/* Output Dwarf format symbol table information from the GNU C compiler. Copyright (C) 1992, 1993, 1995, 1996, 1997, 1998, 2002, 1999, 2000, 2001 Free Software Foundation, Inc. Contributed by Ron Guilmette (rfg@monkeys.com) of Network Computing Devices. This file is part of GCC. GCC is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2, or (at your option) any later version. GCC is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. You should have received a copy of the GNU General Public License along with GCC; see the file COPYING. If not, write to the Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. */ /* Notes on the GNU Implementation of DWARF Debugging Information -------------------------------------------------------------- Last Major Update: Sun Jul 17 08:17:42 PDT 1994 by rfg@segfault.us.com ------------------------------------------------------------ This file describes special and unique aspects of the GNU implementation of the DWARF Version 1 debugging information language, as provided in the GNU version 2.x compiler(s). For general information about the DWARF debugging information language, you should obtain the DWARF version 1.1 specification document (and perhaps also the DWARF version 2 draft specification document) developed by the (now defunct) UNIX International Programming Languages Special Interest Group. To obtain a copy of the DWARF Version 1 and/or DWARF Version 2 specification, visit the web page for the DWARF Version 2 committee, at http://www.eagercon.com/dwarf/dwarf2std.htm The generation of DWARF debugging information by the GNU version 2.x C compiler has now been tested rather extensively for m88k, i386, i860, and Sparc targets. The DWARF output of the GNU C compiler appears to inter- operate well with the standard SVR4 SDB debugger on these kinds of target systems (but of course, there are no guarantees). DWARF 1 generation for the GNU g++ compiler is implemented, but limited. C++ users should definitely use DWARF 2 instead. Future plans for the dwarfout.c module of the GNU compiler(s) includes the addition of full support for GNU FORTRAN. (This should, in theory, be a lot simpler to add than adding support for g++... but we'll see.) Many features of the DWARF version 2 specification have been adapted to (and used in) the GNU implementation of DWARF (version 1). In most of these cases, a DWARF version 2 approach is used in place of (or in addition to) DWARF version 1 stuff simply because it is apparent that DWARF version 1 is not sufficiently expressive to provide the kinds of information which may be necessary to support really robust debugging. In all of these cases however, the use of DWARF version 2 features should not interfere in any way with the interoperability (of GNU compilers) with generally available "classic" (pre version 1) DWARF consumer tools (e.g. SVR4 SDB). The DWARF generation enhancement for the GNU compiler(s) was initially donated to the Free Software Foundation by Network Computing Devices. (Thanks NCD!) Additional development and maintenance of dwarfout.c has been largely supported (i.e. funded) by Intel Corporation. (Thanks Intel!) If you have questions or comments about the DWARF generation feature, please send mail to me . I will be happy to investigate any bugs reported and I may even provide fixes (but of course, I can make no promises). The DWARF debugging information produced by GCC may deviate in a few minor (but perhaps significant) respects from the DWARF debugging information currently produced by other C compilers. A serious attempt has been made however to conform to the published specifications, to existing practice, and to generally accepted norms in the GNU implementation of DWARF. ** IMPORTANT NOTE ** ** IMPORTANT NOTE ** ** IMPORTANT NOTE ** Under normal circumstances, the DWARF information generated by the GNU compilers (in an assembly language file) is essentially impossible for a human being to read. This fact can make it very difficult to debug certain DWARF-related problems. In order to overcome this difficulty, a feature has been added to dwarfout.c (enabled by the -dA option) which causes additional comments to be placed into the assembly language output file, out to the right-hand side of most bits of DWARF material. The comments indicate (far more clearly that the obscure DWARF hex codes do) what is actually being encoded in DWARF. Thus, the -dA option can be highly useful for those who must study the DWARF output from the GNU compilers in detail. --------- (Footnote: Within this file, the term `Debugging Information Entry' will be abbreviated as `DIE'.) Release Notes (aka known bugs) ------------------------------- In one very obscure case involving dynamically sized arrays, the DWARF "location information" for such an array may make it appear that the array has been totally optimized out of existence, when in fact it *must* actually exist. (This only happens when you are using *both* -g *and* -O.) This is due to aggressive dead store elimination in the compiler, and to the fact that the DECL_RTL expressions associated with variables are not always updated to correctly reflect the effects of GCC's aggressive dead store elimination. ------------------------------- When attempting to set a breakpoint at the "start" of a function compiled with -g1, the debugger currently has no way of knowing exactly where the end of the prologue code for the function is. Thus, for most targets, all the debugger can do is to set the breakpoint at the AT_low_pc address for the function. But if you stop there and then try to look at one or more of the formal parameter values, they may not have been "homed" yet, so you may get inaccurate answers (or perhaps even addressing errors). Some people may consider this simply a non-feature, but I consider it a bug, and I hope to provide some GNU-specific attributes (on function DIEs) which will specify the address of the end of the prologue and the address of the beginning of the epilogue in a future release. ------------------------------- It is believed at this time that old bugs relating to the AT_bit_offset values for bit-fields have been fixed. There may still be some very obscure bugs relating to the DWARF description of type `long long' bit-fields for target machines (e.g. 80x86 machines) where the alignment of type `long long' data objects is different from (and less than) the size of a type `long long' data object. Please report any problems with the DWARF description of bit-fields as you would any other GCC bug. (Procedures for bug reporting are given in the GNU C compiler manual.) -------------------------------- At this time, GCC does not know how to handle the GNU C "nested functions" extension. (See the GCC manual for more info on this extension to ANSI C.) -------------------------------- The GNU compilers now represent inline functions (and inlined instances thereof) in exactly the manner described by the current DWARF version 2 (draft) specification. The version 1 specification for handling inline functions (and inlined instances) was known to be brain-damaged (by the PLSIG) when the version 1 spec was finalized, but it was simply too late in the cycle to get it removed before the version 1 spec was formally released to the public (by UI). -------------------------------- At this time, GCC does not generate the kind of really precise information about the exact declared types of entities with signed integral types which is required by the current DWARF draft specification. Specifically, the current DWARF draft specification seems to require that the type of an non-unsigned integral bit-field member of a struct or union type be represented as either a "signed" type or as a "plain" type, depending upon the exact set of keywords that were used in the type specification for the given bit-field member. It was felt (by the UI/PLSIG) that this distinction between "plain" and "signed" integral types could have some significance (in the case of bit-fields) because ANSI C does not constrain the signedness of a plain bit-field, whereas it does constrain the signedness of an explicitly "signed" bit-field. For this reason, the current DWARF specification calls for compilers to produce type information (for *all* integral typed entities... not just bit-fields) which explicitly indicates the signedness of the relevant type to be "signed" or "plain" or "unsigned". Unfortunately, the GNU DWARF implementation is currently incapable of making such distinctions. -------------------------------- Known Interoperability Problems ------------------------------- Although the GNU implementation of DWARF conforms (for the most part) with the current UI/PLSIG DWARF version 1 specification (with many compatible version 2 features added in as "vendor specific extensions" just for good measure) there are a few known cases where GCC's DWARF output can cause some confusion for "classic" (pre version 1) DWARF consumers such as the System V Release 4 SDB debugger. These cases are described in this section. -------------------------------- The DWARF version 1 specification includes the fundamental type codes FT_ext_prec_float, FT_complex, FT_dbl_prec_complex, and FT_ext_prec_complex. Since GNU C is only a C compiler (and since C doesn't provide any "complex" data types) the only one of these fundamental type codes which GCC ever generates is FT_ext_prec_float. This fundamental type code is generated by GCC for the `long double' data type. Unfortunately, due to an apparent bug in the SVR4 SDB debugger, SDB can become very confused wherever any attempt is made to print a variable, parameter, or field whose type was given in terms of FT_ext_prec_float. (Actually, SVR4 SDB fails to understand *any* of the four fundamental type codes mentioned here. This will fact will cause additional problems when there is a GNU FORTRAN front-end.) -------------------------------- In general, it appears that SVR4 SDB is not able to effectively ignore fundamental type codes in the "implementation defined" range. This can cause problems when a program being debugged uses the `long long' data type (or the signed or unsigned varieties thereof) because these types are not defined by ANSI C, and thus, GCC must use its own private fundamental type codes (from the implementation-defined range) to represent these types. -------------------------------- General GNU DWARF extensions ---------------------------- In the current DWARF version 1 specification, no mechanism is specified by which accurate information about executable code from include files can be properly (and fully) described. (The DWARF version 2 specification *does* specify such a mechanism, but it is about 10 times more complicated than it needs to be so I'm not terribly anxious to try to implement it right away.) In the GNU implementation of DWARF version 1, a fully downward-compatible extension has been implemented which permits the GNU compilers to specify which executable lines come from which files. This extension places additional information (about source file names) in GNU-specific sections (which should be totally ignored by all non-GNU DWARF consumers) so that this extended information can be provided (to GNU DWARF consumers) in a way which is totally transparent (and invisible) to non-GNU DWARF consumers (e.g. the SVR4 SDB debugger). The additional information is placed *only* in specialized GNU-specific sections, where it should never even be seen by non-GNU DWARF consumers. To understand this GNU DWARF extension, imagine that the sequence of entries in the .lines section is broken up into several subsections. Each contiguous sequence of .line entries which relates to a sequence of lines (or statements) from one particular file (either a `base' file or an `include' file) could be called a `line entries chunk' (LEC). For each LEC there is one entry in the .debug_srcinfo section. Each normal entry in the .debug_srcinfo section consists of two 4-byte words of data as follows: (1) The starting address (relative to the entire .line section) of the first .line entry in the relevant LEC. (2) The starting address (relative to the entire .debug_sfnames section) of a NUL terminated string representing the relevant filename. (This filename name be either a relative or an absolute filename, depending upon how the given source file was located during compilation.) Obviously, each .debug_srcinfo entry allows you to find the relevant filename, and it also points you to the first .line entry that was generated as a result of having compiled a given source line from the given source file. Each subsequent .line entry should also be assumed to have been produced as a result of compiling yet more lines from the same file. The end of any given LEC is easily found by looking at the first 4-byte pointer in the *next* .debug_srcinfo entry. That next .debug_srcinfo entry points to a new and different LEC, so the preceding LEC (implicitly) must have ended with the last .line section entry which occurs at the 2 1/2 words just before the address given in the first pointer of the new .debug_srcinfo entry. The following picture may help to clarify this feature. Let's assume that `LE' stands for `.line entry'. Also, assume that `* 'stands for a pointer. .line section .debug_srcinfo section .debug_sfnames section ---------------------------------------------------------------- LE <---------------------- * LE * -----------------> "foobar.c" <--- LE | LE | LE <---------------------- * | LE * -----------------> "foobar.h" <| | LE | | LE | | LE <---------------------- * | | LE * -----------------> "inner.h" | | LE | | LE <---------------------- * | | LE * ------------------------------- | LE | LE | LE | LE | LE <---------------------- * | LE * ----------------------------------- LE LE LE In effect, each entry in the .debug_srcinfo section points to *both* a filename (in the .debug_sfnames section) and to the start of a block of consecutive LEs (in the .line section). Note that just like in the .line section, there are specialized first and last entries in the .debug_srcinfo section for each object file. These special first and last entries for the .debug_srcinfo section are very different from the normal .debug_srcinfo section entries. They provide additional information which may be helpful to a debugger when it is interpreting the data in the .debug_srcinfo, .debug_sfnames, and .line sections. The first entry in the .debug_srcinfo section for each compilation unit consists of five 4-byte words of data. The contents of these five words should be interpreted (by debuggers) as follows: (1) The starting address (relative to the entire .line section) of the .line section for this compilation unit. (2) The starting address (relative to the entire .debug_sfnames section) of the .debug_sfnames section for this compilation unit. (3) The starting address (in the execution virtual address space) of the .text section for this compilation unit. (4) The ending address plus one (in the execution virtual address space) of the .text section for this compilation unit. (5) The date/time (in seconds since midnight 1/1/70) at which the compilation of this compilation unit occurred. This value should be interpreted as an unsigned quantity because gcc might be configured to generate a default value of 0xffffffff in this field (in cases where it is desired to have object files created at different times from identical source files be byte-for-byte identical). By default, these timestamps are *not* generated by dwarfout.c (so that object files compiled at different times will be byte-for-byte identical). If you wish to enable this "timestamp" feature however, you can simply place a #define for the symbol `DWARF_TIMESTAMPS' in your target configuration file and then rebuild the GNU compiler(s). Note that the first string placed into the .debug_sfnames section for each compilation unit is the name of the directory in which compilation occurred. This string ends with a `/' (to help indicate that it is the pathname of a directory). Thus, the second word of each specialized initial .debug_srcinfo entry for each compilation unit may be used as a pointer to the (string) name of the compilation directory, and that string may in turn be used to "absolutize" any relative pathnames which may appear later on in the .debug_sfnames section entries for the same compilation unit. The fifth and last word of each specialized starting entry for a compilation unit in the .debug_srcinfo section may (depending upon your configuration) indicate the date/time of compilation, and this may be used (by a debugger) to determine if any of the source files which contributed code to this compilation unit are newer than the object code for the compilation unit itself. If so, the debugger may wish to print an "out-of-date" warning about the compilation unit. The .debug_srcinfo section associated with each compilation will also have a specialized terminating entry. This terminating .debug_srcinfo section entry will consist of the following two 4-byte words of data: (1) The offset, measured from the start of the .line section to the beginning of the terminating entry for the .line section. (2) A word containing the value 0xffffffff. -------------------------------- In the current DWARF version 1 specification, no mechanism is specified by which information about macro definitions and un-definitions may be provided to the DWARF consumer. The DWARF version 2 (draft) specification does specify such a mechanism. That specification was based on the GNU ("vendor specific extension") which provided some support for macro definitions and un-definitions, but the "official" DWARF version 2 (draft) specification mechanism for handling macros and the GNU implementation have diverged somewhat. I plan to update the GNU implementation to conform to the "official" DWARF version 2 (draft) specification as soon as I get time to do that. Note that in the GNU implementation, additional information about macro definitions and un-definitions is *only* provided when the -g3 level of debug-info production is selected. (The default level is -g2 and the plain old -g option is considered to be identical to -g2.) GCC records information about macro definitions and undefinitions primarily in a section called the .debug_macinfo section. Normal entries in the .debug_macinfo section consist of the following three parts: (1) A special "type" byte. (2) A 3-byte line-number/filename-offset field. (3) A NUL terminated string. The interpretation of the second and third parts is dependent upon the value of the leading (type) byte. The type byte may have one of four values depending upon the type of the .debug_macinfo entry which follows. The 1-byte MACINFO type codes presently used, and their meanings are as follows: MACINFO_start A base file or an include file starts here. MACINFO_resume The current base or include file ends here. MACINFO_define A #define directive occurs here. MACINFO_undef A #undef directive occur here. (Note that the MACINFO_... codes mentioned here are simply symbolic names for constants which are defined in the GNU dwarf.h file.) For MACINFO_define and MACINFO_undef entries, the second (3-byte) field contains the number of the source line (relative to the start of the current base source file or the current include files) when the #define or #undef directive appears. For a MACINFO_define entry, the following string field contains the name of the macro which is defined, followed by its definition. Note that the definition is always separated from the name of the macro by at least one whitespace character. For a MACINFO_undef entry, the string which follows the 3-byte line number field contains just the name of the macro which is being undef'ed. For a MACINFO_start entry, the 3-byte field following the type byte contains the offset, relative to the start of the .debug_sfnames section for the current compilation unit, of a string which names the new source file which is beginning its inclusion at this point. Following that 3-byte field, each MACINFO_start entry always contains a zero length NUL terminated string. For a MACINFO_resume entry, the 3-byte field following the type byte contains the line number WITHIN THE INCLUDING FILE at which the inclusion of the current file (whose inclusion ends here) was initiated. Following that 3-byte field, each MACINFO_resume entry always contains a zero length NUL terminated string. Each set of .debug_macinfo entries for each compilation unit is terminated by a special .debug_macinfo entry consisting of a 4-byte zero value followed by a single NUL byte. -------------------------------- In the current DWARF draft specification, no provision is made for providing a separate level of (limited) debugging information necessary to support tracebacks (only) through fully-debugged code (e.g. code in system libraries). A proposal to define such a level was submitted (by me) to the UI/PLSIG. This proposal was rejected by the UI/PLSIG for inclusion into the DWARF version 1 specification for two reasons. First, it was felt (by the PLSIG) that the issues involved in supporting a "traceback only" subset of DWARF were not well understood. Second, and perhaps more importantly, the PLSIG is already having enough trouble agreeing on what it means to be "conforming" to the DWARF specification, and it was felt that trying to specify multiple different *levels* of conformance would only complicate our discussions of this already divisive issue. Nonetheless, the GNU implementation of DWARF provides an abbreviated "traceback only" level of debug-info production for use with fully-debugged "system library" code. This level should only be used for fully debugged system library code, and even then, it should only be used where there is a very strong need to conserve disk space. This abbreviated level of debug-info production can be used by specifying the -g1 option on the compilation command line. -------------------------------- As mentioned above, the GNU implementation of DWARF currently uses the DWARF version 2 (draft) approach for inline functions (and inlined instances thereof). This is used in preference to the version 1 approach because (quite simply) the version 1 approach is highly brain-damaged and probably unworkable. -------------------------------- GNU DWARF Representation of GNU C Extensions to ANSI C ------------------------------------------------------ The file dwarfout.c has been designed and implemented so as to provide some reasonable DWARF representation for each and every declarative construct which is accepted by the GNU C compiler. Since the GNU C compiler accepts a superset of ANSI C, this means that there are some cases in which the DWARF information produced by GCC must take some liberties in improvising DWARF representations for declarations which are only valid in (extended) GNU C. In particular, GNU C provides at least three significant extensions to ANSI C when it comes to declarations. These are (1) inline functions, and (2) dynamic arrays, and (3) incomplete enum types. (See the GCC manual for more information on these GNU extensions to ANSI C.) When used, these GNU C extensions are represented (in the generated DWARF output of GCC) in the most natural and intuitively obvious ways. In the case of inline functions, the DWARF representation is exactly as called for in the DWARF version 2 (draft) specification for an identical function written in C++; i.e. we "reuse" the representation of inline functions which has been defined for C++ to support this GNU C extension. In the case of dynamic arrays, we use the most obvious