should return 0 for success, non-zero for failure.
701 @findex target_stopped_data_address
702 @item target_stopped_data_address (@var{addr_p})
703 If the inferior has some watchpoint that triggered, place the address
704 associated with the watchpoint at the location pointed to by
705 @var{addr_p} and return non-zero. Otherwise, return zero. This
706 is required for data-read and data-access watchpoints. It is
707 not required for data-write watchpoints, but @value{GDBN} uses
708 it to improve handling of those also.
710 @value{GDBN} will only call this method once per watchpoint stop,
711 immediately after calling @code{STOPPED_BY_WATCHPOINT}. If the
712 target's watchpoint indication is sticky, i.e., stays set after
713 resuming, this method should clear it. For instance, the x86 debug
714 control register has sticky triggered flags.
716 @findex target_watchpoint_addr_within_range
717 @item target_watchpoint_addr_within_range (@var{target}, @var{addr}, @var{start}, @var{length})
718 Check whether @var{addr} (as returned by @code{target_stopped_data_address})
719 lies within the hardware-defined watchpoint region described by
720 @var{start} and @var{length}. This only needs to be provided if the
721 granularity of a watchpoint is greater than one byte, i.e., if the
722 watchpoint can also trigger on nearby addresses outside of the watched
725 @findex HAVE_STEPPABLE_WATCHPOINT
726 @item HAVE_STEPPABLE_WATCHPOINT
727 If defined to a non-zero value, it is not necessary to disable a
728 watchpoint to step over it. Like @code{gdbarch_have_nonsteppable_watchpoint},
729 this is usually set when watchpoints trigger at the instruction
730 which will perform an interesting read or write. It should be
731 set if there is a temporary disable bit which allows the processor
732 to step over the interesting instruction without raising the
733 watchpoint exception again.
735 @findex gdbarch_have_nonsteppable_watchpoint
736 @item int gdbarch_have_nonsteppable_watchpoint (@var{gdbarch})
737 If it returns a non-zero value, @value{GDBN} should disable a
738 watchpoint to step the inferior over it. This is usually set when
739 watchpoints trigger at the instruction which will perform an
740 interesting read or write.
742 @findex HAVE_CONTINUABLE_WATCHPOINT
743 @item HAVE_CONTINUABLE_WATCHPOINT
744 If defined to a non-zero value, it is possible to continue the
745 inferior after a watchpoint has been hit. This is usually set
746 when watchpoints trigger at the instruction following an interesting
749 @findex CANNOT_STEP_HW_WATCHPOINTS
750 @item CANNOT_STEP_HW_WATCHPOINTS
751 If this is defined to a non-zero value, @value{GDBN} will remove all
752 watchpoints before stepping the inferior.
754 @findex STOPPED_BY_WATCHPOINT
755 @item STOPPED_BY_WATCHPOINT (@var{wait_status})
756 Return non-zero if stopped by a watchpoint. @var{wait_status} is of
757 the type @code{struct target_waitstatus}, defined by @file{target.h}.
758 Normally, this macro is defined to invoke the function pointed to by
759 the @code{to_stopped_by_watchpoint} member of the structure (of the
760 type @code{target_ops}, defined on @file{target.h}) that describes the
761 target-specific operations; @code{to_stopped_by_watchpoint} ignores
762 the @var{wait_status} argument.
764 @value{GDBN} does not require the non-zero value returned by
765 @code{STOPPED_BY_WATCHPOINT} to be 100% correct, so if a target cannot
766 determine for sure whether the inferior stopped due to a watchpoint,
767 it could return non-zero ``just in case''.
770 @subsection Watchpoints and Threads
771 @cindex watchpoints, with threads
773 @value{GDBN} only supports process-wide watchpoints, which trigger
774 in all threads. @value{GDBN} uses the thread ID to make watchpoints
775 act as if they were thread-specific, but it cannot set hardware
776 watchpoints that only trigger in a specific thread. Therefore, even
777 if the target supports threads, per-thread debug registers, and
778 watchpoints which only affect a single thread, it should set the
779 per-thread debug registers for all threads to the same value. On
780 @sc{gnu}/Linux native targets, this is accomplished by using
781 @code{ALL_LWPS} in @code{target_insert_watchpoint} and
782 @code{target_remove_watchpoint} and by using
783 @code{linux_set_new_thread} to register a handler for newly created
786 @value{GDBN}'s @sc{gnu}/Linux support only reports a single event
787 at a time, although multiple events can trigger simultaneously for
788 multi-threaded programs. When multiple events occur, @file{linux-nat.c}
789 queues subsequent events and returns them the next time the program
790 is resumed. This means that @code{STOPPED_BY_WATCHPOINT} and
791 @code{target_stopped_data_address} only need to consult the current
792 thread's state---the thread indicated by @code{inferior_ptid}. If
793 two threads have hit watchpoints simultaneously, those routines
794 will be called a second time for the second thread.
796 @subsection x86 Watchpoints
797 @cindex x86 debug registers
798 @cindex watchpoints, on x86
800 The 32-bit Intel x86 (a.k.a.@: ia32) processors feature special debug
801 registers designed to facilitate debugging. @value{GDBN} provides a
802 generic library of functions that x86-based ports can use to implement
803 support for watchpoints and hardware-assisted breakpoints. This
804 subsection documents the x86 watchpoint facilities in @value{GDBN}.
806 (At present, the library functions read and write debug registers directly, and are
807 thus only available for native configurations.)
809 To use the generic x86 watchpoint support, a port should do the
813 @findex I386_USE_GENERIC_WATCHPOINTS
815 Define the macro @code{I386_USE_GENERIC_WATCHPOINTS} somewhere in the
816 target-dependent headers.
819 Include the @file{config/i386/nm-i386.h} header file @emph{after}
820 defining @code{I386_USE_GENERIC_WATCHPOINTS}.
823 Add @file{i386-nat.o} to the value of the Make variable
824 @code{NATDEPFILES} (@pxref{Native Debugging, NATDEPFILES}) or
825 @code{TDEPFILES} (@pxref{Target Architecture Definition, TDEPFILES}).
828 Provide implementations for the @code{I386_DR_LOW_*} macros described
829 below. Typically, each macro should call a target-specific function
830 which does the real work.
833 The x86 watchpoint support works by maintaining mirror images of the
834 debug registers. Values are copied between the mirror images and the
835 real debug registers via a set of macros which each target needs to
839 @findex I386_DR_LOW_SET_CONTROL
840 @item I386_DR_LOW_SET_CONTROL (@var{val})
841 Set the Debug Control (DR7) register to the value @var{val}.
843 @findex I386_DR_LOW_SET_ADDR
844 @item I386_DR_LOW_SET_ADDR (@var{idx}, @var{addr})
845 Put the address @var{addr} into the debug register number @var{idx}.
847 @findex I386_DR_LOW_RESET_ADDR
848 @item I386_DR_LOW_RESET_ADDR (@var{idx})
849 Reset (i.e.@: zero out) the address stored in the debug register
852 @findex I386_DR_LOW_GET_STATUS
853 @item I386_DR_LOW_GET_STATUS
854 Return the value of the Debug Status (DR6) register. This value is
855 used immediately after it is returned by
856 @code{I386_DR_LOW_GET_STATUS}, so as to support per-thread status
860 For each one of the 4 debug registers (whose indices are from 0 to 3)
861 that store addresses, a reference count is maintained by @value{GDBN},
862 to allow sharing of debug registers by several watchpoints. This
863 allows users to define several watchpoints that watch the same
864 expression, but with different conditions and/or commands, without
865 wasting debug registers which are in short supply. @value{GDBN}
866 maintains the reference counts internally, targets don't have to do
867 anything to use this feature.
869 The x86 debug registers can each watch a region that is 1, 2, or 4
870 bytes long. The ia32 architecture requires that each watched region
871 be appropriately aligned: 2-byte region on 2-byte boundary, 4-byte
872 region on 4-byte boundary. However, the x86 watchpoint support in
873 @value{GDBN} can watch unaligned regions and regions larger than 4
874 bytes (up to 16 bytes) by allocating several debug registers to watch
875 a single region. This allocation of several registers per a watched
876 region is also done automatically without target code intervention.
878 The generic x86 watchpoint support provides the following API for the
879 @value{GDBN}'s application code:
882 @findex i386_region_ok_for_watchpoint
883 @item i386_region_ok_for_watchpoint (@var{addr}, @var{len})
884 The macro @code{TARGET_REGION_OK_FOR_HW_WATCHPOINT} is set to call
885 this function. It counts the number of debug registers required to
886 watch a given region, and returns a non-zero value if that number is
887 less than 4, the number of debug registers available to x86
890 @findex i386_stopped_data_address
891 @item i386_stopped_data_address (@var{addr_p})
893 @code{target_stopped_data_address} is set to call this function.
895 function examines the breakpoint condition bits in the DR6 Debug
896 Status register, as returned by the @code{I386_DR_LOW_GET_STATUS}
897 macro, and returns the address associated with the first bit that is
900 @findex i386_stopped_by_watchpoint
901 @item i386_stopped_by_watchpoint (void)
902 The macro @code{STOPPED_BY_WATCHPOINT}
903 is set to call this function. The
904 argument passed to @code{STOPPED_BY_WATCHPOINT} is ignored. This
905 function examines the breakpoint condition bits in the DR6 Debug
906 Status register, as returned by the @code{I386_DR_LOW_GET_STATUS}
907 macro, and returns true if any bit is set. Otherwise, false is
910 @findex i386_insert_watchpoint
911 @findex i386_remove_watchpoint
912 @item i386_insert_watchpoint (@var{addr}, @var{len}, @var{type})
913 @itemx i386_remove_watchpoint (@var{addr}, @var{len}, @var{type})
914 Insert or remove a watchpoint. The macros
915 @code{target_insert_watchpoint} and @code{target_remove_watchpoint}
916 are set to call these functions. @code{i386_insert_watchpoint} first
917 looks for a debug register which is already set to watch the same
918 region for the same access types; if found, it just increments the
919 reference count of that debug register, thus implementing debug
920 register sharing between watchpoints. If no such register is found,
921 the function looks for a vacant debug register, sets its mirrored
922 value to @var{addr}, sets the mirrored value of DR7 Debug Control
923 register as appropriate for the @var{len} and @var{type} parameters,
924 and then passes the new values of the debug register and DR7 to the
925 inferior by calling @code{I386_DR_LOW_SET_ADDR} and
926 @code{I386_DR_LOW_SET_CONTROL}. If more than one debug register is
927 required to cover the given region, the above process is repeated for
930 @code{i386_remove_watchpoint} does the opposite: it resets the address
931 in the mirrored value of the debug register and its read/write and
932 length bits in the mirrored value of DR7, then passes these new
933 values to the inferior via @code{I386_DR_LOW_RESET_ADDR} and
934 @code{I386_DR_LOW_SET_CONTROL}. If a register is shared by several
935 watchpoints, each time a @code{i386_remove_watchpoint} is called, it
936 decrements the reference count, and only calls
937 @code{I386_DR_LOW_RESET_ADDR} and @code{I386_DR_LOW_SET_CONTROL} when
938 the count goes to zero.
940 @findex i386_insert_hw_breakpoint
941 @findex i386_remove_hw_breakpoint
942 @item i386_insert_hw_breakpoint (@var{bp_tgt})
943 @itemx i386_remove_hw_breakpoint (@var{bp_tgt})
944 These functions insert and remove hardware-assisted breakpoints. The
945 macros @code{target_insert_hw_breakpoint} and
946 @code{target_remove_hw_breakpoint} are set to call these functions.
947 The argument is a @code{struct bp_target_info *}, as described in
948 the documentation for @code{target_insert_breakpoint}.
949 These functions work like @code{i386_insert_watchpoint} and
950 @code{i386_remove_watchpoint}, respectively, except that they set up
951 the debug registers to watch instruction execution, and each
952 hardware-assisted breakpoint always requires exactly one debug
955 @findex i386_stopped_by_hwbp
956 @item i386_stopped_by_hwbp (void)
957 This function returns non-zero if the inferior has some watchpoint or
958 hardware breakpoint that triggered. It works like
959 @code{i386_stopped_data_address}, except that it doesn't record the
960 address whose watchpoint triggered.
962 @findex i386_cleanup_dregs
963 @item i386_cleanup_dregs (void)
964 This function clears all the reference counts, addresses, and control
965 bits in the mirror images of the debug registers. It doesn't affect
966 the actual debug registers in the inferior process.
973 x86 processors support setting watchpoints on I/O reads or writes.
974 However, since no target supports this (as of March 2001), and since
975 @code{enum target_hw_bp_type} doesn't even have an enumeration for I/O
976 watchpoints, this feature is not yet available to @value{GDBN} running
980 x86 processors can enable watchpoints locally, for the current task
981 only, or globally, for all the tasks. For each debug register,
982 there's a bit in the DR7 Debug Control register that determines
983 whether the associated address is watched locally or globally. The
984 current implementation of x86 watchpoint support in @value{GDBN}
985 always sets watchpoints to be locally enabled, since global
986 watchpoints might interfere with the underlying OS and are probably
987 unavailable in many platforms.
993 In the abstract, a checkpoint is a point in the execution history of
994 the program, which the user may wish to return to at some later time.
996 Internally, a checkpoint is a saved copy of the program state, including
997 whatever information is required in order to restore the program to that
998 state at a later time. This can be expected to include the state of
999 registers and memory, and may include external state such as the state
1000 of open files and devices.
1002 There are a number of ways in which checkpoints may be implemented
1003 in gdb, e.g.@: as corefiles, as forked processes, and as some opaque
1004 method implemented on the target side.
1006 A corefile can be used to save an image of target memory and register
1007 state, which can in principle be restored later --- but corefiles do
1008 not typically include information about external entities such as
1009 open files. Currently this method is not implemented in gdb.
1011 A forked process can save the state of user memory and registers,
1012 as well as some subset of external (kernel) state. This method
1013 is used to implement checkpoints on Linux, and in principle might
1014 be used on other systems.
1016 Some targets, e.g.@: simulators, might have their own built-in
1017 method for saving checkpoints, and gdb might be able to take
1018 advantage of that capability without necessarily knowing any
1019 details of how it is done.
1022 @section Observing changes in @value{GDBN} internals
1023 @cindex observer pattern interface
1024 @cindex notifications about changes in internals
1026 In order to function properly, several modules need to be notified when
1027 some changes occur in the @value{GDBN} internals. Traditionally, these
1028 modules have relied on several paradigms, the most common ones being
1029 hooks and gdb-events. Unfortunately, none of these paradigms was
1030 versatile enough to become the standard notification mechanism in
1031 @value{GDBN}. The fact that they only supported one ``client'' was also
1032 a strong limitation.
1034 A new paradigm, based on the Observer pattern of the @cite{Design
1035 Patterns} book, has therefore been implemented. The goal was to provide
1036 a new interface overcoming the issues with the notification mechanisms
1037 previously available. This new interface needed to be strongly typed,
1038 easy to extend, and versatile enough to be used as the standard
1039 interface when adding new notifications.
1041 See @ref{GDB Observers} for a brief description of the observers
1042 currently implemented in GDB. The rationale for the current
1043 implementation is also briefly discussed.
1045 @node User Interface
1047 @chapter User Interface
1049 @value{GDBN} has several user interfaces, of which the traditional
1050 command-line interface is perhaps the most familiar.
1052 @section Command Interpreter
1054 @cindex command interpreter
1056 The command interpreter in @value{GDBN} is fairly simple. It is designed to
1057 allow for the set of commands to be augmented dynamically, and also
1058 has a recursive subcommand capability, where the first argument to
1059 a command may itself direct a lookup on a different command list.
1061 For instance, the @samp{set} command just starts a lookup on the
1062 @code{setlist} command list, while @samp{set thread} recurses
1063 to the @code{set_thread_cmd_list}.
1067 To add commands in general, use @code{add_cmd}. @code{add_com} adds to
1068 the main command list, and should be used for those commands. The usual
1069 place to add commands is in the @code{_initialize_@var{xyz}} routines at
1070 the ends of most source files.
1072 @findex add_setshow_cmd
1073 @findex add_setshow_cmd_full
1074 To add paired @samp{set} and @samp{show} commands, use
1075 @code{add_setshow_cmd} or @code{add_setshow_cmd_full}. The former is
1076 a slightly simpler interface which is useful when you don't need to
1077 further modify the new command structures, while the latter returns
1078 the new command structures for manipulation.
1080 @cindex deprecating commands
1081 @findex deprecate_cmd
1082 Before removing commands from the command set it is a good idea to
1083 deprecate them for some time. Use @code{deprecate_cmd} on commands or
1084 aliases to set the deprecated flag. @code{deprecate_cmd} takes a
1085 @code{struct cmd_list_element} as it's first argument. You can use the
1086 return value from @code{add_com} or @code{add_cmd} to deprecate the
1087 command immediately after it is created.
1089 The first time a command is used the user will be warned and offered a
1090 replacement (if one exists). Note that the replacement string passed to
1091 @code{deprecate_cmd} should be the full name of the command, i.e., the
1092 entire string the user should type at the command line.
1094 @section UI-Independent Output---the @code{ui_out} Functions
1095 @c This section is based on the documentation written by Fernando
1096 @c Nasser <fnasser@redhat.com>.
1098 @cindex @code{ui_out} functions
1099 The @code{ui_out} functions present an abstraction level for the
1100 @value{GDBN} output code. They hide the specifics of different user
1101 interfaces supported by @value{GDBN}, and thus free the programmer
1102 from the need to write several versions of the same code, one each for
1103 every UI, to produce output.
1105 @subsection Overview and Terminology
1107 In general, execution of each @value{GDBN} command produces some sort
1108 of output, and can even generate an input request.
1110 Output can be generated for the following purposes:
1114 to display a @emph{result} of an operation;
1117 to convey @emph{info} or produce side-effects of a requested
1121 to provide a @emph{notification} of an asynchronous event (including
1122 progress indication of a prolonged asynchronous operation);
1125 to display @emph{error messages} (including warnings);
1128 to show @emph{debug data};
1131 to @emph{query} or prompt a user for input (a special case).
1135 This section mainly concentrates on how to build result output,
1136 although some of it also applies to other kinds of output.
1138 Generation of output that displays the results of an operation
1139 involves one or more of the following:
1143 output of the actual data
1146 formatting the output as appropriate for console output, to make it
1147 easily readable by humans
1150 machine oriented formatting--a more terse formatting to allow for easy
1151 parsing by programs which read @value{GDBN}'s output
1154 annotation, whose purpose is to help legacy GUIs to identify interesting
1158 The @code{ui_out} routines take care of the first three aspects.
1159 Annotations are provided by separate annotation routines. Note that use
1160 of annotations for an interface between a GUI and @value{GDBN} is
1163 Output can be in the form of a single item, which we call a @dfn{field};
1164 a @dfn{list} consisting of identical fields; a @dfn{tuple} consisting of
1165 non-identical fields; or a @dfn{table}, which is a tuple consisting of a
1166 header and a body. In a BNF-like form:
1169 @item <table> @expansion{}
1170 @code{<header> <body>}
1171 @item <header> @expansion{}
1172 @code{@{ <column> @}}
1173 @item <column> @expansion{}
1174 @code{<width> <alignment> <title>}
1175 @item <body> @expansion{}
1180 @subsection General Conventions
1182 Most @code{ui_out} routines are of type @code{void}, the exceptions are
1183 @code{ui_out_stream_new} (which returns a pointer to the newly created
1184 object) and the @code{make_cleanup} routines.
1186 The first parameter is always the @code{ui_out} vector object, a pointer
1187 to a @code{struct ui_out}.
1189 The @var{format} parameter is like in @code{printf} family of functions.
1190 When it is present, there must also be a variable list of arguments
1191 sufficient used to satisfy the @code{%} specifiers in the supplied
1194 When a character string argument is not used in a @code{ui_out} function
1195 call, a @code{NULL} pointer has to be supplied instead.
1198 @subsection Table, Tuple and List Functions
1200 @cindex list output functions
1201 @cindex table output functions
1202 @cindex tuple output functions
1203 This section introduces @code{ui_out} routines for building lists,
1204 tuples and tables. The routines to output the actual data items
1205 (fields) are presented in the next section.
1207 To recap: A @dfn{tuple} is a sequence of @dfn{fields}, each field
1208 containing information about an object; a @dfn{list} is a sequence of
1209 fields where each field describes an identical object.
1211 Use the @dfn{table} functions when your output consists of a list of
1212 rows (tuples) and the console output should include a heading. Use this
1213 even when you are listing just one object but you still want the header.
1215 @cindex nesting level in @code{ui_out} functions
1216 Tables can not be nested. Tuples and lists can be nested up to a
1217 maximum of five levels.
1219 The overall structure of the table output code is something like this:
1234 Here is the description of table-, tuple- and list-related @code{ui_out}
1237 @deftypefun void ui_out_table_begin (struct ui_out *@var{uiout}, int @var{nbrofcols}, int @var{nr_rows}, const char *@var{tblid})
1238 The function @code{ui_out_table_begin} marks the beginning of the output
1239 of a table. It should always be called before any other @code{ui_out}
1240 function for a given table. @var{nbrofcols} is the number of columns in
1241 the table. @var{nr_rows} is the number of rows in the table.
1242 @var{tblid} is an optional string identifying the table. The string
1243 pointed to by @var{tblid} is copied by the implementation of
1244 @code{ui_out_table_begin}, so the application can free the string if it
1245 was @code{malloc}ed.
1247 The companion function @code{ui_out_table_end}, described below, marks
1248 the end of the table's output.
1251 @deftypefun void ui_out_table_header (struct ui_out *@var{uiout}, int @var{width}, enum ui_align @var{alignment}, const char *@var{colhdr})
1252 @code{ui_out_table_header} provides the header information for a single
1253 table column. You call this function several times, one each for every
1254 column of the table, after @code{ui_out_table_begin}, but before
1255 @code{ui_out_table_body}.
1257 The value of @var{width} gives the column width in characters. The
1258 value of @var{alignment} is one of @code{left}, @code{center}, and
1259 @code{right}, and it specifies how to align the header: left-justify,
1260 center, or right-justify it. @var{colhdr} points to a string that
1261 specifies the column header; the implementation copies that string, so
1262 column header strings in @code{malloc}ed storage can be freed after the
1266 @deftypefun void ui_out_table_body (struct ui_out *@var{uiout})
1267 This function delimits the table header from the table body.
1270 @deftypefun void ui_out_table_end (struct ui_out *@var{uiout})
1271 This function signals the end of a table's output. It should be called
1272 after the table body has been produced by the list and field output
1275 There should be exactly one call to @code{ui_out_table_end} for each
1276 call to @code{ui_out_table_begin}, otherwise the @code{ui_out} functions
1277 will signal an internal error.
1280 The output of the tuples that represent the table rows must follow the
1281 call to @code{ui_out_table_body} and precede the call to
1282 @code{ui_out_table_end}. You build a tuple by calling
1283 @code{ui_out_tuple_begin} and @code{ui_out_tuple_end}, with suitable
1284 calls to functions which actually output fields between them.
1286 @deftypefun void ui_out_tuple_begin (struct ui_out *@var{uiout}, const char *@var{id})
1287 This function marks the beginning of a tuple output. @var{id} points
1288 to an optional string that identifies the tuple; it is copied by the
1289 implementation, and so strings in @code{malloc}ed storage can be freed
1293 @deftypefun void ui_out_tuple_end (struct ui_out *@var{uiout})
1294 This function signals an end of a tuple output. There should be exactly
1295 one call to @code{ui_out_tuple_end} for each call to
1296 @code{ui_out_tuple_begin}, otherwise an internal @value{GDBN} error will
1300 @deftypefun struct cleanup *make_cleanup_ui_out_tuple_begin_end (struct ui_out *@var{uiout}, const char *@var{id})
1301 This function first opens the tuple and then establishes a cleanup
1302 (@pxref{Coding, Cleanups}) to close the tuple. It provides a convenient
1303 and correct implementation of the non-portable@footnote{The function
1304 cast is not portable ISO C.} code sequence:
1306 struct cleanup *old_cleanup;
1307 ui_out_tuple_begin (uiout, "...");
1308 old_cleanup = make_cleanup ((void(*)(void *)) ui_out_tuple_end,
1313 @deftypefun void ui_out_list_begin (struct ui_out *@var{uiout}, const char *@var{id})
1314 This function marks the beginning of a list output. @var{id} points to
1315 an optional string that identifies the list; it is copied by the
1316 implementation, and so strings in @code{malloc}ed storage can be freed
1320 @deftypefun void ui_out_list_end (struct ui_out *@var{uiout})
1321 This function signals an end of a list output. There should be exactly
1322 one call to @code{ui_out_list_end} for each call to
1323 @code{ui_out_list_begin}, otherwise an internal @value{GDBN} error will
1327 @deftypefun struct cleanup *make_cleanup_ui_out_list_begin_end (struct ui_out *@var{uiout}, const char *@var{id})
1328 Similar to @code{make_cleanup_ui_out_tuple_begin_end}, this function
1329 opens a list and then establishes cleanup (@pxref{Coding, Cleanups})
1330 that will close the list.
1333 @subsection Item Output Functions
1335 @cindex item output functions
1336 @cindex field output functions
1338 The functions described below produce output for the actual data
1339 items, or fields, which contain information about the object.
1341 Choose the appropriate function accordingly to your particular needs.
1343 @deftypefun void ui_out_field_fmt (struct ui_out *@var{uiout}, char *@var{fldname}, char *@var{format}, ...)
1344 This is the most general output function. It produces the
1345 representation of the data in the variable-length argument list
1346 according to formatting specifications in @var{format}, a
1347 @code{printf}-like format string. The optional argument @var{fldname}
1348 supplies the name of the field. The data items themselves are
1349 supplied as additional arguments after @var{format}.
1351 This generic function should be used only when it is not possible to
1352 use one of the specialized versions (see below).
1355 @deftypefun void ui_out_field_int (struct ui_out *@var{uiout}, const char *@var{fldname}, int @var{value})
1356 This function outputs a value of an @code{int} variable. It uses the
1357 @code{"%d"} output conversion specification. @var{fldname} specifies
1358 the name of the field.
1361 @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})
1362 This function outputs a value of an @code{int} variable. It differs from
1363 @code{ui_out_field_int} in that the caller specifies the desired @var{width} and @var{alignment} of the output.
1364 @var{fldname} specifies
1365 the name of the field.
1368 @deftypefun void ui_out_field_core_addr (struct ui_out *@var{uiout}, const char *@var{fldname}, CORE_ADDR @var{address})
1369 This function outputs an address.
1372 @deftypefun void ui_out_field_string (struct ui_out *@var{uiout}, const char *@var{fldname}, const char *@var{string})
1373 This function outputs a string using the @code{"%s"} conversion
1377 Sometimes, there's a need to compose your output piece by piece using
1378 functions that operate on a stream, such as @code{value_print} or
1379 @code{fprintf_symbol_filtered}. These functions accept an argument of
1380 the type @code{struct ui_file *}, a pointer to a @code{ui_file} object
1381 used to store the data stream used for the output. When you use one
1382 of these functions, you need a way to pass their results stored in a
1383 @code{ui_file} object to the @code{ui_out} functions. To this end,
1384 you first create a @code{ui_stream} object by calling
1385 @code{ui_out_stream_new}, pass the @code{stream} member of that
1386 @code{ui_stream} object to @code{value_print} and similar functions,
1387 and finally call @code{ui_out_field_stream} to output the field you
1388 constructed. When the @code{ui_stream} object is no longer needed,
1389 you should destroy it and free its memory by calling
1390 @code{ui_out_stream_delete}.
1392 @deftypefun struct ui_stream *ui_out_stream_new (struct ui_out *@var{uiout})
1393 This function creates a new @code{ui_stream} object which uses the
1394 same output methods as the @code{ui_out} object whose pointer is
1395 passed in @var{uiout}. It returns a pointer to the newly created
1396 @code{ui_stream} object.
1399 @deftypefun void ui_out_stream_delete (struct ui_stream *@var{streambuf})
1400 This functions destroys a @code{ui_stream} object specified by
1404 @deftypefun void ui_out_field_stream (struct ui_out *@var{uiout}, const char *@var{fieldname}, struct ui_stream *@var{streambuf})
1405 This function consumes all the data accumulated in
1406 @code{streambuf->stream} and outputs it like
1407 @code{ui_out_field_string} does. After a call to
1408 @code{ui_out_field_stream}, the accumulated data no longer exists, but
1409 the stream is still valid and may be used for producing more fields.
1412 @strong{Important:} If there is any chance that your code could bail
1413 out before completing output generation and reaching the point where
1414 @code{ui_out_stream_delete} is called, it is necessary to set up a
1415 cleanup, to avoid leaking memory and other resources. Here's a
1416 skeleton code to do that:
1419 struct ui_stream *mybuf = ui_out_stream_new (uiout);
1420 struct cleanup *old = make_cleanup (ui_out_stream_delete, mybuf);
1425 If the function already has the old cleanup chain set (for other kinds
1426 of cleanups), you just have to add your cleanup to it:
1429 mybuf = ui_out_stream_new (uiout);
1430 make_cleanup (ui_out_stream_delete, mybuf);
1433 Note that with cleanups in place, you should not call
1434 @code{ui_out_stream_delete} directly, or you would attempt to free the
1437 @subsection Utility Output Functions
1439 @deftypefun void ui_out_field_skip (struct ui_out *@var{uiout}, const char *@var{fldname})
1440 This function skips a field in a table. Use it if you have to leave
1441 an empty field without disrupting the table alignment. The argument
1442 @var{fldname} specifies a name for the (missing) filed.
1445 @deftypefun void ui_out_text (struct ui_out *@var{uiout}, const char *@var{string})
1446 This function outputs the text in @var{string} in a way that makes it
1447 easy to be read by humans. For example, the console implementation of
1448 this method filters the text through a built-in pager, to prevent it
1449 from scrolling off the visible portion of the screen.
1451 Use this function for printing relatively long chunks of text around
1452 the actual field data: the text it produces is not aligned according
1453 to the table's format. Use @code{ui_out_field_string} to output a
1454 string field, and use @code{ui_out_message}, described below, to
1455 output short messages.
1458 @deftypefun void ui_out_spaces (struct ui_out *@var{uiout}, int @var{nspaces})
1459 This function outputs @var{nspaces} spaces. It is handy to align the
1460 text produced by @code{ui_out_text} with the rest of the table or
1464 @deftypefun void ui_out_message (struct ui_out *@var{uiout}, int @var{verbosity}, const char *@var{format}, ...)
1465 This function produces a formatted message, provided that the current
1466 verbosity level is at least as large as given by @var{verbosity}. The
1467 current verbosity level is specified by the user with the @samp{set
1468 verbositylevel} command.@footnote{As of this writing (April 2001),
1469 setting verbosity level is not yet implemented, and is always returned
1470 as zero. So calling @code{ui_out_message} with a @var{verbosity}
1471 argument more than zero will cause the message to never be printed.}
1474 @deftypefun void ui_out_wrap_hint (struct ui_out *@var{uiout}, char *@var{indent})
1475 This function gives the console output filter (a paging filter) a hint
1476 of where to break lines which are too long. Ignored for all other
1477 output consumers. @var{indent}, if non-@code{NULL}, is the string to
1478 be printed to indent the wrapped text on the next line; it must remain
1479 accessible until the next call to @code{ui_out_wrap_hint}, or until an
1480 explicit newline is produced by one of the other functions. If
1481 @var{indent} is @code{NULL}, the wrapped text will not be indented.
1484 @deftypefun void ui_out_flush (struct ui_out *@var{uiout})
1485 This function flushes whatever output has been accumulated so far, if
1486 the UI buffers output.
1490 @subsection Examples of Use of @code{ui_out} functions
1492 @cindex using @code{ui_out} functions
1493 @cindex @code{ui_out} functions, usage examples
1494 This section gives some practical examples of using the @code{ui_out}
1495 functions to generalize the old console-oriented code in
1496 @value{GDBN}. The examples all come from functions defined on the
1497 @file{breakpoints.c} file.
1499 This example, from the @code{breakpoint_1} function, shows how to
1502 The original code was:
1505 if (!found_a_breakpoint++)
1507 annotate_breakpoints_headers ();
1510 printf_filtered ("Num ");
1512 printf_filtered ("Type ");
1514 printf_filtered ("Disp ");
1516 printf_filtered ("Enb ");
1520 printf_filtered ("Address ");
1523 printf_filtered ("What\n");
1525 annotate_breakpoints_table ();
1529 Here's the new version:
1532 nr_printable_breakpoints = @dots{};
1535 ui_out_table_begin (ui, 6, nr_printable_breakpoints, "BreakpointTable");
1537 ui_out_table_begin (ui, 5, nr_printable_breakpoints, "BreakpointTable");
1539 if (nr_printable_breakpoints > 0)
1540 annotate_breakpoints_headers ();
1541 if (nr_printable_breakpoints > 0)
1543 ui_out_table_header (uiout, 3, ui_left, "number", "Num"); /* 1 */
1544 if (nr_printable_breakpoints > 0)
1546 ui_out_table_header (uiout, 14, ui_left, "type", "Type"); /* 2 */
1547 if (nr_printable_breakpoints > 0)
1549 ui_out_table_header (uiout, 4, ui_left, "disp", "Disp"); /* 3 */
1550 if (nr_printable_breakpoints > 0)
1552 ui_out_table_header (uiout, 3, ui_left, "enabled", "Enb"); /* 4 */
1555 if (nr_printable_breakpoints > 0)
1557 if (gdbarch_addr_bit (current_gdbarch) <= 32)
1558 ui_out_table_header (uiout, 10, ui_left, "addr", "Address");/* 5 */
1560 ui_out_table_header (uiout, 18, ui_left, "addr", "Address");/* 5 */
1562 if (nr_printable_breakpoints > 0)
1564 ui_out_table_header (uiout, 40, ui_noalign, "what", "What"); /* 6 */
1565 ui_out_table_body (uiout);
1566 if (nr_printable_breakpoints > 0)
1567 annotate_breakpoints_table ();
1570 This example, from the @code{print_one_breakpoint} function, shows how
1571 to produce the actual data for the table whose structure was defined
1572 in the above example. The original code was:
1577 printf_filtered ("%-3d ", b->number);
1579 if ((int)b->type > (sizeof(bptypes)/sizeof(bptypes[0]))
1580 || ((int) b->type != bptypes[(int) b->type].type))
1581 internal_error ("bptypes table does not describe type #%d.",
1583 printf_filtered ("%-14s ", bptypes[(int)b->type].description);
1585 printf_filtered ("%-4s ", bpdisps[(int)b->disposition]);
1587 printf_filtered ("%-3c ", bpenables[(int)b->enable]);
1591 This is the new version:
1595 ui_out_tuple_begin (uiout, "bkpt");
1597 ui_out_field_int (uiout, "number", b->number);
1599 if (((int) b->type > (sizeof (bptypes) / sizeof (bptypes[0])))
1600 || ((int) b->type != bptypes[(int) b->type].type))
1601 internal_error ("bptypes table does not describe type #%d.",
1603 ui_out_field_string (uiout, "type", bptypes[(int)b->type].description);
1605 ui_out_field_string (uiout, "disp", bpdisps[(int)b->disposition]);
1607 ui_out_field_fmt (uiout, "enabled", "%c", bpenables[(int)b->enable]);
1611 This example, also from @code{print_one_breakpoint}, shows how to
1612 produce a complicated output field using the @code{print_expression}
1613 functions which requires a stream to be passed. It also shows how to
1614 automate stream destruction with cleanups. The original code was:
1618 print_expression (b->exp, gdb_stdout);
1624 struct ui_stream *stb = ui_out_stream_new (uiout);
1625 struct cleanup *old_chain = make_cleanup_ui_out_stream_delete (stb);
1628 print_expression (b->exp, stb->stream);
1629 ui_out_field_stream (uiout, "what", local_stream);
1632 This example, also from @code{print_one_breakpoint}, shows how to use
1633 @code{ui_out_text} and @code{ui_out_field_string}. The original code
1638 if (b->dll_pathname == NULL)
1639 printf_filtered ("<any library> ");
1641 printf_filtered ("library \"%s\" ", b->dll_pathname);
1648 if (b->dll_pathname == NULL)
1650 ui_out_field_string (uiout, "what", "<any library>");
1651 ui_out_spaces (uiout, 1);
1655 ui_out_text (uiout, "library \"");
1656 ui_out_field_string (uiout, "what", b->dll_pathname);
1657 ui_out_text (uiout, "\" ");
1661 The following example from @code{print_one_breakpoint} shows how to
1662 use @code{ui_out_field_int} and @code{ui_out_spaces}. The original
1667 if (b->forked_inferior_pid != 0)
1668 printf_filtered ("process %d ", b->forked_inferior_pid);
1675 if (b->forked_inferior_pid != 0)
1677 ui_out_text (uiout, "process ");
1678 ui_out_field_int (uiout, "what", b->forked_inferior_pid);
1679 ui_out_spaces (uiout, 1);
1683 Here's an example of using @code{ui_out_field_string}. The original
1688 if (b->exec_pathname != NULL)
1689 printf_filtered ("program \"%s\" ", b->exec_pathname);
1696 if (b->exec_pathname != NULL)
1698 ui_out_text (uiout, "program \"");
1699 ui_out_field_string (uiout, "what", b->exec_pathname);
1700 ui_out_text (uiout, "\" ");
1704 Finally, here's an example of printing an address. The original code:
1708 printf_filtered ("%s ",
1709 hex_string_custom ((unsigned long) b->address, 8));
1716 ui_out_field_core_addr (uiout, "Address", b->address);
1720 @section Console Printing
1729 @cindex @code{libgdb}
1730 @code{libgdb} 1.0 was an abortive project of years ago. The theory was
1731 to provide an API to @value{GDBN}'s functionality.
1734 @cindex @code{libgdb}
1735 @code{libgdb} 2.0 is an ongoing effort to update @value{GDBN} so that is
1736 better able to support graphical and other environments.
1738 Since @code{libgdb} development is on-going, its architecture is still
1739 evolving. The following components have so far been identified:
1743 Observer - @file{gdb-events.h}.
1745 Builder - @file{ui-out.h}
1747 Event Loop - @file{event-loop.h}
1749 Library - @file{gdb.h}
1752 The model that ties these components together is described below.
1754 @section The @code{libgdb} Model
1756 A client of @code{libgdb} interacts with the library in two ways.
1760 As an observer (using @file{gdb-events}) receiving notifications from
1761 @code{libgdb} of any internal state changes (break point changes, run
1764 As a client querying @code{libgdb} (using the @file{ui-out} builder) to
1765 obtain various status values from @value{GDBN}.
1768 Since @code{libgdb} could have multiple clients (e.g., a GUI supporting
1769 the existing @value{GDBN} CLI), those clients must co-operate when
1770 controlling @code{libgdb}. In particular, a client must ensure that
1771 @code{libgdb} is idle (i.e. no other client is using @code{libgdb})
1772 before responding to a @file{gdb-event} by making a query.
1774 @section CLI support
1776 At present @value{GDBN}'s CLI is very much entangled in with the core of
1777 @code{libgdb}. Consequently, a client wishing to include the CLI in
1778 their interface needs to carefully co-ordinate its own and the CLI's
1781 It is suggested that the client set @code{libgdb} up to be bi-modal
1782 (alternate between CLI and client query modes). The notes below sketch
1787 The client registers itself as an observer of @code{libgdb}.
1789 The client create and install @code{cli-out} builder using its own
1790 versions of the @code{ui-file} @code{gdb_stderr}, @code{gdb_stdtarg} and
1791 @code{gdb_stdout} streams.
1793 The client creates a separate custom @code{ui-out} builder that is only
1794 used while making direct queries to @code{libgdb}.
1797 When the client receives input intended for the CLI, it simply passes it
1798 along. Since the @code{cli-out} builder is installed by default, all
1799 the CLI output in response to that command is routed (pronounced rooted)
1800 through to the client controlled @code{gdb_stdout} et.@: al.@: streams.
1801 At the same time, the client is kept abreast of internal changes by
1802 virtue of being a @code{libgdb} observer.
1804 The only restriction on the client is that it must wait until
1805 @code{libgdb} becomes idle before initiating any queries (using the
1806 client's custom builder).
1808 @section @code{libgdb} components
1810 @subheading Observer - @file{gdb-events.h}
1811 @file{gdb-events} provides the client with a very raw mechanism that can
1812 be used to implement an observer. At present it only allows for one
1813 observer and that observer must, internally, handle the need to delay
1814 the processing of any event notifications until after @code{libgdb} has
1815 finished the current command.
1817 @subheading Builder - @file{ui-out.h}
1818 @file{ui-out} provides the infrastructure necessary for a client to
1819 create a builder. That builder is then passed down to @code{libgdb}
1820 when doing any queries.
1822 @subheading Event Loop - @file{event-loop.h}
1823 @c There could be an entire section on the event-loop
1824 @file{event-loop}, currently non-re-entrant, provides a simple event
1825 loop. A client would need to either plug its self into this loop or,
1826 implement a new event-loop that GDB would use.
1828 The event-loop will eventually be made re-entrant. This is so that
1829 @value{GDBN} can better handle the problem of some commands blocking
1830 instead of returning.
1832 @subheading Library - @file{gdb.h}
1833 @file{libgdb} is the most obvious component of this system. It provides
1834 the query interface. Each function is parameterized by a @code{ui-out}
1835 builder. The result of the query is constructed using that builder
1836 before the query function returns.
1839 @chapter Stack Frames
1842 @cindex call stack frame
1843 A frame is a construct that @value{GDBN} uses to keep track of calling
1844 and called functions.
1846 @cindex unwind frame
1847 @value{GDBN}'s frame model, a fresh design, was implemented with the
1848 need to support @sc{dwarf}'s Call Frame Information in mind. In fact,
1849 the term ``unwind'' is taken directly from that specification.
1850 Developers wishing to learn more about unwinders, are encouraged to
1851 read the @sc{dwarf} specification, available from
1852 @url{http://www.dwarfstd.org}.
1854 @findex frame_register_unwind
1855 @findex get_frame_register
1856 @value{GDBN}'s model is that you find a frame's registers by
1857 ``unwinding'' them from the next younger frame. That is,
1858 @samp{get_frame_register} which returns the value of a register in
1859 frame #1 (the next-to-youngest frame), is implemented by calling frame
1860 #0's @code{frame_register_unwind} (the youngest frame). But then the
1861 obvious question is: how do you access the registers of the youngest
1864 @cindex sentinel frame
1865 @findex get_frame_type
1866 @vindex SENTINEL_FRAME
1867 To answer this question, GDB has the @dfn{sentinel} frame, the
1868 ``-1st'' frame. Unwinding registers from the sentinel frame gives you
1869 the current values of the youngest real frame's registers. If @var{f}
1870 is a sentinel frame, then @code{get_frame_type (@var{f}) @equiv{}
1873 @section Selecting an Unwinder
1875 @findex frame_unwind_prepend_unwinder
1876 @findex frame_unwind_append_unwinder
1877 The architecture registers a list of frame unwinders (@code{struct
1878 frame_unwind}), using the functions
1879 @code{frame_unwind_prepend_unwinder} and
1880 @code{frame_unwind_append_unwinder}. Each unwinder includes a
1881 sniffer. Whenever @value{GDBN} needs to unwind a frame (to fetch the
1882 previous frame's registers or the current frame's ID), it calls
1883 registered sniffers in order to find one which recognizes the frame.
1884 The first time a sniffer returns non-zero, the corresponding unwinder
1885 is assigned to the frame.
1887 @section Unwinding the Frame ID
1890 Every frame has an associated ID, of type @code{struct frame_id}.
1891 The ID includes the stack base and function start address for
1892 the frame. The ID persists through the entire life of the frame,
1893 including while other called frames are running; it is used to
1894 locate an appropriate @code{struct frame_info} from the cache.
1896 Every time the inferior stops, and at various other times, the frame
1897 cache is flushed. Because of this, parts of @value{GDBN} which need
1898 to keep track of individual frames cannot use pointers to @code{struct
1899 frame_info}. A frame ID provides a stable reference to a frame, even
1900 when the unwinder must be run again to generate a new @code{struct
1901 frame_info} for the same frame.
1903 The frame's unwinder's @code{this_id} method is called to find the ID.
1904 Note that this is different from register unwinding, where the next
1905 frame's @code{prev_register} is called to unwind this frame's
1908 Both stack base and function address are required to identify the
1909 frame, because a recursive function has the same function address for
1910 two consecutive frames and a leaf function may have the same stack
1911 address as its caller. On some platforms, a third address is part of
1912 the ID to further disambiguate frames---for instance, on IA-64
1913 the separate register stack address is included in the ID.
1915 An invalid frame ID (@code{null_frame_id}) returned from the
1916 @code{this_id} method means to stop unwinding after this frame.
1918 @section Unwinding Registers
1920 Each unwinder includes a @code{prev_register} method. This method
1921 takes a frame, an associated cache pointer, and a register number.
1922 It returns a @code{struct value *} describing the requested register,
1923 as saved by this frame. This is the value of the register that is
1924 current in this frame's caller.
1926 The returned value must have the same type as the register. It may
1927 have any lvalue type. In most circumstances one of these routines
1928 will generate the appropriate value:
1931 @item frame_unwind_got_optimized
1932 @findex frame_unwind_got_optimized
1933 This register was not saved.
1935 @item frame_unwind_got_register
1936 @findex frame_unwind_got_register
1937 This register was copied into another register in this frame. This
1938 is also used for unchanged registers; they are ``copied'' into the
1941 @item frame_unwind_got_memory
1942 @findex frame_unwind_got_memory
1943 This register was saved in memory.
1945 @item frame_unwind_got_constant
1946 @findex frame_unwind_got_constant
1947 This register was not saved, but the unwinder can compute the previous
1948 value some other way.
1950 @item frame_unwind_got_address
1951 @findex frame_unwind_got_address
1952 Same as @code{frame_unwind_got_constant}, except that the value is a target
1953 address. This is frequently used for the stack pointer, which is not
1954 explicitly saved but has a known offset from this frame's stack
1955 pointer. For architectures with a flat unified address space, this is
1956 generally the same as @code{frame_unwind_got_constant}.
1959 @node Symbol Handling
1961 @chapter Symbol Handling
1963 Symbols are a key part of @value{GDBN}'s operation. Symbols include
1964 variables, functions, and types.
1966 Symbol information for a large program can be truly massive, and
1967 reading of symbol information is one of the major performance
1968 bottlenecks in @value{GDBN}; it can take many minutes to process it
1969 all. Studies have shown that nearly all the time spent is
1970 computational, rather than file reading.
1972 One of the ways for @value{GDBN} to provide a good user experience is
1973 to start up quickly, taking no more than a few seconds. It is simply
1974 not possible to process all of a program's debugging info in that
1975 time, and so we attempt to handle symbols incrementally. For instance,
1976 we create @dfn{partial symbol tables} consisting of only selected
1977 symbols, and only expand them to full symbol tables when necessary.
1979 @section Symbol Reading
1981 @cindex symbol reading
1982 @cindex reading of symbols
1983 @cindex symbol files
1984 @value{GDBN} reads symbols from @dfn{symbol files}. The usual symbol
1985 file is the file containing the program which @value{GDBN} is
1986 debugging. @value{GDBN} can be directed to use a different file for
1987 symbols (with the @samp{symbol-file} command), and it can also read
1988 more symbols via the @samp{add-file} and @samp{load} commands. In
1989 addition, it may bring in more symbols while loading shared
1992 @findex find_sym_fns
1993 Symbol files are initially opened by code in @file{symfile.c} using
1994 the BFD library (@pxref{Support Libraries}). BFD identifies the type
1995 of the file by examining its header. @code{find_sym_fns} then uses
1996 this identification to locate a set of symbol-reading functions.
1998 @findex add_symtab_fns
1999 @cindex @code{sym_fns} structure
2000 @cindex adding a symbol-reading module
2001 Symbol-reading modules identify themselves to @value{GDBN} by calling
2002 @code{add_symtab_fns} during their module initialization. The argument
2003 to @code{add_symtab_fns} is a @code{struct sym_fns} which contains the
2004 name (or name prefix) of the symbol format, the length of the prefix,
2005 and pointers to four functions. These functions are called at various
2006 times to process symbol files whose identification matches the specified
2009 The functions supplied by each module are:
2012 @item @var{xyz}_symfile_init(struct sym_fns *sf)
2014 @cindex secondary symbol file
2015 Called from @code{symbol_file_add} when we are about to read a new
2016 symbol file. This function should clean up any internal state (possibly
2017 resulting from half-read previous files, for example) and prepare to
2018 read a new symbol file. Note that the symbol file which we are reading
2019 might be a new ``main'' symbol file, or might be a secondary symbol file
2020 whose symbols are being added to the existing symbol table.
2022 The argument to @code{@var{xyz}_symfile_init} is a newly allocated
2023 @code{struct sym_fns} whose @code{bfd} field contains the BFD for the
2024 new symbol file being read. Its @code{private} field has been zeroed,
2025 and can be modified as desired. Typically, a struct of private
2026 information will be @code{malloc}'d, and a pointer to it will be placed
2027 in the @code{private} field.
2029 There is no result from @code{@var{xyz}_symfile_init}, but it can call
2030 @code{error} if it detects an unavoidable problem.
2032 @item @var{xyz}_new_init()
2034 Called from @code{symbol_file_add} when discarding existing symbols.
2035 This function needs only handle the symbol-reading module's internal
2036 state; the symbol table data structures visible to the rest of
2037 @value{GDBN} will be discarded by @code{symbol_file_add}. It has no
2038 arguments and no result. It may be called after
2039 @code{@var{xyz}_symfile_init}, if a new symbol table is being read, or
2040 may be called alone if all symbols are simply being discarded.
2042 @item @var{xyz}_symfile_read(struct sym_fns *sf, CORE_ADDR addr, int mainline)
2044 Called from @code{symbol_file_add} to actually read the symbols from a
2045 symbol-file into a set of psymtabs or symtabs.
2047 @code{sf} points to the @code{struct sym_fns} originally passed to
2048 @code{@var{xyz}_sym_init} for possible initialization. @code{addr} is
2049 the offset between the file's specified start address and its true
2050 address in memory. @code{mainline} is 1 if this is the main symbol
2051 table being read, and 0 if a secondary symbol file (e.g., shared library
2052 or dynamically loaded file) is being read.@refill
2055 In addition, if a symbol-reading module creates psymtabs when
2056 @var{xyz}_symfile_read is called, these psymtabs will contain a pointer
2057 to a function @code{@var{xyz}_psymtab_to_symtab}, which can be called
2058 from any point in the @value{GDBN} symbol-handling code.
2061 @item @var{xyz}_psymtab_to_symtab (struct partial_symtab *pst)
2063 Called from @code{psymtab_to_symtab} (or the @code{PSYMTAB_TO_SYMTAB} macro) if
2064 the psymtab has not already been read in and had its @code{pst->symtab}
2065 pointer set. The argument is the psymtab to be fleshed-out into a
2066 symtab. Upon return, @code{pst->readin} should have been set to 1, and
2067 @code{pst->symtab} should contain a pointer to the new corresponding symtab, or
2068 zero if there were no symbols in that part of the symbol file.
2071 @section Partial Symbol Tables
2073 @value{GDBN} has three types of symbol tables:
2076 @cindex full symbol table
2079 Full symbol tables (@dfn{symtabs}). These contain the main
2080 information about symbols and addresses.
2084 Partial symbol tables (@dfn{psymtabs}). These contain enough
2085 information to know when to read the corresponding part of the full
2088 @cindex minimal symbol table
2091 Minimal symbol tables (@dfn{msymtabs}). These contain information
2092 gleaned from non-debugging symbols.
2095 @cindex partial symbol table
2096 This section describes partial symbol tables.
2098 A psymtab is constructed by doing a very quick pass over an executable
2099 file's debugging information. Small amounts of information are
2100 extracted---enough to identify which parts of the symbol table will
2101 need to be re-read and fully digested later, when the user needs the
2102 information. The speed of this pass causes @value{GDBN} to start up very
2103 quickly. Later, as the detailed rereading occurs, it occurs in small
2104 pieces, at various times, and the delay therefrom is mostly invisible to
2106 @c (@xref{Symbol Reading}.)
2108 The symbols that show up in a file's psymtab should be, roughly, those
2109 visible to the debugger's user when the program is not running code from
2110 that file. These include external symbols and types, static symbols and
2111 types, and @code{enum} values declared at file scope.
2113 The psymtab also contains the range of instruction addresses that the
2114 full symbol table would represent.
2116 @cindex finding a symbol
2117 @cindex symbol lookup
2118 The idea is that there are only two ways for the user (or much of the
2119 code in the debugger) to reference a symbol:
2122 @findex find_pc_function
2123 @findex find_pc_line
2125 By its address (e.g., execution stops at some address which is inside a
2126 function in this file). The address will be noticed to be in the
2127 range of this psymtab, and the full symtab will be read in.
2128 @code{find_pc_function}, @code{find_pc_line}, and other
2129 @code{find_pc_@dots{}} functions handle this.
2131 @cindex lookup_symbol
2134 (e.g., the user asks to print a variable, or set a breakpoint on a
2135 function). Global names and file-scope names will be found in the
2136 psymtab, which will cause the symtab to be pulled in. Local names will
2137 have to be qualified by a global name, or a file-scope name, in which
2138 case we will have already read in the symtab as we evaluated the
2139 qualifier. Or, a local symbol can be referenced when we are ``in'' a
2140 local scope, in which case the first case applies. @code{lookup_symbol}
2141 does most of the work here.
2144 The only reason that psymtabs exist is to cause a symtab to be read in
2145 at the right moment. Any symbol that can be elided from a psymtab,
2146 while still causing that to happen, should not appear in it. Since
2147 psymtabs don't have the idea of scope, you can't put local symbols in
2148 them anyway. Psymtabs don't have the idea of the type of a symbol,
2149 either, so types need not appear, unless they will be referenced by
2152 It is a bug for @value{GDBN} to behave one way when only a psymtab has
2153 been read, and another way if the corresponding symtab has been read
2154 in. Such bugs are typically caused by a psymtab that does not contain
2155 all the visible symbols, or which has the wrong instruction address
2158 The psymtab for a particular section of a symbol file (objfile) could be
2159 thrown away after the symtab has been read in. The symtab should always
2160 be searched before the psymtab, so the psymtab will never be used (in a
2161 bug-free environment). Currently, psymtabs are allocated on an obstack,
2162 and all the psymbols themselves are allocated in a pair of large arrays
2163 on an obstack, so there is little to be gained by trying to free them
2164 unless you want to do a lot more work.
2168 @unnumberedsubsec Fundamental Types (e.g., @code{FT_VOID}, @code{FT_BOOLEAN}).
2170 @cindex fundamental types
2171 These are the fundamental types that @value{GDBN} uses internally. Fundamental
2172 types from the various debugging formats (stabs, ELF, etc) are mapped
2173 into one of these. They are basically a union of all fundamental types
2174 that @value{GDBN} knows about for all the languages that @value{GDBN}
2177 @unnumberedsubsec Type Codes (e.g., @code{TYPE_CODE_PTR}, @code{TYPE_CODE_ARRAY}).
2180 Each time @value{GDBN} builds an internal type, it marks it with one
2181 of these types. The type may be a fundamental type, such as
2182 @code{TYPE_CODE_INT}, or a derived type, such as @code{TYPE_CODE_PTR}
2183 which is a pointer to another type. Typically, several @code{FT_*}
2184 types map to one @code{TYPE_CODE_*} type, and are distinguished by
2185 other members of the type struct, such as whether the type is signed
2186 or unsigned, and how many bits it uses.
2188 @unnumberedsubsec Builtin Types (e.g., @code{builtin_type_void}, @code{builtin_type_char}).
2190 These are instances of type structs that roughly correspond to
2191 fundamental types and are created as global types for @value{GDBN} to
2192 use for various ugly historical reasons. We eventually want to
2193 eliminate these. Note for example that @code{builtin_type_int}
2194 initialized in @file{gdbtypes.c} is basically the same as a
2195 @code{TYPE_CODE_INT} type that is initialized in @file{c-lang.c} for
2196 an @code{FT_INTEGER} fundamental type. The difference is that the
2197 @code{builtin_type} is not associated with any particular objfile, and
2198 only one instance exists, while @file{c-lang.c} builds as many
2199 @code{TYPE_CODE_INT} types as needed, with each one associated with
2200 some particular objfile.
2202 @section Object File Formats
2203 @cindex object file formats
2207 @cindex @code{a.out} format
2208 The @code{a.out} format is the original file format for Unix. It
2209 consists of three sections: @code{text}, @code{data}, and @code{bss},
2210 which are for program code, initialized data, and uninitialized data,
2213 The @code{a.out} format is so simple that it doesn't have any reserved
2214 place for debugging information. (Hey, the original Unix hackers used
2215 @samp{adb}, which is a machine-language debugger!) The only debugging
2216 format for @code{a.out} is stabs, which is encoded as a set of normal
2217 symbols with distinctive attributes.
2219 The basic @code{a.out} reader is in @file{dbxread.c}.
2224 The COFF format was introduced with System V Release 3 (SVR3) Unix.
2225 COFF files may have multiple sections, each prefixed by a header. The
2226 number of sections is limited.
2228 The COFF specification includes support for debugging. Although this
2229 was a step forward, the debugging information was woefully limited.
2230 For instance, it was not possible to represent code that came from an
2231 included file. GNU's COFF-using configs often use stabs-type info,
2232 encapsulated in special sections.
2234 The COFF reader is in @file{coffread.c}.
2238 @cindex ECOFF format
2239 ECOFF is an extended COFF originally introduced for Mips and Alpha
2242 The basic ECOFF reader is in @file{mipsread.c}.
2246 @cindex XCOFF format
2247 The IBM RS/6000 running AIX uses an object file format called XCOFF.
2248 The COFF sections, symbols, and line numbers are used, but debugging
2249 symbols are @code{dbx}-style stabs whose strings are located in the
2250 @code{.debug} section (rather than the string table). For more
2251 information, see @ref{Top,,,stabs,The Stabs Debugging Format}.
2253 The shared library scheme has a clean interface for figuring out what
2254 shared libraries are in use, but the catch is that everything which
2255 refers to addresses (symbol tables and breakpoints at least) needs to be
2256 relocated for both shared libraries and the main executable. At least
2257 using the standard mechanism this can only be done once the program has
2258 been run (or the core file has been read).
2262 @cindex PE-COFF format
2263 Windows 95 and NT use the PE (@dfn{Portable Executable}) format for their
2264 executables. PE is basically COFF with additional headers.
2266 While BFD includes special PE support, @value{GDBN} needs only the basic
2272 The ELF format came with System V Release 4 (SVR4) Unix. ELF is
2273 similar to COFF in being organized into a number of sections, but it
2274 removes many of COFF's limitations. Debugging info may be either stabs
2275 encapsulated in ELF sections, or more commonly these days, DWARF.