representational mechanism available; i.e. an array type in which the upper bound of some dimension (usually the first and only dimension) is a variable rather than a constant. (See the DWARF version 1 specification for more details.) In the case of incomplete enum types, such types are represented simply as TAG_enumeration_type DIEs which DO NOT contain either AT_byte_size attributes or AT_element_list attributes. -------------------------------- Future Directions ----------------- The codes, formats, and other paraphernalia necessary to provide proper support for symbolic debugging for the C++ language are still being worked on by the UI/PLSIG. The vast majority of the additions to DWARF which will be needed to completely support C++ have already been hashed out and agreed upon, but a few small issues (e.g. anonymous unions, access declarations) are still being discussed. Also, we in the PLSIG are still discussing whether or not we need to do anything special for C++ templates. (At this time it is not yet clear whether we even need to do anything special for these.) With regard to FORTRAN, the UI/PLSIG has defined what is believed to be a complete and sufficient set of codes and rules for adequately representing all of FORTRAN 77, and most of Fortran 90 in DWARF. While some support for this has been implemented in dwarfout.c, further implementation and testing is needed. GNU DWARF support for other languages (i.e. Pascal and Modula) is a moot issue until there are GNU front-ends for these other languages. As currently defined, DWARF only describes a (binary) language which can be used to communicate symbolic debugging information from a compiler through an assembler and a linker, to a debugger. There is no clear specification of what processing should be (or must be) done by the assembler and/or the linker. Fortunately, the role of the assembler is easily inferred (by anyone knowledgeable about assemblers) just by looking at examples of assembly-level DWARF code. Sadly though, the allowable (or required) processing steps performed by a linker are harder to infer and (perhaps) even harder to agree upon. There are several forms of very useful `post-processing' steps which intelligent linkers *could* (in theory) perform on object files containing DWARF, but any and all such link-time transformations are currently both disallowed and unspecified. In particular, possible link-time transformations of DWARF code which could provide significant benefits include (but are not limited to): Commonization of duplicate DIEs obtained from multiple input (object) files. Cross-compilation type checking based upon DWARF type information for objects and functions. Other possible `compacting' transformations designed to save disk space and to reduce linker & debugger I/O activity. */ #include "config.h" #ifdef DWARF_DEBUGGING_INFO #include "system.h" #include "dwarf.h" #include "tree.h" #include "flags.h" #include "function.h" #include "rtl.h" #include "hard-reg-set.h" #include "insn-config.h" #include "reload.h" #include "output.h" #include "dwarf2asm.h" #include "toplev.h" #include "tm_p.h" #include "debug.h" #include "langhooks.h" /* NOTE: In the comments in this file, many references are made to so called "Debugging Information Entries". For the sake of brevity, this term is abbreviated to `DIE' throughout the remainder of this file. */ /* Note that the implementation of C++ support herein is (as yet) unfinished. If you want to try to complete it, more power to you. */ /* How to start an assembler comment. */ #ifndef ASM_COMMENT_START #define ASM_COMMENT_START ";#" #endif /* How to print out a register name. */ #ifndef PRINT_REG #define PRINT_REG(RTX, CODE, FILE) \ fprintf ((FILE), "%s", reg_names[REGNO (RTX)]) #endif /* Define a macro which returns non-zero for any tagged type which is used (directly or indirectly) in the specification of either some function's return type or some formal parameter of some function. We use this macro when we are operating in "terse" mode to help us know what tagged types have to be represented in Dwarf (even in terse mode) and which ones don't. A flag bit with this meaning really should be a part of the normal GCC ..._TYPE nodes, but at the moment, there is no such bit defined for these nodes. For now, we have to just fake it. It it safe for us to simply return zero for all complete tagged types (which will get forced out anyway if they were used in the specification of some formal or return type) and non-zero for all incomplete tagged types. */ #define TYPE_USED_FOR_FUNCTION(tagged_type) (TYPE_SIZE (tagged_type) == 0) /* Define a macro which returns non-zero for a TYPE_DECL which was implicitly generated for a tagged type. Note that unlike the gcc front end (which generates a NULL named TYPE_DECL node for each complete tagged type, each array type, and each function type node created) the g++ front end generates a _named_ TYPE_DECL node for each tagged type node created. These TYPE_DECLs have DECL_ARTIFICIAL set, so we know not to generate a DW_TAG_typedef DIE for them. */ #define TYPE_DECL_IS_STUB(decl) \ (DECL_NAME (decl) == NULL \ || (DECL_ARTIFICIAL (decl) \ && is_tagged_type (TREE_TYPE (decl)) \ && decl == TYPE_STUB_DECL (TREE_TYPE (decl)))) /* Maximum size (in bytes) of an artificially generated label. */ #define MAX_ARTIFICIAL_LABEL_BYTES 30 /* Structure to keep track of source filenames. */ struct filename_entry { unsigned number; const char * name; }; typedef struct filename_entry filename_entry; /* Pointer to an array of elements, each one having the structure above. */ static filename_entry *filename_table; /* Total number of entries in the table (i.e. array) pointed to by `filename_table'. This is the *total* and includes both used and unused slots. */ static unsigned ft_entries_allocated; /* Number of entries in the filename_table which are actually in use. */ static unsigned ft_entries; /* Size (in elements) of increments by which we may expand the filename table. Actually, a single hunk of space of this size should be enough for most typical programs. */ #define FT_ENTRIES_INCREMENT 64 /* Local pointer to the name of the main input file. Initialized in dwarfout_init. */ static const char *primary_filename; /* Counter to generate unique names for DIEs. */ static unsigned next_unused_dienum = 1; /* Number of the DIE which is currently being generated. */ static unsigned current_dienum; /* Number to use for the special "pubname" label on the next DIE which represents a function or data object defined in this compilation unit which has "extern" linkage. */ static int next_pubname_number = 0; #define NEXT_DIE_NUM pending_sibling_stack[pending_siblings-1] /* Pointer to a dynamically allocated list of pre-reserved and still pending sibling DIE numbers. Note that this list will grow as needed. */ static unsigned *pending_sibling_stack; /* Counter to keep track of the number of pre-reserved and still pending sibling DIE numbers. */ static unsigned pending_siblings; /* The currently allocated size of the above list (expressed in number of list elements). */ static unsigned pending_siblings_allocated; /* Size (in elements) of increments by which we may expand the pending sibling stack. Actually, a single hunk of space of this size should be enough for most typical programs. */ #define PENDING_SIBLINGS_INCREMENT 64 /* Non-zero if we are performing our file-scope finalization pass and if we should force out Dwarf descriptions of any and all file-scope tagged types which are still incomplete types. */ static int finalizing = 0; /* A pointer to the base of a list of pending types which we haven't generated DIEs for yet, but which we will have to come back to later on. */ static tree *pending_types_list; /* Number of elements currently allocated for the pending_types_list. */ static unsigned pending_types_allocated; /* Number of elements of pending_types_list currently in use. */ static unsigned pending_types; /* Size (in elements) of increments by which we may expand the pending types list. Actually, a single hunk of space of this size should be enough for most typical programs. */ #define PENDING_TYPES_INCREMENT 64 /* A pointer to the base of a list of incomplete types which might be completed at some later time. */ static tree *incomplete_types_list; /* Number of elements currently allocated for the incomplete_types_list. */ static unsigned incomplete_types_allocated; /* Number of elements of incomplete_types_list currently in use. */ static unsigned incomplete_types; /* Size (in elements) of increments by which we may expand the incomplete types list. Actually, a single hunk of space of this size should be enough for most typical programs. */ #define INCOMPLETE_TYPES_INCREMENT 64 /* Pointer to an artificial RECORD_TYPE which we create in dwarfout_init. This is used in a hack to help us get the DIEs describing types of formal parameters to come *after* all of the DIEs describing the formal parameters themselves. That's necessary in order to be compatible with what the brain-damaged svr4 SDB debugger requires. */ static tree fake_containing_scope; /* A pointer to the ..._DECL node which we have most recently been working on. We keep this around just in case something about it looks screwy and we want to tell the user what the source coordinates for the actual declaration are. */ static tree dwarf_last_decl; /* A flag indicating that we are emitting the member declarations of a class, so member functions and variables should not be entirely emitted. This is a kludge to avoid passing a second argument to output_*_die. */ static int in_class; /* Forward declarations for functions defined in this file. */ static void dwarfout_init PARAMS ((const char *)); static void dwarfout_finish PARAMS ((const char *)); static void dwarfout_define PARAMS ((unsigned int, const char *)); static void dwarfout_undef PARAMS ((unsigned int, const char *)); static void dwarfout_start_source_file PARAMS ((unsigned, const char *)); static void dwarfout_start_source_file_check PARAMS ((unsigned, const char *)); static void dwarfout_end_source_file PARAMS ((unsigned)); static void dwarfout_end_source_file_check PARAMS ((unsigned)); static void dwarfout_begin_block PARAMS ((unsigned, unsigned)); static void dwarfout_end_block PARAMS ((unsigned, unsigned)); static void dwarfout_end_epilogue PARAMS ((void)); static void dwarfout_source_line PARAMS ((unsigned int, const char *)); static void dwarfout_end_prologue PARAMS ((unsigned int)); static void dwarfout_end_function PARAMS ((unsigned int)); static void dwarfout_function_decl PARAMS ((tree)); static void dwarfout_global_decl PARAMS ((tree)); static void dwarfout_deferred_inline_function PARAMS ((tree)); static void dwarfout_file_scope_decl PARAMS ((tree , int)); static const char *dwarf_tag_name PARAMS ((unsigned)); static const char *dwarf_attr_name PARAMS ((unsigned)); static const char *dwarf_stack_op_name PARAMS ((unsigned)); static const char *dwarf_typemod_name PARAMS ((unsigned)); static const char *dwarf_fmt_byte_name PARAMS ((unsigned)); static const char *dwarf_fund_type_name PARAMS ((unsigned)); static tree decl_ultimate_origin PARAMS ((tree)); static tree block_ultimate_origin PARAMS ((tree)); static tree decl_class_context PARAMS ((tree)); #if 0 static void output_unsigned_leb128 PARAMS ((unsigned long)); static void output_signed_leb128 PARAMS ((long)); #endif static int fundamental_type_code PARAMS ((tree)); static tree root_type_1 PARAMS ((tree, int)); static tree root_type PARAMS ((tree)); static void write_modifier_bytes_1 PARAMS ((tree, int, int, int)); static void write_modifier_bytes PARAMS ((tree, int, int)); static inline int type_is_fundamental PARAMS ((tree)); static void equate_decl_number_to_die_number PARAMS ((tree)); static inline void equate_type_number_to_die_number PARAMS ((tree)); static void output_reg_number PARAMS ((rtx)); static void output_mem_loc_descriptor PARAMS ((rtx)); static void output_loc_descriptor PARAMS ((rtx)); static void output_bound_representation PARAMS ((tree, unsigned, int)); static void output_enumeral_list PARAMS ((tree)); static inline HOST_WIDE_INT ceiling PARAMS ((HOST_WIDE_INT, unsigned int)); static inline tree field_type PARAMS ((tree)); static inline unsigned int simple_type_align_in_bits PARAMS ((tree)); static inline unsigned HOST_WIDE_INT simple_type_size_in_bits PARAMS ((tree)); static HOST_WIDE_INT field_byte_offset PARAMS ((tree)); static inline void sibling_attribute PARAMS ((void)); static void location_attribute PARAMS ((rtx)); static void data_member_location_attribute PARAMS ((tree)); static void const_value_attribute PARAMS ((rtx)); static void location_or_const_value_attribute PARAMS ((tree)); static inline void name_attribute PARAMS ((const char *)); static inline void fund_type_attribute PARAMS ((unsigned)); static void mod_fund_type_attribute PARAMS ((tree, int, int)); static inline void user_def_type_attribute PARAMS ((tree)); static void mod_u_d_type_attribute PARAMS ((tree, int, int)); #ifdef USE_ORDERING_ATTRIBUTE static inline void ordering_attribute PARAMS ((unsigned)); #endif /* defined(USE_ORDERING_ATTRIBUTE) */ static void subscript_data_attribute PARAMS ((tree)); static void byte_size_attribute PARAMS ((tree)); static inline void bit_offset_attribute PARAMS ((tree)); static inline void bit_size_attribute PARAMS ((tree)); static inline void element_list_attribute PARAMS ((tree)); static inline void stmt_list_attribute PARAMS ((const char *)); static inline void low_pc_attribute PARAMS ((const char *)); static inline void high_pc_attribute PARAMS ((const char *)); static inline void body_begin_attribute PARAMS ((const char *)); static inline void body_end_attribute PARAMS ((const char *)); static inline void language_attribute PARAMS ((unsigned)); static inline void member_attribute PARAMS ((tree)); #if 0 static inline void string_length_attribute PARAMS ((tree)); #endif static inline void comp_dir_attribute PARAMS ((const char *)); static inline void sf_names_attribute PARAMS ((const char *)); static inline void src_info_attribute PARAMS ((const char *)); static inline void mac_info_attribute PARAMS ((const char *)); static inline void prototyped_attribute PARAMS ((tree)); static inline void producer_attribute PARAMS ((const char *)); static inline void inline_attribute PARAMS ((tree)); static inline void containing_type_attribute PARAMS ((tree)); static inline void abstract_origin_attribute PARAMS ((tree)); #ifdef DWARF_DECL_COORDINATES static inline void src_coords_attribute PARAMS ((unsigned, unsigned)); #endif /* defined(DWARF_DECL_COORDINATES) */ static inline void pure_or_virtual_attribute PARAMS ((tree)); static void name_and_src_coords_attributes PARAMS ((tree)); static void type_attribute PARAMS ((tree, int, int)); static const char *type_tag PARAMS ((tree)); static inline void dienum_push PARAMS ((void)); static inline void dienum_pop PARAMS ((void)); static inline tree member_declared_type PARAMS ((tree)); static const char *function_start_label PARAMS ((tree)); static void output_array_type_die PARAMS ((void *)); static void output_set_type_die PARAMS ((void *)); #if 0 static void output_entry_point_die PARAMS ((void *)); #endif static void output_inlined_enumeration_type_die PARAMS ((void *)); static void output_inlined_structure_type_die PARAMS ((void *)); static void output_inlined_union_type_die PARAMS ((void *)); static void output_enumeration_type_die PARAMS ((void *)); static void output_formal_parameter_die PARAMS ((void *)); static void output_global_subroutine_die PARAMS ((void *)); static void output_global_variable_die PARAMS ((void *)); static void output_label_die PARAMS ((void *)); static void output_lexical_block_die PARAMS ((void *)); static void output_inlined_subroutine_die PARAMS ((void *)); static void output_local_variable_die PARAMS ((void *)); static void output_member_die PARAMS ((void *)); #if 0 static void output_pointer_type_die PARAMS ((void *)); static void output_reference_type_die PARAMS ((void *)); #endif static void output_ptr_to_mbr_type_die PARAMS ((void *)); static void output_compile_unit_die PARAMS ((void *)); static void output_string_type_die PARAMS ((void *)); static void output_inheritance_die PARAMS ((void *)); static void output_structure_type_die PARAMS ((void *)); static void output_local_subroutine_die PARAMS ((void *)); static void output_subroutine_type_die PARAMS ((void *)); static void output_typedef_die PARAMS ((void *)); static void output_union_type_die PARAMS ((void *)); static void output_unspecified_parameters_die PARAMS ((void *)); static void output_padded_null_die PARAMS ((void *)); static void output_die PARAMS ((void (*)(void *), void *)); static void end_sibling_chain PARAMS ((void)); static void output_formal_types PARAMS ((tree)); static void pend_type PARAMS ((tree)); static int type_ok_for_scope PARAMS ((tree, tree)); static void output_pending_types_for_scope PARAMS ((tree)); static void output_type PARAMS ((tree, tree)); static void output_tagged_type_instantiation PARAMS ((tree)); static void output_block PARAMS ((tree, int)); static void output_decls_for_scope PARAMS ((tree, int)); static void output_decl PARAMS ((tree, tree)); static void shuffle_filename_entry PARAMS ((filename_entry *)); static void generate_new_sfname_entry PARAMS ((void)); static unsigned lookup_filename PARAMS ((const char *)); static void generate_srcinfo_entry PARAMS ((unsigned, unsigned)); static void generate_macinfo_entry PARAMS ((unsigned int, rtx, const char *)); static int is_pseudo_reg PARAMS ((rtx)); static tree type_main_variant PARAMS ((tree)); static int is_tagged_type PARAMS ((tree)); static int is_redundant_typedef PARAMS ((tree)); static void add_incomplete_type PARAMS ((tree)); static void retry_incomplete_types PARAMS ((void)); /* Definitions of defaults for assembler-dependent names of various pseudo-ops and section names. Theses may be overridden in your tm.h file (if necessary) for your particular assembler. The default values provided here correspond to what is expected by "standard" AT&T System V.4 assemblers. */ #ifndef FILE_ASM_OP #define FILE_ASM_OP "\t.file\t" #endif #ifndef SET_ASM_OP #define SET_ASM_OP "\t.set\t" #endif /* Pseudo-ops for pushing the current section onto the section stack (and simultaneously changing to a new section) and for poping back to the section we were in immediately before this one. Note that most svr4 assemblers only maintain a one level stack... you can push all the sections you want, but you can only pop out one level. (The sparc svr4 assembler is an exception to this general rule.) That's OK because we only use at most one level of the section stack herein. */ #ifndef PUSHSECTION_ASM_OP #define PUSHSECTION_ASM_OP "\t.section\t" #endif #ifndef POPSECTION_ASM_OP #define POPSECTION_ASM_OP "\t.previous" #endif /* The default format used by the ASM_OUTPUT_PUSH_SECTION macro (see below) to print the PUSHSECTION_ASM_OP and the section name. The default here works for almost all svr4 assemblers, except for the sparc, where the section name must be enclosed in double quotes. (See sparcv4.h.) */ #ifndef PUSHSECTION_FORMAT #define PUSHSECTION_FORMAT "%s%s\n" #endif #ifndef DEBUG_SECTION #define DEBUG_SECTION ".debug" #endif #ifndef LINE_SECTION #define LINE_SECTION ".line" #endif #ifndef DEBUG_SFNAMES_SECTION #define DEBUG_SFNAMES_SECTION ".debug_sfnames" #endif #ifndef DEBUG_SRCINFO_SECTION #define DEBUG_SRCINFO_SECTION ".debug_srcinfo" #endif #ifndef DEBUG_MACINFO_SECTION #define DEBUG_MACINFO_SECTION ".debug_macinfo" #endif #ifndef DEBUG_PUBNAMES_SECTION #define DEBUG_PUBNAMES_SECTION ".debug_pubnames" #endif #ifndef DEBUG_ARANGES_SECTION #define DEBUG_ARANGES_SECTION ".debug_aranges" #endif #ifndef TEXT_SECTION_NAME #define TEXT_SECTION_NAME ".text" #endif #ifndef DATA_SECTION_NAME #define DATA_SECTION_NAME ".data" #endif #ifndef DATA1_SECTION_NAME #define DATA1_SECTION_NAME ".data1" #endif #ifndef RODATA_SECTION_NAME #define RODATA_SECTION_NAME ".rodata" #endif #ifndef RODATA1_SECTION_NAME #define RODATA1_SECTION_NAME ".rodata1" #endif #ifndef BSS_SECTION_NAME #define BSS_SECTION_NAME ".bss" #endif /* Definitions of defaults for formats and names of various special (artificial) labels which may be generated within this file (when the -g options is used and DWARF_DEBUGGING_INFO is in effect. If necessary, these may be overridden from within your tm.h file, but typically, you should never need to override these. These labels have been hacked (temporarily) so that they all begin with a `.L' sequence so as to appease the stock sparc/svr4 assembler and the stock m88k/svr4 assembler, both of which need to see .L at the start of a label in order to prevent that label from going into the linker symbol table). When I get time, I'll have to fix this the right way so that we will use ASM_GENERATE_INTERNAL_LABEL and ASM_OUTPUT_INTERNAL_LABEL herein, but that will require a rather massive set of changes. For the moment, the following definitions out to produce the right results for all svr4 and svr3 assemblers. -- rfg */ #ifndef TEXT_BEGIN_LABEL #define TEXT_BEGIN_LABEL "*.L_text_b" #endif #ifndef TEXT_END_LABEL #define TEXT_END_LABEL "*.L_text_e" #endif #ifndef DATA_BEGIN_LABEL #define DATA_BEGIN_LABEL "*.L_data_b" #endif #ifndef DATA_END_LABEL #define DATA_END_LABEL "*.L_data_e" #endif #ifndef DATA1_BEGIN_LABEL #define DATA1_BEGIN_LABEL "*.L_data1_b" #endif #ifndef DATA1_END_LABEL #define DATA1_END_LABEL "*.L_data1_e" #endif #ifndef RODATA_BEGIN_LABEL #define RODATA_BEGIN_LABEL "*.L_rodata_b" #endif #ifndef RODATA_END_LABEL #define RODATA_END_LABEL "*.L_rodata_e" #endif #ifndef RODATA1_BEGIN_LABEL #define RODATA1_BEGIN_LABEL "*.L_rodata1_b" #endif #ifndef RODATA1_END_LABEL #define RODATA1_END_LABEL "*.L_rodata1_e" #endif #ifndef BSS_BEGIN_LABEL #define BSS_BEGIN_LABEL "*.L_bss_b" #endif #ifndef BSS_END_LABEL #define BSS_END_LABEL "*.L_bss_e" #endif #ifndef LINE_BEGIN_LABEL #define LINE_BEGIN_LABEL "*.L_line_b" #endif #ifndef LINE_LAST_ENTRY_LABEL #define LINE_LAST_ENTRY_LABEL "*.L_line_last" #endif #ifndef LINE_END_LABEL #define LINE_END_LABEL "*.L_line_e" #endif #ifndef DEBUG_BEGIN_LABEL #define DEBUG_BEGIN_LABEL "*.L_debug_b" #endif #ifndef SFNAMES_BEGIN_LABEL #define SFNAMES_BEGIN_LABEL "*.L_sfnames_b" #endif #ifndef SRCINFO_BEGIN_LABEL #define SRCINFO_BEGIN_LABEL "*.L_srcinfo_b" #endif #ifndef MACINFO_BEGIN_LABEL #define MACINFO_BEGIN_LABEL "*.L_macinfo_b" #endif #ifndef DEBUG_ARANGES_BEGIN_LABEL #define DEBUG_ARANGES_BEGIN_LABEL "*.L_debug_aranges_begin" #endif #ifndef DEBUG_ARANGES_END_LABEL #define DEBUG_ARANGES_END_LABEL "*.L_debug_aranges_end" #endif #ifndef DIE_BEGIN_LABEL_FMT #define DIE_BEGIN_LABEL_FMT "*.L_D%u" #endif #ifndef DIE_END_LABEL_FMT #define DIE_END_LABEL_FMT "*.L_D%u_e" #endif #ifndef PUB_DIE_LABEL_FMT #define PUB_DIE_LABEL_FMT "*.L_P%u" #endif #ifndef BLOCK_BEGIN_LABEL_FMT #define BLOCK_BEGIN_LABEL_FMT "*.L_B%u" #endif #ifndef BLOCK_END_LABEL_FMT #define BLOCK_END_LABEL_FMT "*.L_B%u_e" #endif #ifndef SS_BEGIN_LABEL_FMT #define SS_BEGIN_LABEL_FMT "*.L_s%u" #endif #ifndef SS_END_LABEL_FMT #define SS_END_LABEL_FMT "*.L_s%u_e" #endif #ifndef EE_BEGIN_LABEL_FMT #define EE_BEGIN_LABEL_FMT "*.L_e%u" #endif #ifndef EE_END_LABEL_FMT #define EE_END_LABEL_FMT "*.L_e%u_e" #endif #ifndef MT_BEGIN_LABEL_FMT #define MT_BEGIN_LABEL_FMT "*.L_t%u" #endif #ifndef MT_END_LABEL_FMT #define MT_END_LABEL_FMT "*.L_t%u_e" #endif #ifndef LOC_BEGIN_LABEL_FMT #define LOC_BEGIN_LABEL_FMT "*.L_l%u" #endif #ifndef LOC_END_LABEL_FMT #define LOC_END_LABEL_FMT "*.L_l%u_e" #endif #ifndef BOUND_BEGIN_LABEL_FMT #define BOUND_BEGIN_LABEL_FMT "*.L_b%u_%u_%c" #endif #ifndef BOUND_END_LABEL_FMT #define BOUND_END_LABEL_FMT "*.L_b%u_%u_%c_e" #endif #ifndef BODY_BEGIN_LABEL_FMT #define BODY_BEGIN_LABEL_FMT "*.L_b%u" #endif #ifndef BODY_END_LABEL_FMT #define BODY_END_LABEL_FMT "*.L_b%u_e" #endif #ifndef FUNC_END_LABEL_FMT #define FUNC_END_LABEL_FMT "*.L_f%u_e" #endif #ifndef TYPE_NAME_FMT #define TYPE_NAME_FMT "*.L_T%u" #endif #ifndef DECL_NAME_FMT #define DECL_NAME_FMT "*.L_E%u" #endif #ifndef LINE_CODE_LABEL_FMT #define LINE_CODE_LABEL_FMT "*.L_LC%u" #endif #ifndef SFNAMES_ENTRY_LABEL_FMT #define SFNAMES_ENTRY_LABEL_FMT "*.L_F%u" #endif #ifndef LINE_ENTRY_LABEL_FMT #define LINE_ENTRY_LABEL_FMT "*.L_LE%u" #endif /* Definitions of defaults for various types of primitive assembly language output operations. If necessary, these may be overridden from within your tm.h file, but typically, you shouldn't need to override these. */ #ifndef ASM_OUTPUT_PUSH_SECTION #define ASM_OUTPUT_PUSH_SECTION(FILE, SECTION) \ fprintf ((FILE), PUSHSECTION_FORMAT, PUSHSECTION_ASM_OP, SECTION) #endif #ifndef ASM_OUTPUT_POP_SECTION #define ASM_OUTPUT_POP_SECTION(FILE) \ fprintf ((FILE), "%s\n", POPSECTION_ASM_OP) #endif #ifndef ASM_OUTPUT_DWARF_DELTA2 #define ASM_OUTPUT_DWARF_DELTA2(FILE,LABEL1,LABEL2) \ dw2_asm_output_delta (2, LABEL1, LABEL2, NULL) #endif #ifndef ASM_OUTPUT_DWARF_DELTA4 #define ASM_OUTPUT_DWARF_DELTA4(FILE,LABEL1,LABEL2) \ dw2_asm_output_delta (4, LABEL1, LABEL2, NULL) #endif #ifndef ASM_OUTPUT_DWARF_TAG #define ASM_OUTPUT_DWARF_TAG(FILE,TAG) \ dw2_asm_output_data (2, TAG, "%s", dwarf_tag_name (TAG)); #endif #ifndef ASM_OUTPUT_DWARF_ATTRIBUTE #define ASM_OUTPUT_DWARF_ATTRIBUTE(FILE,ATTR) \ dw2_asm_output_data (2, ATTR, "%s", dwarf_attr_name (ATTR)) #endif #ifndef ASM_OUTPUT_DWARF_STACK_OP #define ASM_OUTPUT_DWARF_STACK_OP(FILE,OP) \ dw2_asm_output_data (1, OP, "%s", dwarf_stack_op_name (OP)) #endif #ifndef ASM_OUTPUT_DWARF_FUND_TYPE #define ASM_OUTPUT_DWARF_FUND_TYPE(FILE,FT) \ dw2_asm_output_data (2, FT, "%s", dwarf_fund_type_name (FT)) #endif #ifndef ASM_OUTPUT_DWARF_FMT_BYTE #define ASM_OUTPUT_DWARF_FMT_BYTE(FILE,FMT) \ dw2_asm_output_data (1, FMT, "%s", dwarf_fmt_byte_name (FMT)); #endif #ifndef ASM_OUTPUT_DWARF_TYPE_MODIFIER #define ASM_OUTPUT_DWARF_TYPE_MODIFIER(FILE,MOD) \ dw2_asm_output_data (1, MOD, "%s", dwarf_typemod_name (MOD)); #endif #ifndef ASM_OUTPUT_DWARF_ADDR #define ASM_OUTPUT_DWARF_ADDR(FILE,LABEL) \ dw2_asm_output_addr (4, LABEL, NULL) #endif #ifndef ASM_OUTPUT_DWARF_ADDR_CONST #define ASM_OUTPUT_DWARF_ADDR_CONST(FILE,RTX) \ dw2_asm_output_addr_rtx (4, RTX, NULL) #endif #ifndef ASM_OUTPUT_DWARF_REF #define ASM_OUTPUT_DWARF_REF(FILE,LABEL) \ dw2_asm_output_addr (4, LABEL, NULL) #endif #ifndef ASM_OUTPUT_DWARF_DATA1 #define ASM_OUTPUT_DWARF_DATA1(FILE,VALUE) \ dw2_asm_output_data (1, VALUE, NULL) #endif #ifndef ASM_OUTPUT_DWARF_DATA2 #define ASM_OUTPUT_DWARF_DATA2(FILE,VALUE) \ dw2_asm_output_data (2, VALUE, NULL) #endif #ifndef ASM_OUTPUT_DWARF_DATA4 #define ASM_OUTPUT_DWARF_DATA4(FILE,VALUE) \ dw2_asm_output_data (4, VALUE, NULL) #endif #ifndef ASM_OUTPUT_DWARF_DATA8 #define ASM_OUTPUT_DWARF_DATA8(FILE,HIGH_VALUE,LOW_VALUE) \ dw2_asm_output_data (8, VALUE, NULL) #endif /* ASM_OUTPUT_DWARF_STRING is defined to output an ascii string, but to NOT issue a trailing newline. We define ASM_OUTPUT_DWARF_STRING_NEWLINE based on whether ASM_OUTPUT_DWARF_STRING is defined or not. If it is defined, we call it, then issue the line feed. If not, we supply a default definition of calling ASM_OUTPUT_ASCII */ #ifndef ASM_OUTPUT_DWARF_STRING #define ASM_OUTPUT_DWARF_STRING_NEWLINE(FILE,P) \ ASM_OUTPUT_ASCII ((FILE), P, strlen (P)+1) #else #define ASM_OUTPUT_DWARF_STRING_NEWLINE(FILE,P) \ ASM_OUTPUT_DWARF_STRING (FILE,P), ASM_OUTPUT_DWARF_STRING (FILE,"\n") #endif /* The debug hooks structure. */ const struct gcc_debug_hooks dwarf_debug_hooks = { dwarfout_init, dwarfout_finish, dwarfout_define, dwarfout_undef, dwarfout_start_source_file_check, dwarfout_end_source_file_check, dwarfout_begin_block, dwarfout_end_block, debug_true_tree, /* ignore_block */ dwarfout_source_line, /* source_line */ dwarfout_source_line, /* begin_prologue */ dwarfout_end_prologue, dwarfout_end_epilogue, debug_nothing_tree, /* begin_function */ dwarfout_end_function, dwarfout_function_decl, dwarfout_global_decl, dwarfout_deferred_inline_function, debug_nothing_tree, /* outlining_inline_function */ debug_nothing_rtx /* label */ }; /************************ general utility functions **************************/ static inline int is_pseudo_reg (rtl) rtx rtl; { return (((GET_CODE (rtl) == REG) && (REGNO (rtl) >= FIRST_PSEUDO_REGISTER)) || ((GET_CODE (rtl) == SUBREG) && (REGNO (SUBREG_REG (rtl)) >= FIRST_PSEUDO_REGISTER))); } static inline tree type_main_variant (type) tree type; { type = TYPE_MAIN_VARIANT (type); /* There really should be only one main variant among any group of variants of a given type (and all of the MAIN_VARIANT values for all members of the group should point to that one type) but sometimes the C front-end messes this up for array types, so we work around that bug here. */ if (TREE_CODE (type) == ARRAY_TYPE) { while (type != TYPE_MAIN_VARIANT (type)) type = TYPE_MAIN_VARIANT (type); } return type; } /* Return non-zero if the given type node represents a tagged type. */ static inline int is_tagged_type (type) tree type; { enum tree_code code = TREE_CODE (type); return (code == RECORD_TYPE || code == UNION_TYPE || code == QUAL_UNION_TYPE || code == ENUMERAL_TYPE); } static const char * dwarf_tag_name (tag) unsigned tag; { switch (tag) { case TAG_padding: return "TAG_padding"; case TAG_array_type: return "TAG_array_type"; case TAG_class_type: return "TAG_class_type"; case TAG_entry_point: return "TAG_entry_point"; case TAG_enumeration_type: return "TAG_enumeration_type"; case TAG_formal_parameter: return "TAG_formal_parameter"; case TAG_global_subroutine: return "TAG_global_subroutine"; case TAG_global_variable: return "TAG_global_variable"; case TAG_label: return "TAG_label"; case TAG_lexical_block: return "TAG_lexical_block"; case TAG_local_variable: return "TAG_local_variable"; case TAG_member: return "TAG_member"; case TAG_pointer_type: return "TAG_pointer_type"; case TAG_reference_type: return "TAG_reference_type"; case TAG_compile_unit: return "TAG_compile_unit"; case TAG_string_type: return "TAG_string_type"; case TAG_structure_type: return "TAG_structure_type"; case TAG_subroutine: return "TAG_subroutine"; case TAG_subroutine_type: return "TAG_subroutine_type"; case TAG_typedef: return "TAG_typedef"; case TAG_union_type: return "TAG_union_type"; case TAG_unspecified_parameters: return "TAG_unspecified_parameters"; case TAG_variant: return "TAG_variant"; case TAG_common_block: return "TAG_common_block"; case TAG_common_inclusion: return "TAG_common_inclusion"; case TAG_inheritance: return "TAG_inheritance"; case TAG_inlined_subroutine: return "TAG_inlined_subroutine"; case TAG_module: return "TAG_module"; case TAG_ptr_to_member_type: return "TAG_ptr_to_member_type"; case TAG_set_type: return "TAG_set_type"; case TAG_subrange_type: return "TAG_subrange_type"; case TAG_with_stmt: return "TAG_with_stmt"; /* GNU extensions. */ case TAG_format_label: return "TAG_format_label"; case TAG_namelist: return "TAG_namelist"; case TAG_function_template: return "TAG_function_template"; case TAG_class_template: return "TAG_class_template"; default: return "TAG_"; } } static const char * dwarf_attr_name (attr) unsigned attr; { switch (attr) { case AT_sibling: return "AT_sibling"; case AT_location: return "AT_location"; case AT_name: return "AT_name"; case AT_fund_type: return "AT_fund_type"; case AT_mod_fund_type: return "AT_mod_fund_type"; case AT_user_def_type: return "AT_user_def_type"; case AT_mod_u_d_type: return "AT_mod_u_d_type"; case AT_ordering: return "AT_ordering"; case AT_subscr_data: return "AT_subscr_data"; case AT_byte_size: return "AT_byte_size"; case AT_bit_offset: return "AT_bit_offset"; case AT_bit_size: return "AT_bit_size"; case AT_element_list: return "AT_element_list"; case AT_stmt_list: return "AT_stmt_list"; case AT_low_pc: return "AT_low_pc"; case AT_high_pc: return "AT_high_pc"; case AT_language: return "AT_language"; case AT_member: return "AT_member"; case AT_discr: return "AT_discr"; case AT_discr_value: return "AT_discr_value"; case AT_string_length: return "AT_string_length"; case AT_common_reference: return "AT_common_reference"; case AT_comp_dir: return "AT_comp_dir"; case AT_const_value_string: return "AT_const_value_string"; case AT_const_value_data2: return "AT_const_value_data2"; case AT_const_value_data4: return "AT_const_value_data4"; case AT_const_value_data8: return "AT_const_value_data8"; case AT_const_value_block2: return "AT_const_value_block2"; case AT_const_value_block4: return "AT_const_value_block4"; case AT_containing_type: return "AT_containing_type"; case AT_default_value_addr: return "AT_default_value_addr"; case AT_default_value_data2: return "AT_default_value_data2"; case AT_default_value_data4: return "AT_default_value_data4"; case AT_default_value_data8: return "AT_default_value_data8"; case AT_default_value_string: return "AT_default_value_string"; case AT_friends: return "AT_friends"; case AT_inline: return "AT_inline"; case AT_is_optional: return "AT_is_optional"; case AT_lower_bound_ref: return "AT_lower_bound_ref"; case AT_lower_bound_data2: return "AT_lower_bound_data2"; case AT_lower_bound_data4: return "AT_lower_bound_data4"; case AT_lower_bound_data8: return "AT_lower_bound_data8"; case AT_private: return "AT_private"; case AT_producer: return "AT_producer"; case AT_program: return "AT_program"; case AT_protected: return "AT_protected"; case AT_prototyped: return "AT_prototyped"; case AT_public: return "AT_public"; case AT_pure_virtual: return "AT_pure_virtual"; case AT_return_addr: return "AT_return_addr"; case AT_abstract_origin: return "AT_abstract_origin"; case AT_start_scope: return "AT_start_scope"; case AT_stride_size: return "AT_stride_size"; case AT_upper_bound_ref: return "AT_upper_bound_ref"; case AT_upper_bound_data2: return "AT_upper_bound_data2"; case AT_upper_bound_data4: return "AT_upper_bound_data4"; case AT_upper_bound_data8: return "AT_upper_bound_data8"; case AT_virtual: return "AT_virtual"; /* GNU extensions */ case AT_sf_names: return "AT_sf_names"; case AT_src_info: return "AT_src_info"; case AT_mac_info: return "AT_mac_info"; case AT_src_coords: return "AT_src_coords"; case AT_body_begin: return "AT_body_begin"; case AT_body_end: return "AT_body_end"; default: return "AT_"; } } static const char * dwarf_stack_op_name (op) unsigned op; { switch (op) { case OP_REG: return "OP_REG"; case OP_BASEREG: return "OP_BASEREG"; case OP_ADDR: return "OP_ADDR"; case OP_CONST: return "OP_CONST"; case OP_DEREF2: return "OP_DEREF2"; case OP_DEREF4: return "OP_DEREF4"; case OP_ADD: return "OP_ADD"; default: return "OP_"; } } static const char * dwarf_typemod_name (mod) unsigned mod; { switch (mod) { case MOD_pointer_to: return "MOD_pointer_to"; case MOD_reference_to: return "MOD_reference_to"; case MOD_const: return "MOD_const"; case MOD_volatile: return "MOD_volatile"; default: return "MOD_"; } } static const char * dwarf_fmt_byte_name (fmt) unsigned fmt; { switch (fmt) { case FMT_FT_C_C: return "FMT_FT_C_C"; case FMT_FT_C_X: return "FMT_FT_C_X"; case FMT_FT_X_C: return "FMT_FT_X_C"; case FMT_FT_X_X: return "FMT_FT_X_X"; case FMT_UT_C_C: return "FMT_UT_C_C"; case FMT_UT_C_X: return "FMT_UT_C_X"; case FMT_UT_X_C: return "FMT_UT_X_C"; case FMT_UT_X_X: return "FMT_UT_X_X"; case FMT_ET: return "FMT_ET"; default: return "FMT_"; } } static const char * dwarf_fund_type_name (ft) unsigned ft; { switch (ft) { case FT_char: return "FT_char"; case FT_signed_char: return "FT_signed_char"; case FT_unsigned_char: return "FT_unsigned_char"; case FT_short: return "FT_short"; case FT_signed_short: return "FT_signed_short"; case FT_unsigned_short: return "FT_unsigned_short"; case FT_integer: return "FT_integer"; case FT_signed_integer: return "FT_signed_integer"; case FT_unsigned_integer: return "FT_unsigned_integer"; case FT_long: return "FT_long"; case FT_signed_long: return "FT_signed_long"; case FT_unsigned_long: return "FT_unsigned_long"; case FT_pointer: return "FT_pointer"; case FT_float: return "FT_float"; case FT_dbl_prec_float: return "FT_dbl_prec_float"; case FT_ext_prec_float: return "FT_ext_prec_float"; case FT_complex: return "FT_complex"; case FT_dbl_prec_complex: return "FT_dbl_prec_complex"; case FT_void: return "FT_void"; case FT_boolean: return "FT_boolean"; case FT_ext_prec_complex: return "FT_ext_prec_complex"; case FT_label: return "FT_label"; /* GNU extensions. */ case FT_long_long: return "FT_long_long"; case FT_signed_long_long: return "FT_signed_long_long"; case FT_unsigned_long_long: return "FT_unsigned_long_long"; case FT_int8: return "FT_int8"; case FT_signed_int8: return "FT_signed_int8"; case FT_unsigned_int8: return "FT_unsigned_int8"; case FT_int16: return "FT_int16"; case FT_signed_int16: return "FT_signed_int16"; case FT_unsigned_int16: return "FT_unsigned_int16"; case FT_int32: return "FT_int32"; case FT_signed_int32: return "FT_signed_int32"; case FT_unsigned_int32: return "FT_unsigned_int32"; case FT_int64: return "FT_int64"; case FT_signed_int64: return "FT_signed_int64"; case FT_unsigned_int64: return "FT_unsigned_int64"; case FT_int128: return "FT_int128"; case FT_signed_int128: return "FT_signed_int128"; case FT_unsigned_int128: return "FT_unsigned_int128"; case FT_real32: return "FT_real32"; case FT_real64: return "FT_real64"; case FT_real96: return "FT_real96"; case FT_real128: return "FT_real128"; default: return "FT_"; } } /* Determine the "ultimate origin" of a decl. The decl may be an inlined instance of an inlined instance of a decl which is local to an inline function, so we have to trace all of the way back through the origin chain to find out what sort of node actually served as the original seed for the given block. */ static tree decl_ultimate_origin (decl) tree decl; { #ifdef ENABLE_CHECKING if (DECL_FROM_INLINE (DECL_ORIGIN (decl))) /* Since the DECL_ABSTRACT_ORIGIN for a DECL is supposed to be the most distant ancestor, this should never happen. */ abort (); #endif return DECL_ABSTRACT_ORIGIN (decl); } /* Determine the "ultimate origin" of a block. The block may be an inlined instance of an inlined instance of a block which is local to an inline function, so we have to trace all of the way back through the origin chain to find out what sort of node actually served as the original seed for the given block. */ static tree block_ultimate_origin (block) tree block; { tree immediate_origin = BLOCK_ABSTRACT_ORIGIN (block); if (immediate_origin == NULL) return NULL; else { tree ret_val; tree lookahead = immediate_origin; do { ret_val = lookahead; lookahead = (TREE_CODE (ret_val) == BLOCK) ? BLOCK_ABSTRACT_ORIGIN (ret_val) : NULL; } while (lookahead != NULL && lookahead != ret_val); return ret_val; } } /* Get the class to which DECL belongs, if any. In g++, the DECL_CONTEXT of a virtual function may refer to a base class, so we check the 'this' parameter. */ static tree decl_class_context (decl) tree decl; { tree context = NULL_TREE; if (TREE_CODE (decl) != FUNCTION_DECL || ! DECL_VINDEX (decl)) context = DECL_CONTEXT (decl); else context = TYPE_MAIN_VARIANT (TREE_TYPE (TREE_VALUE (TYPE_ARG_TYPES (TREE_TYPE (decl))))); if (context && !TYPE_P (context)) context = NULL_TREE; return context; } #if 0 static void output_unsigned_leb128 (value) unsigned long value; { unsigned long orig_value = value; do { unsigned byte = (value & 0x7f); value >>= 7; if (value != 0) /* more bytes to follow */ byte |= 0x80; dw2_asm_output_data (1, byte, "\t%s ULEB128 number - value = %lu", orig_value); } while (value != 0); } static void output_signed_leb128 (value) long value; { long orig_value = value; int negative = (value < 0); int more; do { unsigned byte = (value & 0x7f); value >>= 7; if (negative) value |= 0xfe000000; /* manually sign extend */ if (((value == 0) && ((byte & 0x40) == 0)) || ((value == -1) && ((byte & 0x40) == 1))) more = 0; else { byte |= 0x80; more = 1; } dw2_asm_output_data (1, byte, "\t%s SLEB128 number - value = %ld", orig_value); } while (more); } #endif /**************** utility functions for attribute functions ******************/ /* Given a pointer to a tree node for some type, return a Dwarf fundamental type code for the given type. This routine must only be called for GCC type nodes that correspond to Dwarf fundamental types. The current Dwarf draft specification calls for Dwarf fundamental types to accurately reflect the fact that a given type was either a "plain" integral type or an explicitly "signed" integral type. Unfortunately, we can't always do this, because GCC may already have thrown away the information about the precise way in which the type was originally specified, as in: typedef signed int my_type; struct s { my_type f; }; Since we may be stuck here without enough information to do exactly what is called for in the Dwarf draft specification, we do the best that we can under the circumstances and always use the "plain" integral fundamental type codes for int, short, and long types. That's probably good enough. The additional accuracy called for in the current DWARF draft specification is probably never even useful in practice. */ static int fundamental_type_code (type) tree type; { if (TREE_CODE (type) == ERROR_MARK) return 0; switch (TREE_CODE (type)) { case ERROR_MARK: return FT_void; case VOID_TYPE: return FT_void; case INTEGER_TYPE: /* Carefully distinguish all the standard types of C, without messing up if the language is not C. Note that we check only for the names that contain spaces; other names might occur by coincidence in other languages. */ if (TYPE_NAME (type) != 0 && TREE_CODE (TYPE_NAME (type)) == TYPE_DECL && DECL_NAME (TYPE_NAME (type)) != 0 && TREE_CODE (DECL_NAME (TYPE_NAME (type))) == IDENTIFIER_NODE) { const char *const name = IDENTIFIER_POINTER (DECL_NAME (TYPE_NAME (type))); if (!strcmp (name, "unsigned char")) return FT_unsigned_char; if (!strcmp (name, "signed char")) return FT_signed_char; if (!strcmp (name, "unsigned int")) return FT_unsigned_integer; if (!strcmp (name, "short int")) return FT_short; if (!strcmp (name, "short unsigned int")) return FT_unsigned_short; if (!strcmp (name, "long int")) return FT_long; if (!strcmp (name, "long unsigned int")) return FT_unsigned_long; if (!strcmp (name, "long long int")) return FT_long_long; /* Not grok'ed by svr4 SDB */ if (!strcmp (name, "long long unsigned int")) return FT_unsigned_long_long; /* Not grok'ed by svr4 SDB */ } /* Most integer types will be sorted out above, however, for the sake of special `array index' integer types, the following code is also provided. */ if (TYPE_PRECISION (type) == INT_TYPE_SIZE) return (TREE_UNSIGNED (type) ? FT_unsigned_integer : FT_integer); if (TYPE_PRECISION (type) == LONG_TYPE_SIZE) return (TREE_UNSIGNED (type) ? FT_unsigned_long : FT_long); if (TYPE_PRECISION (type) == LONG_LONG_TYPE_SIZE) return (TREE_UNSIGNED (type) ? FT_unsigned_long_long : FT_long_long); if (TYPE_PRECISION (type) == SHORT_TYPE_SIZE) return (TREE_UNSIGNED (type) ? FT_unsigned_short : FT_short); if (TYPE_PRECISION (type) == CHAR_TYPE_SIZE) return (TREE_UNSIGNED (type) ? FT_unsigned_char : FT_char); if (TYPE_MODE (type) == TImode) return (TREE_UNSIGNED (type) ? FT_unsigned_int128 : FT_int128); /* In C++, __java_boolean is an INTEGER_TYPE with precision == 1 */ if (TYPE_PRECISION (type) == 1) return FT_boolean; abort (); case REAL_TYPE: /* Carefully distinguish all the standard types of C, without messing up if the language is not C. */ if (TYPE_NAME (type) != 0 && TREE_CODE (TYPE_NAME (type)) == TYPE_DECL && DECL_NAME (TYPE_NAME (type)) != 0 && TREE_CODE (DECL_NAME (TYPE_NAME (type))) == IDENTIFIER_NODE) { const char *const name = IDENTIFIER_POINTER (DECL_NAME (TYPE_NAME (type))); /* Note that here we can run afoul of a serious bug in "classic" svr4 SDB debuggers. They don't seem to understand the FT_ext_prec_float type (even though they should). */ if (!strcmp (name, "long double")) return FT_ext_prec_float; } if (TYPE_PRECISION (type) == DOUBLE_TYPE_SIZE) { /* On the SH, when compiling with -m3e or -m4-single-only, both float and double are 32 bits. But since the debugger doesn't know about the subtarget, it always thinks double is 64 bits. So we have to tell the debugger that the type is float to make the output of the 'print' command etc. readable. */ if (DOUBLE_TYPE_SIZE == FLOAT_TYPE_SIZE && FLOAT_TYPE_SIZE == 32) return FT_float; return FT_dbl_prec_float; } if (TYPE_PRECISION (type) == FLOAT_TYPE_SIZE) return FT_float; /* Note that here we can run afoul of a serious bug in "classic" svr4 SDB debuggers. They don't seem to understand the FT_ext_prec_float type (even though they should). */ if (TYPE_PRECISION (type) == LONG_DOUBLE_TYPE_SIZE) return FT_ext_prec_float; abort (); case COMPLEX_TYPE: return FT_complex; /* GNU FORTRAN COMPLEX type. */ case CHAR_TYPE: return FT_char; /* GNU Pascal CHAR type. Not used in C. */ case BOOLEAN_TYPE: return FT_boolean; /* GNU FORTRAN BOOLEAN type. */ default: abort (); /* No other TREE_CODEs are Dwarf fundamental types. */ } return 0; } /* Given a pointer to an arbitrary ..._TYPE tree node, return a pointer to the Dwarf "root" type for the given input type. The Dwarf "root" type of a given type is generally the same as the given type, except that if the given type is a pointer or reference type, then the root type of the given type is the root type of the "basis" type for the pointer or reference type. (This definition of the "root" type is recursive.) Also, the root type of a `const' qualified type or a `volatile' qualified type is the root type of the given type without the qualifiers. */ static tree root_type_1 (type, count) tree type; int count; { /* Give up after searching 1000 levels, in case this is a recursive pointer type. Such types are possible in Ada, but it is not possible to represent them in DWARF1 debug info. */ if (count > 1000) return error_mark_node; switch (TREE_CODE (type)) { case ERROR_MARK: return error_mark_node; case POINTER_TYPE: case REFERENCE_TYPE: return root_type_1 (TREE_TYPE (type), count+1); default: return type; } } static tree root_type (type) tree type; { type = root_type_1 (type, 0); if (type != error_mark_node) type = type_main_variant (type); return type; } /* Given a pointer to an arbitrary ..._TYPE tree node, write out a sequence of zero or more Dwarf "type-modifier" bytes applicable to the type. */ static void write_modifier_bytes_1 (type, decl_const, decl_volatile, count) tree type; int decl_const; int decl_volatile; int count; { if (TREE_CODE (type) == ERROR_MARK) return; /* Give up after searching 1000 levels, in case this is a recursive pointer type. Such types are possible in Ada, but it is not possible to represent them in DWARF1 debug info. */ if (count > 1000) return; if (TYPE_READONLY (type) || decl_const) ASM_OUTPUT_DWARF_TYPE_MODIFIER (asm_out_file, MOD_const); if (TYPE_VOLATILE (type) || decl_volatile) ASM_OUTPUT_DWARF_TYPE_MODIFIER (asm_out_file, MOD_volatile); switch (TREE_CODE (type)) { case POINTER_TYPE: ASM_OUTPUT_DWARF_TYPE_MODIFIER (asm_out_file, MOD_pointer_to); write_modifier_bytes_1 (TREE_TYPE (type), 0, 0, count+1); return; case REFERENCE_TYPE: ASM_OUTPUT_DWARF_TYPE_MODIFIER (asm_out_file, MOD_reference_to); write_modifier_bytes_1 (TREE_TYPE (type), 0, 0, count+1); return; case ERROR_MARK: default: return; } } static void write_modifier_bytes (type, decl_const, decl_volatile) tree type; int decl_const; int decl_volatile; { write_modifier_bytes_1 (type, decl_const, decl_volatile, 0); } /* Given a