2277 The basic ELF reader is in @file{elfread.c}.
2282 SOM is HP's object file and debug format (not to be confused with IBM's
2283 SOM, which is a cross-language ABI).
2285 The SOM reader is in @file{somread.c}.
2287 @section Debugging File Formats
2289 This section describes characteristics of debugging information that
2290 are independent of the object file format.
2294 @cindex stabs debugging info
2295 @code{stabs} started out as special symbols within the @code{a.out}
2296 format. Since then, it has been encapsulated into other file
2297 formats, such as COFF and ELF.
2299 While @file{dbxread.c} does some of the basic stab processing,
2300 including for encapsulated versions, @file{stabsread.c} does
2305 @cindex COFF debugging info
2306 The basic COFF definition includes debugging information. The level
2307 of support is minimal and non-extensible, and is not often used.
2309 @subsection Mips debug (Third Eye)
2311 @cindex ECOFF debugging info
2312 ECOFF includes a definition of a special debug format.
2314 The file @file{mdebugread.c} implements reading for this format.
2316 @c mention DWARF 1 as a formerly-supported format
2320 @cindex DWARF 2 debugging info
2321 DWARF 2 is an improved but incompatible version of DWARF 1.
2323 The DWARF 2 reader is in @file{dwarf2read.c}.
2325 @subsection Compressed DWARF 2
2327 @cindex Compressed DWARF 2 debugging info
2328 Compressed DWARF 2 is not technically a separate debugging format, but
2329 merely DWARF 2 debug information that has been compressed. In this
2330 format, every object-file section holding DWARF 2 debugging
2331 information is compressed and prepended with a header. (The section
2332 is also typically renamed, so a section called @code{.debug_info} in a
2333 DWARF 2 binary would be called @code{.zdebug_info} in a compressed
2334 DWARF 2 binary.) The header is 12 bytes long:
2338 4 bytes: the literal string ``ZLIB''
2340 8 bytes: the uncompressed size of the section, in big-endian byte
2344 The same reader is used for both compressed an normal DWARF 2 info.
2345 Section decompression is done in @code{zlib_decompress_section} in
2346 @file{dwarf2read.c}.
2350 @cindex DWARF 3 debugging info
2351 DWARF 3 is an improved version of DWARF 2.
2355 @cindex SOM debugging info
2356 Like COFF, the SOM definition includes debugging information.
2358 @section Adding a New Symbol Reader to @value{GDBN}
2360 @cindex adding debugging info reader
2361 If you are using an existing object file format (@code{a.out}, COFF, ELF, etc),
2362 there is probably little to be done.
2364 If you need to add a new object file format, you must first add it to
2365 BFD. This is beyond the scope of this document.
2367 You must then arrange for the BFD code to provide access to the
2368 debugging symbols. Generally @value{GDBN} will have to call swapping
2369 routines from BFD and a few other BFD internal routines to locate the
2370 debugging information. As much as possible, @value{GDBN} should not
2371 depend on the BFD internal data structures.
2373 For some targets (e.g., COFF), there is a special transfer vector used
2374 to call swapping routines, since the external data structures on various
2375 platforms have different sizes and layouts. Specialized routines that
2376 will only ever be implemented by one object file format may be called
2377 directly. This interface should be described in a file
2378 @file{bfd/lib@var{xyz}.h}, which is included by @value{GDBN}.
2380 @section Memory Management for Symbol Files
2382 Most memory associated with a loaded symbol file is stored on
2383 its @code{objfile_obstack}. This includes symbols, types,
2384 namespace data, and other information produced by the symbol readers.
2386 Because this data lives on the objfile's obstack, it is automatically
2387 released when the objfile is unloaded or reloaded. Therefore one
2388 objfile must not reference symbol or type data from another objfile;
2389 they could be unloaded at different times.
2391 User convenience variables, et cetera, have associated types. Normally
2392 these types live in the associated objfile. However, when the objfile
2393 is unloaded, those types are deep copied to global memory, so that
2394 the values of the user variables and history items are not lost.
2397 @node Language Support
2399 @chapter Language Support
2401 @cindex language support
2402 @value{GDBN}'s language support is mainly driven by the symbol reader,
2403 although it is possible for the user to set the source language
2406 @value{GDBN} chooses the source language by looking at the extension
2407 of the file recorded in the debug info; @file{.c} means C, @file{.f}
2408 means Fortran, etc. It may also use a special-purpose language
2409 identifier if the debug format supports it, like with DWARF.
2411 @section Adding a Source Language to @value{GDBN}
2413 @cindex adding source language
2414 To add other languages to @value{GDBN}'s expression parser, follow the
2418 @item Create the expression parser.
2420 @cindex expression parser
2421 This should reside in a file @file{@var{lang}-exp.y}. Routines for
2422 building parsed expressions into a @code{union exp_element} list are in
2425 @cindex language parser
2426 Since we can't depend upon everyone having Bison, and YACC produces
2427 parsers that define a bunch of global names, the following lines
2428 @strong{must} be included at the top of the YACC parser, to prevent the
2429 various parsers from defining the same global names:
2432 #define yyparse @var{lang}_parse
2433 #define yylex @var{lang}_lex
2434 #define yyerror @var{lang}_error
2435 #define yylval @var{lang}_lval
2436 #define yychar @var{lang}_char
2437 #define yydebug @var{lang}_debug
2438 #define yypact @var{lang}_pact
2439 #define yyr1 @var{lang}_r1
2440 #define yyr2 @var{lang}_r2
2441 #define yydef @var{lang}_def
2442 #define yychk @var{lang}_chk
2443 #define yypgo @var{lang}_pgo
2444 #define yyact @var{lang}_act
2445 #define yyexca @var{lang}_exca
2446 #define yyerrflag @var{lang}_errflag
2447 #define yynerrs @var{lang}_nerrs
2450 At the bottom of your parser, define a @code{struct language_defn} and
2451 initialize it with the right values for your language. Define an
2452 @code{initialize_@var{lang}} routine and have it call
2453 @samp{add_language(@var{lang}_language_defn)} to tell the rest of @value{GDBN}
2454 that your language exists. You'll need some other supporting variables
2455 and functions, which will be used via pointers from your
2456 @code{@var{lang}_language_defn}. See the declaration of @code{struct
2457 language_defn} in @file{language.h}, and the other @file{*-exp.y} files,
2458 for more information.
2460 @item Add any evaluation routines, if necessary
2462 @cindex expression evaluation routines
2463 @findex evaluate_subexp
2464 @findex prefixify_subexp
2465 @findex length_of_subexp
2466 If you need new opcodes (that represent the operations of the language),
2467 add them to the enumerated type in @file{expression.h}. Add support
2468 code for these operations in the @code{evaluate_subexp} function
2469 defined in the file @file{eval.c}. Add cases
2470 for new opcodes in two functions from @file{parse.c}:
2471 @code{prefixify_subexp} and @code{length_of_subexp}. These compute
2472 the number of @code{exp_element}s that a given operation takes up.
2474 @item Update some existing code
2476 Add an enumerated identifier for your language to the enumerated type
2477 @code{enum language} in @file{defs.h}.
2479 Update the routines in @file{language.c} so your language is included.
2480 These routines include type predicates and such, which (in some cases)
2481 are language dependent. If your language does not appear in the switch
2482 statement, an error is reported.
2484 @vindex current_language
2485 Also included in @file{language.c} is the code that updates the variable
2486 @code{current_language}, and the routines that translate the
2487 @code{language_@var{lang}} enumerated identifier into a printable
2490 @findex _initialize_language
2491 Update the function @code{_initialize_language} to include your
2492 language. This function picks the default language upon startup, so is
2493 dependent upon which languages that @value{GDBN} is built for.
2495 @findex allocate_symtab
2496 Update @code{allocate_symtab} in @file{symfile.c} and/or symbol-reading
2497 code so that the language of each symtab (source file) is set properly.
2498 This is used to determine the language to use at each stack frame level.
2499 Currently, the language is set based upon the extension of the source
2500 file. If the language can be better inferred from the symbol
2501 information, please set the language of the symtab in the symbol-reading
2504 @findex print_subexp
2505 @findex op_print_tab
2506 Add helper code to @code{print_subexp} (in @file{expprint.c}) to handle any new
2507 expression opcodes you have added to @file{expression.h}. Also, add the
2508 printed representations of your operators to @code{op_print_tab}.
2510 @item Add a place of call
2513 Add a call to @code{@var{lang}_parse()} and @code{@var{lang}_error} in
2514 @code{parse_exp_1} (defined in @file{parse.c}).
2516 @item Use macros to trim code
2518 @cindex trimming language-dependent code
2519 The user has the option of building @value{GDBN} for some or all of the
2520 languages. If the user decides to build @value{GDBN} for the language
2521 @var{lang}, then every file dependent on @file{language.h} will have the
2522 macro @code{_LANG_@var{lang}} defined in it. Use @code{#ifdef}s to
2523 leave out large routines that the user won't need if he or she is not
2524 using your language.
2526 Note that you do not need to do this in your YACC parser, since if @value{GDBN}
2527 is not build for @var{lang}, then @file{@var{lang}-exp.tab.o} (the
2528 compiled form of your parser) is not linked into @value{GDBN} at all.
2530 See the file @file{configure.in} for how @value{GDBN} is configured
2531 for different languages.
2533 @item Edit @file{Makefile.in}
2535 Add dependencies in @file{Makefile.in}. Make sure you update the macro
2536 variables such as @code{HFILES} and @code{OBJS}, otherwise your code may
2537 not get linked in, or, worse yet, it may not get @code{tar}red into the
2542 @node Host Definition
2544 @chapter Host Definition
2546 With the advent of Autoconf, it's rarely necessary to have host
2547 definition machinery anymore. The following information is provided,
2548 mainly, as an historical reference.
2550 @section Adding a New Host
2552 @cindex adding a new host
2553 @cindex host, adding
2554 @value{GDBN}'s host configuration support normally happens via Autoconf.
2555 New host-specific definitions should not be needed. Older hosts
2556 @value{GDBN} still use the host-specific definitions and files listed
2557 below, but these mostly exist for historical reasons, and will
2558 eventually disappear.
2561 @item gdb/config/@var{arch}/@var{xyz}.mh
2562 This file is a Makefile fragment that once contained both host and
2563 native configuration information (@pxref{Native Debugging}) for the
2564 machine @var{xyz}. The host configuration information is now handled
2567 Host configuration information included definitions for @code{CC},
2568 @code{SYSV_DEFINE}, @code{XM_CFLAGS}, @code{XM_ADD_FILES},
2569 @code{XM_CLIBS}, @code{XM_CDEPS}, etc.; see @file{Makefile.in}.
2571 New host-only configurations do not need this file.
2575 (Files named @file{gdb/config/@var{arch}/xm-@var{xyz}.h} were once
2576 used to define host-specific macros, but were no longer needed and
2577 have all been removed.)
2579 @subheading Generic Host Support Files
2581 @cindex generic host support
2582 There are some ``generic'' versions of routines that can be used by
2586 @cindex remote debugging support
2587 @cindex serial line support
2589 This contains serial line support for Unix systems. It is included by
2590 default on all Unix-like hosts.
2593 This contains serial pipe support for Unix systems. It is included by
2594 default on all Unix-like hosts.
2597 This contains serial line support for 32-bit programs running under
2598 Windows using MinGW.
2601 This contains serial line support for 32-bit programs running under DOS,
2602 using the DJGPP (a.k.a.@: GO32) execution environment.
2604 @cindex TCP remote support
2606 This contains generic TCP support using sockets. It is included by
2607 default on all Unix-like hosts and with MinGW.
2610 @section Host Conditionals
2612 When @value{GDBN} is configured and compiled, various macros are
2613 defined or left undefined, to control compilation based on the
2614 attributes of the host system. While formerly they could be set in
2615 host-specific header files, at present they can be changed only by
2616 setting @code{CFLAGS} when building, or by editing the source code.
2618 These macros and their meanings (or if the meaning is not documented
2619 here, then one of the source files where they are used is indicated)
2623 @item @value{GDBN}INIT_FILENAME
2624 The default name of @value{GDBN}'s initialization file (normally
2627 @item SIGWINCH_HANDLER
2628 If your host defines @code{SIGWINCH}, you can define this to be the name
2629 of a function to be called if @code{SIGWINCH} is received.
2631 @item SIGWINCH_HANDLER_BODY
2632 Define this to expand into code that will define the function named by
2633 the expansion of @code{SIGWINCH_HANDLER}.
2635 @item CRLF_SOURCE_FILES
2636 @cindex DOS text files
2637 Define this if host files use @code{\r\n} rather than @code{\n} as a
2638 line terminator. This will cause source file listings to omit @code{\r}
2639 characters when printing and it will allow @code{\r\n} line endings of files
2640 which are ``sourced'' by gdb. It must be possible to open files in binary
2641 mode using @code{O_BINARY} or, for fopen, @code{"rb"}.
2643 @item DEFAULT_PROMPT
2645 The default value of the prompt string (normally @code{"(gdb) "}).
2648 @cindex terminal device
2649 The name of the generic TTY device, defaults to @code{"/dev/tty"}.
2652 Substitute for isatty, if not available.
2655 Define this if binary files are opened the same way as text files.
2657 @item CC_HAS_LONG_LONG
2658 @cindex @code{long long} data type
2659 Define this if the host C compiler supports @code{long long}. This is set
2660 by the @code{configure} script.
2662 @item PRINTF_HAS_LONG_LONG
2663 Define this if the host can handle printing of long long integers via
2664 the printf format conversion specifier @code{ll}. This is set by the
2665 @code{configure} script.
2667 @item LSEEK_NOT_LINEAR
2668 Define this if @code{lseek (n)} does not necessarily move to byte number
2669 @code{n} in the file. This is only used when reading source files. It
2670 is normally faster to define @code{CRLF_SOURCE_FILES} when possible.
2673 If defined, this should be one or more tokens, such as @code{volatile},
2674 that can be used in both the declaration and definition of functions to
2675 indicate that they never return. The default is already set correctly
2676 if compiling with GCC. This will almost never need to be defined.
2679 If defined, this should be one or more tokens, such as
2680 @code{__attribute__ ((noreturn))}, that can be used in the declarations
2681 of functions to indicate that they never return. The default is already
2682 set correctly if compiling with GCC. This will almost never need to be
2686 Define this to help placate @code{lint} in some situations.
2689 Define this to override the defaults of @code{__volatile__} or
2694 @node Target Architecture Definition
2696 @chapter Target Architecture Definition
2698 @cindex target architecture definition
2699 @value{GDBN}'s target architecture defines what sort of
2700 machine-language programs @value{GDBN} can work with, and how it works
2703 The target architecture object is implemented as the C structure
2704 @code{struct gdbarch *}. The structure, and its methods, are generated
2705 using the Bourne shell script @file{gdbarch.sh}.
2708 * OS ABI Variant Handling::
2709 * Initialize New Architecture::
2710 * Registers and Memory::
2711 * Pointers and Addresses::
2713 * Raw and Virtual Registers::
2714 * Register and Memory Data::
2715 * Frame Interpretation::
2716 * Inferior Call Setup::
2717 * Compiler Characteristics::
2718 * Target Conditionals::
2719 * Adding a New Target::
2722 @node OS ABI Variant Handling
2723 @section Operating System ABI Variant Handling
2724 @cindex OS ABI variants
2726 @value{GDBN} provides a mechanism for handling variations in OS
2727 ABIs. An OS ABI variant may have influence over any number of
2728 variables in the target architecture definition. There are two major
2729 components in the OS ABI mechanism: sniffers and handlers.
2731 A @dfn{sniffer} examines a file matching a BFD architecture/flavour pair
2732 (the architecture may be wildcarded) in an attempt to determine the
2733 OS ABI of that file. Sniffers with a wildcarded architecture are considered
2734 to be @dfn{generic}, while sniffers for a specific architecture are
2735 considered to be @dfn{specific}. A match from a specific sniffer
2736 overrides a match from a generic sniffer. Multiple sniffers for an
2737 architecture/flavour may exist, in order to differentiate between two
2738 different operating systems which use the same basic file format. The
2739 OS ABI framework provides a generic sniffer for ELF-format files which
2740 examines the @code{EI_OSABI} field of the ELF header, as well as note
2741 sections known to be used by several operating systems.
2743 @cindex fine-tuning @code{gdbarch} structure
2744 A @dfn{handler} is used to fine-tune the @code{gdbarch} structure for the
2745 selected OS ABI. There may be only one handler for a given OS ABI
2746 for each BFD architecture.
2748 The following OS ABI variants are defined in @file{defs.h}:
2752 @findex GDB_OSABI_UNINITIALIZED
2753 @item GDB_OSABI_UNINITIALIZED
2754 Used for struct gdbarch_info if ABI is still uninitialized.
2756 @findex GDB_OSABI_UNKNOWN
2757 @item GDB_OSABI_UNKNOWN
2758 The ABI of the inferior is unknown. The default @code{gdbarch}
2759 settings for the architecture will be used.
2761 @findex GDB_OSABI_SVR4
2762 @item GDB_OSABI_SVR4
2763 UNIX System V Release 4.
2765 @findex GDB_OSABI_HURD
2766 @item GDB_OSABI_HURD
2767 GNU using the Hurd kernel.
2769 @findex GDB_OSABI_SOLARIS
2770 @item GDB_OSABI_SOLARIS
2773 @findex GDB_OSABI_OSF1
2774 @item GDB_OSABI_OSF1
2775 OSF/1, including Digital UNIX and Compaq Tru64 UNIX.
2777 @findex GDB_OSABI_LINUX
2778 @item GDB_OSABI_LINUX
2779 GNU using the Linux kernel.
2781 @findex GDB_OSABI_FREEBSD_AOUT
2782 @item GDB_OSABI_FREEBSD_AOUT
2783 FreeBSD using the @code{a.out} executable format.
2785 @findex GDB_OSABI_FREEBSD_ELF
2786 @item GDB_OSABI_FREEBSD_ELF
2787 FreeBSD using the ELF executable format.
2789 @findex GDB_OSABI_NETBSD_AOUT
2790 @item GDB_OSABI_NETBSD_AOUT
2791 NetBSD using the @code{a.out} executable format.
2793 @findex GDB_OSABI_NETBSD_ELF
2794 @item GDB_OSABI_NETBSD_ELF
2795 NetBSD using the ELF executable format.
2797 @findex GDB_OSABI_OPENBSD_ELF
2798 @item GDB_OSABI_OPENBSD_ELF
2799 OpenBSD using the ELF executable format.
2801 @findex GDB_OSABI_WINCE
2802 @item GDB_OSABI_WINCE
2805 @findex GDB_OSABI_GO32
2806 @item GDB_OSABI_GO32
2809 @findex GDB_OSABI_IRIX
2810 @item GDB_OSABI_IRIX
2813 @findex GDB_OSABI_INTERIX
2814 @item GDB_OSABI_INTERIX
2815 Interix (Posix layer for MS-Windows systems).
2817 @findex GDB_OSABI_HPUX_ELF
2818 @item GDB_OSABI_HPUX_ELF
2819 HP/UX using the ELF executable format.
2821 @findex GDB_OSABI_HPUX_SOM
2822 @item GDB_OSABI_HPUX_SOM
2823 HP/UX using the SOM executable format.
2825 @findex GDB_OSABI_QNXNTO
2826 @item GDB_OSABI_QNXNTO
2829 @findex GDB_OSABI_CYGWIN
2830 @item GDB_OSABI_CYGWIN
2833 @findex GDB_OSABI_AIX
2839 Here are the functions that make up the OS ABI framework:
2841 @deftypefun const char *gdbarch_osabi_name (enum gdb_osabi @var{osabi})
2842 Return the name of the OS ABI corresponding to @var{osabi}.
2845 @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}))
2846 Register the OS ABI handler specified by @var{init_osabi} for the
2847 architecture, machine type and OS ABI specified by @var{arch},
2848 @var{machine} and @var{osabi}. In most cases, a value of zero for the
2849 machine type, which implies the architecture's default machine type,
2853 @deftypefun void gdbarch_register_osabi_sniffer (enum bfd_architecture @var{arch}, enum bfd_flavour @var{flavour}, enum gdb_osabi (*@var{sniffer})(bfd *@var{abfd}))
2854 Register the OS ABI file sniffer specified by @var{sniffer} for the
2855 BFD architecture/flavour pair specified by @var{arch} and @var{flavour}.
2856 If @var{arch} is @code{bfd_arch_unknown}, the sniffer is considered to
2857 be generic, and is allowed to examine @var{flavour}-flavoured files for
2861 @deftypefun enum gdb_osabi gdbarch_lookup_osabi (bfd *@var{abfd})
2862 Examine the file described by @var{abfd} to determine its OS ABI.
2863 The value @code{GDB_OSABI_UNKNOWN} is returned if the OS ABI cannot
2867 @deftypefun void gdbarch_init_osabi (struct gdbarch info @var{info}, struct gdbarch *@var{gdbarch}, enum gdb_osabi @var{osabi})
2868 Invoke the OS ABI handler corresponding to @var{osabi} to fine-tune the
2869 @code{gdbarch} structure specified by @var{gdbarch}. If a handler
2870 corresponding to @var{osabi} has not been registered for @var{gdbarch}'s
2871 architecture, a warning will be issued and the debugging session will continue
2872 with the defaults already established for @var{gdbarch}.
2875 @deftypefun void generic_elf_osabi_sniff_abi_tag_sections (bfd *@var{abfd}, asection *@var{sect}, void *@var{obj})
2876 Helper routine for ELF file sniffers. Examine the file described by
2877 @var{abfd} and look at ABI tag note sections to determine the OS ABI
2878 from the note. This function should be called via
2879 @code{bfd_map_over_sections}.
2882 @node Initialize New Architecture
2883 @section Initializing a New Architecture
2885 Each @code{gdbarch} is associated with a single @sc{bfd} architecture,
2886 via a @code{bfd_arch_@var{arch}} constant. The @code{gdbarch} is
2887 registered by a call to @code{register_gdbarch_init}, usually from
2888 the file's @code{_initialize_@var{filename}} routine, which will
2889 be automatically called during @value{GDBN} startup. The arguments
2890 are a @sc{bfd} architecture constant and an initialization function.
2892 The initialization function has this type:
2895 static struct gdbarch *
2896 @var{arch}_gdbarch_init (struct gdbarch_info @var{info},
2897 struct gdbarch_list *@var{arches})
2900 The @var{info} argument contains parameters used to select the correct
2901 architecture, and @var{arches} is a list of architectures which
2902 have already been created with the same @code{bfd_arch_@var{arch}}
2905 The initialization function should first make sure that @var{info}
2906 is acceptable, and return @code{NULL} if it is not. Then, it should
2907 search through @var{arches} for an exact match to @var{info}, and
2908 return one if found. Lastly, if no exact match was found, it should
2909 create a new architecture based on @var{info} and return it.
2911 Only information in @var{info} should be used to choose the new
2912 architecture. Historically, @var{info} could be sparse, and
2913 defaults would be collected from the first element on @var{arches}.
2914 However, @value{GDBN} now fills in @var{info} more thoroughly,
2915 so new @code{gdbarch} initialization functions should not take
2916 defaults from @var{arches}.
2918 @node Registers and Memory
2919 @section Registers and Memory
2921 @value{GDBN}'s model of the target machine is rather simple.
2922 @value{GDBN} assumes the machine includes a bank of registers and a
2923 block of memory. Each register may have a different size.
2925 @value{GDBN} does not have a magical way to match up with the
2926 compiler's idea of which registers are which; however, it is critical
2927 that they do match up accurately. The only way to make this work is
2928 to get accurate information about the order that the compiler uses,
2929 and to reflect that in the @code{gdbarch_register_name} and related functions.
2931 @value{GDBN} can handle big-endian, little-endian, and bi-endian architectures.
2933 @node Pointers and Addresses
2934 @section Pointers Are Not Always Addresses
2935 @cindex pointer representation
2936 @cindex address representation
2937 @cindex word-addressed machines
2938 @cindex separate data and code address spaces
2939 @cindex spaces, separate data and code address
2940 @cindex address spaces, separate data and code
2941 @cindex code pointers, word-addressed
2942 @cindex converting between pointers and addresses
2943 @cindex D10V addresses
2945 On almost all 32-bit architectures, the representation of a pointer is
2946 indistinguishable from the representation of some fixed-length number
2947 whose value is the byte address of the object pointed to. On such
2948 machines, the words ``pointer'' and ``address'' can be used interchangeably.
2949 However, architectures with smaller word sizes are often cramped for
2950 address space, so they may choose a pointer representation that breaks this
2951 identity, and allows a larger code address space.
2953 @c D10V is gone from sources - more current example?
2955 For example, the Renesas D10V is a 16-bit VLIW processor whose
2956 instructions are 32 bits long@footnote{Some D10V instructions are
2957 actually pairs of 16-bit sub-instructions. However, since you can't
2958 jump into the middle of such a pair, code addresses can only refer to
2959 full 32 bit instructions, which is what matters in this explanation.}.
2960 If the D10V used ordinary byte addresses to refer to code locations,
2961 then the processor would only be able to address 64kb of instructions.
2962 However, since instructions must be aligned on four-byte boundaries, the
2963 low two bits of any valid instruction's byte address are always
2964 zero---byte addresses waste two bits. So instead of byte addresses,
2965 the D10V uses word addresses---byte addresses shifted right two bits---to
2966 refer to code. Thus, the D10V can use 16-bit words to address 256kb of
2969 However, this means that code pointers and data pointers have different
2970 forms on the D10V. The 16-bit word @code{0xC020} refers to byte address
2971 @code{0xC020} when used as a data address, but refers to byte address
2972 @code{0x30080} when used as a code address.
2974 (The D10V also uses separate code and data address spaces, which also
2975 affects the correspondence between pointers and addresses, but we're
2976 going to ignore that here; this example is already too long.)
2978 To cope with architectures like this---the D10V is not the only
2979 one!---@value{GDBN} tries to distinguish between @dfn{addresses}, which are
2980 byte numbers, and @dfn{pointers}, which are the target's representation
2981 of an address of a particular type of data. In the example above,
2982 @code{0xC020} is the pointer, which refers to one of the addresses
2983 @code{0xC020} or @code{0x30080}, depending on the type imposed upon it.
2984 @value{GDBN} provides functions for turning a pointer into an address
2985 and vice versa, in the appropriate way for the current architecture.
2987 Unfortunately, since addresses and pointers are identical on almost all
2988 processors, this distinction tends to bit-rot pretty quickly. Thus,
2989 each time you port @value{GDBN} to an architecture which does
2990 distinguish between pointers and addresses, you'll probably need to
2991 clean up some architecture-independent code.
2993 Here are functions which convert between pointers and addresses:
2995 @deftypefun CORE_ADDR extract_typed_address (void *@var{buf}, struct type *@var{type})
2996 Treat the bytes at @var{buf} as a pointer or reference of type
2997 @var{type}, and return the address it represents, in a manner
2998 appropriate for the current architecture. This yields an address
2999 @value{GDBN} can use to read target memory, disassemble, etc. Note that
3000 @var{buf} refers to a buffer in @value{GDBN}'s memory, not the
3003 For example, if the current architecture is the Intel x86, this function
3004 extracts a little-endian integer of the appropriate length from
3005 @var{buf} and returns it. However, if the current architecture is the
3006 D10V, this function will return a 16-bit integer extracted from
3007 @var{buf}, multiplied by four if @var{type} is a pointer to a function.
3009 If @var{type} is not a pointer or reference type, then this function
3010 will signal an internal error.
3013 @deftypefun CORE_ADDR store_typed_address (void *@var{buf}, struct type *@var{type}, CORE_ADDR @var{addr})
3014 Store the address @var{addr} in @var{buf}, in the proper format for a
3015 pointer of type @var{type} in the current architecture. Note that
3016 @var{buf} refers to a buffer in @value{GDBN}'s memory, not the
3019 For example, if the current architecture is the Intel x86, this function
3020 stores @var{addr} unmodified as a little-endian integer of the
3021 appropriate length in @var{buf}. However, if the current architecture
3022 is the D10V, this function divides @var{addr} by four if @var{type} is
3023 a pointer to a function, and then stores it in @var{buf}.
3025 If @var{type} is not a pointer or reference type, then this function
3026 will signal an internal error.
3029 @deftypefun CORE_ADDR value_as_address (struct value *@var{val})
3030 Assuming that @var{val} is a pointer, return the address it represents,
3031 as appropriate for the current architecture.
3033 This function actually works on integral values, as well as pointers.
3034 For pointers, it performs architecture-specific conversions as
3035 described above for @code{extract_typed_address}.
3038 @deftypefun CORE_ADDR value_from_pointer (struct type *@var{type}, CORE_ADDR @var{addr})
3039 Create and return a value representing a pointer of type @var{type} to
3040 the address @var{addr}, as appropriate for the current architecture.
3041 This function performs architecture-specific conversions as described
3042 above for @code{store_typed_address}.
3045 Here are two functions which architectures can define to indicate the
3046 relationship between pointers and addresses. These have default
3047 definitions, appropriate for architectures on which all pointers are
3048 simple unsigned byte addresses.
3050 @deftypefun CORE_ADDR gdbarch_pointer_to_address (struct gdbarch *@var{current_gdbarch}, struct type *@var{type}, char *@var{buf})
3051 Assume that @var{buf} holds a pointer of type @var{type}, in the
3052 appropriate format for the current architecture. Return the byte
3053 address the pointer refers to.
3055 This function may safely assume that @var{type} is either a pointer or a
3056 C@t{++} reference type.
3059 @deftypefun void gdbarch_address_to_pointer (struct gdbarch *@var{current_gdbarch}, struct type *@var{type}, char *@var{buf}, CORE_ADDR @var{addr})
3060 Store in @var{buf} a pointer of type @var{type} representing the address
3061 @var{addr}, in the appropriate format for the current architecture.
3063 This function may safely assume that @var{type} is either a pointer or a
3064 C@t{++} reference type.
3067 @node Address Classes
3068 @section Address Classes
3069 @cindex address classes
3070 @cindex DW_AT_byte_size
3071 @cindex DW_AT_address_class
3073 Sometimes information about different kinds of addresses is available
3074 via the debug information. For example, some programming environments
3075 define addresses of several different sizes. If the debug information
3076 distinguishes these kinds of address classes through either the size
3077 info (e.g, @code{DW_AT_byte_size} in @w{DWARF 2}) or through an explicit
3078 address class attribute (e.g, @code{DW_AT_address_class} in @w{DWARF 2}), the
3079 following macros should be defined in order to disambiguate these
3080 types within @value{GDBN} as well as provide the added information to
3081 a @value{GDBN} user when printing type expressions.
3083 @deftypefun int gdbarch_address_class_type_flags (struct gdbarch *@var{current_gdbarch}, int @var{byte_size}, int @var{dwarf2_addr_class})
3084 Returns the type flags needed to construct a pointer type whose size
3085 is @var{byte_size} and whose address class is @var{dwarf2_addr_class}.
3086 This function is normally called from within a symbol reader. See
3087 @file{dwarf2read.c}.
3090 @deftypefun char *gdbarch_address_class_type_flags_to_name (struct gdbarch *@var{current_gdbarch}, int @var{type_flags})
3091 Given the type flags representing an address class qualifier, return
3094 @deftypefun int gdbarch_address_class_name_to_type_flags (struct gdbarch *@var{current_gdbarch}, int @var{name}, int *@var{type_flags_ptr})
3095 Given an address qualifier name, set the @code{int} referenced by @var{type_flags_ptr} to the type flags
3096 for that address class qualifier.
3099 Since the need for address classes is rather rare, none of
3100 the address class functions are defined by default. Predicate
3101 functions are provided to detect when they are defined.
3103 Consider a hypothetical architecture in which addresses are normally
3104 32-bits wide, but 16-bit addresses are also supported. Furthermore,
3105 suppose that the @w{DWARF 2} information for this architecture simply
3106 uses a @code{DW_AT_byte_size} value of 2 to indicate the use of one
3107 of these "short" pointers. The following functions could be defined
3108 to implement the address class functions:
3111 somearch_address_class_type_flags (int byte_size,
3112 int dwarf2_addr_class)
3115 return TYPE_FLAG_ADDRESS_CLASS_1;
3121 somearch_address_class_type_flags_to_name (int type_flags)
3123 if (type_flags & TYPE_FLAG_ADDRESS_CLASS_1)
3130 somearch_address_class_name_to_type_flags (char *name,
3131 int *type_flags_ptr)
3133 if (strcmp (name, "short") == 0)
3135 *type_flags_ptr = TYPE_FLAG_ADDRESS_CLASS_1;
3143 The qualifier @code{@@short} is used in @value{GDBN}'s type expressions
3144 to indicate the presence of one of these "short" pointers. E.g, if
3145 the debug information indicates that @code{short_ptr_var} is one of these
3146 short pointers, @value{GDBN} might show the following behavior:
3149 (gdb) ptype short_ptr_var
3150 type = int * @@short
3154 @node Raw and Virtual Registers
3155 @section Raw and Virtual Register Representations
3156 @cindex raw register representation
3157 @cindex virtual register representation
3158 @cindex representations, raw and virtual registers
3160 @emph{Maintainer note: This section is pretty much obsolete. The
3161 functionality described here has largely been replaced by
3162 pseudo-registers and the mechanisms described in @ref{Register and
3163 Memory Data, , Using Different Register and Memory Data
3164 Representations}. See also @uref{http://www.gnu.org/software/gdb/bugs/,
3165 Bug Tracking Database} and
3166 @uref{http://sources.redhat.com/gdb/current/ari/, ARI Index} for more
3167 up-to-date information.}
3169 Some architectures use one representation for a value when it lives in a
3170 register, but use a different representation when it lives in memory.
3171 In @value{GDBN}'s terminology, the @dfn{raw} representation is the one used in
3172 the target registers, and the @dfn{virtual} representation is the one
3173 used in memory, and within @value{GDBN} @code{struct value} objects.
3175 @emph{Maintainer note: Notice that the same mechanism is being used to
3176 both convert a register to a @code{struct value} and alternative
3179 For almost all data types on almost all architectures, the virtual and
3180 raw representations are identical, and no special handling is needed.
3181 However, they do occasionally differ. For example:
3185 The x86 architecture supports an 80-bit @code{long double} type. However, when
3186 we store those values in memory, they occupy twelve bytes: the
3187 floating-point number occupies the first ten, and the final two bytes
3188 are unused. This keeps the values aligned on four-byte boundaries,
3189 allowing more efficient access. Thus, the x86 80-bit floating-point
3190 type is the raw representation, and the twelve-byte loosely-packed
3191 arrangement is the virtual representation.
3194 Some 64-bit MIPS targets present 32-bit registers to @value{GDBN} as 64-bit
3195 registers, with garbage in their upper bits. @value{GDBN} ignores the top 32
3196 bits. Thus, the 64-bit form, with garbage in the upper 32 bits, is the
3197 raw representation, and the trimmed 32-bit representation is the
3198 virtual representation.
3201 In general, the raw representation is determined by the architecture, or
3202 @value{GDBN}'s interface to the architecture, while the virtual representation
3203 can be chosen for @value{GDBN}'s convenience. @value{GDBN}'s register file,
3204 @code{registers}, holds the register contents in raw format, and the
3205 @value{GDBN} remote protocol transmits register values in raw format.
3207 Your architecture may define the following macros to request
3208 conversions between the raw and virtual format:
3210 @deftypefn {Target Macro} int REGISTER_CONVERTIBLE (int @var{reg})
3211 Return non-zero if register number @var{reg}'s value needs different raw
3212 and virtual formats.
3214 You should not use @code{REGISTER_CONVERT_TO_VIRTUAL} for a register
3215 unless this macro returns a non-zero value for that register.
3218 @deftypefn {Target Macro} void REGISTER_CONVERT_TO_VIRTUAL (int @var{reg}, struct type *@var{type}, char *@var{from}, char *@var{to})
3219 Convert the value of register number @var{reg} to @var{type}, which
3220 should always be @code{gdbarch_register_type (@var{reg})}. The buffer
3221 at @var{from} holds the register's value in raw format; the macro should
3222 convert the value to virtual format, and place it at @var{to}.
3224 Note that @code{REGISTER_CONVERT_TO_VIRTUAL} and
3225 @code{REGISTER_CONVERT_TO_RAW} take their @var{reg} and @var{type}
3226 arguments in different orders.
3228 You should only use @code{REGISTER_CONVERT_TO_VIRTUAL} with registers
3229 for which the @code{REGISTER_CONVERTIBLE} macro returns a non-zero
3233 @deftypefn {Target Macro} void REGISTER_CONVERT_TO_RAW (struct type *@var{type}, int @var{reg}, char *@var{from}, char *@var{to})
3234 Convert the value of register number @var{reg} to @var{type}, which
3235 should always be @code{gdbarch_register_type (@var{reg})}. The buffer
3236 at @var{from} holds the register's value in raw format; the macro should
3237 convert the value to virtual format, and place it at @var{to}.
3239 Note that REGISTER_CONVERT_TO_VIRTUAL and REGISTER_CONVERT_TO_RAW take
3240 their @var{reg} and @var{type} arguments in different orders.
3244 @node Register and Memory Data
3245 @section Using Different Register and Memory Data Representations
3246 @cindex register representation
3247 @cindex memory representation
3248 @cindex representations, register and memory
3249 @cindex register data formats, converting
3250 @cindex @code{struct value}, converting register contents to
3252 @emph{Maintainer's note: The way GDB manipulates registers is undergoing
3253 significant change. Many of the macros and functions referred to in this
3254 section are likely to be subject to further revision. See
3255 @uref{http://sources.redhat.com/gdb/current/ari/, A.R. Index} and
3256 @uref{http://www.gnu.org/software/gdb/bugs, Bug Tracking Database} for
3257 further information. cagney/2002-05-06.}
3259 Some architectures can represent a data object in a register using a
3260 form that is different to the objects more normal memory representation.
3266 The Alpha architecture can represent 32 bit integer values in
3267 floating-point registers.
3270 The x86 architecture supports 80-bit floating-point registers. The
3271 @code{long double} data type occupies 96 bits in memory but only 80 bits
3272 when stored in a register.
3276 In general, the register representation of a data type is determined by
3277 the architecture, or @value{GDBN}'s interface to the architecture, while
3278 the memory representation is determined by the Application Binary
3281 For almost all data types on almost all architectures, the two
3282 representations are identical, and no special handling is needed.
3283 However, they do occasionally differ. Your architecture may define the
3284 following macros to request conversions between the register and memory
3285 representations of a data type:
3287 @deftypefun int gdbarch_convert_register_p (struct gdbarch *@var{gdbarch}, int @var{reg})
3288 Return non-zero if the representation of a data value stored in this
3289 register may be different to the representation of that same data value
3290 when stored in memory.
3292 When non-zero, the macros @code{gdbarch_register_to_value} and
3293 @code{value_to_register} are used to perform any necessary conversion.
3295 This function should return zero for the register's native type, when
3296 no conversion is necessary.
3299 @deftypefun void gdbarch_register_to_value (struct gdbarch *@var{gdbarch}, int @var{reg}, struct type *@var{type}, char *@var{from}, char *@var{to})
3300 Convert the value of register number @var{reg} to a data object of type
3301 @var{type}. The buffer at @var{from} holds the register's value in raw
3302 format; the converted value should be placed in the buffer at @var{to}.
3304 Note that @code{gdbarch_register_to_value} and @code{gdbarch_value_to_register}
3305 take their @var{reg} and @var{type} arguments in different orders.
3307 You should only use @code{gdbarch_register_to_value} with registers for which
3308 the @code{gdbarch_convert_register_p} function returns a non-zero value.
3311 @deftypefun void gdbarch_value_to_register (struct gdbarch *@var{gdbarch}, struct type *@var{type}, int @var{reg}, char *@var{from}, char *@var{to})
3312 Convert a data value of type @var{type} to register number @var{reg}'
3315 Note that @code{gdbarch_register_to_value} and @code{gdbarch_value_to_register}
3316 take their @var{reg} and @var{type} arguments in different orders.
3318 You should only use @code{gdbarch_value_to_register} with registers for which
3319 the @code{gdbarch_convert_register_p} function returns a non-zero value.
3322 @node Frame Interpretation
3323 @section Frame Interpretation
3325 @node Inferior Call Setup
3326 @section Inferior Call Setup
3328 @node Compiler Characteristics
3329 @section Compiler Characteristics
3331 @node Target Conditionals
3332 @section Target Conditionals
3334 This section describes the macros and functions that you can use to define the
3339 @item CORE_ADDR gdbarch_addr_bits_remove (@var{gdbarch}, @var{addr})
3340 @findex gdbarch_addr_bits_remove
3341 If a raw machine instruction address includes any bits that are not
3342 really part of the address, then this function is used to zero those bits in
3343 @var{addr}. This is only used for addresses of instructions, and even then not
3346 For example, the two low-order bits of the PC on the Hewlett-Packard PA
3347 2.0 architecture contain the privilege level of the corresponding
3348 instruction. Since instructions must always be aligned on four-byte
3349 boundaries, the processor masks out these bits to generate the actual
3350 address of the instruction. @code{gdbarch_addr_bits_remove} would then for
3351 example look like that:
3353 arch_addr_bits_remove (CORE_ADDR addr)
3355 return (addr &= ~0x3);
3359 @item int address_class_name_to_type_flags (@var{gdbarch}, @var{name}, @var{type_flags_ptr})
3360 @findex address_class_name_to_type_flags
3361 If @var{name} is a valid address class qualifier name, set the @code{int}
3362 referenced by @var{type_flags_ptr} to the mask representing the qualifier
3363 and return 1. If @var{name} is not a valid address class qualifier name,
3366 The value for @var{type_flags_ptr} should be one of
3367 @code{TYPE_FLAG_ADDRESS_CLASS_1}, @code{TYPE_FLAG_ADDRESS_CLASS_2}, or
3368 possibly some combination of these values or'd together.
3369 @xref{Target Architecture Definition, , Address Classes}.
3371 @item int address_class_name_to_type_flags_p (@var{gdbarch})
3372 @findex address_class_name_to_type_flags_p
3373 Predicate which indicates whether @code{address_class_name_to_type_flags}
3376 @item int gdbarch_address_class_type_flags (@var{gdbarch}, @var{byte_size}, @var{dwarf2_addr_class})
3377 @findex gdbarch_address_class_type_flags
3378 Given a pointers byte size (as described by the debug information) and
3379 the possible @code{DW_AT_address_class} value, return the type flags
3380 used by @value{GDBN} to represent this address class. The value
3381 returned should be one of @code{TYPE_FLAG_ADDRESS_CLASS_1},
3382 @code{TYPE_FLAG_ADDRESS_CLASS_2}, or possibly some combination of these
3383 values or'd together.
3384 @xref{Target Architecture Definition, , Address Classes}.
3386 @item int gdbarch_address_class_type_flags_p (@var{gdbarch})
3387 @findex gdbarch_address_class_type_flags_p
3388 Predicate which indicates whether @code{gdbarch_address_class_type_flags_p} has
3391 @item const char *gdbarch_address_class_type_flags_to_name (@var{gdbarch}, @var{type_flags})
3392 @findex gdbarch_address_class_type_flags_to_name
3393 Return the name of the address class qualifier associated with the type
3394 flags given by @var{type_flags}.
3396 @item int gdbarch_address_class_type_flags_to_name_p (@var{gdbarch})
3397 @findex gdbarch_address_class_type_flags_to_name_p
3398 Predicate which indicates whether @code{gdbarch_address_class_type_flags_to_name} has been defined.
3399 @xref{Target Architecture Definition, , Address Classes}.
3401 @item void gdbarch_address_to_pointer (@var{gdbarch}, @var{type}, @var{buf}, @var{addr})
3402 @findex gdbarch_address_to_pointer
3403 Store in @var{buf} a pointer of type @var{type} representing the address
3404 @var{addr}, in the appropriate format for the current architecture.
3405 This function may safely assume that @var{type} is either a pointer or a
3406 C@t{++} reference type.
3407 @xref{Target Architecture Definition, , Pointers Are Not Always Addresses}.
3409 @item int gdbarch_believe_pcc_promotion (@var{gdbarch})
3410 @findex gdbarch_believe_pcc_promotion
3411 Used to notify if the compiler promotes a @code{short} or @code{char}
3412 parameter to an @code{int}, but still reports the parameter as its
3413 original type, rather than the promoted type.
3415 @item gdbarch_bits_big_endian (@var{gdbarch})
3416 @findex gdbarch_bits_big_endian
3417 This is used if the numbering of bits in the targets does @strong{not} match
3418 the endianness of the target byte order. A value of 1 means that the bits
3419 are numbered in a big-endian bit order, 0 means little-endian.
3421 @item set_gdbarch_bits_big_endian (@var{gdbarch}, @var{bits_big_endian})
3422 @findex set_gdbarch_bits_big_endian
3423 Calling set_gdbarch_bits_big_endian with a value of 1 indicates that the
3424 bits in the target are numbered in a big-endian bit order, 0 indicates
3429 This is the character array initializer for the bit pattern to put into
3430 memory where a breakpoint is set. Although it's common to use a trap
3431 instruction for a breakpoint, it's not required; for instance, the bit
3432 pattern could be an invalid instruction. The breakpoint must be no
3433 longer than the shortest instruction of the architecture.
3435 @code{BREAKPOINT} has been deprecated in favor of
3436 @code{gdbarch_breakpoint_from_pc}.
3438 @item BIG_BREAKPOINT
3439 @itemx LITTLE_BREAKPOINT
3440 @findex LITTLE_BREAKPOINT
3441 @findex BIG_BREAKPOINT
3442 Similar to BREAKPOINT, but used for bi-endian targets.
3444 @code{BIG_BREAKPOINT} and @code{LITTLE_BREAKPOINT} have been deprecated in
3445 favor of @code{gdbarch_breakpoint_from_pc}.
3447 @item const gdb_byte *gdbarch_breakpoint_from_pc (@var{gdbarch}, @var{pcptr}, @var{lenptr})
3448 @findex gdbarch_breakpoint_from_pc
3449 @anchor{gdbarch_breakpoint_from_pc} Use the program counter to determine the
3450 contents and size of a breakpoint instruction. It returns a pointer to
3451 a string of bytes that encode a breakpoint instruction, stores the
3452 length of the string to @code{*@var{lenptr}}, and adjusts the program
3453 counter (if necessary) to point to the actual memory location where the
3454 breakpoint should be inserted.
3456 Although it is common to use a trap instruction for a breakpoint, it's
3457 not required; for instance, the bit pattern could be an invalid
3458 instruction. The breakpoint must be no longer than the shortest
3459 instruction of the architecture.
3461 Replaces all the other @var{BREAKPOINT} macros.
3463 @item int gdbarch_memory_insert_breakpoint (@var{gdbarch}, @var{bp_tgt})
3464 @itemx gdbarch_memory_remove_breakpoint (@var{gdbarch}, @var{bp_tgt})
3465 @findex gdbarch_memory_remove_breakpoint
3466 @findex gdbarch_memory_insert_breakpoint
3467 Insert or remove memory based breakpoints. Reasonable defaults
3468 (@code{default_memory_insert_breakpoint} and
3469 @code{default_memory_remove_breakpoint} respectively) have been
3470 provided so that it is not necessary to set these for most
3471 architectures. Architectures which may want to set
3472 @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
3473 conventional manner.
3475 It may also be desirable (from an efficiency standpoint) to define
3476 custom breakpoint insertion and removal routines if
3477 @code{gdbarch_breakpoint_from_pc} needs to read the target's memory for some
3480 @item CORE_ADDR gdbarch_adjust_breakpoint_address (@var{gdbarch}, @var{bpaddr})
3481 @findex gdbarch_adjust_breakpoint_address
3482 @cindex breakpoint address adjusted
3483 Given an address at which a breakpoint is desired, return a breakpoint
3484 address adjusted to account for architectural constraints on
3485 breakpoint placement. This method is not needed by most targets.
3487 The FR-V target (see @file{frv-tdep.c}) requires this method.
3488 The FR-V is a VLIW architecture in which a number of RISC-like
3489 instructions are grouped (packed) together into an aggregate
3490 instruction or instruction bundle. When the processor executes
3491 one of these bundles, the component instructions are executed
3494 In the course of optimization, the compiler may group instructions
3495 from distinct source statements into the same bundle. The line number
3496 information associated with one of the latter statements will likely
3497 refer to some instruction other than the first one in the bundle. So,
3498 if the user attempts to place a breakpoint on one of these latter
3499 statements, @value{GDBN} must be careful to @emph{not} place the break
3500 instruction on any instruction other than the first one in the bundle.
3501 (Remember though that the instructions within a bundle execute
3502 in parallel, so the @emph{first} instruction is the instruction
3503 at the lowest address and has nothing to do with execution order.)
3505 The FR-V's @code{gdbarch_adjust_breakpoint_address} method will adjust a
3506 breakpoint's address by scanning backwards for the beginning of
3507 the bundle, returning the address of the bundle.
3509 Since the adjustment of a breakpoint may significantly alter a user's
3510 expectation, @value{GDBN} prints a warning when an adjusted breakpoint
3511 is initially set and each time that that breakpoint is hit.
3513 @item int gdbarch_call_dummy_location (@var{gdbarch})
3514 @findex gdbarch_call_dummy_location
3515 See the file @file{inferior.h}.
3517 This method has been replaced by @code{gdbarch_push_dummy_code}
3518 (@pxref{gdbarch_push_dummy_code}).
3520 @item int gdbarch_cannot_fetch_register (@var{gdbarch}, @var{regum})
3521 @findex gdbarch_cannot_fetch_register
3522 This function should return nonzero if @var{regno} cannot be fetched
3523 from an inferior process. This is only relevant if
3524 @code{FETCH_INFERIOR_REGISTERS} is not defined.
3526 @item int gdbarch_cannot_store_register (@var{gdbarch}, @var{regnum})
3527 @findex gdbarch_cannot_store_register
3528 This function should return nonzero if @var{regno} should not be
3529 written to the target. This is often the case for program counters,
3530 status words, and other special registers. This function returns 0 as
3531 default so that @value{GDBN} will assume that all registers may be written.
3533 @item int gdbarch_convert_register_p (@var{gdbarch}, @var{regnum}, struct type *@var{type})
3534 @findex gdbarch_convert_register_p
3535 Return non-zero if register @var{regnum} represents data values of type
3536 @var{type} in a non-standard form.
3537 @xref{Target Architecture Definition, , Using Different Register and Memory Data Representations}.
3539 @item CORE_ADDR gdbarch_decr_pc_after_break (@var{gdbarch})
3540 @findex gdbarch_decr_pc_after_break
3541 This function shall return the amount by which to decrement the PC after the
3542 program encounters a breakpoint. This is often the number of bytes in
3543 @code{BREAKPOINT}, though not always. For most targets this value will be 0.
3545 @item DISABLE_UNSETTABLE_BREAK (@var{addr})
3546 @findex DISABLE_UNSETTABLE_BREAK
3547 If defined, this should evaluate to 1 if @var{addr} is in a shared
3548 library in which breakpoints cannot be set and so should be disabled.
3550 @item void gdbarch_print_float_info (@var{gdbarch}, @var{file}, @var{frame}, @var{args})
3551 @findex gdbarch_print_float_info
3552 If defined, then the @samp{info float} command will print information about
3553 the processor's floating point unit.
3555 @item void gdbarch_print_registers_info (@var{gdbarch}, @var{frame}, @var{regnum}, @var{all})
3556 @findex gdbarch_print_registers_info
3557 If defined, pretty print the value of the register @var{regnum} for the
3558 specified @var{frame}. If the value of @var{regnum} is -1, pretty print
3559 either all registers (@var{all} is non zero) or a select subset of
3560 registers (@var{all} is zero).
3562 The default method prints one register per line, and if @var{all} is
3563 zero omits floating-point registers.
3565 @item int gdbarch_print_vector_info (@var{gdbarch}, @var{file}, @var{frame}, @var{args})
3566 @findex gdbarch_print_vector_info
3567 If defined, then the @samp{info vector} command will call this function
3568 to print information about the processor's vector unit.
3570 By default, the @samp{info vector} command will print all vector
3571 registers (the register's type having the vector attribute).
3573 @item int gdbarch_dwarf2_reg_to_regnum (@var{gdbarch}, @var{dwarf2_regnr})
3574 @findex gdbarch_dwarf2_reg_to_regnum
3575 Convert DWARF2 register number @var{dwarf2_regnr} into @value{GDBN} regnum.
3576 If not defined, no conversion will be performed.
3578 @item int gdbarch_ecoff_reg_to_regnum (@var{gdbarch}, @var{ecoff_regnr})
3579 @findex gdbarch_ecoff_reg_to_regnum
3580 Convert ECOFF register number @var{ecoff_regnr} into @value{GDBN} regnum. If
3581 not defined, no conversion will be performed.
3583 @item CORE_ADDR frame_align (@var{gdbarch}, @var{address})
3584 @anchor{frame_align}
3586 Define this to adjust @var{address} so that it meets the alignment
3587 requirements for the start of a new stack frame. A stack frame's
3588 alignment requirements are typically stronger than a target processors
3589 stack alignment requirements.
3591 This function is used to ensure that, when creating a dummy frame, both
3592 the initial stack pointer and (if needed) the address of the return
3593 value are correctly aligned.
3595 This function always adjusts the address in the direction of stack
3598 By default, no frame based stack alignment is performed.
3600 @item int gdbarch_frame_red_zone_size (@var{gdbarch})
3601 @findex gdbarch_frame_red_zone_size
3602 The number of bytes, beyond the innermost-stack-address, reserved by the
3603 @sc{abi}. A function is permitted to use this scratch area (instead of
3604 allocating extra stack space).
3606 When performing an inferior function call, to ensure that it does not
3607 modify this area, @value{GDBN} adjusts the innermost-stack-address by
3608 @var{gdbarch_frame_red_zone_size} bytes before pushing parameters onto the
3611 By default, zero bytes are allocated. The value must be aligned
3612 (@pxref{frame_align}).
3614 The @sc{amd64} (nee x86-64) @sc{abi} documentation refers to the
3615 @emph{red zone} when describing this scratch area.
3618 @code{FRAME_FIND_SAVED_REGS} is deprecated.
3620 @item int gdbarch_frame_num_args (@var{gdbarch}, @var{frame})
3621 @findex gdbarch_frame_num_args
3622 For the frame described by @var{frame} return the number of arguments that
3623 are being passed. If the number of arguments is not known, return
3626 @item CORE_ADDR gdbarch_unwind_pc (@var{next_frame})
3627 @findex gdbarch_unwind_pc
3628 @anchor{gdbarch_unwind_pc} Return the instruction address, in
3629 @var{next_frame}'s caller, at which execution will resume after
3630 @var{next_frame} returns. This is commonly referred to as the return address.
3632 The implementation, which must be frame agnostic (work with any frame),
3633 is typically no more than:
3637 pc = frame_unwind_register_unsigned (next_frame, S390_PC_REGNUM);
3638 return gdbarch_addr_bits_remove (gdbarch, pc);
3643 @item CORE_ADDR gdbarch_unwind_sp (@var{gdbarch}, @var{next_frame})
3644 @findex gdbarch_unwind_sp
3645 @anchor{gdbarch_unwind_sp} Return the frame's inner most stack address. This is
3646 commonly referred to as the frame's @dfn{stack pointer}.
3648 The implementation, which must be frame agnostic (work with any frame),
3649 is typically no more than:
3653 sp = frame_unwind_register_unsigned (next_frame, S390_SP_REGNUM);
3654 return gdbarch_addr_bits_remove (gdbarch, sp);
3658 @xref{TARGET_READ_SP}, which this method replaces.
3660 @item GCC_COMPILED_FLAG_SYMBOL
3661 @itemx GCC2_COMPILED_FLAG_SYMBOL
3662 @findex GCC2_COMPILED_FLAG_SYMBOL
3663 @findex GCC_COMPILED_FLAG_SYMBOL
3664 If defined, these are the names of the symbols that @value{GDBN} will
3665 look for to detect that GCC compiled the file. The default symbols
3666 are @code{gcc_compiled.} and @code{gcc2_compiled.},
3667 respectively. (Currently only defined for the Delta 68.)
3669 @item gdbarch_get_longjmp_target
3670 @findex gdbarch_get_longjmp_target
3671 This function determines the target PC address that @code{longjmp}
3672 will jump to, assuming that we have just stopped at a @code{longjmp}
3673 breakpoint. It takes a @code{CORE_ADDR *} as argument, and stores the
3674 target PC value through this pointer. It examines the current state
3675 of the machine as needed, typically by using a manually-determined
3676 offset into the @code{jmp_buf}. (While we might like to get the offset
3677 from the target's @file{jmpbuf.h}, that header file cannot be assumed
3678 to be available when building a cross-debugger.)
3680 @item DEPRECATED_IBM6000_TARGET
3681 @findex DEPRECATED_IBM6000_TARGET
3682 Shows that we are configured for an IBM RS/6000 system. This
3683 conditional should be eliminated (FIXME) and replaced by
3684 feature-specific macros. It was introduced in haste and we are
3685 repenting at leisure.
3687 @item I386_USE_GENERIC_WATCHPOINTS
3688 An x86-based target can define this to use the generic x86 watchpoint
3689 support; see @ref{Algorithms, I386_USE_GENERIC_WATCHPOINTS}.
3691 @item int gdbarch_inner_than (@var{gdbarch}, @var{lhs}, @var{rhs})
3692 @findex gdbarch_inner_than
3693 Returns non-zero if stack address @var{lhs} is inner than (nearer to the
3694 stack top) stack address @var{rhs}. Let the function return
3695 @w{@code{lhs < rhs}} if the target's stack grows downward in memory, or
3696 @w{@code{lhs > rsh}} if the stack grows upward.
3698 @item gdbarch_in_function_epilogue_p (@var{gdbarch}, @var{addr})
3699 @findex gdbarch_in_function_epilogue_p
3700 Returns non-zero if the given @var{addr} is in the epilogue of a function.
3701 The epilogue of a function is defined as the part of a function where
3702 the stack frame of the function already has been destroyed up to the
3703 final `return from function call' instruction.
3705 @item int gdbarch_in_solib_return_trampoline (@var{gdbarch}, @var{pc}, @var{name})
3706 @findex gdbarch_in_solib_return_trampoline
3707 Define this function to return nonzero if the program is stopped in the
3708 trampoline that returns from a shared library.
3710 @item target_so_ops.in_dynsym_resolve_code (@var{pc})
3711 @findex in_dynsym_resolve_code
3712 Define this to return nonzero if the program is stopped in the
3715 @item SKIP_SOLIB_RESOLVER (@var{pc})
3716 @findex SKIP_SOLIB_RESOLVER
3717 Define this to evaluate to the (nonzero) address at which execution
3718 should continue to get past the dynamic linker's symbol resolution
3719 function. A zero value indicates that it is not important or necessary
3720 to set a breakpoint to get through the dynamic linker and that single
3721 stepping will suffice.