pointer to an arbitrary ..._TYPE tree node, return non-zero if the given input type is a Dwarf "fundamental" type. Otherwise return zero. */ static inline int type_is_fundamental (type) tree type; { switch (TREE_CODE (type)) { case ERROR_MARK: case VOID_TYPE: case INTEGER_TYPE: case REAL_TYPE: case COMPLEX_TYPE: case BOOLEAN_TYPE: case CHAR_TYPE: return 1; case SET_TYPE: case ARRAY_TYPE: case RECORD_TYPE: case UNION_TYPE: case QUAL_UNION_TYPE: case ENUMERAL_TYPE: case FUNCTION_TYPE: case METHOD_TYPE: case POINTER_TYPE: case REFERENCE_TYPE: case FILE_TYPE: case OFFSET_TYPE: case LANG_TYPE: case VECTOR_TYPE: return 0; default: abort (); } return 0; } /* Given a pointer to some ..._DECL tree node, generate an assembly language equate directive which will associate a symbolic name with the current DIE. The name used is an artificial label generated from the DECL_UID number associated with the given decl node. The name it gets equated to is the symbolic label that we (previously) output at the start of the DIE that we are currently generating. Calling this function while generating some "decl related" form of DIE makes it possible to later refer to the DIE which represents the given decl simply by re-generating the symbolic name from the ..._DECL node's UID number. */ static void equate_decl_number_to_die_number (decl) tree decl; { /* In the case where we are generating a DIE for some ..._DECL node which represents either some inline function declaration or some entity declared within an inline function declaration/definition, setup a symbolic name for the current DIE so that we have a name for this DIE that we can easily refer to later on within AT_abstract_origin attributes. */ char decl_label[MAX_ARTIFICIAL_LABEL_BYTES]; char die_label[MAX_ARTIFICIAL_LABEL_BYTES]; sprintf (decl_label, DECL_NAME_FMT, DECL_UID (decl)); sprintf (die_label, DIE_BEGIN_LABEL_FMT, current_dienum); ASM_OUTPUT_DEF (asm_out_file, decl_label, die_label); } /* Given a pointer to some ..._TYPE tree node, generate an assembly language equate directive which will associate a symbolic name with the current DIE. The name used is an artificial label generated from the TYPE_UID number associated with the given type node. The name it gets equated to is the symbolic label that we (previously) output at the start of the DIE that we are currently generating. Calling this function while generating some "type related" form of DIE makes it easy to later refer to the DIE which represents the given type simply by re-generating the alternative name from the ..._TYPE node's UID number. */ static inline void equate_type_number_to_die_number (type) tree type; { char type_label[MAX_ARTIFICIAL_LABEL_BYTES]; char die_label[MAX_ARTIFICIAL_LABEL_BYTES]; /* We are generating a DIE to represent the main variant of this type (i.e the type without any const or volatile qualifiers) so in order to get the equate to come out right, we need to get the main variant itself here. */ type = type_main_variant (type); sprintf (type_label, TYPE_NAME_FMT, TYPE_UID (type)); sprintf (die_label, DIE_BEGIN_LABEL_FMT, current_dienum); ASM_OUTPUT_DEF (asm_out_file, type_label, die_label); } static void output_reg_number (rtl) rtx rtl; { unsigned regno = REGNO (rtl); if (regno >= DWARF_FRAME_REGISTERS) { warning_with_decl (dwarf_last_decl, "internal regno botch: `%s' has regno = %d\n", regno); regno = 0; } dw2_assemble_integer (4, GEN_INT (DBX_REGISTER_NUMBER (regno))); if (flag_debug_asm) { fprintf (asm_out_file, "\t%s ", ASM_COMMENT_START); PRINT_REG (rtl, 0, asm_out_file); } fputc ('\n', asm_out_file); } /* The following routine is a nice and simple transducer. It converts the RTL for a variable or parameter (resident in memory) into an equivalent Dwarf representation of a mechanism for getting the address of that same variable onto the top of a hypothetical "address evaluation" stack. When creating memory location descriptors, we are effectively trans- forming the RTL for a memory-resident object into its Dwarf postfix expression equivalent. This routine just recursively descends an RTL tree, turning it into Dwarf postfix code as it goes. */ static void output_mem_loc_descriptor (rtl) rtx rtl; { /* Note that for a dynamically sized array, the location we will generate a description of here will be the lowest numbered location which is actually within the array. That's *not* necessarily the same as the zeroth element of the array. */ #ifdef ASM_SIMPLIFY_DWARF_ADDR rtl = ASM_SIMPLIFY_DWARF_ADDR (rtl); #endif switch (GET_CODE (rtl)) { case SUBREG: /* The case of a subreg may arise when we have a local (register) variable or a formal (register) parameter which doesn't quite fill up an entire register. For now, just assume that it is legitimate to make the Dwarf info refer to the whole register which contains the given subreg. */ rtl = SUBREG_REG (rtl); /* Drop thru. */ case REG: /* Whenever a register number forms a part of the description of the method for calculating the (dynamic) address of a memory resident object, DWARF rules require the register number to be referred to as a "base register". This distinction is not based in any way upon what category of register the hardware believes the given register belongs to. This is strictly DWARF terminology we're dealing with here. Note that in cases where the location of a memory-resident data object could be expressed as: OP_ADD (OP_BASEREG (basereg), OP_CONST (0)) the actual DWARF location descriptor that we generate may just be OP_BASEREG (basereg). This may look deceptively like the object in question was allocated to a register (rather than in memory) so DWARF consumers need to be aware of the subtle distinction between OP_REG and OP_BASEREG. */ ASM_OUTPUT_DWARF_STACK_OP (asm_out_file, OP_BASEREG); output_reg_number (rtl); break; case MEM: output_mem_loc_descriptor (XEXP (rtl, 0)); ASM_OUTPUT_DWARF_STACK_OP (asm_out_file, OP_DEREF4); break; case CONST: case SYMBOL_REF: ASM_OUTPUT_DWARF_STACK_OP (asm_out_file, OP_ADDR); ASM_OUTPUT_DWARF_ADDR_CONST (asm_out_file, rtl); break; case PLUS: output_mem_loc_descriptor (XEXP (rtl, 0)); output_mem_loc_descriptor (XEXP (rtl, 1)); ASM_OUTPUT_DWARF_STACK_OP (asm_out_file, OP_ADD); break; case CONST_INT: ASM_OUTPUT_DWARF_STACK_OP (asm_out_file, OP_CONST); ASM_OUTPUT_DWARF_DATA4 (asm_out_file, INTVAL (rtl)); break; case MULT: /* If a pseudo-reg is optimized away, it is possible for it to be replaced with a MEM containing a multiply. Use a GNU extension to describe it. */ output_mem_loc_descriptor (XEXP (rtl, 0)); output_mem_loc_descriptor (XEXP (rtl, 1)); ASM_OUTPUT_DWARF_STACK_OP (asm_out_file, OP_MULT); break; default: abort (); } } /* Output a proper Dwarf location descriptor for a variable or parameter which is either allocated in a register or in a memory location. For a register, we just generate an OP_REG and the register number. For a memory location we provide a Dwarf postfix expression describing how to generate the (dynamic) address of the object onto the address stack. */ static void output_loc_descriptor (rtl) rtx rtl; { switch (GET_CODE (rtl)) { case SUBREG: /* The case of a subreg may arise when we have a local (register) variable or a formal (register) parameter which doesn't quite fill up an entire register. For now, just assume that it is legitimate to make the Dwarf info refer to the whole register which contains the given subreg. */ rtl = SUBREG_REG (rtl); /* Drop thru. */ case REG: ASM_OUTPUT_DWARF_STACK_OP (asm_out_file, OP_REG); output_reg_number (rtl); break; case MEM: output_mem_loc_descriptor (XEXP (rtl, 0)); break; default: abort (); /* Should never happen */ } } /* Given a tree node describing an array bound (either lower or upper) output a representation for that bound. */ static void output_bound_representation (bound, dim_num, u_or_l) tree bound; unsigned dim_num; /* For multi-dimensional arrays. */ char u_or_l; /* Designates upper or lower bound. */ { switch (TREE_CODE (bound)) { case ERROR_MARK: return; /* All fixed-bounds are represented by INTEGER_CST nodes. */ case INTEGER_CST: if (host_integerp (bound, 0)) ASM_OUTPUT_DWARF_DATA4 (asm_out_file, tree_low_cst (bound, 0)); break; default: /* Dynamic bounds may be represented by NOP_EXPR nodes containing SAVE_EXPR nodes, in which case we can do something, or as an expression, which we cannot represent. */ { char begin_label[MAX_ARTIFICIAL_LABEL_BYTES]; char end_label[MAX_ARTIFICIAL_LABEL_BYTES]; sprintf (begin_label, BOUND_BEGIN_LABEL_FMT, current_dienum, dim_num, u_or_l); sprintf (end_label, BOUND_END_LABEL_FMT, current_dienum, dim_num, u_or_l); ASM_OUTPUT_DWARF_DELTA2 (asm_out_file, end_label, begin_label); ASM_OUTPUT_LABEL (asm_out_file, begin_label); /* If optimization is turned on, the SAVE_EXPRs that describe how to access the upper bound values are essentially bogus. They only describe (at best) how to get at these values at the points in the generated code right after they have just been computed. Worse yet, in the typical case, the upper bound values will not even *be* computed in the optimized code, so these SAVE_EXPRs are entirely bogus. In order to compensate for this fact, we check here to see if optimization is enabled, and if so, we effectively create an empty location description for the (unknown and unknowable) upper bound. This should not cause too much trouble for existing (stupid?) debuggers because they have to deal with empty upper bounds location descriptions anyway in order to be able to deal with incomplete array types. Of course an intelligent debugger (GDB?) should be able to comprehend that a missing upper bound specification in a array type used for a storage class `auto' local array variable indicates that the upper bound is both unknown (at compile- time) and unknowable (at run-time) due to optimization. */ if (! optimize) { while (TREE_CODE (bound) == NOP_EXPR || TREE_CODE (bound) == CONVERT_EXPR) bound = TREE_OPERAND (bound, 0); if (TREE_CODE (bound) == SAVE_EXPR && SAVE_EXPR_RTL (bound)) output_loc_descriptor (eliminate_regs (SAVE_EXPR_RTL (bound), 0, NULL_RTX)); } ASM_OUTPUT_LABEL (asm_out_file, end_label); } break; } } /* Recursive function to output a sequence of value/name pairs for enumeration constants in reversed order. This is called from enumeration_type_die. */ static void output_enumeral_list (link) tree link; { if (link) { output_enumeral_list (TREE_CHAIN (link)); if (host_integerp (TREE_VALUE (link), 0)) ASM_OUTPUT_DWARF_DATA4 (asm_out_file, tree_low_cst (TREE_VALUE (link), 0)); ASM_OUTPUT_DWARF_STRING_NEWLINE (asm_out_file, IDENTIFIER_POINTER (TREE_PURPOSE (link))); } } /* Given an unsigned value, round it up to the lowest multiple of `boundary' which is not less than the value itself. */ static inline HOST_WIDE_INT ceiling (value, boundary) HOST_WIDE_INT value; unsigned int boundary; { return (((value + boundary - 1) / boundary) * boundary); } /* Given a pointer to what is assumed to be a FIELD_DECL node, return a pointer to the declared type for the relevant field variable, or return `integer_type_node' if the given node turns out to be an ERROR_MARK node. */ static inline tree field_type (decl) tree decl; { tree type; if (TREE_CODE (decl) == ERROR_MARK) return integer_type_node; type = DECL_BIT_FIELD_TYPE (decl); if (type == NULL) type = TREE_TYPE (decl); return type; } /* Given a pointer to a tree node, assumed to be some kind of a ..._TYPE node, return the alignment in bits for the type, or else return BITS_PER_WORD if the node actually turns out to be an ERROR_MARK node. */ static inline unsigned int simple_type_align_in_bits (type) tree type; { return (TREE_CODE (type) != ERROR_MARK) ? TYPE_ALIGN (type) : BITS_PER_WORD; } /* Given a pointer to a tree node, assumed to be some kind of a ..._TYPE node, return the size in bits for the type if it is a constant, or else return the alignment for the type if the type's size is not constant, or else return BITS_PER_WORD if the type actually turns out to be an ERROR_MARK node. */ static inline unsigned HOST_WIDE_INT simple_type_size_in_bits (type) tree type; { tree type_size_tree; if (TREE_CODE (type) == ERROR_MARK) return BITS_PER_WORD; type_size_tree = TYPE_SIZE (type); if (type_size_tree == NULL_TREE) return 0; if (! host_integerp (type_size_tree, 1)) return TYPE_ALIGN (type); return tree_low_cst (type_size_tree, 1); } /* Given a pointer to what is assumed to be a FIELD_DECL node, compute and return the byte offset of the lowest addressed byte of the "containing object" for the given FIELD_DECL, or return 0 if we are unable to deter- mine what that offset is, either because the argument turns out to be a pointer to an ERROR_MARK node, or because the offset is actually variable. (We can't handle the latter case just yet.) */ static HOST_WIDE_INT field_byte_offset (decl) tree decl; { unsigned int type_align_in_bytes; unsigned int type_align_in_bits; unsigned HOST_WIDE_INT type_size_in_bits; HOST_WIDE_INT object_offset_in_align_units; HOST_WIDE_INT object_offset_in_bits; HOST_WIDE_INT object_offset_in_bytes; tree type; tree field_size_tree; HOST_WIDE_INT bitpos_int; HOST_WIDE_INT deepest_bitpos; unsigned HOST_WIDE_INT field_size_in_bits; if (TREE_CODE (decl) == ERROR_MARK) return 0; if (TREE_CODE (decl) != FIELD_DECL) abort (); type = field_type (decl); field_size_tree = DECL_SIZE (decl); /* The size could be unspecified if there was an error, or for a flexible array member. */ if (! field_size_tree) field_size_tree = bitsize_zero_node; /* We cannot yet cope with fields whose positions or sizes are variable, so for now, when we see such things, we simply return 0. Someday, we may be able to handle such cases, but it will be damn difficult. */ if (! host_integerp (bit_position (decl), 0) || ! host_integerp (field_size_tree, 1)) return 0; bitpos_int = int_bit_position (decl); field_size_in_bits = tree_low_cst (field_size_tree, 1); type_size_in_bits = simple_type_size_in_bits (type); type_align_in_bits = simple_type_align_in_bits (type); type_align_in_bytes = type_align_in_bits / BITS_PER_UNIT; /* Note that the GCC front-end doesn't make any attempt to keep track of the starting bit offset (relative to the start of the containing structure type) of the hypothetical "containing object" for a bit- field. Thus, when computing the byte offset value for the start of the "containing object" of a bit-field, we must deduce this infor- mation on our own. This can be rather tricky to do in some cases. For example, handling the following structure type definition when compiling for an i386/i486 target (which only aligns long long's to 32-bit boundaries) can be very tricky: struct S { int field1; long long field2:31; }; Fortunately, there is a simple rule-of-thumb which can be used in such cases. When compiling for an i386/i486, GCC will allocate 8 bytes for the structure shown above. It decides to do this based upon one simple rule for bit-field allocation. Quite simply, GCC allocates each "con- taining object" for each bit-field at the first (i.e. lowest addressed) legitimate alignment boundary (based upon the required minimum alignment for the declared type of the field) which it can possibly use, subject to the condition that there is still enough available space remaining in the containing object (when allocated at the selected point) to fully accommodate all of the bits of the bit-field itself. This simple rule makes it obvious why GCC allocates 8 bytes for each object of the structure type shown above. When looking for a place to allocate the "containing object" for `field2', the compiler simply tries to allocate a 64-bit "containing object" at each successive 32-bit boundary (starting at zero) until it finds a place to allocate that 64- bit field such that at least 31 contiguous (and previously unallocated) bits remain within that selected 64 bit field. (As it turns out, for the example above, the compiler finds that it is OK to allocate the "containing object" 64-bit field at bit-offset zero within the structure type.) Here we attempt to work backwards from the limited set of facts we're given, and we try to deduce from those facts, where GCC must have believed that the containing object started (within the structure type). The value we deduce is then used (by the callers of this routine) to generate AT_location and AT_bit_offset attributes for fields (both bit-fields and, in the case of AT_location, regular fields as well). */ /* Figure out the bit-distance from the start of the structure to the "deepest" bit of the bit-field. */ deepest_bitpos = bitpos_int + field_size_in_bits; /* This is the tricky part. Use some fancy footwork to deduce where the lowest addressed bit of the containing object must be. */ object_offset_in_bits = ceiling (deepest_bitpos, type_align_in_bits) - type_size_in_bits; /* Compute the offset of the containing object in "alignment units". */ object_offset_in_align_units = object_offset_in_bits / type_align_in_bits; /* Compute the offset of the containing object in bytes. */ object_offset_in_bytes = object_offset_in_align_units * type_align_in_bytes; /* The above code assumes that the field does not cross an alignment boundary. This can happen if PCC_BITFIELD_TYPE_MATTERS is not defined, or if the structure is packed. If this happens, then we get an object which starts after the bitfield, which means that the bit offset is negative. Gdb fails when given negative bit offsets. We avoid this by recomputing using the first bit of the bitfield. This will give us an object which does not completely contain the bitfield, but it will be aligned, and it will contain the first bit of the bitfield. However, only do this for a BYTES_BIG_ENDIAN target. For a ! BYTES_BIG_ENDIAN target, bitpos_int + field_size_in_bits is the first first bit of the bitfield. If we recompute using bitpos_int + 1 below, then we end up computing the object byte offset for the wrong word of the desired bitfield, which in turn causes the field offset to be negative in bit_offset_attribute. */ if (BYTES_BIG_ENDIAN && object_offset_in_bits > bitpos_int) { deepest_bitpos = bitpos_int + 1; object_offset_in_bits = ceiling (deepest_bitpos, type_align_in_bits) - type_size_in_bits; object_offset_in_align_units = (object_offset_in_bits / type_align_in_bits); object_offset_in_bytes = (object_offset_in_align_units * type_align_in_bytes); } return object_offset_in_bytes; } /****************************** attributes *********************************/ /* The following routines are responsible for writing out the various types of Dwarf attributes (and any following data bytes associated with them). These routines are listed in order based on the numerical codes of their associated attributes. */ /* Generate an AT_sibling attribute. */ static inline void sibling_attribute () { char label[MAX_ARTIFICIAL_LABEL_BYTES]; ASM_OUTPUT_DWARF_ATTRIBUTE (asm_out_file, AT_sibling); sprintf (label, DIE_BEGIN_LABEL_FMT, NEXT_DIE_NUM); ASM_OUTPUT_DWARF_REF (asm_out_file, label); } /* Output the form of location attributes suitable for whole variables and whole parameters. Note that the location attributes for struct fields are generated by the routine `data_member_location_attribute' below. */ static void location_attribute (rtl) rtx rtl; { char begin_label[MAX_ARTIFICIAL_LABEL_BYTES]; char end_label[MAX_ARTIFICIAL_LABEL_BYTES]; ASM_OUTPUT_DWARF_ATTRIBUTE (asm_out_file, AT_location); sprintf (begin_label, LOC_BEGIN_LABEL_FMT, current_dienum); sprintf (end_label, LOC_END_LABEL_FMT, current_dienum); ASM_OUTPUT_DWARF_DELTA2 (asm_out_file, end_label, begin_label); ASM_OUTPUT_LABEL (asm_out_file, begin_label); /* Handle a special case. If we are about to output a location descriptor for a variable or parameter which has been optimized out of existence, don't do that. Instead we output a zero-length location descriptor value as part of the location attribute. A variable which has been optimized out of existence will have a DECL_RTL value which denotes a pseudo-reg. Currently, in some rare cases, variables can have DECL_RTL values which look like (MEM (REG pseudo-reg#)). These cases are due to bugs elsewhere in the compiler. We treat such cases as if the variable(s) in question had been optimized out of existence. Note that in all cases where we wish to express the fact that a variable has been optimized out of existence, we do not simply suppress the generation of the entire location attribute because the absence of a location attribute in certain kinds of DIEs is used to indicate something else entirely... i.e. that the DIE represents an object declaration, but not a definition. So saith the PLSIG. */ if (! is_pseudo_reg (rtl) && (GET_CODE (rtl) != MEM || ! is_pseudo_reg (XEXP (rtl, 0)))) output_loc_descriptor (rtl); ASM_OUTPUT_LABEL (asm_out_file, end_label); } /* Output the specialized form of location attribute used for data members of struct and union types. In the special case of a FIELD_DECL node which represents a bit-field, the "offset" part of this special location descriptor must indicate the distance in bytes from the lowest-addressed byte of the containing struct or union type to the lowest-addressed byte of the "containing object" for the bit-field. (See the `field_byte_offset' function above.) For any given bit-field, the "containing object" is a hypothetical object (of some integral or enum type) within which the given bit-field lives. The type of this hypothetical "containing object" is always the same as the declared type of the individual bit-field itself (for GCC anyway... the DWARF spec doesn't actually mandate this). Note that it is the size (in bytes) of the hypothetical "containing object" which will be given in the AT_byte_size attribute for this bit-field. (See the `byte_size_attribute' function below.) It is also used when calculating the value of the AT_bit_offset attribute. (See the `bit_offset_attribute' function below.) */ static void data_member_location_attribute (t) tree t; { unsigned object_offset_in_bytes; char begin_label[MAX_ARTIFICIAL_LABEL_BYTES]; char end_label[MAX_ARTIFICIAL_LABEL_BYTES]; if (TREE_CODE (t) == TREE_VEC) object_offset_in_bytes = tree_low_cst (BINFO_OFFSET (t), 0); else object_offset_in_bytes = field_byte_offset (t); ASM_OUTPUT_DWARF_ATTRIBUTE (asm_out_file, AT_location); sprintf (begin_label, LOC_BEGIN_LABEL_FMT, current_dienum); sprintf (end_label, LOC_END_LABEL_FMT, current_dienum); ASM_OUTPUT_DWARF_DELTA2 (asm_out_file, end_label, begin_label); ASM_OUTPUT_LABEL (asm_out_file, begin_label); ASM_OUTPUT_DWARF_STACK_OP (asm_out_file, OP_CONST); ASM_OUTPUT_DWARF_DATA4 (asm_out_file, object_offset_in_bytes); ASM_OUTPUT_DWARF_STACK_OP (asm_out_file, OP_ADD); ASM_OUTPUT_LABEL (asm_out_file, end_label); } /* Output an AT_const_value attribute for a variable or a parameter which does not have a "location" either in memory or in a register. These things can arise in GNU C when a constant is passed as an actual parameter to an inlined function. They can also arise in C++ where declared constants do not necessarily get memory "homes". */ static void const_value_attribute (rtl) rtx rtl; { char begin_label[MAX_ARTIFICIAL_LABEL_BYTES]; char end_label[MAX_ARTIFICIAL_LABEL_BYTES]; ASM_OUTPUT_DWARF_ATTRIBUTE (asm_out_file, AT_const_value_block4); sprintf (begin_label, LOC_BEGIN_LABEL_FMT, current_dienum); sprintf (end_label, LOC_END_LABEL_FMT, current_dienum); ASM_OUTPUT_DWARF_DELTA4 (asm_out_file, end_label, begin_label); ASM_OUTPUT_LABEL (asm_out_file, begin_label); switch (GET_CODE (rtl)) { case CONST_INT: /* Note that a CONST_INT rtx could represent either an integer or a floating-point constant. A CONST_INT is used whenever the constant will fit into a single word. In all such cases, the original mode of the constant value is