3723 @item CORE_ADDR gdbarch_integer_to_address (@var{gdbarch}, @var{type}, @var{buf})
3724 @findex gdbarch_integer_to_address
3725 @cindex converting integers to addresses
3726 Define this when the architecture needs to handle non-pointer to address
3727 conversions specially. Converts that value to an address according to
3728 the current architectures conventions.
3730 @emph{Pragmatics: When the user copies a well defined expression from
3731 their source code and passes it, as a parameter, to @value{GDBN}'s
3732 @code{print} command, they should get the same value as would have been
3733 computed by the target program. Any deviation from this rule can cause
3734 major confusion and annoyance, and needs to be justified carefully. In
3735 other words, @value{GDBN} doesn't really have the freedom to do these
3736 conversions in clever and useful ways. It has, however, been pointed
3737 out that users aren't complaining about how @value{GDBN} casts integers
3738 to pointers; they are complaining that they can't take an address from a
3739 disassembly listing and give it to @code{x/i}. Adding an architecture
3740 method like @code{gdbarch_integer_to_address} certainly makes it possible for
3741 @value{GDBN} to ``get it right'' in all circumstances.}
3743 @xref{Target Architecture Definition, , Pointers Are Not Always
3746 @item CORE_ADDR gdbarch_pointer_to_address (@var{gdbarch}, @var{type}, @var{buf})
3747 @findex gdbarch_pointer_to_address
3748 Assume that @var{buf} holds a pointer of type @var{type}, in the
3749 appropriate format for the current architecture. Return the byte
3750 address the pointer refers to.
3751 @xref{Target Architecture Definition, , Pointers Are Not Always Addresses}.
3753 @item void gdbarch_register_to_value(@var{gdbarch}, @var{frame}, @var{regnum}, @var{type}, @var{fur})
3754 @findex gdbarch_register_to_value
3755 Convert the raw contents of register @var{regnum} into a value of type
3757 @xref{Target Architecture Definition, , Using Different Register and Memory Data Representations}.
3759 @item register_reggroup_p (@var{gdbarch}, @var{regnum}, @var{reggroup})
3760 @findex register_reggroup_p
3761 @cindex register groups
3762 Return non-zero if register @var{regnum} is a member of the register
3763 group @var{reggroup}.
3765 By default, registers are grouped as follows:
3768 @item float_reggroup
3769 Any register with a valid name and a floating-point type.
3770 @item vector_reggroup
3771 Any register with a valid name and a vector type.
3772 @item general_reggroup
3773 Any register with a valid name and a type other than vector or
3774 floating-point. @samp{float_reggroup}.
3776 @itemx restore_reggroup
3778 Any register with a valid name.
3781 @item struct type *register_type (@var{gdbarch}, @var{reg})
3782 @findex register_type
3783 If defined, return the type of register @var{reg}.
3784 @xref{Target Architecture Definition, , Raw and Virtual Register
3787 @item REGISTER_CONVERT_TO_VIRTUAL(@var{reg}, @var{type}, @var{from}, @var{to})
3788 @findex REGISTER_CONVERT_TO_VIRTUAL
3789 Convert the value of register @var{reg} from its raw form to its virtual
3791 @xref{Target Architecture Definition, , Raw and Virtual Register Representations}.
3793 @item REGISTER_CONVERT_TO_RAW(@var{type}, @var{reg}, @var{from}, @var{to})
3794 @findex REGISTER_CONVERT_TO_RAW
3795 Convert the value of register @var{reg} from its virtual form to its raw
3797 @xref{Target Architecture Definition, , Raw and Virtual Register Representations}.
3799 @item const struct regset *regset_from_core_section (struct gdbarch * @var{gdbarch}, const char * @var{sect_name}, size_t @var{sect_size})
3800 @findex regset_from_core_section
3801 Return the appropriate register set for a core file section with name
3802 @var{sect_name} and size @var{sect_size}.
3804 @item SOFTWARE_SINGLE_STEP_P()
3805 @findex SOFTWARE_SINGLE_STEP_P
3806 Define this as 1 if the target does not have a hardware single-step
3807 mechanism. The macro @code{SOFTWARE_SINGLE_STEP} must also be defined.
3809 @item SOFTWARE_SINGLE_STEP(@var{signal}, @var{insert_breakpoints_p})
3810 @findex SOFTWARE_SINGLE_STEP
3811 A function that inserts or removes (depending on
3812 @var{insert_breakpoints_p}) breakpoints at each possible destinations of
3813 the next instruction. See @file{sparc-tdep.c} and @file{rs6000-tdep.c}
3816 @item set_gdbarch_sofun_address_maybe_missing (@var{gdbarch}, @var{set})
3817 @findex set_gdbarch_sofun_address_maybe_missing
3818 Somebody clever observed that, the more actual addresses you have in the
3819 debug information, the more time the linker has to spend relocating
3820 them. So whenever there's some other way the debugger could find the
3821 address it needs, you should omit it from the debug info, to make
3824 Calling @code{set_gdbarch_sofun_address_maybe_missing} with a non-zero
3825 argument @var{set} indicates that a particular set of hacks of this sort
3826 are in use, affecting @code{N_SO} and @code{N_FUN} entries in stabs-format
3827 debugging information. @code{N_SO} stabs mark the beginning and ending
3828 addresses of compilation units in the text segment. @code{N_FUN} stabs
3829 mark the starts and ends of functions.
3831 In this case, @value{GDBN} assumes two things:
3835 @code{N_FUN} stabs have an address of zero. Instead of using those
3836 addresses, you should find the address where the function starts by
3837 taking the function name from the stab, and then looking that up in the
3838 minsyms (the linker/assembler symbol table). In other words, the stab
3839 has the name, and the linker/assembler symbol table is the only place
3840 that carries the address.
3843 @code{N_SO} stabs have an address of zero, too. You just look at the
3844 @code{N_FUN} stabs that appear before and after the @code{N_SO} stab, and
3845 guess the starting and ending addresses of the compilation unit from them.
3848 @item int gdbarch_pc_regnum (@var{gdbarch})
3849 @findex gdbarch_pc_regnum
3850 If the program counter is kept in a register, then let this function return
3851 the number (greater than or equal to zero) of that register.
3853 This should only need to be defined if @code{gdbarch_read_pc} and
3854 @code{gdbarch_write_pc} are not defined.
3856 @item int gdbarch_stabs_argument_has_addr (@var{gdbarch}, @var{type})
3857 @findex gdbarch_stabs_argument_has_addr
3858 @anchor{gdbarch_stabs_argument_has_addr} Define this function to return
3859 nonzero if a function argument of type @var{type} is passed by reference
3862 @item PROCESS_LINENUMBER_HOOK
3863 @findex PROCESS_LINENUMBER_HOOK
3864 A hook defined for XCOFF reading.
3866 @item gdbarch_ps_regnum (@var{gdbarch}
3867 @findex gdbarch_ps_regnum
3868 If defined, this function returns the number of the processor status
3870 (This definition is only used in generic code when parsing "$ps".)
3872 @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})
3873 @findex gdbarch_push_dummy_call
3874 @anchor{gdbarch_push_dummy_call} Define this to push the dummy frame's call to
3875 the inferior function onto the stack. In addition to pushing @var{nargs}, the
3876 code should push @var{struct_addr} (when @var{struct_return} is non-zero), and
3877 the return address (@var{bp_addr}).
3879 @var{function} is a pointer to a @code{struct value}; on architectures that use
3880 function descriptors, this contains the function descriptor value.
3882 Returns the updated top-of-stack pointer.
3884 @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})
3885 @findex gdbarch_push_dummy_code
3886 @anchor{gdbarch_push_dummy_code} Given a stack based call dummy, push the
3887 instruction sequence (including space for a breakpoint) to which the
3888 called function should return.
3890 Set @var{bp_addr} to the address at which the breakpoint instruction
3891 should be inserted, @var{real_pc} to the resume address when starting
3892 the call sequence, and return the updated inner-most stack address.
3894 By default, the stack is grown sufficient to hold a frame-aligned
3895 (@pxref{frame_align}) breakpoint, @var{bp_addr} is set to the address
3896 reserved for that breakpoint, and @var{real_pc} set to @var{funaddr}.
3898 This method replaces @w{@code{gdbarch_call_dummy_location (@var{gdbarch})}}.
3900 @item const char *gdbarch_register_name (@var{gdbarch}, @var{regnr})
3901 @findex gdbarch_register_name
3902 Return the name of register @var{regnr} as a string. May return @code{NULL}
3903 to indicate that @var{regnr} is not a valid register.
3905 @item int gdbarch_sdb_reg_to_regnum (@var{gdbarch}, @var{sdb_regnr})
3906 @findex gdbarch_sdb_reg_to_regnum
3907 Use this function to convert sdb register @var{sdb_regnr} into @value{GDBN}
3908 regnum. If not defined, no conversion will be done.
3910 @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})
3911 @findex gdbarch_return_value
3912 @anchor{gdbarch_return_value} Given a function with a return-value of
3913 type @var{rettype}, return which return-value convention that function
3916 @value{GDBN} currently recognizes two function return-value conventions:
3917 @code{RETURN_VALUE_REGISTER_CONVENTION} where the return value is found
3918 in registers; and @code{RETURN_VALUE_STRUCT_CONVENTION} where the return
3919 value is found in memory and the address of that memory location is
3920 passed in as the function's first parameter.
3922 If the register convention is being used, and @var{writebuf} is
3923 non-@code{NULL}, also copy the return-value in @var{writebuf} into
3926 If the register convention is being used, and @var{readbuf} is
3927 non-@code{NULL}, also copy the return value from @var{regcache} into
3928 @var{readbuf} (@var{regcache} contains a copy of the registers from the
3929 just returned function).
3931 @emph{Maintainer note: This method replaces separate predicate, extract,
3932 store methods. By having only one method, the logic needed to determine
3933 the return-value convention need only be implemented in one place. If
3934 @value{GDBN} were written in an @sc{oo} language, this method would
3935 instead return an object that knew how to perform the register
3936 return-value extract and store.}
3938 @emph{Maintainer note: This method does not take a @var{gcc_p}
3939 parameter, and such a parameter should not be added. If an architecture
3940 that requires per-compiler or per-function information be identified,
3941 then the replacement of @var{rettype} with @code{struct value}
3942 @var{function} should be pursued.}
3944 @emph{Maintainer note: The @var{regcache} parameter limits this methods
3945 to the inner most frame. While replacing @var{regcache} with a
3946 @code{struct frame_info} @var{frame} parameter would remove that
3947 limitation there has yet to be a demonstrated need for such a change.}
3949 @item void gdbarch_skip_permanent_breakpoint (@var{gdbarch}, @var{regcache})
3950 @findex gdbarch_skip_permanent_breakpoint
3951 Advance the inferior's PC past a permanent breakpoint. @value{GDBN} normally
3952 steps over a breakpoint by removing it, stepping one instruction, and
3953 re-inserting the breakpoint. However, permanent breakpoints are
3954 hardwired into the inferior, and can't be removed, so this strategy
3955 doesn't work. Calling @code{gdbarch_skip_permanent_breakpoint} adjusts the
3956 processor's state so that execution will resume just after the breakpoint.
3957 This function does the right thing even when the breakpoint is in the delay slot
3958 of a branch or jump.
3960 @item CORE_ADDR gdbarch_skip_prologue (@var{gdbarch}, @var{ip})
3961 @findex gdbarch_skip_prologue
3962 A function that returns the address of the ``real'' code beyond the
3963 function entry prologue found at @var{ip}.
3965 @item CORE_ADDR gdbarch_skip_trampoline_code (@var{gdbarch}, @var{frame}, @var{pc})
3966 @findex gdbarch_skip_trampoline_code
3967 If the target machine has trampoline code that sits between callers and
3968 the functions being called, then define this function to return a new PC
3969 that is at the start of the real function.
3971 @item int gdbarch_sp_regnum (@var{gdbarch})
3972 @findex gdbarch_sp_regnum
3973 If the stack-pointer is kept in a register, then use this function to return
3974 the number (greater than or equal to zero) of that register, or -1 if
3975 there is no such register.
3977 @item int gdbarch_deprecated_fp_regnum (@var{gdbarch})
3978 @findex gdbarch_deprecated_fp_regnum
3979 If the frame pointer is in a register, use this function to return the
3980 number of that register.
3982 @item int gdbarch_stab_reg_to_regnum (@var{gdbarch}, @var{stab_regnr})
3983 @findex gdbarch_stab_reg_to_regnum
3984 Use this function to convert stab register @var{stab_regnr} into @value{GDBN}
3985 regnum. If not defined, no conversion will be done.
3987 @item SYMBOL_RELOADING_DEFAULT
3988 @findex SYMBOL_RELOADING_DEFAULT
3989 The default value of the ``symbol-reloading'' variable. (Never defined in
3992 @item TARGET_CHAR_BIT
3993 @findex TARGET_CHAR_BIT
3994 Number of bits in a char; defaults to 8.
3996 @item int gdbarch_char_signed (@var{gdbarch})
3997 @findex gdbarch_char_signed
3998 Non-zero if @code{char} is normally signed on this architecture; zero if
3999 it should be unsigned.
4001 The ISO C standard requires the compiler to treat @code{char} as
4002 equivalent to either @code{signed char} or @code{unsigned char}; any
4003 character in the standard execution set is supposed to be positive.
4004 Most compilers treat @code{char} as signed, but @code{char} is unsigned
4005 on the IBM S/390, RS6000, and PowerPC targets.
4007 @item int gdbarch_double_bit (@var{gdbarch})
4008 @findex gdbarch_double_bit
4009 Number of bits in a double float; defaults to @w{@code{8 * TARGET_CHAR_BIT}}.
4011 @item int gdbarch_float_bit (@var{gdbarch})
4012 @findex gdbarch_float_bit
4013 Number of bits in a float; defaults to @w{@code{4 * TARGET_CHAR_BIT}}.
4015 @item int gdbarch_int_bit (@var{gdbarch})
4016 @findex gdbarch_int_bit
4017 Number of bits in an integer; defaults to @w{@code{4 * TARGET_CHAR_BIT}}.
4019 @item int gdbarch_long_bit (@var{gdbarch})
4020 @findex gdbarch_long_bit
4021 Number of bits in a long integer; defaults to @w{@code{4 * TARGET_CHAR_BIT}}.
4023 @item int gdbarch_long_double_bit (@var{gdbarch})
4024 @findex gdbarch_long_double_bit
4025 Number of bits in a long double float;
4026 defaults to @w{@code{2 * gdbarch_double_bit (@var{gdbarch})}}.
4028 @item int gdbarch_long_long_bit (@var{gdbarch})
4029 @findex gdbarch_long_long_bit
4030 Number of bits in a long long integer; defaults to
4031 @w{@code{2 * gdbarch_long_bit (@var{gdbarch})}}.
4033 @item int gdbarch_ptr_bit (@var{gdbarch})
4034 @findex gdbarch_ptr_bit
4035 Number of bits in a pointer; defaults to
4036 @w{@code{gdbarch_int_bit (@var{gdbarch})}}.
4038 @item int gdbarch_short_bit (@var{gdbarch})
4039 @findex gdbarch_short_bit
4040 Number of bits in a short integer; defaults to @w{@code{2 * TARGET_CHAR_BIT}}.
4042 @item CORE_ADDR gdbarch_read_pc (@var{gdbarch}, @var{regcache})
4043 @findex gdbarch_read_pc
4044 @itemx gdbarch_write_pc (@var{gdbarch}, @var{regcache}, @var{val})
4045 @findex gdbarch_write_pc
4046 @anchor{gdbarch_write_pc}
4047 @itemx TARGET_READ_SP
4048 @findex TARGET_READ_SP
4049 @itemx TARGET_READ_FP
4050 @findex TARGET_READ_FP
4051 @findex gdbarch_read_pc
4052 @findex gdbarch_write_pc
4055 @anchor{TARGET_READ_SP} These change the behavior of @code{gdbarch_read_pc},
4056 @code{gdbarch_write_pc}, and @code{read_sp}. For most targets, these may be
4057 left undefined. @value{GDBN} will call the read and write register
4058 functions with the relevant @code{_REGNUM} argument.
4060 These macros and functions are useful when a target keeps one of these
4061 registers in a hard to get at place; for example, part in a segment register
4062 and part in an ordinary register.
4064 @xref{gdbarch_unwind_sp}, which replaces @code{TARGET_READ_SP}.
4066 @item void gdbarch_virtual_frame_pointer (@var{gdbarch}, @var{pc}, @var{frame_regnum}, @var{frame_offset})
4067 @findex gdbarch_virtual_frame_pointer
4068 Returns a @code{(@var{register}, @var{offset})} pair representing the virtual
4069 frame pointer in use at the code address @var{pc}. If virtual frame
4070 pointers are not used, a default definition simply returns
4071 @code{gdbarch_deprecated_fp_regnum} (or @code{gdbarch_sp_regnum}, if
4072 no frame pointer is defined), with an offset of zero.
4074 @c need to explain virtual frame pointers, they are recorded in agent expressions
4077 @item TARGET_HAS_HARDWARE_WATCHPOINTS
4078 If non-zero, the target has support for hardware-assisted
4079 watchpoints. @xref{Algorithms, watchpoints}, for more details and
4080 other related macros.
4082 @item int gdbarch_print_insn (@var{gdbarch}, @var{vma}, @var{info})
4083 @findex gdbarch_print_insn
4084 This is the function used by @value{GDBN} to print an assembly
4085 instruction. It prints the instruction at address @var{vma} in
4086 debugged memory and returns the length of the instruction, in bytes.
4087 This usually points to a function in the @code{opcodes} library
4088 (@pxref{Support Libraries, ,Opcodes}). @var{info} is a structure (of
4089 type @code{disassemble_info}) defined in the header file
4090 @file{include/dis-asm.h}, and used to pass information to the
4091 instruction decoding routine.
4093 @item frame_id gdbarch_dummy_id (@var{gdbarch}, @var{frame})
4094 @findex gdbarch_dummy_id
4095 @anchor{gdbarch_dummy_id} Given @var{frame} return a @w{@code{struct
4096 frame_id}} that uniquely identifies an inferior function call's dummy
4097 frame. The value returned must match the dummy frame stack value
4098 previously saved by @code{call_function_by_hand}.
4100 @item void gdbarch_value_to_register (@var{gdbarch}, @var{frame}, @var{type}, @var{buf})
4101 @findex gdbarch_value_to_register
4102 Convert a value of type @var{type} into the raw contents of a register.
4103 @xref{Target Architecture Definition, , Using Different Register and Memory Data Representations}.
4107 Motorola M68K target conditionals.
4111 Define this to be the 4-bit location of the breakpoint trap vector. If
4112 not defined, it will default to @code{0xf}.
4114 @item REMOTE_BPT_VECTOR
4115 Defaults to @code{1}.
4117 @item const char *gdbarch_name_of_malloc (@var{gdbarch})
4118 @findex gdbarch_name_of_malloc
4119 A string containing the name of the function to call in order to
4120 allocate some memory in the inferior. The default value is "malloc".
4124 @node Adding a New Target
4125 @section Adding a New Target
4127 @cindex adding a target
4128 The following files add a target to @value{GDBN}:
4132 @item gdb/config/@var{arch}/@var{ttt}.mt
4133 Contains a Makefile fragment specific to this target. Specifies what
4134 object files are needed for target @var{ttt}, by defining
4135 @samp{TDEPFILES=@dots{}} and @samp{TDEPLIBS=@dots{}}.
4137 You can also define @samp{TM_CLIBS} and @samp{TM_CDEPS}, but these are
4138 now deprecated, replaced by autoconf, and may go away in future
4139 versions of @value{GDBN}.
4141 @item gdb/@var{ttt}-tdep.c
4142 Contains any miscellaneous code required for this target machine. On
4143 some machines it doesn't exist at all.
4145 @item gdb/@var{arch}-tdep.c
4146 @itemx gdb/@var{arch}-tdep.h
4147 This is required to describe the basic layout of the target machine's
4148 processor chip (registers, stack, etc.). It can be shared among many
4149 targets that use the same processor architecture.
4153 (Target header files such as
4154 @file{gdb/config/@var{arch}/tm-@var{ttt}.h},
4155 @file{gdb/config/@var{arch}/tm-@var{arch}.h}, and
4156 @file{config/tm-@var{os}.h} are no longer used.)
4158 @node Target Descriptions
4159 @chapter Target Descriptions
4160 @cindex target descriptions
4162 The target architecture definition (@pxref{Target Architecture Definition})
4163 contains @value{GDBN}'s hard-coded knowledge about an architecture. For
4164 some platforms, it is handy to have more flexible knowledge about a specific
4165 instance of the architecture---for instance, a processor or development board.
4166 @dfn{Target descriptions} provide a mechanism for the user to tell @value{GDBN}
4167 more about what their target supports, or for the target to tell @value{GDBN}
4170 For details on writing, automatically supplying, and manually selecting
4171 target descriptions, see @ref{Target Descriptions, , , gdb,
4172 Debugging with @value{GDBN}}. This section will cover some related
4173 topics about the @value{GDBN} internals.
4176 * Target Descriptions Implementation::
4177 * Adding Target Described Register Support::
4180 @node Target Descriptions Implementation
4181 @section Target Descriptions Implementation
4182 @cindex target descriptions, implementation
4184 Before @value{GDBN} connects to a new target, or runs a new program on
4185 an existing target, it discards any existing target description and
4186 reverts to a default gdbarch. Then, after connecting, it looks for a
4187 new target description by calling @code{target_find_description}.
4189 A description may come from a user specified file (XML), the remote
4190 @samp{qXfer:features:read} packet (also XML), or from any custom
4191 @code{to_read_description} routine in the target vector. For instance,
4192 the remote target supports guessing whether a MIPS target is 32-bit or
4193 64-bit based on the size of the @samp{g} packet.
4195 If any target description is found, @value{GDBN} creates a new gdbarch
4196 incorporating the description by calling @code{gdbarch_update_p}. Any
4197 @samp{<architecture>} element is handled first, to determine which
4198 architecture's gdbarch initialization routine is called to create the
4199 new architecture. Then the initialization routine is called, and has
4200 a chance to adjust the constructed architecture based on the contents
4201 of the target description. For instance, it can recognize any
4202 properties set by a @code{to_read_description} routine. Also
4203 see @ref{Adding Target Described Register Support}.
4205 @node Adding Target Described Register Support
4206 @section Adding Target Described Register Support
4207 @cindex target descriptions, adding register support
4209 Target descriptions can report additional registers specific to an
4210 instance of the target. But it takes a little work in the architecture
4211 specific routines to support this.
4213 A target description must either have no registers or a complete
4214 set---this avoids complexity in trying to merge standard registers
4215 with the target defined registers. It is the architecture's
4216 responsibility to validate that a description with registers has
4217 everything it needs. To keep architecture code simple, the same
4218 mechanism is used to assign fixed internal register numbers to
4221 If @code{tdesc_has_registers} returns 1, the description contains
4222 registers. The architecture's @code{gdbarch_init} routine should:
4227 Call @code{tdesc_data_alloc} to allocate storage, early, before
4228 searching for a matching gdbarch or allocating a new one.
4231 Use @code{tdesc_find_feature} to locate standard features by name.
4234 Use @code{tdesc_numbered_register} and @code{tdesc_numbered_register_choices}
4235 to locate the expected registers in the standard features.
4238 Return @code{NULL} if a required feature is missing, or if any standard
4239 feature is missing expected registers. This will produce a warning that
4240 the description was incomplete.
4243 Free the allocated data before returning, unless @code{tdesc_use_registers}
4247 Call @code{set_gdbarch_num_regs} as usual, with a number higher than any
4248 fixed number passed to @code{tdesc_numbered_register}.
4251 Call @code{tdesc_use_registers} after creating a new gdbarch, before
4256 After @code{tdesc_use_registers} has been called, the architecture's
4257 @code{register_name}, @code{register_type}, and @code{register_reggroup_p}
4258 routines will not be called; that information will be taken from
4259 the target description. @code{num_regs} may be increased to account
4260 for any additional registers in the description.
4262 Pseudo-registers require some extra care:
4267 Using @code{tdesc_numbered_register} allows the architecture to give
4268 constant register numbers to standard architectural registers, e.g.@:
4269 as an @code{enum} in @file{@var{arch}-tdep.h}. But because
4270 pseudo-registers are always numbered above @code{num_regs},
4271 which may be increased by the description, constant numbers
4272 can not be used for pseudos. They must be numbered relative to
4273 @code{num_regs} instead.
4276 The description will not describe pseudo-registers, so the
4277 architecture must call @code{set_tdesc_pseudo_register_name},
4278 @code{set_tdesc_pseudo_register_type}, and
4279 @code{set_tdesc_pseudo_register_reggroup_p} to supply routines
4280 describing pseudo registers. These routines will be passed
4281 internal register numbers, so the same routines used for the
4282 gdbarch equivalents are usually suitable.
4287 @node Target Vector Definition
4289 @chapter Target Vector Definition
4290 @cindex target vector
4292 The target vector defines the interface between @value{GDBN}'s
4293 abstract handling of target systems, and the nitty-gritty code that
4294 actually exercises control over a process or a serial port.
4295 @value{GDBN} includes some 30-40 different target vectors; however,
4296 each configuration of @value{GDBN} includes only a few of them.
4299 * Managing Execution State::
4300 * Existing Targets::
4303 @node Managing Execution State
4304 @section Managing Execution State
4305 @cindex execution state
4307 A target vector can be completely inactive (not pushed on the target
4308 stack), active but not running (pushed, but not connected to a fully
4309 manifested inferior), or completely active (pushed, with an accessible
4310 inferior). Most targets are only completely inactive or completely
4311 active, but some support persistent connections to a target even
4312 when the target has exited or not yet started.
4314 For example, connecting to the simulator using @code{target sim} does
4315 not create a running program. Neither registers nor memory are
4316 accessible until @code{run}. Similarly, after @code{kill}, the
4317 program can not continue executing. But in both cases @value{GDBN}
4318 remains connected to the simulator, and target-specific commands
4319 are directed to the simulator.
4321 A target which only supports complete activation should push itself
4322 onto the stack in its @code{to_open} routine (by calling
4323 @code{push_target}), and unpush itself from the stack in its
4324 @code{to_mourn_inferior} routine (by calling @code{unpush_target}).
4326 A target which supports both partial and complete activation should
4327 still call @code{push_target} in @code{to_open}, but not call
4328 @code{unpush_target} in @code{to_mourn_inferior}. Instead, it should
4329 call either @code{target_mark_running} or @code{target_mark_exited}
4330 in its @code{to_open}, depending on whether the target is fully active
4331 after connection. It should also call @code{target_mark_running} any
4332 time the inferior becomes fully active (e.g.@: in
4333 @code{to_create_inferior} and @code{to_attach}), and
4334 @code{target_mark_exited} when the inferior becomes inactive (in
4335 @code{to_mourn_inferior}). The target should also make sure to call
4336 @code{target_mourn_inferior} from its @code{to_kill}, to return the
4337 target to inactive state.
4339 @node Existing Targets
4340 @section Existing Targets
4343 @subsection File Targets
4345 Both executables and core files have target vectors.
4347 @subsection Standard Protocol and Remote Stubs
4349 @value{GDBN}'s file @file{remote.c} talks a serial protocol to code
4350 that runs in the target system. @value{GDBN} provides several sample
4351 @dfn{stubs} that can be integrated into target programs or operating
4352 systems for this purpose; they are named @file{@var{cpu}-stub.c}. Many
4353 operating systems, embedded targets, emulators, and simulators already
4354 have a GDB stub built into them, and maintenance of the remote
4355 protocol must be careful to preserve compatibility.
4357 The @value{GDBN} user's manual describes how to put such a stub into
4358 your target code. What follows is a discussion of integrating the
4359 SPARC stub into a complicated operating system (rather than a simple
4360 program), by Stu Grossman, the author of this stub.
4362 The trap handling code in the stub assumes the following upon entry to
4367 %l1 and %l2 contain pc and npc respectively at the time of the trap;
4373 you are in the correct trap window.
4376 As long as your trap handler can guarantee those conditions, then there
4377 is no reason why you shouldn't be able to ``share'' traps with the stub.
4378 The stub has no requirement that it be jumped to directly from the
4379 hardware trap vector. That is why it calls @code{exceptionHandler()},
4380 which is provided by the external environment. For instance, this could
4381 set up the hardware traps to actually execute code which calls the stub
4382 first, and then transfers to its own trap handler.
4384 For the most point, there probably won't be much of an issue with
4385 ``sharing'' traps, as the traps we use are usually not used by the kernel,
4386 and often indicate unrecoverable error conditions. Anyway, this is all
4387 controlled by a table, and is trivial to modify. The most important
4388 trap for us is for @code{ta 1}. Without that, we can't single step or
4389 do breakpoints. Everything else is unnecessary for the proper operation
4390 of the debugger/stub.
4392 From reading the stub, it's probably not obvious how breakpoints work.
4393 They are simply done by deposit/examine operations from @value{GDBN}.
4395 @subsection ROM Monitor Interface
4397 @subsection Custom Protocols
4399 @subsection Transport Layer
4401 @subsection Builtin Simulator
4404 @node Native Debugging
4406 @chapter Native Debugging
4407 @cindex native debugging
4409 Several files control @value{GDBN}'s configuration for native support:
4413 @item gdb/config/@var{arch}/@var{xyz}.mh
4414 Specifies Makefile fragments needed by a @emph{native} configuration on
4415 machine @var{xyz}. In particular, this lists the required
4416 native-dependent object files, by defining @samp{NATDEPFILES=@dots{}}.
4417 Also specifies the header file which describes native support on
4418 @var{xyz}, by defining @samp{NAT_FILE= nm-@var{xyz}.h}. You can also
4419 define @samp{NAT_CFLAGS}, @samp{NAT_ADD_FILES}, @samp{NAT_CLIBS},
4420 @samp{NAT_CDEPS}, etc.; see @file{Makefile.in}.
4422 @emph{Maintainer's note: The @file{.mh} suffix is because this file
4423 originally contained @file{Makefile} fragments for hosting @value{GDBN}
4424 on machine @var{xyz}. While the file is no longer used for this
4425 purpose, the @file{.mh} suffix remains. Perhaps someone will
4426 eventually rename these fragments so that they have a @file{.mn}
4429 @item gdb/config/@var{arch}/nm-@var{xyz}.h
4430 (@file{nm.h} is a link to this file, created by @code{configure}). Contains C
4431 macro definitions describing the native system environment, such as
4432 child process control and core file support.
4434 @item gdb/@var{xyz}-nat.c
4435 Contains any miscellaneous C code required for this native support of
4436 this machine. On some machines it doesn't exist at all.
4439 There are some ``generic'' versions of routines that can be used by
4440 various systems. These can be customized in various ways by macros
4441 defined in your @file{nm-@var{xyz}.h} file. If these routines work for
4442 the @var{xyz} host, you can just include the generic file's name (with
4443 @samp{.o}, not @samp{.c}) in @code{NATDEPFILES}.
4445 Otherwise, if your machine needs custom support routines, you will need
4446 to write routines that perform the same functions as the generic file.
4447 Put them into @file{@var{xyz}-nat.c}, and put @file{@var{xyz}-nat.o}
4448 into @code{NATDEPFILES}.
4452 This contains the @emph{target_ops vector} that supports Unix child
4453 processes on systems which use ptrace and wait to control the child.
4456 This contains the @emph{target_ops vector} that supports Unix child
4457 processes on systems which use /proc to control the child.
4460 This does the low-level grunge that uses Unix system calls to do a ``fork
4461 and exec'' to start up a child process.
4464 This is the low level interface to inferior processes for systems using
4465 the Unix @code{ptrace} call in a vanilla way.
4468 @section Native core file Support
4469 @cindex native core files
4472 @findex fetch_core_registers
4473 @item core-aout.c::fetch_core_registers()
4474 Support for reading registers out of a core file. This routine calls
4475 @code{register_addr()}, see below. Now that BFD is used to read core
4476 files, virtually all machines should use @code{core-aout.c}, and should
4477 just provide @code{fetch_core_registers} in @code{@var{xyz}-nat.c} (or
4478 @code{REGISTER_U_ADDR} in @code{nm-@var{xyz}.h}).
4480 @item core-aout.c::register_addr()
4481 If your @code{nm-@var{xyz}.h} file defines the macro
4482 @code{REGISTER_U_ADDR(addr, blockend, regno)}, it should be defined to
4483 set @code{addr} to the offset within the @samp{user} struct of @value{GDBN}
4484 register number @code{regno}. @code{blockend} is the offset within the
4485 ``upage'' of @code{u.u_ar0}. If @code{REGISTER_U_ADDR} is defined,
4486 @file{core-aout.c} will define the @code{register_addr()} function and
4487 use the macro in it. If you do not define @code{REGISTER_U_ADDR}, but
4488 you are using the standard @code{fetch_core_registers()}, you will need
4489 to define your own version of @code{register_addr()}, put it into your
4490 @code{@var{xyz}-nat.c} file, and be sure @code{@var{xyz}-nat.o} is in
4491 the @code{NATDEPFILES} list. If you have your own
4492 @code{fetch_core_registers()}, you may not need a separate
4493 @code{register_addr()}. Many custom @code{fetch_core_registers()}
4494 implementations simply locate the registers themselves.@refill
4497 When making @value{GDBN} run native on a new operating system, to make it
4498 possible to debug core files, you will need to either write specific
4499 code for parsing your OS's core files, or customize
4500 @file{bfd/trad-core.c}. First, use whatever @code{#include} files your
4501 machine uses to define the struct of registers that is accessible
4502 (possibly in the u-area) in a core file (rather than
4503 @file{machine/reg.h}), and an include file that defines whatever header
4504 exists on a core file (e.g., the u-area or a @code{struct core}). Then
4505 modify @code{trad_unix_core_file_p} to use these values to set up the
4506 section information for the data segment, stack segment, any other
4507 segments in the core file (perhaps shared library contents or control
4508 information), ``registers'' segment, and if there are two discontiguous
4509 sets of registers (e.g., integer and float), the ``reg2'' segment. This
4510 section information basically delimits areas in the core file in a
4511 standard way, which the section-reading routines in BFD know how to seek
4514 Then back in @value{GDBN}, you need a matching routine called
4515 @code{fetch_core_registers}. If you can use the generic one, it's in
4516 @file{core-aout.c}; if not, it's in your @file{@var{xyz}-nat.c} file.
4517 It will be passed a char pointer to the entire ``registers'' segment,
4518 its length, and a zero; or a char pointer to the entire ``regs2''
4519 segment, its length, and a 2. The routine should suck out the supplied
4520 register values and install them into @value{GDBN}'s ``registers'' array.
4522 If your system uses @file{/proc} to control processes, and uses ELF
4523 format core files, then you may be able to use the same routines for
4524 reading the registers out of processes and out of core files.
4532 @section shared libraries
4534 @section Native Conditionals
4535 @cindex native conditionals
4537 When @value{GDBN} is configured and compiled, various macros are
4538 defined or left undefined, to control compilation when the host and
4539 target systems are the same. These macros should be defined (or left
4540 undefined) in @file{nm-@var{system}.h}.
4544 @item CHILD_PREPARE_TO_STORE
4545 @findex CHILD_PREPARE_TO_STORE
4546 If the machine stores all registers at once in the child process, then
4547 define this to ensure that all values are correct. This usually entails
4548 a read from the child.
4550 [Note that this is incorrectly defined in @file{xm-@var{system}.h} files
4553 @item FETCH_INFERIOR_REGISTERS
4554 @findex FETCH_INFERIOR_REGISTERS
4555 Define this if the native-dependent code will provide its own routines
4556 @code{fetch_inferior_registers} and @code{store_inferior_registers} in
4557 @file{@var{host}-nat.c}. If this symbol is @emph{not} defined, and
4558 @file{infptrace.c} is included in this configuration, the default
4559 routines in @file{infptrace.c} are used for these functions.
4561 @item int gdbarch_fp0_regnum (@var{gdbarch})
4562 @findex gdbarch_fp0_regnum
4563 This functions normally returns the number of the first floating
4564 point register, if the machine has such registers. As such, it would
4565 appear only in target-specific code. However, @file{/proc} support uses this
4566 to decide whether floats are in use on this target.
4568 @item int gdbarch_get_longjmp_target (@var{gdbarch})
4569 @findex gdbarch_get_longjmp_target
4570 This function determines the target PC address that @code{longjmp} will jump to,
4571 assuming that we have just stopped at a longjmp breakpoint. It takes a
4572 @code{CORE_ADDR *} as argument, and stores the target PC value through this
4573 pointer. It examines the current state of the machine as needed.
4575 @item I386_USE_GENERIC_WATCHPOINTS
4576 An x86-based machine can define this to use the generic x86 watchpoint
4577 support; see @ref{Algorithms, I386_USE_GENERIC_WATCHPOINTS}.
4579 @item ONE_PROCESS_WRITETEXT
4580 @findex ONE_PROCESS_WRITETEXT
4581 Define this to be able to, when a breakpoint insertion fails, warn the
4582 user that another process may be running with the same executable.
4585 @findex PROC_NAME_FMT
4586 Defines the format for the name of a @file{/proc} device. Should be
4587 defined in @file{nm.h} @emph{only} in order to override the default
4588 definition in @file{procfs.c}.
4590 @item SOLIB_ADD (@var{filename}, @var{from_tty}, @var{targ}, @var{readsyms})
4592 Define this to expand into an expression that will cause the symbols in
4593 @var{filename} to be added to @value{GDBN}'s symbol table. If
4594 @var{readsyms} is zero symbols are not read but any necessary low level
4595 processing for @var{filename} is still done.
4597 @item SOLIB_CREATE_INFERIOR_HOOK
4598 @findex SOLIB_CREATE_INFERIOR_HOOK
4599 Define this to expand into any shared-library-relocation code that you
4600 want to be run just after the child process has been forked.
4602 @item START_INFERIOR_TRAPS_EXPECTED
4603 @findex START_INFERIOR_TRAPS_EXPECTED
4604 When starting an inferior, @value{GDBN} normally expects to trap
4606 the shell execs, and once when the program itself execs. If the actual
4607 number of traps is something other than 2, then define this macro to
4608 expand into the number expected.
4612 @node Support Libraries
4614 @chapter Support Libraries
4619 BFD provides support for @value{GDBN} in several ways:
4622 @item identifying executable and core files
4623 BFD will identify a variety of file types, including a.out, coff, and
4624 several variants thereof, as well as several kinds of core files.
4626 @item access to sections of files
4627 BFD parses the file headers to determine the names, virtual addresses,
4628 sizes, and file locations of all the various named sections in files
4629 (such as the text section or the data section). @value{GDBN} simply
4630 calls BFD to read or write section @var{x} at byte offset @var{y} for
4633 @item specialized core file support
4634 BFD provides routines to determine the failing command name stored in a
4635 core file, the signal with which the program failed, and whether a core
4636 file matches (i.e.@: could be a core dump of) a particular executable
4639 @item locating the symbol information
4640 @value{GDBN} uses an internal interface of BFD to determine where to find the
4641 symbol information in an executable file or symbol-file. @value{GDBN} itself
4642 handles the reading of symbols, since BFD does not ``understand'' debug
4643 symbols, but @value{GDBN} uses BFD's cached information to find the symbols,
4648 @cindex opcodes library
4650 The opcodes library provides @value{GDBN}'s disassembler. (It's a separate
4651 library because it's also used in binutils, for @file{objdump}).
4654 @cindex readline library
4655 The @code{readline} library provides a set of functions for use by applications
4656 that allow users to edit command lines as they are typed in.
4659 @cindex @code{libiberty} library
4661 The @code{libiberty} library provides a set of functions and features
4662 that integrate and improve on functionality found in modern operating
4663 systems. Broadly speaking, such features can be divided into three
4664 groups: supplemental functions (functions that may be missing in some
4665 environments and operating systems), replacement functions (providing
4666 a uniform and easier to use interface for commonly used standard
4667 functions), and extensions (which provide additional functionality
4668 beyond standard functions).
4670 @value{GDBN} uses various features provided by the @code{libiberty}
4671 library, for instance the C@t{++} demangler, the @acronym{IEEE}
4672 floating format support functions, the input options parser
4673 @samp{getopt}, the @samp{obstack} extension, and other functions.
4675 @subsection @code{obstacks} in @value{GDBN}
4676 @cindex @code{obstacks}
4678 The obstack mechanism provides a convenient way to allocate and free
4679 chunks of memory. Each obstack is a pool of memory that is managed
4680 like a stack. Objects (of any nature, size and alignment) are
4681 allocated and freed in a @acronym{LIFO} fashion on an obstack (see
4682 @code{libiberty}'s documentation for a more detailed explanation of
4685 The most noticeable use of the @code{obstacks} in @value{GDBN} is in
4686 object files. There is an obstack associated with each internal
4687 representation of an object file. Lots of things get allocated on
4688 these @code{obstacks}: dictionary entries, blocks, blockvectors,
4689 symbols, minimal symbols, types, vectors of fundamental types, class
4690 fields of types, object files section lists, object files section
4691 offset lists, line tables, symbol tables, partial symbol tables,
4692 string tables, symbol table private data, macros tables, debug
4693 information sections and entries, import and export lists (som),
4694 unwind information (hppa), dwarf2 location expressions data. Plus
4695 various strings such as directory names strings, debug format strings,
4698 An essential and convenient property of all data on @code{obstacks} is
4699 that memory for it gets allocated (with @code{obstack_alloc}) at
4700 various times during a debugging session, but it is released all at
4701 once using the @code{obstack_free} function. The @code{obstack_free}
4702 function takes a pointer to where in the stack it must start the
4703 deletion from (much like the cleanup chains have a pointer to where to
4704 start the cleanups). Because of the stack like structure of the
4705 @code{obstacks}, this allows to free only a top portion of the
4706 obstack. There are a few instances in @value{GDBN} where such thing
4707 happens. Calls to @code{obstack_free} are done after some local data
4708 is allocated to the obstack. Only the local data is deleted from the
4709 obstack. Of course this assumes that nothing between the
4710 @code{obstack_alloc} and the @code{obstack_free} allocates anything
4711 else on the same obstack. For this reason it is best and safest to
4712 use temporary @code{obstacks}.
4714 Releasing the whole obstack is also not safe per se. It is safe only
4715 under the condition that we know the @code{obstacks} memory is no
4716 longer needed. In @value{GDBN} we get rid of the @code{obstacks} only
4717 when we get rid of the whole objfile(s), for instance upon reading a
4721 @cindex regular expressions library
4732 @item SIGN_EXTEND_CHAR
4734 @item SWITCH_ENUM_BUG
4743 @section Array Containers
4744 @cindex Array Containers
4747 Often it is necessary to manipulate a dynamic array of a set of
4748 objects. C forces some bookkeeping on this, which can get cumbersome
4749 and repetitive. The @file{vec.h} file contains macros for defining
4750 and using a typesafe vector type. The functions defined will be
4751 inlined when compiling, and so the abstraction cost should be zero.
4752 Domain checks are added to detect programming errors.
4754 An example use would be an array of symbols or section information.
4755 The array can be grown as symbols are read in (or preallocated), and
4756 the accessor macros provided keep care of all the necessary
4757 bookkeeping. Because the arrays are type safe, there is no danger of
4758 accidentally mixing up the contents. Think of these as C++ templates,
4759 but implemented in C.
4761 Because of the different behavior of structure objects, scalar objects
4762 and of pointers, there are three flavors of vector, one for each of
4763 these variants. Both the structure object and pointer variants pass
4764 pointers to objects around --- in the former case the pointers are
4765 stored into the vector and in the latter case the pointers are
4766 dereferenced and the objects copied into the vector. The scalar
4767 object variant is suitable for @code{int}-like objects, and the vector
4768 elements are returned by value.
4770 There are both @code{index} and @code{iterate} accessors. The iterator
4771 returns a boolean iteration condition and updates the iteration
4772 variable passed by reference. Because the iterator will be inlined,
4773 the address-of can be optimized away.
4775 The vectors are implemented using the trailing array idiom, thus they
4776 are not resizeable without changing the address of the vector object
4777 itself. This means you cannot have variables or fields of vector type
4778 --- always use a pointer to a vector. The one exception is the final
4779 field of a structure, which could be a vector type. You will have to
4780 use the @code{embedded_size} & @code{embedded_init} calls to create
4781 such objects, and they will probably not be resizeable (so don't use
4782 the @dfn{safe} allocation variants). The trailing array idiom is used
4783 (rather than a pointer to an array of data), because, if we allow
4784 @code{NULL} to also represent an empty vector, empty vectors occupy
4785 minimal space in the structure containing them.
4787 Each operation that increases the number of active elements is
4788 available in @dfn{quick} and @dfn{safe} variants. The former presumes
4789 that there is sufficient allocated space for the operation to succeed
4790 (it dies if there is not). The latter will reallocate the vector, if
4791 needed. Reallocation causes an exponential increase in vector size.
4792 If you know you will be adding N elements, it would be more efficient
4793 to use the reserve operation before adding the elements with the
4794 @dfn{quick} operation. This will ensure there are at least as many
4795 elements as you ask for, it will exponentially increase if there are
4796 too few spare slots. If you want reserve a specific number of slots,
4797 but do not want the exponential increase (for instance, you know this
4798 is the last allocation), use a negative number for reservation. You
4799 can also create a vector of a specific size from the get go.
4801 You should prefer the push and pop operations, as they append and
4802 remove from the end of the vector. If you need to remove several items
4803 in one go, use the truncate operation. The insert and remove
4804 operations allow you to change elements in the middle of the vector.
4805 There are two remove operations, one which preserves the element
4806 ordering @code{ordered_remove}, and one which does not
4807 @code{unordered_remove}. The latter function copies the end element
4808 into the removed slot, rather than invoke a memmove operation. The
4809 @code{lower_bound} function will determine where to place an item in
4810 the array using insert that will maintain sorted order.
4812 If you need to directly manipulate a vector, then the @code{address}
4813 accessor will return the address of the start of the vector. Also the
4814 @code{space} predicate will tell you whether there is spare capacity in the
4815 vector. You will not normally need to use these two functions.
4817 Vector types are defined using a
4818 @code{DEF_VEC_@{O,P,I@}(@var{typename})} macro. Variables of vector
4819 type are declared using a @code{VEC(@var{typename})} macro. The
4820 characters @code{O}, @code{P} and @code{I} indicate whether
4821 @var{typename} is an object (@code{O}), pointer (@code{P}) or integral
4822 (@code{I}) type. Be careful to pick the correct one, as you'll get an
4823 awkward and inefficient API if you use the wrong one. There is a
4824 check, which results in a compile-time warning, for the @code{P} and
4825 @code{I} versions, but there is no check for the @code{O} versions, as
4826 that is not possible in plain C.
4828 An example of their use would be,
4831 DEF_VEC_P(tree); // non-managed tree vector.
4834 VEC(tree) *v; // A (pointer to) a vector of tree pointers.
4837 struct my_struct *s;
4839 if (VEC_length(tree, s->v)) @{ we have some contents @}
4840 VEC_safe_push(tree, s->v, decl); // append some decl onto the end
4841 for (ix = 0; VEC_iterate(tree, s->v, ix, elt); ix++)
4842 @{ do something with elt @}
4846 The @file{vec.h} file provides details on how to invoke the various
4847 accessors provided. They are enumerated here:
4851 Return the number of items in the array,
4854 Return true if the array has no elements.
4858 Return the last or arbitrary item in the array.
4861 Access an array element and indicate whether the array has been
4866 Create and destroy an array.
4868 @item VEC_embedded_size
4869 @itemx VEC_embedded_init
4870 Helpers for embedding an array as the final element of another struct.
4876 Return the amount of free space in an array.
4879 Ensure a certain amount of free space.
4881 @item VEC_quick_push
4882 @itemx VEC_safe_push
4883 Append to an array, either assuming the space is available, or making
4887 Remove the last item from an array.
4890 Remove several items from the end of an array.
4893 Add several items to the end of an array.
4896 Overwrite an item in the array.
4898 @item VEC_quick_insert
4899 @itemx VEC_safe_insert
4900 Insert an item into the middle of the array. Either the space must
4901 already exist, or the space is created.
4903 @item VEC_ordered_remove
4904 @itemx VEC_unordered_remove
4905 Remove an item from the array, preserving order or not.
4907 @item VEC_block_remove
4908 Remove a set of items from the array.
4911 Provide the address of the first element.
4913 @item VEC_lower_bound
4914 Binary search the array.
4924 This chapter covers topics that are lower-level than the major
4925 algorithms of @value{GDBN}.
4930 Cleanups are a structured way to deal with things that need to be done
4933 When your code does something (e.g., @code{xmalloc} some memory, or
4934 @code{open} a file) that needs to be undone later (e.g., @code{xfree}
4935 the memory or @code{close} the file), it can make a cleanup. The
4936 cleanup will be done at some future point: when the command is finished
4937 and control returns to the top level; when an error occurs and the stack
4938 is unwound; or when your code decides it's time to explicitly perform
4939 cleanups. Alternatively you can elect to discard the cleanups you
4945 @item struct cleanup *@var{old_chain};
4946 Declare a variable which will hold a cleanup chain handle.
4948 @findex make_cleanup
4949 @item @var{old_chain} = make_cleanup (@var{function}, @var{arg});
4950 Make a cleanup which will cause @var{function} to be called with
4951 @var{arg} (a @code{char *}) later. The result, @var{old_chain}, is a
4952 handle that can later be passed to @code{do_cleanups} or
4953 @code{discard_cleanups}. Unless you are going to call
4954 @code{do_cleanups} or @code{discard_cleanups}, you can ignore the result
4955 from @code{make_cleanup}.
4958 @item do_cleanups (@var{old_chain});
4959 Do all cleanups added to the chain since the corresponding
4960 @code{make_cleanup} call was made.
4962 @findex discard_cleanups
4963 @item discard_cleanups (@var{old_chain});
4964 Same as @code{do_cleanups} except that it just removes the cleanups from
4965 the chain and does not call the specified functions.
4968 Cleanups are implemented as a chain. The handle returned by
4969 @code{make_cleanups} includes the cleanup passed to the call and any
4970 later cleanups appended to the chain (but not yet discarded or
4974 make_cleanup (a, 0);
4976 struct cleanup *old = make_cleanup (b, 0);
4984 will call @code{c()} and @code{b()} but will not call @code{a()}. The
4985 cleanup that calls @code{a()} will remain in the cleanup chain, and will
4986 be done later unless otherwise discarded.@refill
4988 Your function should explicitly do or discard the cleanups it creates.
4989 Failing to do this leads to non-deterministic behavior since the caller
4990 will arbitrarily do or discard your functions cleanups. This need leads
4991 to two common cleanup styles.
4993 The first style is try/finally. Before it exits, your code-block calls
4994 @code{do_cleanups} with the old cleanup chain and thus ensures that your
4995 code-block's cleanups are always performed. For instance, the following
4996 code-segment avoids a memory leak problem (even when @code{error} is
4997 called and a forced stack unwind occurs) by ensuring that the
4998 @code{xfree} will always be called:
5001 struct cleanup *old = make_cleanup (null_cleanup, 0);
5002 data = xmalloc (sizeof blah);
5003 make_cleanup (xfree, data);
5008 The second style is try/except. Before it exits, your code-block calls
5009 @code{discard_cleanups} with the old cleanup chain and thus ensures that
5010 any created cleanups are not performed. For instance, the following
5011 code segment, ensures that the file will be closed but only if there is
5015 FILE *file = fopen ("afile", "r");
5016 struct cleanup *old = make_cleanup (close_file, file);
5018 discard_cleanups (old);
5022 Some functions, e.g., @code{fputs_filtered()} or @code{error()}, specify
5023 that they ``should not be called when cleanups are not in place''. This
5024 means that any actions you need to reverse in the case of an error or
5025 interruption must be on the cleanup chain before you call these
5026 functions, since they might never return to your code (they
5027 @samp{longjmp} instead).
5029 @section Per-architecture module data
5030 @cindex per-architecture module data
5031 @cindex multi-arch data
5032 @cindex data-pointer, per-architecture/per-module
5034 The multi-arch framework includes a mechanism for adding module
5035 specific per-architecture data-pointers to the @code{struct gdbarch}
5036 architecture object.
5038 A module registers one or more per-architecture data-pointers using:
5040 @deftypefun struct gdbarch_data *gdbarch_data_register_pre_init (gdbarch_data_pre_init_ftype *@var{pre_init})
5041 @var{pre_init} is used to, on-demand, allocate an initial value for a
5042 per-architecture data-pointer using the architecture's obstack (passed
5043 in as a parameter). Since @var{pre_init} can be called during
5044 architecture creation, it is not parameterized with the architecture.
5045 and must not call modules that use per-architecture data.
5048 @deftypefun struct gdbarch_data *gdbarch_data_register_post_init (gdbarch_data_post_init_ftype *@var{post_init})
5049 @var{post_init} is used to obtain an initial value for a
5050 per-architecture data-pointer @emph{after}. Since @var{post_init} is
5051 always called after architecture creation, it both receives the fully
5052 initialized architecture and is free to call modules that use
5053 per-architecture data (care needs to be taken to ensure that those
5054 other modules do not try to call back to this module as that will
5055 create in cycles in the initialization call graph).
5058 These functions return a @code{struct gdbarch_data} that is used to
5059 identify the per-architecture data-pointer added for that module.
5061 The per-architecture data-pointer is accessed using the function:
5063 @deftypefun void *gdbarch_data (struct gdbarch *@var{gdbarch}, struct gdbarch_data *@var{data_handle})
5064 Given the architecture @var{arch} and module data handle
5065 @var{data_handle} (returned by @code{gdbarch_data_register_pre_init}
5066 or @code{gdbarch_data_register_post_init}), this function returns the
5067 current value of the per-architecture data-pointer. If the data
5068 pointer is @code{NULL}, it is first initialized by calling the
5069 corresponding @var{pre_init} or @var{post_init} method.
5072 The examples below assume the following definitions:
5075 struct nozel @{ int total; @};
5076 static struct gdbarch_data *nozel_handle;
5079 A module can extend the architecture vector, adding additional
5080 per-architecture data, using the @var{pre_init} method. The module's
5081 per-architecture data is then initialized during architecture
5084 In the below, the module's per-architecture @emph{nozel} is added. An
5085 architecture can specify its nozel by calling @code{set_gdbarch_nozel}
5086 from @code{gdbarch_init}.
5090 nozel_pre_init (struct obstack *obstack)
5092 struct nozel *data = OBSTACK_ZALLOC (obstack, struct nozel);
5099 set_gdbarch_nozel (struct gdbarch *gdbarch, int total)
5101 struct nozel *data = gdbarch_data (gdbarch, nozel_handle);
5102 data->total = nozel;
5106 A module can on-demand create architecture dependant data structures
5107 using @code{post_init}.
5109 In the below, the nozel's total is computed on-demand by
5110 @code{nozel_post_init} using information obtained from the
5115 nozel_post_init (struct gdbarch *gdbarch)
5117 struct nozel *data = GDBARCH_OBSTACK_ZALLOC (gdbarch, struct nozel);
5118 nozel->total = gdbarch@dots{} (gdbarch);
5125 nozel_total (struct gdbarch *gdbarch)
5127 struct nozel *data = gdbarch_data (gdbarch, nozel_handle);
5132 @section Wrapping Output Lines
5133 @cindex line wrap in output
5136 Output that goes through @code{printf_filtered} or @code{fputs_filtered}
5137 or @code{fputs_demangled} needs only to have calls to @code{wrap_here}
5138 added in places that would be good breaking points. The utility
5139 routines will take care of actually wrapping if the line width is
5142 The argument to @code{wrap_here} is an indentation string which is
5143 printed @emph{only} if the line breaks there. This argument is saved
5144 away and used later. It must remain valid until the next call to
5145 @code{wrap_here} or until a newline has been printed through the
5146 @code{*_filtered} functions. Don't pass in a local variable and then
5149 It is usually best to call @code{wrap_here} after printing a comma or
5150 space. If you call it before printing a space, make sure that your
5151 indentation properly accounts for the leading space that will print if
5152 the line wraps there.
5154 Any function or set of functions that produce filtered output must
5155 finish by printing a newline, to flush the wrap buffer, before switching
5156 to unfiltered (@code{printf}) output. Symbol reading routines that
5157 print warnings are a good example.
5159 @section @value{GDBN} Coding Standards
5160 @cindex coding standards
5162 @value{GDBN} follows the GNU coding standards, as described in
5163 @file{etc/standards.texi}. This file is also available for anonymous
5164 FTP from GNU archive sites. @value{GDBN} takes a strict interpretation
5165 of the standard; in general, when the GNU standard recommends a practice
5166 but does not require it, @value{GDBN} requires it.
5168 @value{GDBN} follows an additional set of coding standards specific to
5169 @value{GDBN}, as described in the following sections.
5174 @value{GDBN} assumes an ISO/IEC 9899:1990 (a.k.a.@: ISO C90) compliant
5177 @value{GDBN} does not assume an ISO C or POSIX compliant C library.
5180 @subsection Memory Management
5182 @value{GDBN} does not use the functions @code{malloc}, @code{realloc},
5183 @code{calloc}, @code{free} and @code{asprintf}.
5185 @value{GDBN} uses the functions @code{xmalloc}, @code{xrealloc} and
5186 @code{xcalloc} when allocating memory. Unlike @code{malloc} et.al.@:
5187 these functions do not return when the memory pool is empty. Instead,
5188 they unwind the stack using cleanups. These functions return
5189 @code{NULL} when requested to allocate a chunk of memory of size zero.
5191 @emph{Pragmatics: By using these functions, the need to check every
5192 memory allocation is removed. These functions provide portable
5195 @value{GDBN} does not use the function @code{free}.
5197 @value{GDBN} uses the function @code{xfree} to return memory to the
5198 memory pool. Consistent with ISO-C, this function ignores a request to
5199 free a @code{NULL} pointer.
5201 @emph{Pragmatics: On some systems @code{free} fails when passed a
5202 @code{NULL} pointer.}
5204 @value{GDBN} can use the non-portable function @code{alloca} for the
5205 allocation of small temporary values (such as strings).
5207 @emph{Pragmatics: This function is very non-portable. Some systems
5208 restrict the memory being allocated to no more than a few kilobytes.}
5210 @value{GDBN} uses the string function @code{xstrdup} and the print
5211 function @code{xstrprintf}.
5213 @emph{Pragmatics: @code{asprintf} and @code{strdup} can fail. Print
5214 functions such as @code{sprintf} are very prone to buffer overflow
5218 @subsection Compiler Warnings
5219 @cindex compiler warnings
5221 With few exceptions, developers should avoid the configuration option
5222 @samp{--disable-werror} when building @value{GDBN}. The exceptions
5223 are listed in the file @file{gdb/MAINTAINERS}. The default, when
5224 building with @sc{gcc}, is @samp{--enable-werror}.
5226 This option causes @value{GDBN} (when built using GCC) to be compiled
5227 with a carefully selected list of compiler warning flags. Any warnings
5228 from those flags are treated as errors.
5230 The current list of warning flags includes:
5234 Recommended @sc{gcc} warnings.
5236 @item -Wdeclaration-after-statement
5238 @sc{gcc} 3.x (and later) and @sc{c99} allow declarations mixed with
5239 code, but @sc{gcc} 2.x and @sc{c89} do not.
5241 @item -Wpointer-arith
5243 @item -Wformat-nonliteral
5244 Non-literal format strings, with a few exceptions, are bugs - they
5245 might contain unintended user-supplied format specifiers.
5246 Since @value{GDBN} uses the @code{format printf} attribute on all
5247 @code{printf} like functions this checks not just @code{printf} calls
5248 but also calls to functions such as @code{fprintf_unfiltered}.
5250 @item -Wno-pointer-sign
5251 In version 4.0, GCC began warning about pointer argument passing or
5252 assignment even when the source and destination differed only in
5253 signedness. However, most @value{GDBN} code doesn't distinguish
5254 carefully between @code{char} and @code{unsigned char}. In early 2006
5255 the @value{GDBN} developers decided correcting these warnings wasn't
5256 worth the time it would take.
5258 @item -Wno-unused-parameter
5259 Due to the way that @value{GDBN} is implemented many functions have
5260 unused parameters. Consequently this warning is avoided. The macro
5261 @code{ATTRIBUTE_UNUSED} is not used as it leads to false negatives ---
5262 it is not an error to have @code{ATTRIBUTE_UNUSED} on a parameter that
5267 @itemx -Wno-char-subscripts
5268 These are warnings which might be useful for @value{GDBN}, but are
5269 currently too noisy to enable with @samp{-Werror}.
5273 @subsection Formatting
5275 @cindex source code formatting
5276 The standard GNU recommendations for formatting must be followed
5279 A function declaration should not have its name in column zero. A
5280 function definition should have its name in column zero.
5284 static void foo (void);
5292 @emph{Pragmatics: This simplifies scripting. Function definitions can
5293 be found using @samp{^function-name}.}
5295 There must be a space between a function or macro name and the opening
5296 parenthesis of its argument list (except for macro definitions, as
5297 required by C). There must not be a space after an open paren/bracket
5298 or before a close paren/bracket.
5300 While additional whitespace is generally helpful for reading, do not use
5301 more than one blank line to separate blocks, and avoid adding whitespace
5302 after the end of a program line (as of 1/99, some 600 lines had
5303 whitespace after the semicolon). Excess whitespace causes difficulties
5304 for @code{diff} and @code{patch} utilities.
5306 Pointers are declared using the traditional K&R C style:
5320 @subsection Comments
5322 @cindex comment formatting
5323 The standard GNU requirements on comments must be followed strictly.
5325 Block comments must appear in the following form, with no @code{/*}- or
5326 @code{*/}-only lines, and no leading @code{*}:
5329 /* Wait for control to return from inferior to debugger. If inferior
5330 gets a signal, we may decide to start it up again instead of
5331 returning. That is why there is a loop in this function. When
5332 this function actually returns it means the inferior should be left
5333 stopped and @value{GDBN} should read more commands. */
5336 (Note that this format is encouraged by Emacs; tabbing for a multi-line
5337 comment works correctly, and @kbd{M-q} fills the block consistently.)
5339 Put a blank line between the block comments preceding function or
5340 variable definitions, and the definition itself.
5342 In general, put function-body comments on lines by themselves, rather
5343 than trying to fit them into the 20 characters left at the end of a
5344 line, since either the comment or the code will inevitably get longer
5345 than will fit, and then somebody will have to move it anyhow.
5349 @cindex C data types
5350 Code must not depend on the sizes of C data types, the format of the
5351 host's floating point numbers, the alignment of anything, or the order
5352 of evaluation of expressions.
5354 @cindex function usage
5355 Use functions freely. There are only a handful of compute-bound areas
5356 in @value{GDBN} that might be affected by the overhead of a function
5357 call, mainly in symbol reading. Most of @value{GDBN}'s performance is
5358 limited by the target interface (whether serial line or system call).
5360 However, use functions with moderation. A thousand one-line functions
5361 are just as hard to understand as a single thousand-line function.
5363 @emph{Macros are bad, M'kay.}
5364 (But if you have to use a macro, make sure that the macro arguments are
5365 protected with parentheses.)
5369 Declarations like @samp{struct foo *} should be used in preference to
5370 declarations like @samp{typedef struct foo @{ @dots{} @} *foo_ptr}.
5373 @subsection Function Prototypes
5374 @cindex function prototypes
5376 Prototypes must be used when both @emph{declaring} and @emph{defining}
5377 a function. Prototypes for @value{GDBN} functions must include both the
5378 argument type and name, with the name matching that used in the actual
5379 function definition.
5381 All external functions should have a declaration in a header file that
5382 callers include, except for @code{_initialize_*} functions, which must
5383 be external so that @file{init.c} construction works, but shouldn't be
5384 visible to random source files.
5386 Where a source file needs a forward declaration of a static function,
5387 that declaration must appear in a block near the top of the source file.
5390 @subsection Internal Error Recovery
5392 During its execution, @value{GDBN} can encounter two types of errors.
5393 User errors and internal errors. User errors include not only a user
5394 entering an incorrect command but also problems arising from corrupt
5395 object files and system errors when interacting with the target.
5396 Internal errors include situations where @value{GDBN} has detected, at
5397 run time, a corrupt or erroneous situation.
5399 When reporting an internal error, @value{GDBN} uses
5400 @code{internal_error} and @code{gdb_assert}.
5402 @value{GDBN} must not call @code{abort} or @code{assert}.
5404 @emph{Pragmatics: There is no @code{internal_warning} function. Either
5405 the code detected a user error, recovered from it and issued a
5406 @code{warning} or the code failed to correctly recover from the user
5407 error and issued an @code{internal_error}.}
5409 @subsection File Names
5411 Any file used when building the core of @value{GDBN} must be in lower
5412 case. Any file used when building the core of @value{GDBN} must be 8.3
5413 unique. These requirements apply to both source and generated files.
5415 @emph{Pragmatics: The core of @value{GDBN} must be buildable on many
5416 platforms including DJGPP and MacOS/HFS. Every time an unfriendly file
5417 is introduced to the build process both @file{Makefile.in} and
5418 @file{configure.in} need to be modified accordingly. Compare the
5419 convoluted conversion process needed to transform @file{COPYING} into
5420 @file{copying.c} with the conversion needed to transform
5421 @file{version.in} into @file{version.c}.}
5423 Any file non 8.3 compliant file (that is not used when building the core
5424 of @value{GDBN}) must be added to @file{gdb/config/djgpp/fnchange.lst}.
5426 @emph{Pragmatics: This is clearly a compromise.}
5428 When @value{GDBN} has a local version of a system header file (ex
5429 @file{string.h}) the file name based on the POSIX header prefixed with
5430 @file{gdb_} (@file{gdb_string.h}). These headers should be relatively
5431 independent: they should use only macros defined by @file{configure},
5432 the compiler, or the host; they should include only system headers; they
5433 should refer only to system types. They may be shared between multiple
5434 programs, e.g.@: @value{GDBN} and @sc{gdbserver}.
5436 For other files @samp{-} is used as the separator.
5439 @subsection Include Files
5441 A @file{.c} file should include @file{defs.h} first.
5443 A @file{.c} file should directly include the @code{.h} file of every
5444 declaration and/or definition it directly refers to. It cannot rely on
5447 A @file{.h} file should directly include the @code{.h} file of every
5448 declaration and/or definition it directly refers to. It cannot rely on
5449 indirect inclusion. Exception: The file @file{defs.h} does not need to
5450 be directly included.
5452 An external declaration should only appear in one include file.
5454 An external declaration should never appear in a @code{.c} file.
5455 Exception: a declaration for the @code{_initialize} function that
5456 pacifies @option{-Wmissing-declaration}.
5458 A @code{typedef} definition should only appear in one include file.
5460 An opaque @code{struct} declaration can appear in multiple @file{.h}
5461 files. Where possible, a @file{.h} file should use an opaque
5462 @code{struct} declaration instead of an include.
5464 All @file{.h} files should be wrapped in:
5467 #ifndef INCLUDE_FILE_NAME_H
5468 #define INCLUDE_FILE_NAME_H
5474 @subsection Clean Design and Portable Implementation
5477 In addition to getting the syntax right, there's the little question of
5478 semantics. Some things are done in certain ways in @value{GDBN} because long
5479 experience has shown that the more obvious ways caused various kinds of
5482 @cindex assumptions about targets
5483 You can't assume the byte order of anything that comes from a target
5484 (including @var{value}s, object files, and instructions). Such things
5485 must be byte-swapped using @code{SWAP_TARGET_AND_HOST} in
5486 @value{GDBN}, or one of the swap routines defined in @file{bfd.h},
5487 such as @code{bfd_get_32}.
5489 You can't assume that you know what interface is being used to talk to
5490 the target system. All references to the target must go through the
5491 current @code{target_ops} vector.
5493 You can't assume that the host and target machines are the same machine
5494 (except in the ``native'' support modules). In particular, you can't
5495 assume that the target machine's header files will be available on the
5496 host machine. Target code must bring along its own header files --
5497 written from scratch or explicitly donated by their owner, to avoid
5501 Insertion of new @code{#ifdef}'s will be frowned upon. It's much better
5502 to write the code portably than to conditionalize it for various
5505 @cindex system dependencies
5506 New @code{#ifdef}'s which test for specific compilers or manufacturers
5507 or operating systems are unacceptable. All @code{#ifdef}'s should test
5508 for features. The information about which configurations contain which
5509 features should be segregated into the configuration files. Experience
5510 has proven far too often that a feature unique to one particular system
5511 often creeps into other systems; and that a conditional based on some
5512 predefined macro for your current system will become worthless over
5513 time, as new versions of your system come out that behave differently
5514 with regard to this feature.
5516 Adding code that handles specific architectures, operating systems,
5517 target interfaces, or hosts, is not acceptable in generic code.
5519 @cindex portable file name handling
5520 @cindex file names, portability
5521 One particularly notorious area where system dependencies tend to
5522 creep in is handling of file names. The mainline @value{GDBN} code
5523 assumes Posix semantics of file names: absolute file names begin with
5524 a forward slash @file{/}, slashes are used to separate leading
5525 directories, case-sensitive file names. These assumptions are not
5526 necessarily true on non-Posix systems such as MS-Windows. To avoid
5527 system-dependent code where you need to take apart or construct a file
5528 name, use the following portable macros:
5531 @findex HAVE_DOS_BASED_FILE_SYSTEM
5532 @item HAVE_DOS_BASED_FILE_SYSTEM
5533 This preprocessing symbol is defined to a non-zero value on hosts
5534 whose filesystems belong to the MS-DOS/MS-Windows family. Use this
5535 symbol to write conditional code which should only be compiled for
5538 @findex IS_DIR_SEPARATOR
5539 @item IS_DIR_SEPARATOR (@var{c})
5540 Evaluates to a non-zero value if @var{c} is a directory separator
5541 character. On Unix and GNU/Linux systems, only a slash @file{/} is
5542 such a character, but on Windows, both @file{/} and @file{\} will
5545 @findex IS_ABSOLUTE_PATH
5546 @item IS_ABSOLUTE_PATH (@var{file})
5547 Evaluates to a non-zero value if @var{file} is an absolute file name.
5548 For Unix and GNU/Linux hosts, a name which begins with a slash
5549 @file{/} is absolute. On DOS and Windows, @file{d:/foo} and
5550 @file{x:\bar} are also absolute file names.
5552 @findex FILENAME_CMP
5553 @item FILENAME_CMP (@var{f1}, @var{f2})
5554 Calls a function which compares file names @var{f1} and @var{f2} as
5555 appropriate for the underlying host filesystem. For Posix systems,
5556 this simply calls @code{strcmp}; on case-insensitive filesystems it
5557 will call @code{strcasecmp} instead.
5559 @findex DIRNAME_SEPARATOR
5560 @item DIRNAME_SEPARATOR
5561 Evaluates to a character which separates directories in
5562 @code{PATH}-style lists, typically held in environment variables.
5563 This character is @samp{:} on Unix, @samp{;} on DOS and Windows.
5565 @findex SLASH_STRING
5567 This evaluates to a constant string you should use to produce an
5568 absolute filename from leading directories and the file's basename.
5569 @code{SLASH_STRING} is @code{"/"} on most systems, but might be
5570 @code{"\\"} for some Windows-based ports.
5573 In addition to using these macros, be sure to use portable library
5574 functions whenever possible. For example, to extract a directory or a
5575 basename part from a file name, use the @code{dirname} and
5576 @code{basename} library functions (available in @code{libiberty} for
5577 platforms which don't provide them), instead of searching for a slash
5578 with @code{strrchr}.
5580 Another way to generalize @value{GDBN} along a particular interface is with an
5581 attribute struct. For example, @value{GDBN} has been generalized to handle
5582 multiple kinds of remote interfaces---not by @code{#ifdef}s everywhere, but
5583 by defining the @code{target_ops} structure and having a current target (as
5584 well as a stack of targets below it, for memory references). Whenever
5585 something needs to be done that depends on which remote interface we are
5586 using, a flag in the current target_ops structure is tested (e.g.,
5587 @code{target_has_stack}), or a function is called through a pointer in the
5588 current target_ops structure. In this way, when a new remote interface
5589 is added, only one module needs to be touched---the one that actually
5590 implements the new remote interface. Other examples of
5591 attribute-structs are BFD access to multiple kinds of object file
5592 formats, or @value{GDBN}'s access to multiple source languages.
5594 Please avoid duplicating code. For example, in @value{GDBN} 3.x all
5595 the code interfacing between @code{ptrace} and the rest of
5596 @value{GDBN} was duplicated in @file{*-dep.c}, and so changing
5597 something was very painful. In @value{GDBN} 4.x, these have all been
5598 consolidated into @file{infptrace.c}. @file{infptrace.c} can deal
5599 with variations between systems the same way any system-independent
5600 file would (hooks, @code{#if defined}, etc.), and machines which are
5601 radically different don't need to use @file{infptrace.c} at all.
5603 All debugging code must be controllable using the @samp{set debug
5604 @var{module}} command. Do not use @code{printf} to print trace
5605 messages. Use @code{fprintf_unfiltered(gdb_stdlog, ...}. Do not use
5606 @code{#ifdef DEBUG}.
5611 @chapter Porting @value{GDBN}
5612 @cindex porting to new machines
5614 Most of the work in making @value{GDBN} compile on a new machine is in
5615 specifying the configuration of the machine. This is done in a
5616 dizzying variety of header files and configuration scripts, which we
5617 hope to make more sensible soon. Let's say your new host is called an
5618 @var{xyz} (e.g., @samp{sun4}), and its full three-part configuration
5619 name is @code{@var{arch}-@var{xvend}-@var{xos}} (e.g.,
5620 @samp{sparc-sun-sunos4}). In particular:
5624 In the top level directory, edit @file{config.sub} and add @var{arch},
5625 @var{xvend}, and @var{xos} to the lists of supported architectures,
5626 vendors, and operating systems near the bottom of the file. Also, add
5627 @var{xyz} as an alias that maps to
5628 @code{@var{arch}-@var{xvend}-@var{xos}}. You can test your changes by
5632 ./config.sub @var{xyz}
5639 ./config.sub @code{@var{arch}-@var{xvend}-@var{xos}}
5643 which should both respond with @code{@var{arch}-@var{xvend}-@var{xos}}
5644 and no error messages.
5647 You need to port BFD, if that hasn't been done already. Porting BFD is
5648 beyond the scope of this manual.
5651 To configure @value{GDBN} itself, edit @file{gdb/configure.host} to recognize
5652 your system and set @code{gdb_host} to @var{xyz}, and (unless your
5653 desired target is already available) also edit @file{gdb/configure.tgt},
5654 setting @code{gdb_target} to something appropriate (for instance,
5657 @emph{Maintainer's note: Work in progress. The file
5658 @file{gdb/configure.host} originally needed to be modified when either a
5659 new native target or a new host machine was being added to @value{GDBN}.
5660 Recent changes have removed this requirement. The file now only needs
5661 to be modified when adding a new native configuration. This will likely
5662 changed again in the future.}
5665 Finally, you'll need to specify and define @value{GDBN}'s host-, native-, and
5666 target-dependent @file{.h} and @file{.c} files used for your
5670 @node Versions and Branches
5671 @chapter Versions and Branches
5675 @value{GDBN}'s version is determined by the file
5676 @file{gdb/version.in} and takes one of the following forms:
5679 @item @var{major}.@var{minor}
5680 @itemx @var{major}.@var{minor}.@var{patchlevel}
5681 an official release (e.g., 6.2 or 6.2.1)
5682 @item @var{major}.@var{minor}.@var{patchlevel}.@var{YYYY}@var{MM}@var{DD}
5683 a snapshot taken at @var{YYYY}-@var{MM}-@var{DD}-gmt (e.g.,
5684 6.1.50.20020302, 6.1.90.20020304, or 6.1.0.20020308)
5685 @item @var{major}.@var{minor}.@var{patchlevel}.@var{YYYY}@var{MM}@var{DD}-cvs
5686 a @sc{cvs} check out drawn on @var{YYYY}-@var{MM}-@var{DD} (e.g.,
5687 6.1.50.20020302-cvs, 6.1.90.20020304-cvs, or 6.1.0.20020308-cvs)
5688 @item @var{major}.@var{minor}.@var{patchlevel}.@var{YYYY}@var{MM}@var{DD} (@var{vendor})
5689 a vendor specific release of @value{GDBN}, that while based on@*
5690 @var{major}.@var{minor}.@var{patchlevel}.@var{YYYY}@var{MM}@var{DD},
5691 may include additional changes
5694 @value{GDBN}'s mainline uses the @var{major} and @var{minor} version
5695 numbers from the most recent release branch, with a @var{patchlevel}
5696 of 50. At the time each new release branch is created, the mainline's
5697 @var{major} and @var{minor} version numbers are updated.
5699 @value{GDBN}'s release branch is similar. When the branch is cut, the
5700 @var{patchlevel} is changed from 50 to 90. As draft releases are
5701 drawn from the branch, the @var{patchlevel} is incremented. Once the
5702 first release (@var{major}.@var{minor}) has been made, the
5703 @var{patchlevel} is set to 0 and updates have an incremented
5706 For snapshots, and @sc{cvs} check outs, it is also possible to
5707 identify the @sc{cvs} origin:
5710 @item @var{major}.@var{minor}.50.@var{YYYY}@var{MM}@var{DD}
5711 drawn from the @sc{head} of mainline @sc{cvs} (e.g., 6.1.50.20020302)
5712 @item @var{major}.@var{minor}.90.@var{YYYY}@var{MM}@var{DD}
5713 @itemx @var{major}.@var{minor}.91.@var{YYYY}@var{MM}@var{DD} @dots{}
5714 drawn from a release branch prior to the release (e.g.,
5716 @item @var{major}.@var{minor}.0.@var{YYYY}@var{MM}@var{DD}
5717 @itemx @var{major}.@var{minor}.1.@var{YYYY}@var{MM}@var{DD} @dots{}
5718 drawn from a release branch after the release (e.g., 6.2.0.20020308)
5721 If the previous @value{GDBN} version is 6.1 and the current version is
5722 6.2, then, substituting 6 for @var{major} and 1 or 2 for @var{minor},
5723 here's an illustration of a typical sequence:
5730 +--------------------------.
5733 6.2.50.20020303-cvs 6.1.90 (draft #1)
5735 6.2.50.20020304-cvs 6.1.90.20020304-cvs
5737 6.2.50.20020305-cvs 6.1.91 (draft #2)
5739 6.2.50.20020306-cvs 6.1.91.20020306-cvs
5741 6.2.50.20020307-cvs 6.2 (release)
5743 6.2.50.20020308-cvs 6.2.0.20020308-cvs
5745 6.2.50.20020309-cvs 6.2.1 (update)
5747 6.2.50.20020310-cvs <branch closed>
5751 +--------------------------.
5754 6.3.50.20020312-cvs 6.2.90 (draft #1)
5758 @section Release Branches
5759 @cindex Release Branches
5761 @value{GDBN} draws a release series (6.2, 6.2.1, @dots{}) from a
5762 single release branch, and identifies that branch using the @sc{cvs}
5766 gdb_@var{major}_@var{minor}-@var{YYYY}@var{MM}@var{DD}-branchpoint
5767 gdb_@var{major}_@var{minor}-branch
5768 gdb_@var{major}_@var{minor}-@var{YYYY}@var{MM}@var{DD}-release
5771 @emph{Pragmatics: To help identify the date at which a branch or
5772 release is made, both the branchpoint and release tags include the
5773 date that they are cut (@var{YYYY}@var{MM}@var{DD}) in the tag. The
5774 branch tag, denoting the head of the branch, does not need this.}
5776 @section Vendor Branches
5777 @cindex vendor branches
5779 To avoid version conflicts, vendors are expected to modify the file
5780 @file{gdb/version.in} to include a vendor unique alphabetic identifier
5781 (an official @value{GDBN} release never uses alphabetic characters in
5782 its version identifier). E.g., @samp{6.2widgit2}, or @samp{6.2 (Widgit
5785 @section Experimental Branches
5786 @cindex experimental branches
5788 @subsection Guidelines
5790 @value{GDBN} permits the creation of branches, cut from the @sc{cvs}
5791 repository, for experimental development. Branches make it possible
5792 for developers to share preliminary work, and maintainers to examine
5793 significant new developments.
5795 The following are a set of guidelines for creating such branches:
5799 @item a branch has an owner
5800 The owner can set further policy for a branch, but may not change the
5801 ground rules. In particular, they can set a policy for commits (be it
5802 adding more reviewers or deciding who can commit).
5804 @item all commits are posted
5805 All changes committed to a branch shall also be posted to
5806 @email{gdb-patches@@sources.redhat.com, the @value{GDBN} patches
5807 mailing list}. While commentary on such changes are encouraged, people
5808 should remember that the changes only apply to a branch.
5810 @item all commits are covered by an assignment
5811 This ensures that all changes belong to the Free Software Foundation,
5812 and avoids the possibility that the branch may become contaminated.
5814 @item a branch is focused
5815 A focused branch has a single objective or goal, and does not contain
5816 unnecessary or irrelevant changes. Cleanups, where identified, being
5817 be pushed into the mainline as soon as possible.
5819 @item a branch tracks mainline
5820 This keeps the level of divergence under control. It also keeps the
5821 pressure on developers to push cleanups and other stuff into the
5824 @item a branch shall contain the entire @value{GDBN} module
5825 The @value{GDBN} module @code{gdb} should be specified when creating a
5826 branch (branches of individual files should be avoided). @xref{Tags}.
5828 @item a branch shall be branded using @file{version.in}
5829 The file @file{gdb/version.in} shall be modified so that it identifies
5830 the branch @var{owner} and branch @var{name}, e.g.,
5831 @samp{6.2.50.20030303_owner_name} or @samp{6.2 (Owner Name)}.
5838 To simplify the identification of @value{GDBN} branches, the following
5839 branch tagging convention is strongly recommended:
5843 @item @var{owner}_@var{name}-@var{YYYYMMDD}-branchpoint
5844 @itemx @var{owner}_@var{name}-@var{YYYYMMDD}-branch
5845 The branch point and corresponding branch tag. @var{YYYYMMDD} is the
5846 date that the branch was created. A branch is created using the
5847 sequence: @anchor{experimental branch tags}
5849 cvs rtag @var{owner}_@var{name}-@var{YYYYMMDD}-branchpoint gdb
5850 cvs rtag -b -r @var{owner}_@var{name}-@var{YYYYMMDD}-branchpoint \
5851 @var{owner}_@var{name}-@var{YYYYMMDD}-branch gdb
5854 @item @var{owner}_@var{name}-@var{yyyymmdd}-mergepoint
5855 The tagged point, on the mainline, that was used when merging the branch
5856 on @var{yyyymmdd}. To merge in all changes since the branch was cut,
5857 use a command sequence like:
5859 cvs rtag @var{owner}_@var{name}-@var{yyyymmdd}-mergepoint gdb
5861 -j@var{owner}_@var{name}-@var{YYYYMMDD}-branchpoint
5862 -j@var{owner}_@var{name}-@var{yyyymmdd}-mergepoint
5865 Similar sequences can be used to just merge in changes since the last
5871 For further information on @sc{cvs}, see
5872 @uref{http://www.gnu.org/software/cvs/, Concurrent Versions System}.
5874 @node Start of New Year Procedure
5875 @chapter Start of New Year Procedure
5876 @cindex new year procedure
5878 At the start of each new year, the following actions should be performed:
5882 Rotate the ChangeLog file
5884 The current @file{ChangeLog} file should be renamed into
5885 @file{ChangeLog-YYYY} where YYYY is the year that has just passed.
5886 A new @file{ChangeLog} file should be created, and its contents should
5887 contain a reference to the previous ChangeLog. The following should
5888 also be preserved at the end of the new ChangeLog, in order to provide
5889 the appropriate settings when editing this file with Emacs:
5895 version-control: never
5900 Add an entry for the newly created ChangeLog file (@file{ChangeLog-YYYY})
5901 in @file{gdb/config/djgpp/fnchange.lst}.
5904 Update the copyright year in the startup message
5906 Update the copyright year in file @file{top.c}, function
5907 @code{print_gdb_version}.
5910 Add the new year in the copyright notices of all source and documentation
5911 files. This can be done semi-automatically by running the @code{copyright.sh}
5912 script. This script requires Emacs 22 or later to be installed.
5918 @chapter Releasing @value{GDBN}
5919 @cindex making a new release of gdb
5921 @section Branch Commit Policy
5923 The branch commit policy is pretty slack. @value{GDBN} releases 5.0,
5924 5.1 and 5.2 all used the below:
5928 The @file{gdb/MAINTAINERS} file still holds.
5930 Don't fix something on the branch unless/until it is also fixed in the
5931 trunk. If this isn't possible, mentioning it in the @file{gdb/PROBLEMS}
5932 file is better than committing a hack.
5934 When considering a patch for the branch, suggested criteria include:
5935 Does it fix a build? Does it fix the sequence @kbd{break main; run}
5936 when debugging a static binary?
5938 The further a change is from the core of @value{GDBN}, the less likely
5939 the change will worry anyone (e.g., target specific code).
5941 Only post a proposal to change the core of @value{GDBN} after you've
5942 sent individual bribes to all the people listed in the
5943 @file{MAINTAINERS} file @t{;-)}
5946 @emph{Pragmatics: Provided updates are restricted to non-core
5947 functionality there is little chance that a broken change will be fatal.
5948 This means that changes such as adding a new architectures or (within
5949 reason) support for a new host are considered acceptable.}
5952 @section Obsoleting code
5954 Before anything else, poke the other developers (and around the source
5955 code) to see if there is anything that can be removed from @value{GDBN}
5956 (an old target, an unused file).
5958 Obsolete code is identified by adding an @code{OBSOLETE} prefix to every
5959 line. Doing this means that it is easy to identify something that has
5960 been obsoleted when greping through the sources.
5962 The process is done in stages --- this is mainly to ensure that the
5963 wider @value{GDBN} community has a reasonable opportunity to respond.
5964 Remember, everything on the Internet takes a week.
5968 Post the proposal on @email{gdb@@sources.redhat.com, the GDB mailing
5969 list} Creating a bug report to track the task's state, is also highly
5974 Post the proposal on @email{gdb-announce@@sources.redhat.com, the GDB
5975 Announcement mailing list}.
5979 Go through and edit all relevant files and lines so that they are
5980 prefixed with the word @code{OBSOLETE}.
5982 Wait until the next GDB version, containing this obsolete code, has been
5985 Remove the obsolete code.
5989 @emph{Maintainer note: While removing old code is regrettable it is
5990 hopefully better for @value{GDBN}'s long term development. Firstly it
5991 helps the developers by removing code that is either no longer relevant
5992 or simply wrong. Secondly since it removes any history associated with
5993 the file (effectively clearing the slate) the developer has a much freer
5994 hand when it comes to fixing broken files.}
5998 @section Before the Branch
6000 The most important objective at this stage is to find and fix simple
6001 changes that become a pain to track once the branch is created. For
6002 instance, configuration problems that stop @value{GDBN} from even
6003 building. If you can't get the problem fixed, document it in the
6004 @file{gdb/PROBLEMS} file.
6006 @subheading Prompt for @file{gdb/NEWS}
6008 People always forget. Send a post reminding them but also if you know
6009 something interesting happened add it yourself. The @code{schedule}
6010 script will mention this in its e-mail.
6012 @subheading Review @file{gdb/README}
6014 Grab one of the nightly snapshots and then walk through the
6015 @file{gdb/README} looking for anything that can be improved. The
6016 @code{schedule} script will mention this in its e-mail.
6018 @subheading Refresh any imported files.
6020 A number of files are taken from external repositories. They include:
6024 @file{texinfo/texinfo.tex}
6026 @file{config.guess} et.@: al.@: (see the top-level @file{MAINTAINERS}
6029 @file{etc/standards.texi}, @file{etc/make-stds.texi}
6032 @subheading Check the ARI
6034 @uref{http://sources.redhat.com/gdb/ari,,A.R.I.} is an @code{awk} script
6035 (Awk Regression Index ;-) that checks for a number of errors and coding
6036 conventions. The checks include things like using @code{malloc} instead
6037 of @code{xmalloc} and file naming problems. There shouldn't be any
6040 @subsection Review the bug data base
6042 Close anything obviously fixed.
6044 @subsection Check all cross targets build
6046 The targets are listed in @file{gdb/MAINTAINERS}.
6049 @section Cut the Branch
6051 @subheading Create the branch
6056 $ V=`echo $v | sed 's/\./_/g'`
6057 $ D=`date -u +%Y-%m-%d`
6060 $ echo cvs -f -d :ext:sources.redhat.com:/cvs/src rtag \
6061 -D $D-gmt gdb_$V-$D-branchpoint insight
6062 cvs -f -d :ext:sources.redhat.com:/cvs/src rtag
6063 -D 2002-03-03-gmt gdb_5_2-2002-03-03-branchpoint insight
6066 $ echo cvs -f -d :ext:sources.redhat.com:/cvs/src rtag \
6067 -b -r gdb_$V-$D-branchpoint gdb_$V-branch insight
6068 cvs -f -d :ext:sources.redhat.com:/cvs/src rtag \
6069 -b -r gdb_5_2-2002-03-03-branchpoint gdb_5_2-branch insight
6077 By using @kbd{-D YYYY-MM-DD-gmt}, the branch is forced to an exact
6080 The trunk is first tagged so that the branch point can easily be found.
6082 Insight, which includes @value{GDBN}, is tagged at the same time.
6084 @file{version.in} gets bumped to avoid version number conflicts.
6086 The reading of @file{.cvsrc} is disabled using @file{-f}.
6089 @subheading Update @file{version.in}
6094 $ V=`echo $v | sed 's/\./_/g'`
6098 $ echo cvs -f -d :ext:sources.redhat.com:/cvs/src co \
6099 -r gdb_$V-branch src/gdb/version.in
6100 cvs -f -d :ext:sources.redhat.com:/cvs/src co
6101 -r gdb_5_2-branch src/gdb/version.in
6103 U src/gdb/version.in
6105 $ echo $u.90-0000-00-00-cvs > version.in
6107 5.1.90-0000-00-00-cvs
6108 $ cvs -f commit version.in
6113 @file{0000-00-00} is used as a date to pump prime the version.in update
6116 @file{.90} and the previous branch version are used as fairly arbitrary
6117 initial branch version number.
6121 @subheading Update the web and news pages
6125 @subheading Tweak cron to track the new branch
6127 The file @file{gdbadmin/cron/crontab} contains gdbadmin's cron table.
6128 This file needs to be updated so that:
6132 A daily timestamp is added to the file @file{version.in}.
6134 The new branch is included in the snapshot process.
6138 See the file @file{gdbadmin/cron/README} for how to install the updated
6141 The file @file{gdbadmin/ss/README} should also be reviewed to reflect
6142 any changes. That file is copied to both the branch/ and current/
6143 snapshot directories.
6146 @subheading Update the NEWS and README files
6148 The @file{NEWS} file needs to be updated so that on the branch it refers
6149 to @emph{changes in the current release} while on the trunk it also
6150 refers to @emph{changes since the current release}.
6152 The @file{README} file needs to be updated so that it refers to the
6155 @subheading Post the branch info
6157 Send an announcement to the mailing lists:
6161 @email{gdb-announce@@sources.redhat.com, GDB Announcement mailing list}
6163 @email{gdb@@sources.redhat.com, GDB Discussion mailing list} and
6164 @email{gdb-testers@@sources.redhat.com, GDB Testers mailing list}
6167 @emph{Pragmatics: The branch creation is sent to the announce list to
6168 ensure that people people not subscribed to the higher volume discussion
6171 The announcement should include:
6177 How to check out the branch using CVS.
6179 The date/number of weeks until the release.
6181 The branch commit policy still holds.
6184 @section Stabilize the branch
6186 Something goes here.
6188 @section Create a Release
6190 The process of creating and then making available a release is broken
6191 down into a number of stages. The first part addresses the technical
6192 process of creating a releasable tar ball. The later stages address the
6193 process of releasing that tar ball.
6195 When making a release candidate just the first section is needed.
6197 @subsection Create a release candidate
6199 The objective at this stage is to create a set of tar balls that can be
6200 made available as a formal release (or as a less formal release
6203 @subsubheading Freeze the branch
6205 Send out an e-mail notifying everyone that the branch is frozen to
6206 @email{gdb-patches@@sources.redhat.com}.
6208 @subsubheading Establish a few defaults.
6213 $ t=/sourceware/snapshot-tmp/gdbadmin-tmp
6215 /sourceware/snapshot-tmp/gdbadmin-tmp/gdb_5_2-branch/5.2
6219 /sourceware/snapshot-tmp/gdbadmin-tmp/gdb_5_2-branch/5.2
6221 /home/gdbadmin/bin/autoconf
6230 Check the @code{autoconf} version carefully. You want to be using the
6231 version taken from the @file{binutils} snapshot directory, which can be
6232 found at @uref{ftp://sources.redhat.com/pub/binutils/}. It is very
6233 unlikely that a system installed version of @code{autoconf} (e.g.,
6234 @file{/usr/bin/autoconf}) is correct.
6237 @subsubheading Check out the relevant modules:
6240 $ for m in gdb insight
6242 ( mkdir -p $m && cd $m && cvs -q -f -d /cvs/src co -P -r $b $m )
6252 The reading of @file{.cvsrc} is disabled (@file{-f}) so that there isn't
6253 any confusion between what is written here and what your local
6254 @code{cvs} really does.
6257 @subsubheading Update relevant files.
6263 Major releases get their comments added as part of the mainline. Minor
6264 releases should probably mention any significant bugs that were fixed.
6266 Don't forget to include the @file{ChangeLog} entry.
6269 $ emacs gdb/src/gdb/NEWS
6274 $ cp gdb/src/gdb/NEWS insight/src/gdb/NEWS
6275 $ cp gdb/src/gdb/ChangeLog insight/src/gdb/ChangeLog
6280 You'll need to update:
6292 $ emacs gdb/src/gdb/README
6297 $ cp gdb/src/gdb/README insight/src/gdb/README
6298 $ cp gdb/src/gdb/ChangeLog insight/src/gdb/ChangeLog
6301 @emph{Maintainer note: Hopefully the @file{README} file was reviewed
6302 before the initial branch was cut so just a simple substitute is needed
6305 @emph{Maintainer note: Other projects generate @file{README} and
6306 @file{INSTALL} from the core documentation. This might be worth
6309 @item gdb/version.in
6312 $ echo $v > gdb/src/gdb/version.in
6313 $ cat gdb/src/gdb/version.in
6315 $ emacs gdb/src/gdb/version.in
6318 ... Bump to version ...
6320 $ cp gdb/src/gdb/version.in insight/src/gdb/version.in
6321 $ cp gdb/src/gdb/ChangeLog insight/src/gdb/ChangeLog
6326 @subsubheading Do the dirty work
6328 This is identical to the process used to create the daily snapshot.
6331 $ for m in gdb insight
6333 ( cd $m/src && gmake -f src-release $m.tar )
6337 If the top level source directory does not have @file{src-release}
6338 (@value{GDBN} version 5.3.1 or earlier), try these commands instead:
6341 $ for m in gdb insight
6343 ( cd $m/src && gmake -f Makefile.in $m.tar )
6347 @subsubheading Check the source files
6349 You're looking for files that have mysteriously disappeared.
6350 @kbd{distclean} has the habit of deleting files it shouldn't. Watch out
6351 for the @file{version.in} update @kbd{cronjob}.
6354 $ ( cd gdb/src && cvs -f -q -n update )
6358 @dots{} lots of generated files @dots{}
6363 @dots{} lots of generated files @dots{}
6368 @emph{Don't worry about the @file{gdb.info-??} or
6369 @file{gdb/p-exp.tab.c}. They were generated (and yes @file{gdb.info-1}
6370 was also generated only something strange with CVS means that they
6371 didn't get suppressed). Fixing it would be nice though.}
6373 @subsubheading Create compressed versions of the release
6379 gdb/ gdb-5.2.tar insight/ insight-5.2.tar
6380 $ for m in gdb insight
6382 bzip2 -v -9 -c $m-$v.tar > $m-$v.tar.bz2
6383 gzip -v -9 -c $m-$v.tar > $m-$v.tar.gz
6393 A pipe such as @kbd{bunzip2 < xxx.bz2 | gzip -9 > xxx.gz} is not since,
6394 in that mode, @code{gzip} does not know the name of the file and, hence,
6395 can not include it in the compressed file. This is also why the release
6396 process runs @code{tar} and @code{bzip2} as separate passes.
6399 @subsection Sanity check the tar ball
6401 Pick a popular machine (Solaris/PPC?) and try the build on that.
6404 $ bunzip2 < gdb-5.2.tar.bz2 | tar xpf -
6409 $ ./gdb/gdb ./gdb/gdb
6413 Breakpoint 1 at 0x80732bc: file main.c, line 734.
6415 Starting program: /tmp/gdb-5.2/gdb/gdb
6417 Breakpoint 1, main (argc=1, argv=0xbffff8b4) at main.c:734
6418 734 catch_errors (captured_main, &args, "", RETURN_MASK_ALL);
6420 $1 = @{argc = 136426532, argv = 0x821b7f0@}
6424 @subsection Make a release candidate available
6426 If this is a release candidate then the only remaining steps are:
6430 Commit @file{version.in} and @file{ChangeLog}
6432 Tweak @file{version.in} (and @file{ChangeLog} to read
6433 @var{L}.@var{M}.@var{N}-0000-00-00-cvs so that the version update
6434 process can restart.
6436 Make the release candidate available in
6437 @uref{ftp://sources.redhat.com/pub/gdb/snapshots/branch}
6439 Notify the relevant mailing lists ( @email{gdb@@sources.redhat.com} and
6440 @email{gdb-testers@@sources.redhat.com} that the candidate is available.
6443 @subsection Make a formal release available
6445 (And you thought all that was required was to post an e-mail.)
6447 @subsubheading Install on sware
6449 Copy the new files to both the release and the old release directory:
6452 $ cp *.bz2 *.gz ~ftp/pub/gdb/old-releases/
6453 $ cp *.bz2 *.gz ~ftp/pub/gdb/releases
6457 Clean up the releases directory so that only the most recent releases
6458 are available (e.g. keep 5.2 and 5.2.1 but remove 5.1):
6461 $ cd ~ftp/pub/gdb/releases
6466 Update the file @file{README} and @file{.message} in the releases
6473 $ ln README .message
6476 @subsubheading Update the web pages.
6480 @item htdocs/download/ANNOUNCEMENT
6481 This file, which is posted as the official announcement, includes:
6484 General announcement.
6486 News. If making an @var{M}.@var{N}.1 release, retain the news from
6487 earlier @var{M}.@var{N} release.
6492 @item htdocs/index.html
6493 @itemx htdocs/news/index.html
6494 @itemx htdocs/download/index.html
6495 These files include:
6498 Announcement of the most recent release.
6500 News entry (remember to update both the top level and the news directory).
6502 These pages also need to be regenerate using @code{index.sh}.
6504 @item download/onlinedocs/
6505 You need to find the magic command that is used to generate the online
6506 docs from the @file{.tar.bz2}. The best way is to look in the output
6507 from one of the nightly @code{cron} jobs and then just edit accordingly.
6511 $ ~/ss/update-web-docs \
6512 ~ftp/pub/gdb/releases/gdb-5.2.tar.bz2 \
6514 /www/sourceware/htdocs/gdb/download/onlinedocs \
6519 Just like the online documentation. Something like:
6522 $ /bin/sh ~/ss/update-web-ari \
6523 ~ftp/pub/gdb/releases/gdb-5.2.tar.bz2 \
6525 /www/sourceware/htdocs/gdb/download/ari \
6531 @subsubheading Shadow the pages onto gnu
6533 Something goes here.
6536 @subsubheading Install the @value{GDBN} tar ball on GNU
6538 At the time of writing, the GNU machine was @kbd{gnudist.gnu.org} in
6539 @file{~ftp/gnu/gdb}.
6541 @subsubheading Make the @file{ANNOUNCEMENT}
6543 Post the @file{ANNOUNCEMENT} file you created above to:
6547 @email{gdb-announce@@sources.redhat.com, GDB Announcement mailing list}
6549 @email{info-gnu@@gnu.org, General GNU Announcement list} (but delay it a
6550 day or so to let things get out)
6552 @email{bug-gdb@@gnu.org, GDB Bug Report mailing list}
6557 The release is out but you're still not finished.
6559 @subsubheading Commit outstanding changes
6561 In particular you'll need to commit any changes to:
6565 @file{gdb/ChangeLog}
6567 @file{gdb/version.in}
6574 @subsubheading Tag the release
6579 $ d=`date -u +%Y-%m-%d`
6582 $ ( cd insight/src/gdb && cvs -f -q update )
6583 $ ( cd insight/src && cvs -f -q tag gdb_5_2-$d-release )
6586 Insight is used since that contains more of the release than
6589 @subsubheading Mention the release on the trunk
6591 Just put something in the @file{ChangeLog} so that the trunk also
6592 indicates when the release was made.
6594 @subsubheading Restart @file{gdb/version.in}
6596 If @file{gdb/version.in} does not contain an ISO date such as
6597 @kbd{2002-01-24} then the daily @code{cronjob} won't update it. Having
6598 committed all the release changes it can be set to
6599 @file{5.2.0_0000-00-00-cvs} which will restart things (yes the @kbd{_}
6600 is important - it affects the snapshot process).
6602 Don't forget the @file{ChangeLog}.
6604 @subsubheading Merge into trunk
6606 The files committed to the branch may also need changes merged into the
6609 @subsubheading Revise the release schedule
6611 Post a revised release schedule to @email{gdb@@sources.redhat.com, GDB
6612 Discussion List} with an updated announcement. The schedule can be
6613 generated by running:
6616 $ ~/ss/schedule `date +%s` schedule
6620 The first parameter is approximate date/time in seconds (from the epoch)
6621 of the most recent release.
6623 Also update the schedule @code{cronjob}.
6625 @section Post release
6627 Remove any @code{OBSOLETE} code.
6634 The testsuite is an important component of the @value{GDBN} package.
6635 While it is always worthwhile to encourage user testing, in practice
6636 this is rarely sufficient; users typically use only a small subset of
6637 the available commands, and it has proven all too common for a change
6638 to cause a significant regression that went unnoticed for some time.
6640 The @value{GDBN} testsuite uses the DejaGNU testing framework. The
6641 tests themselves are calls to various @code{Tcl} procs; the framework
6642 runs all the procs and summarizes the passes and fails.
6644 @section Using the Testsuite
6646 @cindex running the test suite
6647 To run the testsuite, simply go to the @value{GDBN} object directory (or to the
6648 testsuite's objdir) and type @code{make check}. This just sets up some
6649 environment variables and invokes DejaGNU's @code{runtest} script. While
6650 the testsuite is running, you'll get mentions of which test file is in use,
6651 and a mention of any unexpected passes or fails. When the testsuite is
6652 finished, you'll get a summary that looks like this:
6657 # of expected passes 6016
6658 # of unexpected failures 58
6659 # of unexpected successes 5
6660 # of expected failures 183
6661 # of unresolved testcases 3
6662 # of untested testcases 5
6665 To run a specific test script, type:
6667 make check RUNTESTFLAGS='@var{tests}'
6669 where @var{tests} is a list of test script file names, separated by
6672 The ideal test run consists of expected passes only; however, reality
6673 conspires to keep us from this ideal. Unexpected failures indicate
6674 real problems, whether in @value{GDBN} or in the testsuite. Expected
6675 failures are still failures, but ones which have been decided are too
6676 hard to deal with at the time; for instance, a test case might work
6677 everywhere except on AIX, and there is no prospect of the AIX case
6678 being fixed in the near future. Expected failures should not be added
6679 lightly, since you may be masking serious bugs in @value{GDBN}.
6680 Unexpected successes are expected fails that are passing for some
6681 reason, while unresolved and untested cases often indicate some minor
6682 catastrophe, such as the compiler being unable to deal with a test
6685 When making any significant change to @value{GDBN}, you should run the
6686 testsuite before and after the change, to confirm that there are no
6687 regressions. Note that truly complete testing would require that you
6688 run the testsuite with all supported configurations and a variety of
6689 compilers; however this is more than really necessary. In many cases
6690 testing with a single configuration is sufficient. Other useful
6691 options are to test one big-endian (Sparc) and one little-endian (x86)
6692 host, a cross config with a builtin simulator (powerpc-eabi,
6693 mips-elf), or a 64-bit host (Alpha).
6695 If you add new functionality to @value{GDBN}, please consider adding
6696 tests for it as well; this way future @value{GDBN} hackers can detect
6697 and fix their changes that break the functionality you added.
6698 Similarly, if you fix a bug that was not previously reported as a test
6699 failure, please add a test case for it. Some cases are extremely
6700 difficult to test, such as code that handles host OS failures or bugs
6701 in particular versions of compilers, and it's OK not to try to write
6702 tests for all of those.
6704 DejaGNU supports separate build, host, and target machines. However,
6705 some @value{GDBN} test scripts do not work if the build machine and
6706 the host machine are not the same. In such an environment, these scripts
6707 will give a result of ``UNRESOLVED'', like this:
6710 UNRESOLVED: gdb.base/example.exp: This test script does not work on a remote host.
6713 @section Testsuite Organization
6715 @cindex test suite organization
6716 The testsuite is entirely contained in @file{gdb/testsuite}. While the
6717 testsuite includes some makefiles and configury, these are very minimal,
6718 and used for little besides cleaning up, since the tests themselves
6719 handle the compilation of the programs that @value{GDBN} will run. The file
6720 @file{testsuite/lib/gdb.exp} contains common utility procs useful for
6721 all @value{GDBN} tests, while the directory @file{testsuite/config} contains
6722 configuration-specific files, typically used for special-purpose
6723 definitions of procs like @code{gdb_load} and @code{gdb_start}.
6725 The tests themselves are to be found in @file{testsuite/gdb.*} and
6726 subdirectories of those. The names of the test files must always end
6727 with @file{.exp}. DejaGNU collects the test files by wildcarding
6728 in the test directories, so both subdirectories and individual files
6729 get chosen and run in alphabetical order.
6731 The following table lists the main types of subdirectories and what they
6732 are for. Since DejaGNU finds test files no matter where they are
6733 located, and since each test file sets up its own compilation and
6734 execution environment, this organization is simply for convenience and
6739 This is the base testsuite. The tests in it should apply to all
6740 configurations of @value{GDBN} (but generic native-only tests may live here).
6741 The test programs should be in the subset of C that is valid K&R,
6742 ANSI/ISO, and C@t{++} (@code{#ifdef}s are allowed if necessary, for instance
6745 @item gdb.@var{lang}
6746 Language-specific tests for any language @var{lang} besides C. Examples are
6747 @file{gdb.cp} and @file{gdb.java}.
6749 @item gdb.@var{platform}
6750 Non-portable tests. The tests are specific to a specific configuration
6751 (host or target), such as HP-UX or eCos. Example is @file{gdb.hp}, for
6754 @item gdb.@var{compiler}
6755 Tests specific to a particular compiler. As of this writing (June
6756 1999), there aren't currently any groups of tests in this category that
6757 couldn't just as sensibly be made platform-specific, but one could
6758 imagine a @file{gdb.gcc}, for tests of @value{GDBN}'s handling of GCC
6761 @item gdb.@var{subsystem}
6762 Tests that exercise a specific @value{GDBN} subsystem in more depth. For
6763 instance, @file{gdb.disasm} exercises various disassemblers, while
6764 @file{gdb.stabs} tests pathways through the stabs symbol reader.
6767 @section Writing Tests
6768 @cindex writing tests
6770 In many areas, the @value{GDBN} tests are already quite comprehensive; you
6771 should be able to copy existing tests to handle new cases.
6773 You should try to use @code{gdb_test} whenever possible, since it
6774 includes cases to handle all the unexpected errors that might happen.
6775 However, it doesn't cost anything to add new test procedures; for
6776 instance, @file{gdb.base/exprs.exp} defines a @code{test_expr} that
6777 calls @code{gdb_test} multiple times.
6779 Only use @code{send_gdb} and @code{gdb_expect} when absolutely
6780 necessary. Even if @value{GDBN} has several valid responses to
6781 a command, you can use @code{gdb_test_multiple}. Like @code{gdb_test},
6782 @code{gdb_test_multiple} recognizes internal errors and unexpected
6785 Do not write tests which expect a literal tab character from @value{GDBN}.
6786 On some operating systems (e.g.@: OpenBSD) the TTY layer expands tabs to
6787 spaces, so by the time @value{GDBN}'s output reaches expect the tab is gone.
6789 The source language programs do @emph{not} need to be in a consistent
6790 style. Since @value{GDBN} is used to debug programs written in many different
6791 styles, it's worth having a mix of styles in the testsuite; for
6792 instance, some @value{GDBN} bugs involving the display of source lines would
6793 never manifest themselves if the programs used GNU coding style
6800 Check the @file{README} file, it often has useful information that does not
6801 appear anywhere else in the directory.
6804 * Getting Started:: Getting started working on @value{GDBN}
6805 * Debugging GDB:: Debugging @value{GDBN} with itself
6808 @node Getting Started,,, Hints
6810 @section Getting Started
6812 @value{GDBN} is a large and complicated program, and if you first starting to
6813 work on it, it can be hard to know where to start. Fortunately, if you
6814 know how to go about it, there are ways to figure out what is going on.
6816 This manual, the @value{GDBN} Internals manual, has information which applies
6817 generally to many parts of @value{GDBN}.
6819 Information about particular functions or data structures are located in
6820 comments with those functions or data structures. If you run across a
6821 function or a global variable which does not have a comment correctly
6822 explaining what is does, this can be thought of as a bug in @value{GDBN}; feel
6823 free to submit a bug report, with a suggested comment if you can figure
6824 out what the comment should say. If you find a comment which is
6825 actually wrong, be especially sure to report that.
6827 Comments explaining the function of macros defined in host, target, or
6828 native dependent files can be in several places. Sometimes they are
6829 repeated every place the macro is defined. Sometimes they are where the
6830 macro is used. Sometimes there is a header file which supplies a
6831 default definition of the macro, and the comment is there. This manual
6832 also documents all the available macros.
6833 @c (@pxref{Host Conditionals}, @pxref{Target
6834 @c Conditionals}, @pxref{Native Conditionals}, and @pxref{Obsolete
6837 Start with the header files. Once you have some idea of how
6838 @value{GDBN}'s internal symbol tables are stored (see @file{symtab.h},
6839 @file{gdbtypes.h}), you will find it much easier to understand the
6840 code which uses and creates those symbol tables.
6842 You may wish to process the information you are getting somehow, to
6843 enhance your understanding of it. Summarize it, translate it to another
6844 language, add some (perhaps trivial or non-useful) feature to @value{GDBN}, use
6845 the code to predict what a test case would do and write the test case
6846 and verify your prediction, etc. If you are reading code and your eyes
6847 are starting to glaze over, this is a sign you need to use a more active
6850 Once you have a part of @value{GDBN} to start with, you can find more
6851 specifically the part you are looking for by stepping through each
6852 function with the @code{next} command. Do not use @code{step} or you
6853 will quickly get distracted; when the function you are stepping through
6854 calls another function try only to get a big-picture understanding
6855 (perhaps using the comment at the beginning of the function being
6856 called) of what it does. This way you can identify which of the
6857 functions being called by the function you are stepping through is the
6858 one which you are interested in. You may need to examine the data
6859 structures generated at each stage, with reference to the comments in
6860 the header files explaining what the data structures are supposed to
6863 Of course, this same technique can be used if you are just reading the
6864 code, rather than actually stepping through it. The same general
6865 principle applies---when the code you are looking at calls something
6866 else, just try to understand generally what the code being called does,
6867 rather than worrying about all its details.
6869 @cindex command implementation
6870 A good place to start when tracking down some particular area is with
6871 a command which invokes that feature. Suppose you want to know how
6872 single-stepping works. As a @value{GDBN} user, you know that the
6873 @code{step} command invokes single-stepping. The command is invoked
6874 via command tables (see @file{command.h}); by convention the function
6875 which actually performs the command is formed by taking the name of
6876 the command and adding @samp{_command}, or in the case of an
6877 @code{info} subcommand, @samp{_info}. For example, the @code{step}
6878 command invokes the @code{step_command} function and the @code{info
6879 display} command invokes @code{display_info}. When this convention is
6880 not followed, you might have to use @code{grep} or @kbd{M-x
6881 tags-search} in emacs, or run @value{GDBN} on itself and set a
6882 breakpoint in @code{execute_command}.
6884 @cindex @code{bug-gdb} mailing list
6885 If all of the above fail, it may be appropriate to ask for information
6886 on @code{bug-gdb}. But @emph{never} post a generic question like ``I was
6887 wondering if anyone could give me some tips about understanding
6888 @value{GDBN}''---if we had some magic secret we would put it in this manual.
6889 Suggestions for improving the manual are always welcome, of course.
6891 @node Debugging GDB,,,Hints
6893 @section Debugging @value{GDBN} with itself
6894 @cindex debugging @value{GDBN}
6896 If @value{GDBN} is limping on your machine, this is the preferred way to get it
6897 fully functional. Be warned that in some ancient Unix systems, like
6898 Ultrix 4.2, a program can't be running in one process while it is being
6899 debugged in another. Rather than typing the command @kbd{@w{./gdb
6900 ./gdb}}, which works on Suns and such, you can copy @file{gdb} to
6901 @file{gdb2} and then type @kbd{@w{./gdb ./gdb2}}.
6903 When you run @value{GDBN} in the @value{GDBN} source directory, it will read a
6904 @file{.gdbinit} file that sets up some simple things to make debugging
6905 gdb easier. The @code{info} command, when executed without a subcommand
6906 in a @value{GDBN} being debugged by gdb, will pop you back up to the top level
6907 gdb. See @file{.gdbinit} for details.
6909 If you use emacs, you will probably want to do a @code{make TAGS} after
6910 you configure your distribution; this will put the machine dependent
6911 routines for your local machine where they will be accessed first by
6914 Also, make sure that you've either compiled @value{GDBN} with your local cc, or
6915 have run @code{fixincludes} if you are compiling with gcc.
6917 @section Submitting Patches
6919 @cindex submitting patches
6920 Thanks for thinking of offering your changes back to the community of
6921 @value{GDBN} users. In general we like to get well designed enhancements.
6922 Thanks also for checking in advance about the best way to transfer the
6925 The @value{GDBN} maintainers will only install ``cleanly designed'' patches.
6926 This manual summarizes what we believe to be clean design for @value{GDBN}.
6928 If the maintainers don't have time to put the patch in when it arrives,
6929 or if there is any question about a patch, it goes into a large queue
6930 with everyone else's patches and bug reports.
6932 @cindex legal papers for code contributions
6933 The legal issue is that to incorporate substantial changes requires a
6934 copyright assignment from you and/or your employer, granting ownership
6935 of the changes to the Free Software Foundation. You can get the
6936 standard documents for doing this by sending mail to @code{gnu@@gnu.org}
6937 and asking for it. We recommend that people write in "All programs
6938 owned by the Free Software Foundation" as "NAME OF PROGRAM", so that
6939 changes in many programs (not just @value{GDBN}, but GAS, Emacs, GCC,
6941 contributed with only one piece of legalese pushed through the
6942 bureaucracy and filed with the FSF. We can't start merging changes until
6943 this paperwork is received by the FSF (their rules, which we follow
6944 since we maintain it for them).
6946 Technically, the easiest way to receive changes is to receive each
6947 feature as a small context diff or unidiff, suitable for @code{patch}.
6948 Each message sent to me should include the changes to C code and
6949 header files for a single feature, plus @file{ChangeLog} entries for
6950 each directory where files were modified, and diffs for any changes
6951 needed to the manuals (@file{gdb/doc/gdb.texinfo} or
6952 @file{gdb/doc/gdbint.texinfo}). If there are a lot of changes for a
6953 single feature, they can be split down into multiple messages.
6955 In this way, if we read and like the feature, we can add it to the
6956 sources with a single patch command, do some testing, and check it in.
6957 If you leave out the @file{ChangeLog}, we have to write one. If you leave
6958 out the doc, we have to puzzle out what needs documenting. Etc., etc.
6960 The reason to send each change in a separate message is that we will not
6961 install some of the changes. They'll be returned to you with questions
6962 or comments. If we're doing our job correctly, the message back to you
6963 will say what you have to fix in order to make the change acceptable.
6964 The reason to have separate messages for separate features is so that
6965 the acceptable changes can be installed while one or more changes are
6966 being reworked. If multiple features are sent in a single message, we
6967 tend to not put in the effort to sort out the acceptable changes from
6968 the unacceptable, so none of the features get installed until all are
6971 If this sounds painful or authoritarian, well, it is. But we get a lot
6972 of bug reports and a lot of patches, and many of them don't get
6973 installed because we don't have the time to finish the job that the bug
6974 reporter or the contributor could have done. Patches that arrive
6975 complete, working, and well designed, tend to get installed on the day
6976 they arrive. The others go into a queue and get installed as time
6977 permits, which, since the maintainers have many demands to meet, may not
6978 be for quite some time.
6980 Please send patches directly to
6981 @email{gdb-patches@@sources.redhat.com, the @value{GDBN} maintainers}.
6983 @section Build Script
6985 @cindex build script
6987 The script @file{gdb_buildall.sh} builds @value{GDBN} with flag
6988 @option{--enable-targets=all} set. This builds @value{GDBN} with all supported
6989 targets activated. This helps testing @value{GDBN} when doing changes that
6990 affect more than one architecture and is much faster than using
6991 @file{gdb_mbuild.sh}.
6993 After building @value{GDBN} the script checks which architectures are
6994 supported and then switches the current architecture to each of those to get
6995 information about the architecture. The test results are stored in log files
6996 in the directory the script was called from.
6998 @include observer.texi