wiped out, and the CONST_INT rtx is assigned VOIDmode. Since we no longer have precise mode information for these constants, we always just output them using 4 bytes. */ ASM_OUTPUT_DWARF_DATA4 (asm_out_file, (unsigned) INTVAL (rtl)); break; case CONST_DOUBLE: /* Note that a CONST_DOUBLE rtx could represent either an integer or a floating-point constant. A CONST_DOUBLE is used whenever the constant requires more than one word in order to be adequately represented. In all such cases, the original mode of the constant value is preserved as the mode of the CONST_DOUBLE rtx, but for simplicity we always just output CONST_DOUBLEs using 8 bytes. */ ASM_OUTPUT_DWARF_DATA8 (asm_out_file, (unsigned int) CONST_DOUBLE_HIGH (rtl), (unsigned int) CONST_DOUBLE_LOW (rtl)); break; case CONST_STRING: ASM_OUTPUT_DWARF_STRING_NEWLINE (asm_out_file, XSTR (rtl, 0)); break; case SYMBOL_REF: case LABEL_REF: case CONST: ASM_OUTPUT_DWARF_ADDR_CONST (asm_out_file, rtl); break; case PLUS: /* In cases where an inlined instance of an inline function is passed the address of an `auto' variable (which is local to the caller) we can get a situation where the DECL_RTL of the artificial local variable (for the inlining) which acts as a stand-in for the corresponding formal parameter (of the inline function) will look like (plus:SI (reg:SI FRAME_PTR) (const_int ...)). This is not exactly a compile-time constant expression, but it isn't the address of the (artificial) local variable either. Rather, it represents the *value* which the artificial local variable always has during its lifetime. We currently have no way to represent such quasi-constant values in Dwarf, so for now we just punt and generate an AT_const_value attribute with form FORM_BLOCK4 and a length of zero. */ break; default: abort (); /* No other kinds of rtx should be possible here. */ } ASM_OUTPUT_LABEL (asm_out_file, end_label); } /* Generate *either* an AT_location attribute or else an AT_const_value data attribute for a variable or a parameter. We generate the AT_const_value attribute only in those cases where the given variable or parameter does not have a true "location" either in memory or in a register. This can happen (for example) when a constant is passed as an actual argument in a call to an inline function. (It's possible that these things can crop up in other ways also.) Note that one type of constant value which can be passed into an inlined function is a constant pointer. This can happen for example if an actual argument in an inlined function call evaluates to a compile-time constant address. */ static void location_or_const_value_attribute (decl) tree decl; { rtx rtl; if (TREE_CODE (decl) == ERROR_MARK) return; if ((TREE_CODE (decl) != VAR_DECL) && (TREE_CODE (decl) != PARM_DECL)) { /* Should never happen. */ abort (); return; } /* Here we have to decide where we are going to say the parameter "lives" (as far as the debugger is concerned). We only have a couple of choices. GCC provides us with DECL_RTL and with DECL_INCOMING_RTL. DECL_RTL normally indicates where the parameter lives during most of the activa- tion of the function. If optimization is enabled however, this could be either NULL or else a pseudo-reg. Both of those cases indicate that the parameter doesn't really live anywhere (as far as the code generation parts of GCC are concerned) during most of the function's activation. That will happen (for example) if the parameter is never referenced within the function. We could just generate a location descriptor here for all non-NULL non-pseudo values of DECL_RTL and ignore all of the rest, but we can be a little nicer than that if we also consider DECL_INCOMING_RTL in cases where DECL_RTL is NULL or is a pseudo-reg. Note however that we can only get away with using DECL_INCOMING_RTL as a backup substitute for DECL_RTL in certain limited cases. In cases where DECL_ARG_TYPE(decl) indicates the same type as TREE_TYPE(decl) we can be sure that the parameter was passed using the same type as it is declared to have within the function, and that its DECL_INCOMING_RTL points us to a place where a value of that type is passed. In cases where DECL_ARG_TYPE(decl) and TREE_TYPE(decl) are different types however, we cannot (in general) use DECL_INCOMING_RTL as a backup substitute for DECL_RTL because in these cases, DECL_INCOMING_RTL points us to a value of some type which is *different* from the type of the parameter itself. Thus, if we tried to use DECL_INCOMING_RTL to generate a location attribute in such cases, the debugger would end up (for example) trying to fetch a `float' from a place which actually contains the first part of a `double'. That would lead to really incorrect and confusing output at debug-time, and we don't want that now do we? So in general, we DO NOT use DECL_INCOMING_RTL as a backup for DECL_RTL in cases where DECL_ARG_TYPE(decl) != TREE_TYPE(decl). There are a couple of cute exceptions however. On little-endian machines we can get away with using DECL_INCOMING_RTL even when DECL_ARG_TYPE(decl) is not the same as TREE_TYPE(decl) but only when DECL_ARG_TYPE(decl) is an integral type which is smaller than TREE_TYPE(decl). These cases arise when (on a little-endian machine) a non-prototyped function has a parameter declared to be of type `short' or `char'. In such cases, TREE_TYPE(decl) will be `short' or `char', DECL_ARG_TYPE(decl) will be `int', and DECL_INCOMING_RTL will point to the lowest-order byte of the passed `int' value. If the debugger then uses that address to fetch a `short' or a `char' (on a little-endian machine) the result will be the correct data, so we allow for such exceptional cases below. Note that our goal here is to describe the place where the given formal parameter lives during most of the function's activation (i.e. between the end of the prologue and the start of the epilogue). We'll do that as best as we can. Note however that if the given formal parameter is modified sometime during the execution of the function, then a stack backtrace (at debug-time) will show the function as having been called with the *new* value rather than the value which was originally passed in. This happens rarely enough that it is not a major problem, but it *is* a problem, and I'd like to fix it. A future version of dwarfout.c may generate two additional attributes for any given TAG_formal_parameter DIE which will describe the "passed type" and the "passed location" for the given formal parameter in addition to the attributes we now generate to indicate the "declared type" and the "active location" for each parameter. This additional set of attributes could be used by debuggers for stack backtraces. Separately, note that sometimes DECL_RTL can be NULL and DECL_INCOMING_RTL can be NULL also. This happens (for example) for inlined-instances of inline function formal parameters which are never referenced. This really shouldn't be happening. All PARM_DECL nodes should get valid non-NULL DECL_INCOMING_RTL values, but integrate.c doesn't currently generate these values for inlined instances of inline function parameters, so when we see such cases, we are just out-of-luck for the time being (until integrate.c gets fixed). */ /* Use DECL_RTL as the "location" unless we find something better. */ rtl = DECL_RTL (decl); if (TREE_CODE (decl) == PARM_DECL) if (rtl == NULL_RTX || is_pseudo_reg (rtl)) { /* This decl represents a formal parameter which was optimized out. */ tree declared_type = type_main_variant (TREE_TYPE (decl)); tree passed_type = type_main_variant (DECL_ARG_TYPE (decl)); /* Note that DECL_INCOMING_RTL may be NULL in here, but we handle *all* cases where (rtl == NULL_RTX) just below. */ if (declared_type == passed_type) rtl = DECL_INCOMING_RTL (decl); else if (! BYTES_BIG_ENDIAN) if (TREE_CODE (declared_type) == INTEGER_TYPE) /* NMS WTF? */ if (TYPE_SIZE (declared_type) <= TYPE_SIZE (passed_type)) rtl = DECL_INCOMING_RTL (decl); } if (rtl == NULL_RTX) return; rtl = eliminate_regs (rtl, 0, NULL_RTX); #ifdef LEAF_REG_REMAP if (current_function_uses_only_leaf_regs) leaf_renumber_regs_insn (rtl); #endif switch (GET_CODE (rtl)) { case ADDRESSOF: /* The address of a variable that was optimized away; don't emit anything. */ break; case CONST_INT: case CONST_DOUBLE: case CONST_STRING: case SYMBOL_REF: case LABEL_REF: case CONST: case PLUS: /* DECL_RTL could be (plus (reg ...) (const_int ...)) */ const_value_attribute (rtl); break; case MEM: case REG: case SUBREG: location_attribute (rtl); break; case CONCAT: /* ??? CONCAT is used for complex variables, which may have the real part stored in one place and the imag part stored somewhere else. DWARF1 has no way to describe a variable that lives in two different places, so we just describe where the first part lives, and hope that the second part is stored after it. */ location_attribute (XEXP (rtl, 0)); break; default: abort (); /* Should never happen. */ } } /* Generate an AT_name attribute given some string value to be included as the value of the attribute. */ static inline void name_attribute (name_string) const char *name_string; { if (name_string && *name_string) { ASM_OUTPUT_DWARF_ATTRIBUTE (asm_out_file, AT_name); ASM_OUTPUT_DWARF_STRING_NEWLINE (asm_out_file, name_string); } } static inline void fund_type_attribute (ft_code) unsigned ft_code; { ASM_OUTPUT_DWARF_ATTRIBUTE (asm_out_file, AT_fund_type); ASM_OUTPUT_DWARF_FUND_TYPE (asm_out_file, ft_code); } static void mod_fund_type_attribute (type, decl_const, decl_volatile) tree type; int decl_const; int decl_volatile; { char begin_label[MAX_ARTIFICIAL_LABEL_BYTES]; char end_label[MAX_ARTIFICIAL_LABEL_BYTES]; ASM_OUTPUT_DWARF_ATTRIBUTE (asm_out_file, AT_mod_fund_type); sprintf (begin_label, MT_BEGIN_LABEL_FMT, current_dienum); sprintf (end_label, MT_END_LABEL_FMT, current_dienum); ASM_OUTPUT_DWARF_DELTA2 (asm_out_file, end_label, begin_label); ASM_OUTPUT_LABEL (asm_out_file, begin_label); write_modifier_bytes (type, decl_const, decl_volatile); ASM_OUTPUT_DWARF_FUND_TYPE (asm_out_file, fundamental_type_code (root_type (type))); ASM_OUTPUT_LABEL (asm_out_file, end_label); } static inline void user_def_type_attribute (type) tree type; { char ud_type_name[MAX_ARTIFICIAL_LABEL_BYTES]; ASM_OUTPUT_DWARF_ATTRIBUTE (asm_out_file, AT_user_def_type); sprintf (ud_type_name, TYPE_NAME_FMT, TYPE_UID (type)); ASM_OUTPUT_DWARF_REF (asm_out_file, ud_type_name); } static void mod_u_d_type_attribute (type, decl_const, decl_volatile) tree type; int decl_const; int decl_volatile; { char begin_label[MAX_ARTIFICIAL_LABEL_BYTES]; char end_label[MAX_ARTIFICIAL_LABEL_BYTES]; char ud_type_name[MAX_ARTIFICIAL_LABEL_BYTES]; ASM_OUTPUT_DWARF_ATTRIBUTE (asm_out_file, AT_mod_u_d_type); sprintf (begin_label, MT_BEGIN_LABEL_FMT, current_dienum); sprintf (end_label, MT_END_LABEL_FMT, current_dienum); ASM_OUTPUT_DWARF_DELTA2 (asm_out_file, end_label, begin_label); ASM_OUTPUT_LABEL (asm_out_file, begin_label); write_modifier_bytes (type, decl_const, decl_volatile); sprintf (ud_type_name, TYPE_NAME_FMT, TYPE_UID (root_type (type))); ASM_OUTPUT_DWARF_REF (asm_out_file, ud_type_name); ASM_OUTPUT_LABEL (asm_out_file, end_label); } #ifdef USE_ORDERING_ATTRIBUTE static inline void ordering_attribute (ordering) unsigned ordering; { ASM_OUTPUT_DWARF_ATTRIBUTE (asm_out_file, AT_ordering); ASM_OUTPUT_DWARF_DATA2 (asm_out_file, ordering); } #endif /* defined(USE_ORDERING_ATTRIBUTE) */ /* Note that the block of subscript information for an array type also includes information about the element type of type given array type. */ static void subscript_data_attribute (type) tree type; { unsigned dimension_number; char begin_label[MAX_ARTIFICIAL_LABEL_BYTES]; char end_label[MAX_ARTIFICIAL_LABEL_BYTES]; ASM_OUTPUT_DWARF_ATTRIBUTE (asm_out_file, AT_subscr_data); sprintf (begin_label, SS_BEGIN_LABEL_FMT, current_dienum); sprintf (end_label, SS_END_LABEL_FMT, current_dienum); ASM_OUTPUT_DWARF_DELTA2 (asm_out_file, end_label, begin_label); ASM_OUTPUT_LABEL (asm_out_file, begin_label); /* The GNU compilers represent multidimensional array types as sequences of one dimensional array types whose element types are themselves array types. Here we squish that down, so that each multidimensional array type gets only one array_type DIE in the Dwarf debugging info. The draft Dwarf specification say that we are allowed to do this kind of compression in C (because there is no difference between an array or arrays and a multidimensional array in C) but for other source languages (e.g. Ada) we probably shouldn't do this. */ for (dimension_number = 0; TREE_CODE (type) == ARRAY_TYPE; type = TREE_TYPE (type), dimension_number++) { tree domain = TYPE_DOMAIN (type); /* Arrays come in three flavors. Unspecified bounds, fixed bounds, and (in GNU C only) variable bounds. Handle all three forms here. */ if (domain) { /* We have an array type with specified bounds. */ tree lower = TYPE_MIN_VALUE (domain); tree upper = TYPE_MAX_VALUE (domain); /* Handle only fundamental types as index types for now. */ if (! type_is_fundamental (domain)) abort (); /* Output the representation format byte for this dimension. */ ASM_OUTPUT_DWARF_FMT_BYTE (asm_out_file, FMT_CODE (1, TREE_CODE (lower) == INTEGER_CST, upper && TREE_CODE (upper) == INTEGER_CST)); /* Output the index type for this dimension. */ ASM_OUTPUT_DWARF_FUND_TYPE (asm_out_file, fundamental_type_code (domain)); /* Output the representation for the lower bound. */ output_bound_representation (lower, dimension_number, 'l'); /* Output the representation for the upper bound. */ if (upper) output_bound_representation (upper, dimension_number, 'u'); else ASM_OUTPUT_DWARF_DATA2 (asm_out_file, 0); } else { /* We have an array type with an unspecified length. For C and C++ we can assume that this really means that (a) the index type is an integral type, and (b) the lower bound is zero. Note that Dwarf defines the representation of an unspecified (upper) bound as being a zero-length location description. */ /* Output the array-bounds format byte. */ ASM_OUTPUT_DWARF_FMT_BYTE (asm_out_file, FMT_FT_C_X); /* Output the (assumed) index type. */ ASM_OUTPUT_DWARF_FUND_TYPE (asm_out_file, FT_integer); /* Output the (assumed) lower bound (constant) value. */ ASM_OUTPUT_DWARF_DATA4 (asm_out_file, 0); /* Output the (empty) location description for the upper bound. */ ASM_OUTPUT_DWARF_DATA2 (asm_out_file, 0); } } /* Output the prefix byte that says that the element type is coming up. */ ASM_OUTPUT_DWARF_FMT_BYTE (asm_out_file, FMT_ET); /* Output a representation of the type of the elements of this array type. */ type_attribute (type, 0, 0); ASM_OUTPUT_LABEL (asm_out_file, end_label); } static void byte_size_attribute (tree_node) tree tree_node; { unsigned size; ASM_OUTPUT_DWARF_ATTRIBUTE (asm_out_file, AT_byte_size); switch (TREE_CODE (tree_node)) { case ERROR_MARK: size = 0; break; case ENUMERAL_TYPE: case RECORD_TYPE: case UNION_TYPE: case QUAL_UNION_TYPE: case ARRAY_TYPE: size = int_size_in_bytes (tree_node); break; case FIELD_DECL: /* For a data member of a struct or union, the AT_byte_size is generally given as the number of bytes normally allocated for an object of the *declared* type of the member itself. This is true even for bit-fields. */ size = simple_type_size_in_bits (field_type (tree_node)) / BITS_PER_UNIT; break; default: abort (); } /* Note that `size' might be -1 when we get to this point. If it is, that indicates that the byte size of the entity in question is variable. We have no good way of expressing this fact in Dwarf at the present time, so just let the -1 pass on through. */ ASM_OUTPUT_DWARF_DATA4 (asm_out_file, size); } /* For a FIELD_DECL node which represents a bit-field, output an attribute which specifies the distance in bits from the highest order bit of the "containing object" for the bit-field to the highest order bit of the bit-field itself. For any given bit-field, the "containing object" is a hypothetical object (of some integral or enum type) within which the given bit-field lives. The type of this hypothetical "containing object" is always the same as the declared type of the individual bit-field itself. The determination of the exact location of the "containing object" for a bit-field is rather complicated. It's handled by the `field_byte_offset' function (above). Note that it is the size (in bytes) of the hypothetical "containing object" which will be given in the AT_byte_size attribute for this bit-field. (See `byte_size_attribute' above.) */ static inline void bit_offset_attribute (decl) tree decl; { HOST_WIDE_INT object_offset_in_bytes = field_byte_offset (decl); tree type = DECL_BIT_FIELD_TYPE (decl); HOST_WIDE_INT bitpos_int; HOST_WIDE_INT highest_order_object_bit_offset; HOST_WIDE_INT highest_order_field_bit_offset; HOST_WIDE_INT bit_offset; /* Must be a bit field. */ if (!type || TREE_CODE (decl) != FIELD_DECL) abort (); /* We can't yet handle bit-fields whose offsets or sizes are variable, so if we encounter such things, just return without generating any attribute whatsoever. */ if (! host_integerp (bit_position (decl), 0) || ! host_integerp (DECL_SIZE (decl), 1)) return; bitpos_int = int_bit_position (decl); /* Note that the bit offset is always the distance (in bits) from the highest-order bit of the "containing object" to the highest-order bit of the bit-field itself. Since the "high-order end" of any object or field is different on big-endian and little-endian machines, the computation below must take account of these differences. */ highest_order_object_bit_offset = object_offset_in_bytes * BITS_PER_UNIT; highest_order_field_bit_offset = bitpos_int; if (! BYTES_BIG_ENDIAN) { highest_order_field_bit_offset += tree_low_cst (DECL_SIZE (decl), 1); highest_order_object_bit_offset += simple_type_size_in_bits (type); } bit_offset = (! BYTES_BIG_ENDIAN ? highest_order_object_bit_offset - highest_order_field_bit_offset : highest_order_field_bit_offset - highest_order_object_bit_offset); ASM_OUTPUT_DWARF_ATTRIBUTE (asm_out_file, AT_bit_offset); ASM_OUTPUT_DWARF_DATA2 (asm_out_file, bit_offset); } /* For a FIELD_DECL node which represents a bit field, output an attribute which specifies the length in bits of the given field. */ static inline void bit_size_attribute (decl) tree decl; { /* Must be a field and a bit field. */ if (TREE_CODE (decl) != FIELD_DECL || ! DECL_BIT_FIELD_TYPE (decl)) abort (); if (host_integerp (DECL_SIZE (decl), 1)) { ASM_OUTPUT_DWARF_ATTRIBUTE (asm_out_file, AT_bit_size); ASM_OUTPUT_DWARF_DATA4 (asm_out_file, tree_low_cst (DECL_SIZE (decl), 1)); } } /* The following routine outputs the `element_list' attribute for enumeration type DIEs. The element_lits attribute includes the names and values of all of the enumeration constants associated with the given enumeration type. */ static inline void element_list_attribute (element) tree element; { char begin_label[MAX_ARTIFICIAL_LABEL_BYTES]; char end_label[MAX_ARTIFICIAL_LABEL_BYTES]; ASM_OUTPUT_DWARF_ATTRIBUTE (asm_out_file, AT_element_list); sprintf (begin_label, EE_BEGIN_LABEL_FMT, current_dienum); sprintf (end_label, EE_END_LABEL_FMT, current_dienum); ASM_OUTPUT_DWARF_DELTA4 (asm_out_file, end_label, begin_label); ASM_OUTPUT_LABEL (asm_out_file, begin_label); /* Here we output a list of value/name pairs for each enumeration constant defined for this enumeration type (as required), but we do it in REVERSE order. The order is the one required by the draft #5 Dwarf specification published by the UI/PLSIG. */ output_enumeral_list (element); /* Recursively output the whole list. */ ASM_OUTPUT_LABEL (asm_out_file, end_label); } /* Generate an AT_stmt_list attribute. These are normally present only in DIEs with a TAG_compile_unit tag. */ static inline void stmt_list_attribute (label) const char *label; { ASM_OUTPUT_DWARF_ATTRIBUTE (asm_out_file, AT_stmt_list); /* Don't use ASM_OUTPUT_DWARF_DATA4 here. */ ASM_OUTPUT_DWARF_ADDR (asm_out_file, label); } /* Generate an AT_low_pc attribute for a label DIE, a lexical_block DIE or for a subroutine DIE. */ static inline void low_pc_attribute (asm_low_label) const char *asm_low_label; { ASM_OUTPUT_DWARF_ATTRIBUTE (asm_out_file, AT_low_pc); ASM_OUTPUT_DWARF_ADDR (asm_out_file, asm_low_label); } /* Generate an AT_high_pc attribute for a lexical_block DIE or for a subroutine DIE. */ static inline void high_pc_attribute (asm_high_label) const char *asm_high_label; { ASM_OUTPUT_DWARF_ATTRIBUTE (asm_out_file, AT_high_pc); ASM_OUTPUT_DWARF_ADDR (asm_out_file, asm_high_label); } /* Generate an AT_body_begin attribute for a subroutine DIE. */ static inline void body_begin_attribute (asm_begin_label) const char *asm_begin_label; { ASM_OUTPUT_DWARF_ATTRIBUTE (asm_out_file, AT_body_begin); ASM_OUTPUT_DWARF_ADDR (asm_out_file, asm_begin_label); } /* Generate an AT_body_end attribute for a subroutine DIE. */ static inline void body_end_attribute (asm_end_label) const char *asm_end_label; { ASM_OUTPUT_DWARF_ATTRIBUTE (asm_out_file, AT_body_end); ASM_OUTPUT_DWARF_ADDR (asm_out_file, asm_end_label); } /* Generate an AT_language attribute given a LANG value. These attributes are used only within TAG_compile_unit DIEs. */ static inline void language_attribute (language_code) unsigned language_code; { ASM_OUTPUT_DWARF_ATTRIBUTE (asm_out_file, AT_language); ASM_OUTPUT_DWARF_DATA4 (asm_out_file, language_code); } static inline void member_attribute (context) tree context; { char label[MAX_ARTIFICIAL_LABEL_BYTES]; /* Generate this attribute only for members in C++. */ if (context != NULL && is_tagged_type (context)) { ASM_OUTPUT_DWARF_ATTRIBUTE (asm_out_file, AT_member); sprintf (label, TYPE_NAME_FMT, TYPE_UID (context)); ASM_OUTPUT_DWARF_REF (asm_out_file, label); } } #if 0 #ifndef SL_BEGIN_LABEL_FMT #define SL_BEGIN_LABEL_FMT "*.L_sl%u" #endif #ifndef SL_END_LABEL_FMT #define SL_END_LABEL_FMT "*.L_sl%u_e" #endif static inline void string_length_attribute (upper_bound) tree upper_bound; { char begin_label[MAX_ARTIFICIAL_LABEL_BYTES]; char end_label[MAX_ARTIFICIAL_LABEL_BYTES]; ASM_OUTPUT_DWARF_ATTRIBUTE (asm_out_file, AT_string_length); sprintf (begin_label, SL_BEGIN_LABEL_FMT, current_dienum); sprintf (end_label, SL_END_LABEL_FMT, current_dienum); ASM_OUTPUT_DWARF_DELTA2 (asm_out_file, end_label, begin_label); ASM_OUTPUT_LABEL (asm_out_file, begin_label); output_bound_representation (upper_bound, 0, 'u'); ASM_OUTPUT_LABEL (asm_out_file, end_label); } #endif static inline void comp_dir_attribute (dirname) const char *dirname; { ASM_OUTPUT_DWARF_ATTRIBUTE (asm_out_file, AT_comp_dir); ASM_OUTPUT_DWARF_STRING_NEWLINE (asm_out_file, dirname); } static inline void sf_names_attribute (sf_names_start_label) const char *sf_names_start_label; { ASM_OUTPUT_DWARF_ATTRIBUTE (asm_out_file, AT_sf_names); /* Don't use ASM_OUTPUT_DWARF_DATA4 here. */ ASM_OUTPUT_DWARF_ADDR (asm_out_file, sf_names_start_label); } static inline void src_info_attribute (src_info_start_label) const char *src_info_start_label; { ASM_OUTPUT_DWARF_ATTRIBUTE (asm_out_file, AT_src_info); /* Don't use ASM_OUTPUT_DWARF_DATA4 here. */ ASM_OUTPUT_DWARF_ADDR (asm_out_file, src_info_start_label); } static inline void mac_info_attribute (mac_info_start_label) const char *mac_info_start_label; { ASM_OUTPUT_DWARF_ATTRIBUTE (asm_out_file, AT_mac_info); /* Don't use ASM_OUTPUT_DWARF_DATA4 here. */ ASM_OUTPUT_DWARF_ADDR (asm_out_file, mac_info_start_label); } static inline void prototyped_attribute (func_type) tree func_type; { if ((strcmp (lang_hooks.name, "GNU C") == 0) && (TYPE_ARG_TYPES (func_type) != NULL)) { ASM_OUTPUT_DWARF_ATTRIBUTE (asm_out_file, AT_prototyped); ASM_OUTPUT_DWARF_STRING_NEWLINE (asm_out_file, ""); } } static inline void producer_attribute (producer) const char *producer; { ASM_OUTPUT_DWARF_ATTRIBUTE (asm_out_file, AT_producer); ASM_OUTPUT_DWARF_STRING_NEWLINE (asm_out_file, producer); } static inline void inline_attribute (decl) tree decl; { if (DECL_INLINE (decl)) { ASM_OUTPUT_DWARF_ATTRIBUTE (asm_out_file, AT_inline); ASM_OUTPUT_DWARF_STRING_NEWLINE (asm_out_file, ""); } } static inline void containing_type_attribute (containing_type) tree containing_type; { char label[MAX_ARTIFICIAL_LABEL_BYTES]; ASM_OUTPUT_DWARF_ATTRIBUTE (asm_out_file, AT_containing_type); sprintf (label, TYPE_NAME_FMT, TYPE_UID (containing_type)); ASM_OUTPUT_DWARF_REF (asm_out_file, label); } static inline void abstract_origin_attribute (origin) tree origin; { char label[MAX_ARTIFICIAL_LABEL_BYTES]; ASM_OUTPUT_DWARF_ATTRIBUTE (asm_out_file, AT_abstract_origin); switch (TREE_CODE_CLASS (TREE_CODE (origin))) { case 'd': sprintf (label, DECL_NAME_FMT, DECL_UID (origin)); break; case 't': sprintf (label, TYPE_NAME_FMT, TYPE_UID (origin)); break; default: abort (); /* Should never happen. */ } ASM_OUTPUT_DWARF_REF (asm_out_file, label); } #ifdef DWARF_DECL_COORDINATES static inline void src_coords_attribute (src_fileno, src_lineno) unsigned src_fileno; unsigned src_lineno; { ASM_OUTPUT_DWARF_ATTRIBUTE (asm_out_file, AT_src_coords); ASM_OUTPUT_DWARF_DATA2 (asm_out_file, src_fileno); ASM_OUTPUT_DWARF_DATA2 (asm_out_file, src_lineno); } #endif /* defined(DWARF_DECL_COORDINATES) */ static inline void pure_or_virtual_attribute (func_decl) tree func_decl; { if (DECL_VIRTUAL_P (func_decl)) { #if 0 /* DECL_ABSTRACT_VIRTUAL_P is C++-specific. */ if (DECL_ABSTRACT_VIRTUAL_P (func_decl)) ASM_OUTPUT_DWARF_ATTRIBUTE (asm_out_file, AT_pure_virtual); else #endif ASM_OUTPUT_DWARF_ATTRIBUTE (asm_out_file, AT_virtual); ASM_OUTPUT_DWARF_STRING_NEWLINE (asm_out_file, ""); } } /************************* end of attributes *****************************/ /********************* utility routines for DIEs *************************/ /* Output an AT_name attribute and an AT_src_coords attribute for the given decl, but only if it actually has a name. */ static void name_and_src_coords_attributes (decl) tree decl; { tree decl_name = DECL_NAME (decl); if (decl_name && IDENTIFIER_POINTER (decl_name)) { name_attribute (IDENTIFIER_POINTER (decl_name)); #ifdef DWARF_DECL_COORDINATES { register unsigned file_index; /* This is annoying, but we have to pop out of the .debug section for a moment while we call `lookup_filename' because calling it may cause a temporary switch into the .debug_sfnames section and most svr4 assemblers are not smart enough to be able to nest section switches to any depth greater than one. Note that we also can't skirt this issue by delaying all output to the .debug_sfnames section unit the end of compilation because that would cause us to have inter-section forward references and Fred Fish sez that m68k/svr4 assemblers botch those. */ ASM_OUTPUT_POP_SECTION (asm_out_file); file_index = lookup_filename (DECL_SOURCE_FILE (decl)); ASM_OUTPUT_PUSH_SECTION (asm_out_file, DEBUG_SECTION); src_coords_attribute (file_index, DECL_SOURCE_LINE (decl)); } #endif /* defined(DWARF_DECL_COORDINATES) */ } } /* Many forms of DIEs contain a "type description" part. The following routine writes out these "type descriptor" parts. */ static void type_attribute (type, decl_const, decl_volatile) tree type; int decl_const; int decl_volatile; { enum tree_code code = TREE_CODE (type); int root_type_modified; if (code == ERROR_MARK) return; /* Handle a special case. For functions whose return type is void, we generate *no* type attribute. (Note that no object may have type `void', so this only applies to function return types. */ if (code == VOID_TYPE) return; /* If this is a subtype, find the underlying type. Eventually, this should write out the appropriate subtype info. */ while ((code == INTEGER_TYPE || code == REAL_TYPE) && TREE_TYPE (type) != 0) type = TREE_TYPE (type), code = TREE_CODE (type); root_type_modified = (code == POINTER_TYPE || code == REFERENCE_TYPE || decl_const || decl_volatile || TYPE_READONLY (type) || TYPE_VOLATILE (type)); if (type_is_fundamental (root_type (type))) { if (root_type_modified) mod_fund_type_attribute (type, decl_const, decl_volatile); else fund_type_attribute (fundamental_type_code (type)); } else { if (root_type_modified) mod_u_d_type_attribute (type, decl_const, decl_volatile); else /* We have to get the type_main_variant here (and pass that to the `user_def_type_attribute' routine) because the ..._TYPE node we have might simply be a *copy* of some original type node (where the copy was created to help us keep track of typedef names) and that copy might have a different TYPE_UID from the original ..._TYPE node. (Note that when `equate_type_number_to_die_number' is labeling a given type DIE for future reference, it always and only creates labels for DIEs representing *main variants*, and it never even knows about non-main-variants.) */ user_def_type_attribute (type_main_variant (type)); } } /* Given a tree pointer to a struct, class, union, or enum type node, return a pointer to the (string) tag name for the given type, or zero if the type was declared without a tag. */ static const char * type_tag (type) tree type; { const char *name = 0; if (TYPE_NAME (type) != 0) { tree t = 0; /* Find the IDENTIFIER_NODE for the type name. */ if (TREE_CODE (TYPE_NAME (type)) == IDENTIFIER_NODE) t = TYPE_NAME (type); /* The g++ front end makes the TYPE_NAME of *each* tagged type point to a TYPE_DECL node, regardless of whether or not a `typedef' was involved. */ else if (TREE_CODE (TYPE_NAME (type)) == TYPE_DECL && ! DECL_IGNORED_P (TYPE_NAME (type))) t = DECL_NAME (TYPE_NAME (type)); /* Now get the name as a string, or invent one. */ if (t != 0) name = IDENTIFIER_POINTER (t); } return (name == 0 || *name == '\0') ? 0 : name; } static inline void dienum_push () { /* Start by checking if the pending_sibling_stack needs to be expanded. If necessary, expand it. */ if (pending_siblings == pending_siblings_allocated) { pending_siblings_allocated += PENDING_SIBLINGS_INCREMENT; pending_sibling_stack = (unsigned *) xrealloc (pending_sibling_stack, pending_siblings_allocated * sizeof(unsigned)); } pending_siblings++; NEXT_DIE_NUM = next_unused_dienum++; } /* Pop the sibling stack so that the most recently pushed DIEnum becomes the NEXT_DIE_NUM. */ static inline void dienum_pop () { pending_siblings--; } static inline tree member_declared_type (member) tree member; { return (DECL_BIT_FIELD_TYPE (member)) ? DECL_BIT_FIELD_TYPE (member) : TREE_TYPE (member); } /* Get the function's label, as described by its RTL. This may be different from the DECL_NAME name used in the source file. */ static const char * function_start_label (decl) tree decl; { rtx x; const char *fnname; x = DECL_RTL (decl); if (GET_CODE (x) != MEM) abort (); x = XEXP (x, 0); if (GET_CODE (x) != SYMBOL_REF) abort (); fnname = XSTR (x, 0); return fnname; } /******************************* DIEs ************************************/ /* Output routines for individual types of DIEs. */ /* Note that every type of DIE (except a null DIE) gets a sibling. */ static void output_array_type_die (arg) void *arg; { tree type = arg; ASM_OUTPUT_DWARF_TAG (asm_out_file, TAG_array_type); sibling_attribute (); equate_type_number_to_die_number (type); member_attribute (TYPE_CONTEXT (type)); /* I believe that we can default the array ordering. SDB will probably do the right things even if AT_ordering is not present. It's not even an issue until we start to get into multidimensional arrays anyway. If SDB is ever caught doing the Wrong Thing for multi- dimensional arrays, then we'll have to put the AT_ordering attribute back in. (But if and when we find out that we need to put these in, we will only do so for multidimensional arrays. After all, we don't want to waste space in the .debug section now do we?) */ #ifdef USE_ORDERING_ATTRIBUTE ordering_attribute (ORD_row_major); #endif /* defined(USE_ORDERING_ATTRIBUTE) */ subscript_data_attribute (type); } static void output_set_type_die (arg) void *arg; { tree type = arg; ASM_OUTPUT_DWARF_TAG (asm_out_file, TAG_set_type); sibling_attribute (); equate_type_number_to_die_number (type); member_attribute (TYPE_CONTEXT (type)); type_attribute (TREE_TYPE (type), 0, 0); } #if 0 /* Implement this when there is a GNU FORTRAN or GNU Ada front end. */ static void output_entry_point_die (arg) void *arg; { tree decl = arg; tree origin = decl_ultimate_origin (decl); ASM_OUTPUT_DWARF_TAG (asm_out_file, TAG_entry_point); sibling_attribute (); dienum_push (); if (origin != NULL) abstract_origin_attribute (origin); else { name_and_src_coords_attributes (decl); member_attribute (DECL_CONTEXT (decl)); type_attribute (TREE_TYPE (TREE_TYPE (decl)), 0, 0); } if (DECL_ABSTRACT (decl)) equate_decl_number_to_die_number (decl); else low_pc_attribute (function_start_label (decl)); } #endif /* Output a DIE to represent an inlined instance of an enumeration type. */ static void output_inlined_enumeration_type_die (arg) void *arg; { tree type = arg; ASM_OUTPUT_DWARF_TAG (asm_out_file, TAG_enumeration_type); sibling_attribute (); if (!TREE_ASM_WRITTEN (type)) abort (); abstract_origin_attribute (type); } /* Output a DIE to represent an inlined instance of a structure type. */ static void output_inlined_structure_type_die (arg) void *arg; { tree type = arg; ASM_OUTPUT_DWARF_TAG (asm_out_file, TAG_structure_type); sibling_attribute (); if (!TREE_ASM_WRITTEN (type)) abort (); abstract_origin_attribute (type); } /* Output a DIE to represent an inlined instance of a union type. */ static void output_inlined_union_type_die (arg) void *arg; { tree type = arg; ASM_OUTPUT_DWARF_TAG (asm_out_file, TAG_union_type); sibling_attribute (); if (!TREE_ASM_WRITTEN (type)) abort (); abstract_origin_attribute (type); } /* Output a DIE to represent an enumeration type. Note that these DIEs include all of the information about the enumeration values also. This information is encoded into the element_list attribute. */ static void output_enumeration_type_die (arg) void *arg; { tree type = arg; ASM_OUTPUT_DWARF_TAG (asm_out_file, TAG_enumeration_type); sibling_attribute (); equate_type_number_to_die_number (type); name_attribute (type_tag (type)); member_attribute (TYPE_CONTEXT (type)); /* Handle a GNU C/C++ extension, i.e. incomplete enum types. If the given enum type is incomplete, do not generate the AT_byte_size attribute or the AT_element_list attribute. */ if (COMPLETE_TYPE_P (type)) { byte_size_attribute (type); element_list_attribute (TYPE_FIELDS (type)); } } /* Output a DIE to represent either a real live formal parameter decl or to represent just the type of some formal parameter position in some function type. Note that this routine is a bit unusual because its argument may be a ..._DECL node (i.e. either a PARM_DECL or perhaps a VAR_DECL which represents an inlining of some PARM_DECL) or else some sort of a ..._TYPE node. If it's the former then this function is being called to output a DIE to represent a formal parameter object (or some inlining thereof). If it's the latter, then this function is only being called to output a TAG_formal_parameter DIE to stand as a placeholder for some formal argument type of some subprogram type. */ static void output_formal_parameter_die (arg) void *arg; { tree node = arg; ASM_OUTPUT_DWARF_TAG (asm_out_file, TAG_formal_parameter); sibling_attribute (); switch (TREE_CODE_CLASS (TREE_CODE (node))) { case 'd': /* We were called with some kind of a ..._DECL node. */ { register tree origin = decl_ultimate_origin (node); if (origin != NULL) abstract_origin_attribute (origin); else { name_and_src_coords_attributes (node); type_attribute (TREE_TYPE (node), TREE_READONLY (node), TREE_THIS_VOLATILE (node)); } if (DECL_ABSTRACT (node)) equate_decl_number_to_die_number (node); else location_or_const_value_attribute (node); } break; case 't': /* We were called with some kind of a ..._TYPE node. */ type_attribute (node, 0, 0); break; default: abort (); /* Should never happen. */ } } /* Output a DIE to represent a declared function (either file-scope or block-local) which has "external linkage" (according to ANSI-C). */ static void output_global_subroutine_die (arg) void *arg; { tree decl = arg; tree origin = decl_ultimate_origin (decl); ASM_OUTPUT_DWARF_TAG (asm_out_file, TAG_global_subroutine); sibling_attribute (); dienum_push (); if (origin != NULL) abstract_origin_attribute (origin); else { tree type = TREE_TYPE (decl); name_and_src_coords_attributes (decl); inline_attribute (decl); prototyped_attribute (type); member_attribute (DECL_CONTEXT (decl)); type_attribute (TREE_TYPE (type), 0, 0); pure_or_virtual_attribute (decl); } if (DECL_ABSTRACT (decl)) equate_decl_number_to_die_number (decl); else { if (! DECL_EXTERNAL (decl) && ! in_class && decl == current_function_decl) { char label[MAX_ARTIFICIAL_LABEL_BYTES]; low_pc_attribute (function_start_label (decl)); sprintf (label, FUNC_END_LABEL_FMT, current_function_funcdef_no); high_pc_attribute (label); if (use_gnu_debug_info_extensions) { sprintf (label, BODY_BEGIN_LABEL_FMT, current_function_funcdef_no); body_begin_attribute (label); sprintf (label, BODY_END_LABEL_FMT, current_function_funcdef_no); body_end_attribute (label); } } } } /* Output a DIE to represent a declared data object (either file-scope or block-local) which has "external linkage" (according to ANSI-C). */ static void output_global_variable_die (arg) void *arg; { tree decl = arg; tree origin = decl_ultimate_origin (decl); ASM_OUTPUT_DWARF_TAG (asm_out_file, TAG_global_variable); sibling_attribute (); if (origin != NULL) abstract_origin_attribute (origin); else { name_and_src_coords_attributes (decl); member_attribute (DECL_CONTEXT (decl)); type_attribute (TREE_TYPE (decl), TREE_READONLY (decl), TREE_THIS_VOLATILE (decl)); } if (DECL_ABSTRACT (decl)) equate_decl_number_to_die_number (decl); else { if (! DECL_EXTERNAL (decl) && ! in_class && current_function_decl == decl_function_context (decl)) location_or_const_value_attribute (decl); } } static void output_label_die (arg) void *arg; { tree decl = arg; tree origin = decl_ultimate_origin (decl); ASM_OUTPUT_DWARF_TAG (asm_out_file, TAG_label); sibling_attribute (); if (origin != NULL) abstract_origin_attribute (origin); else name_and_src_coords_attributes (decl); if (DECL_ABSTRACT (decl)) equate_decl_number_to_die_number (decl); else { rtx insn = DECL_RTL (decl); /* Deleted labels are programmer specified labels which have been eliminated because of various optimisations. We still emit them here so that it is possible to put breakpoints on them. */ if (GET_CODE (insn) == CODE_LABEL || ((GET_CODE (insn) == NOTE && NOTE_LINE_NUMBER (insn) == NOTE_INSN_DELETED_LABEL))) { char label[MAX_ARTIFICIAL_LABEL_BYTES]; /* When optimization is enabled (via -O) some parts of the compiler (e.g. jump.c and cse.c) may try to delete CODE_LABEL insns which represent source-level labels which were explicitly declared by the user. This really shouldn't be happening though, so catch it if it ever does happen. */ if (INSN_DELETED_P (insn)) abort (); /* Should never happen. */ ASM_GENERATE_INTERNAL_LABEL (label, "L", CODE_LABEL_NUMBER (insn)); low_pc_attribute (label); } } } static void output_lexical_block_die (arg) void *arg; { tree stmt = arg; ASM_OUTPUT_DWARF_TAG (asm_out_file, TAG_lexical_block); sibling_attribute (); dienum_push (); if (! BLOCK_ABSTRACT (stmt)) { char begin_label[MAX_ARTIFICIAL_LABEL_BYTES]; char end_label[MAX_ARTIFICIAL_LABEL_BYTES]; sprintf (begin_label, BLOCK_BEGIN_LABEL_FMT, BLOCK_NUMBER (stmt)); low_pc_attribute (begin_label); sprintf (end_label, BLOCK_END_LABEL_FMT, BLOCK_NUMBER (stmt)); high_pc_attribute (end_label); } } static void output_inlined_subroutine_die (arg) void *arg; { tree stmt = arg; ASM_OUTPUT_DWARF_TAG (asm_out_file, TAG_inlined_subroutine); sibling_attribute (); dienum_push (); abstract_origin_attribute (block_ultimate_origin (stmt)); if (! BLOCK_ABSTRACT (stmt)) { char begin_label[MAX_ARTIFICIAL_LABEL_BYTES]; char end_label[MAX_ARTIFICIAL_LABEL_BYTES]; sprintf (begin_label, BLOCK_BEGIN_LABEL_FMT, BLOCK_NUMBER (stmt)); low_pc_attribute (begin_label); sprintf (end_label, BLOCK_END_LABEL_FMT, BLOCK_NUMBER (stmt)); high_pc_attribute (end_label); } } /* Output a DIE to represent a declared data object (either file-scope or block-local) which has "internal linkage" (according to ANSI-C). */ static void output_local_variable_die (arg) void *arg; { tree decl = arg; tree origin = decl_ultimate_origin (decl); ASM_OUTPUT_DWARF_TAG (asm_out_file, TAG_local_variable); sibling_attribute (); if (origin != NULL) abstract_origin_attribute (origin); else { name_and_src_coords_attributes (decl); member_attribute (DECL_CONTEXT (decl)); type_attribute (TREE_TYPE (decl), TREE_READONLY (decl), TREE_THIS_VOLATILE (decl)); } if (DECL_ABSTRACT (decl)) equate_decl_number_to_die_number (decl); else location_or_const_value_attribute (decl); } static void output_member_die (arg) void *arg; { tree decl = arg; ASM_OUTPUT_DWARF_TAG (asm_out_file, TAG_member); sibling_attribute (); name_and_src_coords_attributes (decl); member_attribute (DECL_CONTEXT (decl)); type_attribute (member_declared_type (decl), TREE_READONLY (decl), TREE_THIS_VOLATILE (decl)); if (DECL_BIT_FIELD_TYPE (decl)) /* If this is a bit field... */ { byte_size_attribute (decl); bit_size_attribute (decl); bit_offset_attribute (decl); } data_member_location_attribute (decl); } #if 0 /* Don't generate either pointer_type DIEs or reference_type DIEs. Use modified types instead. We keep this code here just in case these types of DIEs may be needed to represent certain things in other languages (e.g. Pascal) someday. */ static void output_pointer_type_die (arg) void *arg; { tree type = arg; ASM_OUTPUT_DWARF_TAG (asm_out_file, TAG_pointer_type); sibling_attribute (); equate_type_number_to_die_number (type); member_attribute (TYPE_CONTEXT (type)); type_attribute (TREE_TYPE (type), 0, 0); } static void output_reference_type_die (arg) void *arg; { tree type = arg; ASM_OUTPUT_DWARF_TAG (asm_out_file, TAG_reference_type); sibling_attribute (); equate_type_number_to_die_number (type); member_attribute (TYPE_CONTEXT (type)); type_attribute (TREE_TYPE (type), 0, 0); } #endif static void output_ptr_to_mbr_type_die (arg) void *arg; { tree type = arg; ASM_OUTPUT_DWARF_TAG (asm_out_file, TAG_ptr_to_member_type); sibling_attribute (); equate_type_number_to_die_number (type); member_attribute (TYPE_CONTEXT (type)); containing_type_attribute (TYPE_OFFSET_BASETYPE (type)); type_attribute (TREE_TYPE (type), 0, 0); } static void output_compile_unit_die (arg) void *arg; { const char *main_input_filename = arg; const char *language_string = lang_hooks.name; ASM_OUTPUT_DWARF_TAG (asm_out_file, TAG_compile_unit); sibling_attribute (); dienum_push (); name_attribute (main_input_filename); { char producer[250]; sprintf (producer, "%s %s", language_string, version_string); producer_attribute (producer); } if (strcmp (language_string, "GNU C++") == 0) language_attribute (LANG_C_PLUS_PLUS); else if (strcmp (language_string, "GNU Ada") == 0) language_attribute (LANG_ADA83); else if (strcmp (language_string, "GNU F77") == 0) language_attribute (LANG_FORTRAN77); else if (strcmp (language_string, "GNU Pascal") == 0) language_attribute (LANG_PASCAL83); else if (strcmp (language_string, "GNU Java") == 0) language_attribute (LANG_JAVA); else language_attribute (LANG_C89); low_pc_attribute (TEXT_BEGIN_LABEL); high_pc_attribute (TEXT_END_LABEL); if (debug_info_level >= DINFO_LEVEL_NORMAL) stmt_list_attribute (LINE_BEGIN_LABEL); { const char *wd = getpwd (); if (wd) comp_dir_attribute (wd); } if (debug_info_level >= DINFO_LEVEL_NORMAL && use_gnu_debug_info_extensions) { sf_names_attribute (SFNAMES_BEGIN_LABEL); src_info_attribute (SRCINFO_BEGIN_LABEL); if (debug_info_level >= DINFO_LEVEL_VERBOSE) mac_info_attribute (MACINFO_BEGIN_LABEL); } } static void output_string_type_die (arg) void *arg; { tree type = arg; ASM_OUTPUT_DWARF_TAG (asm_out_file, TAG_string_type); sibling_attribute (); equate_type_number_to_die_number (type); member_attribute (TYPE_CONTEXT (type)); /* this is a fixed length string */ byte_size_attribute (type); } static void output_inheritance_die (arg) void *arg; { tree binfo = arg; ASM_OUTPUT_DWARF_TAG (asm_out_file, TAG_inheritance); sibling_attribute (); type_attribute (BINFO_TYPE (binfo), 0, 0); data_member_location_attribute (binfo); if (TREE_VIA_VIRTUAL (binfo)) { ASM_OUTPUT_DWARF_ATTRIBUTE (asm_out_file, AT_virtual); ASM_OUTPUT_DWARF_STRING_NEWLINE (asm_out_file, ""); } if (TREE_VIA_PUBLIC (binfo)) { ASM_OUTPUT_DWARF_ATTRIBUTE (asm_out_file, AT_public); ASM_OUTPUT_DWARF_STRING_NEWLINE (asm_out_file, ""); } else if (TREE_VIA_PROTECTED (binfo)) { ASM_OUTPUT_DWARF_ATTRIBUTE (asm_out_file, AT_protected); ASM_OUTPUT_DWARF_STRING_NEWLINE (asm_out_file, ""); } } static void output_structure_type_die (arg) void *arg; { tree type = arg; ASM_OUTPUT_DWARF_TAG (asm_out_file, TAG_structure_type); sibling_attribute (); equate_type_number_to_die_number (type); name_attribute (type_tag (type)); member_attribute (TYPE_CONTEXT (type)); /* If this type has been completed, then give it a byte_size attribute and prepare to give a list of members. Otherwise, don't do either of these things. In the latter case, we will not be generating a list of members (since we don't have any idea what they might be for an incomplete type). */ if (COMPLETE_TYPE_P (type)) { dienum_push (); byte_size_attribute (type); } } /* Output a DIE to represent a declared function (either file-scope or block-local) which has "internal linkage" (according to ANSI-C). */ static void output_local_subroutine_die (arg) void *arg; { tree decl = arg; tree origin = decl_ultimate_origin (decl); ASM_OUTPUT_DWARF_TAG (asm_out_file, TAG_subroutine); sibling_attribute (); dienum_push (); if (origin != NULL) abstract_origin_attribute (origin); else { tree type = TREE_TYPE (decl); name_and_src_coords_attributes (decl); inline_attribute (decl); prototyped_attribute (type); member_attribute (DECL_CONTEXT (decl)); type_attribute (TREE_TYPE (type), 0, 0); pure_or_virtual_attribute (decl); } if (DECL_ABSTRACT (decl)) equate_decl_number_to_die_number (decl); else { /* Avoid getting screwed up in cases where a function was declared static but where no definition was ever given for it. */ if (TREE_ASM_WRITTEN (decl)) { char label[MAX_ARTIFICIAL_LABEL_BYTES]; low_pc_attribute (function_start_label (decl)); sprintf (label, FUNC_END_LABEL_FMT, current_function_funcdef_no); high_pc_attribute (label); if (use_gnu_debug_info_extensions) { sprintf (label, BODY_BEGIN_LABEL_FMT, current_function_funcdef_no); body_begin_attribute (label); sprintf (label, BODY_END_LABEL_FMT, current_function_funcdef_no); body_end_attribute (label); } } } } static void output_subroutine_type_die (arg) void *arg; { tree type = arg; tree return_type = TREE_TYPE (type); ASM_OUTPUT_DWARF_TAG (asm_out_file, TAG_subroutine_type); sibling_attribute (); dienum_push (); equate_type_number_to_die_number (type); prototyped_attribute (type); member_attribute (TYPE_CONTEXT (type)); type_attribute (return_type, 0, 0); } static void output_typedef_die (arg) void *arg; { tree decl = arg; tree origin = decl_ultimate_origin (decl); ASM_OUTPUT_DWARF_TAG (asm_out_file, TAG_typedef); sibling_attribute (); if (origin != NULL) abstract_origin_attribute (origin); else { name_and_src_coords_attributes (decl); member_attribute (DECL_CONTEXT (decl)); type_attribute (TREE_TYPE (decl), TREE_READONLY (decl), TREE_THIS_VOLATILE (decl)); } if (DECL_ABSTRACT (decl)) equate_decl_number_to_die_number (decl); } static void output_union_type_die (arg) void *arg; { tree type = arg; ASM_OUTPUT_DWARF_TAG (asm_out_file, TAG_union_type); sibling_attribute (); equate_type_number_to_die_number (type); name_attribute (type_tag (type)); member_attribute (TYPE_CONTEXT (type)); /* If this type has been completed, then give it a byte_size attribute and prepare to give a list of members. Otherwise, don't do either of these things. In the latter case, we will not be generating a list of members (since we don't have any idea what they might be for an incomplete type). */ if (COMPLETE_TYPE_P (type)) { dienum_push (); byte_size_attribute (type); } } /* Generate a special type of DIE used as a stand-in for a trailing ellipsis at the end of an (ANSI prototyped) formal parameters list. */ static void output_unspecified_parameters_die (arg) void *arg; { tree decl_or_type = arg; ASM_OUTPUT_DWARF_TAG (asm_out_file, TAG_unspecified_parameters); sibling_attribute (); /* This kludge is here only for the sake of being compatible with what the USL CI5 C compiler does. The specification of Dwarf Version 1 doesn't say that TAG_unspecified_parameters DIEs should contain any attributes other than the AT_sibling attribute, but they are certainly allowed to contain additional attributes, and the CI5 compiler generates AT_name, AT_fund_type, and AT_location attributes within TAG_unspecified_parameters DIEs which appear in the child lists for DIEs representing function definitions, so we do likewise here. */ if (TREE_CODE (decl_or_type) == FUNCTION_DECL && DECL_INITIAL (decl_or_type)) { name_attribute ("..."); fund_type_attribute (FT_pointer); /* location_attribute (?); */ } } static void output_padded_null_die (arg) void *arg ATTRIBUTE_UNUSED; { ASM_OUTPUT_ALIGN (asm_out_file, 2); /* 2**2 == 4 */ } /*************************** end of DIEs *********************************/ /* Generate some type of DIE. This routine generates the generic outer wrapper stuff which goes around all types of DIE's (regardless of their TAGs. All forms of DIEs start with a DIE-specific label, followed by a DIE-length word, followed by the guts of the DIE itself. After the guts of the DIE, there must always be a terminator label for the DIE. */ static void output_die (die_specific_output_function, param) void (*die_specific_output_function) PARAMS ((void *)); void *param; { char begin_label[MAX_ARTIFICIAL_LABEL_BYTES]; char end_label[MAX_ARTIFICIAL_LABEL_BYTES]; current_dienum = NEXT_DIE_NUM; NEXT_DIE_NUM = next_unused_dienum; sprintf (begin_label, DIE_BEGIN_LABEL_FMT, current_dienum); sprintf (end_label, DIE_END_LABEL_FMT, current_dienum); /* Write a label which will act as the name for the start of this DIE. */ ASM_OUTPUT_LABEL (asm_out_file, begin_label); /* Write the DIE-length word. */ ASM_OUTPUT_DWARF_DELTA4 (asm_out_file, end_label, begin_label); /* Fill in the guts of the DIE. */ next_unused_dienum++; die_specific_output_function (param); /* Write a label which will act as the name for the end of this DIE. */ ASM_OUTPUT_LABEL (asm_out_file, end_label); } static void end_sibling_chain () { char begin_label[MAX_ARTIFICIAL_LABEL_BYTES]; current_dienum = NEXT_DIE_NUM; NEXT_DIE_NUM = next_unused_dienum; sprintf (begin_label, DIE_BEGIN_LABEL_FMT, current_dienum); /* Write a label which will act as the name for the start of this DIE. */ ASM_OUTPUT_LABEL (asm_out_file, begin_label); /* Write the DIE-length word. */ ASM_OUTPUT_DWARF_DATA4 (asm_out_file, 4); dienum_pop (); } /* Generate a list of nameless TAG_formal_parameter DIEs (and perhaps a TAG_unspecified_parameters DIE) to represent the types of the formal parameters as specified in some function type specification (except for those which appear as part of a function *definition*). Note that we must be careful here to output all of the parameter DIEs *before* we output any DIEs needed to represent the types of the formal parameters. This keeps svr4 SDB happy because it (incorrectly) thinks that the first non-parameter DIE it sees ends the formal parameter list. */ static void output_formal_types (function_or_method_type) tree function_or_method_type; { tree link; tree formal_type = NULL; tree first_parm_type = TYPE_ARG_TYPES (function_or_method_type); /* Set TREE_ASM_WRITTEN while processing the parameters, lest we get bogus recursion when outputting tagged types local to a function declaration. */ int save_asm_written = TREE_ASM_WRITTEN (function_or_method_type); TREE_ASM_WRITTEN (function_or_method_type) = 1; /* In the case where we are generating a formal types list for a C++ non-static member function type, skip over the first thing on the TYPE_ARG_TYPES list because it only represents the type of the hidden `this pointer'. The debugger should be able to figure out (without being explicitly told) that this non-static member function type takes a `this pointer' and should be able to figure what the type of that hidden parameter is from the AT_member attribute of the parent TAG_subroutine_type DIE. */ if (TREE_CODE (function_or_method_type) == METHOD_TYPE) first_parm_type = TREE_CHAIN (first_parm_type); /* Make our first pass over the list of formal parameter types and output a TAG_formal_parameter DIE for each one. */ for (link = first_parm_type; link; link = TREE_CHAIN (link)) { formal_type = TREE_VALUE (link); if (formal_type == void_type_node) break; /* Output a (nameless) DIE to represent the formal parameter itself. */ output_die (output_formal_parameter_die, formal_type); } /* If this function type has an ellipsis, add a TAG_unspecified_parameters DIE to the end of the parameter list. */ if (formal_type != void_type_node) output_die (output_unspecified_parameters_die, function_or_method_type); /* Make our second (and final) pass over the list of formal parameter types and output DIEs to represent those types (as necessary). */ for (link = TYPE_ARG_TYPES (function_or_method_type); link; link = TREE_CHAIN (link)) { formal_type = TREE_VALUE (link); if (formal_type == void_type_node) break; output_type (formal_type, function_or_method_type); } TREE_ASM_WRITTEN (function_or_method_type) = save_asm_written; } /* Remember a type in the pending_types_list. */ static void pend_type (type) tree type; { if (pending_types == pending_types_allocated) { pending_types_allocated += PENDING_TYPES_INCREMENT; pending_types_list = (tree *) xrealloc (pending_types_list, sizeof (tree) * pending_types_allocated); } pending_types_list[pending_types++] = type; /* Mark the pending type as having been output already (even though it hasn't been). This prevents the type from being added to the pending_types_list more than once. */ TREE_ASM_WRITTEN (type) = 1; } /* Return non-zero if it is legitimate to output DIEs to represent a given type while we are generating the list of child DIEs for some DIE (e.g. a function or lexical block DIE) associated with a given scope. See the comments within the function for a description of when it is considered legitimate to output DIEs for various kinds of types. Note that TYPE_CONTEXT(type) may be NULL (to indicate global scope) or it may point to a BLOCK node (for types local to a block), or to a FUNCTION_DECL node (for types local to the heading of some function definition), or to a FUNCTION_TYPE node (for types local to the prototyped parameter list of a function type specification), or to a RECORD_TYPE, UNION_TYPE, or QUAL_UNION_TYPE node (in the case of C++ nested types). The `scope' parameter should likewise be NULL or should point to a BLOCK node, a FUNCTION_DECL node, a FUNCTION_TYPE node, a RECORD_TYPE node, a UNION_TYPE node, or a QUAL_UNION_TYPE node. This function is used only for deciding when to "pend" and when to "un-pend" types to/from the pending_types_list. Note that we sometimes make use of this "type pending" feature in a rather twisted way to temporarily delay the production of DIEs for the types of formal parameters. (We do this just to make svr4 SDB happy.) It order to delay the production of DIEs representing types of formal parameters, callers of this function supply `fake_containing_scope' as the `scope' parameter to this function. Given that fake_containing_scope is a tagged type which is *not* the containing scope for *any* other type, the desired effect is achieved, i.e. output of DIEs representing types is temporarily suspended, and any type DIEs which would have otherwise been output are instead placed onto the pending_types_list. Later on, we force these (temporarily pended) types to be output simply by calling `output_pending_types_for_scope' with an actual argument equal to the true scope of the types we temporarily pended. */ static inline int type_ok_for_scope (type, scope) tree type; tree scope; { /* Tagged types (i.e. struct, union, and enum types) must always be output only in the scopes where they actually belong (or else the scoping of their own tag names and the scoping of their member names will be incorrect). Non-tagged-types on the other hand can generally be output anywhere, except that svr4 SDB really doesn't want to see them nested within struct or union types, so here we say it is always OK to immediately output any such a (non-tagged) type, so long as we are not within such a context. Note that the only kinds of non-tagged types which we will be dealing with here (for C and C++ anyway) will be array types and function types. */ return is_tagged_type (type) ? (TYPE_CONTEXT (type) == scope /* Ignore namespaces for the moment. */ || (scope == NULL_TREE && TREE_CODE (TYPE_CONTEXT (type)) == NAMESPACE_DECL) || (scope == NULL_TREE && is_tagged_type (TYPE_CONTEXT (type)) && TREE_ASM_WRITTEN (TYPE_CONTEXT (type)))) : (scope == NULL_TREE || ! is_tagged_type (scope)); } /* Output any pending types (from the pending_types list) which we can output now (taking into account the scope that we are working on now). For each type output, remove the given type from the pending_types_list *before* we try to output it. Note that we have to process the list in beginning-to-end order, because the call made here to output_type may cause yet more types to be added to the end of the list, and we may have to output some of them too. */ static void output_pending_types_for_scope (containing_scope) tree containing_scope; { unsigned i; for (i = 0; i < pending_types; ) { tree type = pending_types_list[i]; if (type_ok_for_scope (type, containing_scope)) { tree *mover; tree *limit; pending_types--; limit = &pending_types_list[pending_types]; for (mover = &pending_types_list[i]; mover < limit; mover++) *mover = *(mover+1); /* Un-mark the type as having been output already (because it hasn't been, really). Then call output_type to generate a Dwarf representation of it. */ TREE_ASM_WRITTEN (type) = 0; output_type (type, containing_scope); /* Don't increment the loop counter in this case because we have shifted all of the subsequent pending types down one element in the pending_types_list array. */ } else i++; } } /* Remember a type in the incomplete_types_list. */ static void add_incomplete_type (type) tree type; { if (incomplete_types == incomplete_types_allocated) { incomplete_types_allocated += INCOMPLETE_TYPES_INCREMENT; incomplete_types_list = (tree *) xrealloc (incomplete_types_list, sizeof (tree) * incomplete_types_allocated); } incomplete_types_list[incomplete_types++] = type; } /* Walk through the list of incomplete types again, trying once more to emit full debugging info for them. */ static void retry_incomplete_types () { tree type; finalizing = 1; while (incomplete_types) { --incomplete_types; type = incomplete_types_list[incomplete_types]; output_type (type, NULL_TREE); } } static void output_type (type, containing_scope) tree type; tree containing_scope; { if (type == 0 || type == error_mark_node) return; /* We are going to output a DIE to represent the unqualified version of this type (i.e. without any const or volatile qualifiers) so get the main variant (i.e. the unqualified version) of this type now. */ type = type_main_variant (type); if (TREE_ASM_WRITTEN (type)) { if (finalizing && AGGREGATE_TYPE_P (type)) { tree member; /* Some of our nested types might not have been defined when we were written out before; force them out now. */ for (member = TYPE_FIELDS (type); member; member = TREE_CHAIN (member)) if (TREE_CODE (member) == TYPE_DECL && ! TREE_ASM_WRITTEN (TREE_TYPE (member))) output_type (TREE_TYPE (member), containing_scope); } return; } /* If this is a nested type whose containing class hasn't been written out yet, writing it out will cover this one, too. */ if (TYPE_CONTEXT (type) && TYPE_P (TYPE_CONTEXT (type)) && ! TREE_ASM_WRITTEN (TYPE_CONTEXT (type))) { output_type (TYPE_CONTEXT (type), containing_scope); return; } /* Don't generate any DIEs for this type now unless it is OK to do so (based upon what `type_ok_for_scope' tells us). */ if (! type_ok_for_scope (type, containing_scope)) { pend_type (type); return; } switch (TREE_CODE (type)) { case ERROR_MARK: break; case VECTOR_TYPE: output_type (TYPE_DEBUG_REPRESENTATION_TYPE (type), containing_scope); break; case POINTER_TYPE: case REFERENCE_TYPE: /* Prevent infinite recursion in cases where this is a recursive type. Recursive types are possible in Ada. */ TREE_ASM_WRITTEN (type) = 1; /* For these types, all that is required is that we output a DIE (or a set of DIEs) to represent the "basis" type. */ output_type (TREE_TYPE (type), containing_scope); break; case OFFSET_TYPE: /* This code is used for C++ pointer-to-data-member types. */ /* Output a description of the relevant class type. */ output_type (TYPE_OFFSET_BASETYPE (type), containing_scope); /* Output a description of the type of the object pointed to. */ output_type (TREE_TYPE (type), containing_scope); /* Now output a DIE to represent this pointer-to-data-member type itself. */ output_die (output_ptr_to_mbr_type_die, type); break; case SET_TYPE: output_type (TYPE_DOMAIN (type), containing_scope); output_die (output_set_type_die, type); break; case FILE_TYPE: output_type (TREE_TYPE (type), containing_scope); abort (); /* No way to represent these in Dwarf yet! */ break; case FUNCTION_TYPE: /* Force out return type (in case it wasn't forced out already). */ output_type (TREE_TYPE (type), containing_scope); output_die (output_subroutine_type_die, type); output_formal_types (type); end_sibling_chain (); break; case METHOD_TYPE: /* Force out return type (in case it wasn't forced out already). */ output_type (TREE_TYPE (type), containing_scope); output_die (output_subroutine_type_die, type); output_formal_types (type); end_sibling_chain (); break; case ARRAY_TYPE: if (TYPE_STRING_FLAG (type) && TREE_CODE(TREE_TYPE(type)) == CHAR_TYPE) { output_type (TREE_TYPE (type), containing_scope); output_die (output_string_type_die, type); } else { tree element_type; element_type = TREE_TYPE (type); while (TREE_CODE (element_type) == ARRAY_TYPE) element_type = TREE_TYPE (element_type); output_type (element_type, containing_scope); output_die (output_array_type_die, type); } break; case ENUMERAL_TYPE: case RECORD_TYPE: case UNION_TYPE: case QUAL_UNION_TYPE: /* For a non-file-scope tagged type, we can always go ahead and output a Dwarf description of this type right now, even if the type in question is still incomplete, because if this local type *was* ever completed anywhere within its scope, that complete definition would already have been attached to this RECORD_TYPE, UNION_TYPE, QUAL_UNION_TYPE or ENUMERAL_TYPE node by the time we reach this point. That's true because of the way the front-end does its processing of file-scope declarations (of functions and class types) within which other types might be nested. The C and C++ front-ends always gobble up such "local scope" things en-mass before they try to output *any* debugging information for any of the stuff contained inside them and thus, we get the benefit here of what is (in effect) a pre-resolution of forward references to tagged types in local scopes. Note however that for file-scope tagged types we cannot assume that such pre-resolution of forward references has taken place. A given file-scope tagged type may appear to be incomplete when we reach this point, but it may yet be given a full definition (at file-scope) later on during compilation. In order to avoid generating a premature (and possibly incorrect) set of Dwarf DIEs for such (as yet incomplete) file-scope tagged types, we generate nothing at all for as-yet incomplete file-scope tagged types here unless we are making our special "finalization" pass for file-scope things at the very end of compilation. At that time, we will certainly know as much about each file-scope tagged type as we are ever going to know, so at that point in time, we can safely generate correct Dwarf descriptions for these file- scope tagged types. */ if (!COMPLETE_TYPE_P (type) && (TYPE_CONTEXT (type) == NULL || AGGREGATE_TYPE_P (TYPE_CONTEXT (type)) || TREE_CODE (TYPE_CONTEXT (type)) == NAMESPACE_DECL) && !finalizing) { /* We don't need to do this for function-local types. */ if (! decl_function_context (TYPE_STUB_DECL (type))) add_incomplete_type (type); return; /* EARLY EXIT! Avoid setting TREE_ASM_WRITTEN. */ } /* Prevent infinite recursion in cases where the type of some member of this type is expressed in terms of this type itself. */ TREE_ASM_WRITTEN (type) = 1; /* Output a DIE to represent the tagged type itself. */ switch (TREE_CODE (type)) { case ENUMERAL_TYPE: output_die (output_enumeration_type_die, type); return; /* a special case -- nothing left to do so just return */ case RECORD_TYPE: output_die (output_structure_type_die, type); break; case UNION_TYPE: case QUAL_UNION_TYPE: output_die (output_union_type_die, type); break; default: abort (); /* Should never happen. */ } /* If this is not an incomplete type, output descriptions of each of its members. Note that as we output the DIEs necessary to represent the members of this record or union type, we will also be trying to output DIEs to represent the *types* of those members. However the `output_type' function (above) will specifically avoid generating type DIEs for member types *within* the list of member DIEs for this (containing) type except for those types (of members) which are explicitly marked as also being members of this (containing) type themselves. The g++ front- end can force any given type to be treated as a member of some other (containing) type by setting the TYPE_CONTEXT of the given (member) type to point to the TREE node representing the appropriate (containing) type. */ if (COMPLETE_TYPE_P (type)) { /* First output info about the base classes. */ if (TYPE_BINFO (type) && TYPE_BINFO_BASETYPES (type)) { register tree bases = TYPE_BINFO_BASETYPES (type); register int n_bases = TREE_VEC_LENGTH (bases); register int i; for (i = 0; i < n_bases; i++) { tree binfo = TREE_VEC_ELT (bases, i); output_type (BINFO_TYPE (binfo), containing_scope); output_die (output_inheritance_die, binfo); } } ++in_class; { tree normal_member; /* Now output info about the data members and type members. */ for (normal_member = TYPE_FIELDS (type); normal_member; normal_member = TREE_CHAIN (normal_member)) output_decl (normal_member, type); } { tree func_member; /* Now output info about the function members (if any). */ for (func_member = TYPE_METHODS (type); func_member; func_member = TREE_CHAIN (func_member)) { /* Don't include clones in the member list. */ if (DECL_ABSTRACT_ORIGIN (func_member)) continue; output_decl (func_member, type); } } --in_class; /* RECORD_TYPEs, UNION_TYPEs, and QUAL_UNION_TYPEs are themselves scopes (at least in C++) so we must now output any nested pending types which are local just to this type. */ output_pending_types_for_scope (type); end_sibling_chain (); /* Terminate member chain. */ } break; case VOID_TYPE: case INTEGER_TYPE: case REAL_TYPE: case COMPLEX_TYPE: case BOOLEAN_TYPE: case CHAR_TYPE: break; /* No DIEs needed for fundamental types. */ case LANG_TYPE: /* No Dwarf representation currently defined. */ break; default: abort (); } TREE_ASM_WRITTEN (type) = 1; } static void output_tagged_type_instantiation (type) tree type; { if (type == 0 || type == error_mark_node) return; /* We are going to output a DIE to represent the unqualified version of this type (i.e. without any const or volatile qualifiers) so make sure that we have the main variant (i.e. the unqualified version) of this type now. */ if (type != type_main_variant (type)) abort (); if (!TREE_ASM_WRITTEN (type)) abort (); switch (TREE_CODE (type)) { case ERROR_MARK: break; case ENUMERAL_TYPE: output_die (output_inlined_enumeration_type_die, type); break; case RECORD_TYPE: output_die (output_inlined_structure_type_die, type); break; case UNION_TYPE: case QUAL_UNION_TYPE: output_die (output_inlined_union_type_die, type); break; default: abort (); /* Should never happen. */ } } /* Output a TAG_lexical_block DIE followed by DIEs to represent all of the things which are local to the given block. */ static void output_block (stmt, depth) tree stmt; int depth; { int must_output_die = 0; tree origin; enum tree_code origin_code; /* Ignore blocks never really used to make RTL. */ if (! stmt || ! TREE_USED (stmt) || (!TREE_ASM_WRITTEN (stmt) && !BLOCK_ABSTRACT (stmt))) return; /* Determine the "ultimate origin" of this block. This block may be an inlined instance of an inlined instance of inline function, so we have to trace all of the way back through the origin chain to find out what sort of node actually served as the original seed for the creation of the current block. */ origin = block_ultimate_origin (stmt); origin_code = (origin != NULL) ? TREE_CODE (origin) : ERROR_MARK; /* Determine if we need to output any Dwarf DIEs at all to represent this block. */ if (origin_code == FUNCTION_DECL) /* The outer scopes for inlinings *must* always be represented. We generate TAG_inlined_subroutine DIEs for them. (See below.) */ must_output_die = 1; else { /* In the case where the current block represents an inlining of the "body block" of an inline function, we must *NOT* output any DIE for this block because we have already output a DIE to represent the whole inlined function scope and the "body block" of any function doesn't really represent a different scope according to ANSI C rules. So we check here to make sure that this block does not represent a "body block inlining" before trying to set the `must_output_die' flag. */ if (! is_body_block (origin ? origin : stmt)) { /* Determine if this block directly contains any "significant" local declarations which we will need to output DIEs for. */ if (debug_info_level > DINFO_LEVEL_TERSE) /* We are not in terse mode so *any* local declaration counts as being a "significant" one. */ must_output_die = (BLOCK_VARS (stmt) != NULL); else { tree decl; /* We are in terse mode, so only local (nested) function definitions count as "significant" local declarations. */ for (decl = BLOCK_VARS (stmt); decl; decl = TREE_CHAIN (decl)) if (TREE_CODE (decl) == FUNCTION_DECL && DECL_INITIAL (decl)) { must_output_die = 1; break; } } } } /* It would be a waste of space to generate a Dwarf TAG_lexical_block DIE for any block which contains no significant local declarations at all. Rather, in such cases we just call `output_decls_for_scope' so that any needed Dwarf info for any sub-blocks will get properly generated. Note that in terse mode, our definition of what constitutes a "significant" local declaration gets restricted to include only inlined function instances and local (nested) function definitions. */ if (origin_code == FUNCTION_DECL && BLOCK_ABSTRACT (stmt)) /* We don't care about an abstract inlined subroutine. */; else if (must_output_die) { output_die ((origin_code == FUNCTION_DECL) ? output_inlined_subroutine_die : output_lexical_block_die, stmt); output_decls_for_scope (stmt, depth); end_sibling_chain (); } else output_decls_for_scope (stmt, depth); } /* Output all of the decls declared within a given scope (also called a `binding contour') and (recursively) all of it's sub-blocks. */ static void output_decls_for_scope (stmt, depth) tree stmt; int depth; { /* Ignore blocks never really used to make RTL. */ if (! stmt || ! TREE_USED (stmt)) return; /* Output the DIEs to represent all of the data objects, functions, typedefs, and tagged types declared directly within this block but not within any nested sub-blocks. */ { tree decl; for (decl = BLOCK_VARS (stmt); decl; decl = TREE_CHAIN (decl)) output_decl (decl, stmt); } output_pending_types_for_scope (stmt); /* Output the DIEs to represent all sub-blocks (and the items declared therein) of this block. */ { tree subblocks; for (subblocks = BLOCK_SUBBLOCKS (stmt); subblocks; subblocks = BLOCK_CHAIN (subblocks)) output_block (subblocks, depth + 1); } } /* Is this a typedef we can avoid emitting? */ static inline int is_redundant_typedef (decl) tree decl; { if (TYPE_DECL_IS_STUB (decl)) return 1; if (DECL_ARTIFICIAL (decl) && DECL_CONTEXT (decl) && is_tagged_type (DECL_CONTEXT (decl)) && TREE_CODE (TYPE_NAME (DECL_CONTEXT (decl))) == TYPE_DECL && DECL_NAME (decl) == DECL_NAME (TYPE_NAME (DECL_CONTEXT (decl)))) /* Also ignore the artificial member typedef for the class name. */ return 1; return 0; } /* Output Dwarf .debug information for a decl described by DECL. */ static void output_decl (decl, containing_scope) tree decl; tree containing_scope; { /* Make a note of the decl node we are going to be working on. We may need to give the user the source coordinates of where it appeared in case we notice (later on) that something about it looks screwy. */ dwarf_last_decl = decl; if (TREE_CODE (decl) == ERROR_MARK) return; /* If a structure is declared within an initialization, e.g. as the operand of a sizeof, then it will not have a name. We don't want to output a DIE for it, as the tree nodes are in the temporary obstack */ if ((TREE_CODE (TREE_TYPE (decl)) == RECORD_TYPE || TREE_CODE (TREE_TYPE (decl)) == UNION_TYPE) && ((DECL_NAME (decl) == 0 && TYPE_NAME (TREE_TYPE (decl)) == 0) || (TYPE_FIELDS (TREE_TYPE (decl)) && (TREE_CODE (TYPE_FIELDS (TREE_TYPE (decl))) == ERROR_MARK)))) return; /* If this ..._DECL node is marked to be ignored, then ignore it. */ if (DECL_IGNORED_P (decl)) return; switch (TREE_CODE (decl)) { case CONST_DECL: /* The individual enumerators of an enum type get output when we output the Dwarf representation of the relevant enum type itself. */ break; case FUNCTION_DECL: /* If we are in terse mode, don't output any DIEs to represent mere function declarations. Also, if we are conforming to the DWARF version 1 specification, don't output DIEs for mere function declarations. */ if (DECL_INITIAL (decl) == NULL_TREE) #if (DWARF_VERSION > 1) if (debug_info_level <= DINFO_LEVEL_TERSE) #endif break; /* Before we describe the FUNCTION_DECL itself, make sure that we have described its return type. */ output_type (TREE_TYPE (TREE_TYPE (decl)), containing_scope); { /* And its containing type. */ register tree origin = decl_class_context (decl); if (origin) output_type (origin, containing_scope); } /* If we're emitting an out-of-line copy of an inline function, set up to refer to the abstract instance emitted from dwarfout_deferred_inline_function. */ if (DECL_INLINE (decl) && ! DECL_ABSTRACT (decl) && ! (containing_scope && TYPE_P (containing_scope))) set_decl_origin_self (decl); /* If the following DIE will represent a function definition for a function with "extern" linkage, output a special "pubnames" DIE label just ahead of the actual DIE. A reference to this label was already generated in the .debug_pubnames section sub-entry for this function definition. */ if (TREE_PUBLIC (decl)) { char label[MAX_ARTIFICIAL_LABEL_BYTES]; sprintf (label, PUB_DIE_LABEL_FMT, next_pubname_number++); ASM_OUTPUT_LABEL (asm_out_file, label); } /* Now output a DIE to represent the function itself. */ output_die (TREE_PUBLIC (decl) || DECL_EXTERNAL (decl) ? output_global_subroutine_die : output_local_subroutine_die, decl); /* Now output descriptions of the arguments for this function. This gets (unnecessarily?) complex because of the fact that the DECL_ARGUMENT list for a FUNCTION_DECL doesn't indicate cases where there was a trailing `...' at the end of the formal parameter list. In order to find out if there was a trailing ellipsis or not, we must instead look at the type associated with the FUNCTION_DECL. This will be a node of type FUNCTION_TYPE. If the chain of type nodes hanging off of this FUNCTION_TYPE node ends with a void_type_node then there should *not* be an ellipsis at the end. */ /* In the case where we are describing a mere function declaration, all we need to do here (and all we *can* do here) is to describe the *types* of its formal parameters. */ if (decl != current_function_decl || in_class) output_formal_types (TREE_TYPE (decl)); else { /* Generate DIEs to represent all known formal parameters */ tree arg_decls = DECL_ARGUMENTS (decl); tree parm; /* WARNING! Kludge zone ahead! Here we have a special hack for svr4 SDB compatibility. Instead of passing the current FUNCTION_DECL node as the second parameter (i.e. the `containing_scope' parameter) to `output_decl' (as we ought to) we instead pass a pointer to our own private fake_containing_scope node. That node is a RECORD_TYPE node which NO OTHER TYPE may ever actually be a member of. This pointer will ultimately get passed into `output_type' as its `containing_scope' parameter. `Output_type' will then perform its part in the hack... i.e. it will pend the type of the formal parameter onto the pending_types list. Later on, when we are done generating the whole sequence of formal parameter DIEs for this function definition, we will un-pend all previously pended types of formal parameters for this function definition. This whole kludge prevents any type DIEs from being mixed in with the formal parameter DIEs. That's good because svr4 SDB believes that the list of formal parameter DIEs for a function ends wherever the first non-formal-parameter DIE appears. Thus, we have to keep the formal parameter DIEs segregated. They must all appear (consecutively) at the start of the list of children for the DIE representing the function definition. Then (and only then) may we output any additional DIEs needed to represent the types of these formal parameters. */ /* When generating DIEs, generate the unspecified_parameters DIE instead if we come across the arg "__builtin_va_alist" */ for (parm = arg_decls; parm; parm = TREE_CHAIN (parm)) if (TREE_CODE (parm) == PARM_DECL) { if (DECL_NAME(parm) && !strcmp(IDENTIFIER_POINTER(DECL_NAME(parm)), "__builtin_va_alist") ) output_die (output_unspecified_parameters_die, decl); else output_decl (parm, fake_containing_scope); } /* Now that we have finished generating all of the DIEs to represent the formal parameters themselves, force out any DIEs needed to represent their types. We do this simply by un-pending all previously pended types which can legitimately go into the chain of children DIEs for the current FUNCTION_DECL. */ output_pending_types_for_scope (decl); /* Decide whether we need an unspecified_parameters DIE at the end. There are 2 more cases to do this for: 1) the ansi ... declaration - this is detectable when the end of the arg list is not a void_type_node 2) an unprototyped function declaration (not a definition). This just means that we have no info about the parameters at all. */ { tree fn_arg_types = TYPE_ARG_TYPES (TREE_TYPE (decl)); if (fn_arg_types) { /* this is the prototyped case, check for ... */ if (TREE_VALUE (tree_last (fn_arg_types)) != void_type_node) output_die (output_unspecified_parameters_die, decl); } else { /* this is unprototyped, check for undefined (just declaration) */ if (!DECL_INITIAL (decl)) output_die (output_unspecified_parameters_die, decl); } } /* Output Dwarf info for all of the stuff within the body of the function (if it has one - it may be just a declaration). */ { tree outer_scope = DECL_INITIAL (decl); if (outer_scope && TREE_CODE (outer_scope) != ERROR_MARK) { /* Note that here, `outer_scope' is a pointer to the outermost BLOCK node created to represent a function. This outermost BLOCK actually represents the outermost binding contour for the function, i.e. the contour in which the function's formal parameters and labels get declared. Curiously, it appears that the front end doesn't actually put the PARM_DECL nodes for the current function onto the BLOCK_VARS list for this outer scope. (They are strung off of the DECL_ARGUMENTS list for the function instead.) The BLOCK_VARS list for the `outer_scope' does provide us with a list of the LABEL_DECL nodes for the function however, and we output DWARF info for those here. Just within the `outer_scope' there will be a BLOCK node representing the function's outermost pair of curly braces, and any blocks used for the base and member initializers of a C++ constructor function. */ output_decls_for_scope (outer_scope, 0); /* Finally, force out any pending types which are local to the outermost block of this function definition. These will all have a TYPE_CONTEXT which points to the FUNCTION_DECL node itself. */ output_pending_types_for_scope (decl); } } } /* Generate a terminator for the list of stuff `owned' by this function. */ end_sibling_chain (); break; case TYPE_DECL: /* If we are in terse mode, don't generate any DIEs to represent any actual typedefs. Note that even when we are in terse mode, we must still output DIEs to represent those tagged types which are used (directly or indirectly) in the specification of either a return type or a formal parameter type of some function. */ if (debug_info_level <= DINFO_LEVEL_TERSE) if (! TYPE_DECL_IS_STUB (decl) || (! TYPE_USED_FOR_FUNCTION (TREE_TYPE (decl)) && ! in_class)) return; /* In the special case of a TYPE_DECL node representing the declaration of some type tag, if the given TYPE_DECL is marked as having been instantiated from some other (original) TYPE_DECL node (e.g. one which was generated within the original definition of an inline function) we have to generate a special (abbreviated) TAG_structure_type, TAG_union_type, or TAG_enumeration-type DIE here. */ if (TYPE_DECL_IS_STUB (decl) && DECL_ABSTRACT_ORIGIN (decl)) { output_tagged_type_instantiation (TREE_TYPE (decl)); return; } output_type (TREE_TYPE (decl), containing_scope); if (! is_redundant_typedef (decl)) /* Output a DIE to represent the typedef itself. */ output_die (output_typedef_die, decl); break; case LABEL_DECL: if (debug_info_level >= DINFO_LEVEL_NORMAL) output_die (output_label_die, decl); break; case VAR_DECL: /* If we are conforming to the DWARF version 1 specification, don't generated any DIEs to represent mere external object declarations. */ #if (DWARF_VERSION <= 1) if (DECL_EXTERNAL (decl) && ! TREE_PUBLIC (decl)) break; #endif /* If we are in terse mode, don't generate any DIEs to represent any variable declarations or definitions. */ if (debug_info_level <= DINFO_LEVEL_TERSE) break; /* Output any DIEs that are needed to specify the type of this data object. */ output_type (TREE_TYPE (decl), containing_scope); { /* And its containing type. */ register tree origin = decl_class_context (decl); if (origin) output_type (origin, containing_scope); } /* If the following DIE will represent a data object definition for a data object with "extern" linkage, output a special "pubnames" DIE label just ahead of the actual DIE. A reference to this label was already generated in the .debug_pubnames section sub-entry for this data object definition. */ if (TREE_PUBLIC (decl) && ! DECL_ABSTRACT (decl)) { char label[MAX_ARTIFICIAL_LABEL_BYTES]; sprintf (label, PUB_DIE_LABEL_FMT, next_pubname_number++); ASM_OUTPUT_LABEL (asm_out_file, label); } /* Now output the DIE to represent the data object itself. This gets complicated because of the possibility that the VAR_DECL really represents an inlined instance of a formal parameter for an inline function. */ { void (*func) PARAMS ((void *)); register tree origin = decl_ultimate_origin (decl); if (origin != NULL && TREE_CODE (origin) == PARM_DECL) func = output_formal_parameter_die; else { if (TREE_PUBLIC (decl) || DECL_EXTERNAL (decl)) func = output_global_variable_die; else func = output_local_variable_die; } output_die (func, decl); } break; case FIELD_DECL: /* Ignore the nameless fields that are used to skip bits. */ if (DECL_NAME (decl) != 0) { output_type (member_declared_type (decl), containing_scope); output_die (output_member_die, decl); } break; case PARM_DECL: /* Force out the type of this formal, if it was not forced out yet. Note that here we can run afoul of a bug in "classic" svr4 SDB. It should be able to grok the presence of type DIEs within a list of TAG_formal_parameter DIEs, but it doesn't. */ output_type (TREE_TYPE (decl), containing_scope); output_die (output_formal_parameter_die, decl); break; case NAMESPACE_DECL: /* Ignore for now. */ break; default: abort (); } } /* Output debug information for a function. */ static void dwarfout_function_decl (decl) tree decl; { dwarfout_file_scope_decl (decl, 0); } /* Debug information for a global DECL. Called from toplev.c after compilation proper has finished. */ static void dwarfout_global_decl (decl) tree decl; { /* Output DWARF information for file-scope tentative data object declarations, file-scope (extern) function declarations (which had no corresponding body) and file-scope tagged type declarations and definitions which have not yet been forced out. */ if (TREE_CODE (decl) != FUNCTION_DECL || !DECL_INITIAL (decl)) dwarfout_file_scope_decl (decl, 1); } /* DECL is an inline function, whose body is present, but which is not being output at this point. (We're putting that off until we need to do it.) */ static void dwarfout_deferred_inline_function (decl) tree decl; { /* Generate the DWARF info for the "abstract" instance of a function which we may later generate inlined and/or out-of-line instances of. */ if ((DECL_INLINE (decl) || DECL_ABSTRACT (decl)) && ! DECL_ABSTRACT_ORIGIN (decl)) { /* The front-end may not have set CURRENT_FUNCTION_DECL, but the DWARF code expects it to be set in this case. Intuitively, DECL is the function we just finished defining, so setting CURRENT_FUNCTION_DECL is sensible. */ tree saved_cfd = current_function_decl; int was_abstract = DECL_ABSTRACT (decl); current_function_decl = decl; /* Let the DWARF code do its work. */ set_decl_abstract_flags (decl, 1); dwarfout_file_scope_decl (decl, 0); if (! was_abstract) set_decl_abstract_flags (decl, 0); /* Reset CURRENT_FUNCTION_DECL. */ current_function_decl = saved_cfd; } } static void dwarfout_file_scope_decl (decl, set_finalizing) tree decl; int set_finalizing; { if (TREE_CODE (decl) == ERROR_MARK) return; /* If this ..._DECL node is marked to be ignored, then ignore it. */ if (DECL_IGNORED_P (decl)) return; switch (TREE_CODE (decl)) { case FUNCTION_DECL: /* Ignore this FUNCTION_DECL if it refers to a builtin declaration of a builtin function. Explicit programmer-supplied declarations of these same functions should NOT be ignored however. */ if (DECL_EXTERNAL (decl) && DECL_FUNCTION_CODE (decl)) return; /* What we would really like to do here is to filter out all mere file-scope declarations of file-scope functions which are never referenced later within this translation unit (and keep all of ones that *are* referenced later on) but we aren't clairvoyant, so we have no idea which functions will be referenced in the future (i.e. later on within the current translation unit). So here we just ignore all file-scope function declarations which are not also definitions. If and when the debugger needs to know something about these functions, it will have to hunt around and find the DWARF information associated with the *definition* of the function. Note that we can't just check `DECL_EXTERNAL' to find out which FUNCTION_DECL nodes represent definitions and which ones represent mere declarations. We have to check `DECL_INITIAL' instead. That's because the C front-end supports some weird semantics for "extern inline" function definitions. These can get inlined within the current translation unit (an thus, we need to generate DWARF info for their abstract instances so that the DWARF info for the concrete inlined instances can have something to refer to) but the compiler never generates any out-of-lines instances of such things (despite the fact that they *are* definitions). The important point is that the C front-end marks these "extern inline" functions as DECL_EXTERNAL, but we need to generate DWARF for them anyway. Note that the C++ front-end also plays some similar games for inline function definitions appearing within include files which also contain `#pragma interface' pragmas. */ if (DECL_INITIAL (decl) == NULL_TREE) return; if (TREE_PUBLIC (decl) && ! DECL_EXTERNAL (decl) && ! DECL_ABSTRACT (decl)) { char label[MAX_ARTIFICIAL_LABEL_BYTES]; /* Output a .debug_pubnames entry for a public function defined in this compilation unit. */ fputc ('\n', asm_out_file); ASM_OUTPUT_PUSH_SECTION (asm_out_file, DEBUG_PUBNAMES_SECTION); sprintf (label, PUB_DIE_LABEL_FMT, next_pubname_number); ASM_OUTPUT_DWARF_ADDR (asm_out_file, label); ASM_OUTPUT_DWARF_STRING_NEWLINE (asm_out_file, IDENTIFIER_POINTER (DECL_NAME (decl))); ASM_OUTPUT_POP_SECTION (asm_out_file); } break; case VAR_DECL: /* Ignore this VAR_DECL if it refers to a file-scope extern data object declaration and if the declaration was never even referenced from within this entire compilation unit. We suppress these DIEs in order to save space in the .debug section (by eliminating entries which are probably useless). Note that we must not suppress block-local extern declarations (whether used or not) because that would screw-up the debugger's name lookup mechanism and cause it to miss things which really ought to be in scope at a given point. */ if (DECL_EXTERNAL (decl) && !TREE_USED (decl)) return; if (TREE_PUBLIC (decl) && ! DECL_EXTERNAL (decl) && GET_CODE (DECL_RTL (decl)) == MEM && ! DECL_ABSTRACT (decl)) { char label[MAX_ARTIFICIAL_LABEL_BYTES]; if (debug_info_level >= DINFO_LEVEL_NORMAL) { /* Output a .debug_pubnames entry for a public variable defined in this compilation unit. */ fputc ('\n', asm_out_file); ASM_OUTPUT_PUSH_SECTION (asm_out_file, DEBUG_PUBNAMES_SECTION); sprintf (label, PUB_DIE_LABEL_FMT, next_pubname_number); ASM_OUTPUT_DWARF_ADDR (asm_out_file, label); ASM_OUTPUT_DWARF_STRING_NEWLINE (asm_out_file, IDENTIFIER_POINTER (DECL_NAME (decl))); ASM_OUTPUT_POP_SECTION (asm_out_file); } if (DECL_INITIAL (decl) == NULL) { /* Output a .debug_aranges entry for a public variable which is tentatively defined in this compilation unit. */ fputc ('\n', asm_out_file); ASM_OUTPUT_PUSH_SECTION (asm_out_file, DEBUG_ARANGES_SECTION); ASM_OUTPUT_DWARF_ADDR (asm_out_file, IDENTIFIER_POINTER (DECL_ASSEMBLER_NAME (decl))); ASM_OUTPUT_DWARF_DATA4 (asm_out_file, (unsigned) int_size_in_bytes (TREE_TYPE (decl))); ASM_OUTPUT_POP_SECTION (asm_out_file); } } /* If we are in terse mode, don't generate any DIEs to represent any variable declarations or definitions. */ if (debug_info_level <= DINFO_LEVEL_TERSE) return; break; case TYPE_DECL: /* Don't bother trying to generate any DIEs to represent any of the normal built-in types for the language we are compiling, except in cases where the types in question are *not* DWARF fundamental types. We make an exception in the case of non-fundamental types for the sake of objective C (and perhaps C++) because the GNU front-ends for these languages may in fact create certain "built-in" types which are (for example) RECORD_TYPEs. In such cases, we really need to output these (non-fundamental) types because other DIEs may contain references to them. */ /* Also ignore language dependent types here, because they are probably also built-in types. If we didn't ignore them, then we would get references to undefined labels because output_type doesn't support them. So, for now, we need to ignore them to avoid assembler errors. */ /* ??? This code is different than the equivalent code in dwarf2out.c. The dwarf2out.c code is probably more correct. */ if (DECL_SOURCE_LINE (decl) == 0 && (type_is_fundamental (TREE_TYPE (decl)) || TREE_CODE (TREE_TYPE (decl)) == LANG_TYPE)) return; /* If we are in terse mode, don't generate any DIEs to represent any actual typedefs. Note that even when we are in terse mode, we must still output DIEs to represent those tagged types which are used (directly or indirectly) in the specification of either a return type or a formal parameter type of some function. */ if (debug_info_level <= DINFO_LEVEL_TERSE) if (! TYPE_DECL_IS_STUB (decl) || ! TYPE_USED_FOR_FUNCTION (TREE_TYPE (decl))) return; break; default: return; } fputc ('\n', asm_out_file); ASM_OUTPUT_PUSH_SECTION (asm_out_file, DEBUG_SECTION); finalizing = set_finalizing; output_decl (decl, NULL_TREE); /* NOTE: The call above to `output_decl' may have caused one or more file-scope named types (i.e. tagged types) to be placed onto the pending_types_list. We have to get those types off of that list at some point, and this is the perfect time to do it. If we didn't take them off now, they might still be on the list when cc1 finally exits. That might be OK if it weren't for the fact that when we put types onto the pending_types_list, we set the TREE_ASM_WRITTEN flag for these types, and that causes them never to be output unless `output_pending_types_for_scope' takes them off of the list and un-sets their TREE_ASM_WRITTEN flags. */ output_pending_types_for_scope (NULL_TREE); /* The above call should have totally emptied the pending_types_list if this is not a nested function or class. If this is a nested type, then the remaining pending_types will be emitted when the containing type is handled. */ if (! DECL_CONTEXT (decl)) { if (pending_types != 0) abort (); } ASM_OUTPUT_POP_SECTION (asm_out_file); } /* Output a marker (i.e. a label) for the beginning of the generated code for a lexical block. */ static void dwarfout_begin_block (line, blocknum) unsigned int line ATTRIBUTE_UNUSED; unsigned int blocknum; { char label[MAX_ARTIFICIAL_LABEL_BYTES]; function_section (current_function_decl); sprintf (label, BLOCK_BEGIN_LABEL_FMT, blocknum); ASM_OUTPUT_LABEL (asm_out_file, label); } /* Output a marker (i.e. a label) for the end of the generated code for a lexical block. */ static void dwarfout_end_block (line, blocknum) unsigned int line ATTRIBUTE_UNUSED; unsigned int blocknum; { char label[MAX_ARTIFICIAL_LABEL_BYTES]; function_section (current_function_decl); sprintf (label, BLOCK_END_LABEL_FMT, blocknum); ASM_OUTPUT_LABEL (asm_out_file, label); } /* Output a marker (i.e. a label) for the point in the generated code where the real body of the function begins (after parameters have been moved to their home locations). */ static void dwarfout_end_prologue (line) unsigned int line ATTRIBUTE_UNUSED; { char label[MAX_ARTIFICIAL_LABEL_BYTES]; if (! use_gnu_debug_info_extensions) return; function_section (current_function_decl); sprintf (label, BODY_BEGIN_LABEL_FMT, current_function_funcdef_no); ASM_OUTPUT_LABEL (asm_out_file, label); } /* Output a marker (i.e. a label) for the point in the generated code where the real body of the function ends (just before the epilogue code). */ static void dwarfout_end_function (line) unsigned int line ATTRIBUTE_UNUSED; { char label[MAX_ARTIFICIAL_LABEL_BYTES]; if (! use_gnu_debug_info_extensions) return; function_section (current_function_decl); sprintf (label, BODY_END_LABEL_FMT, current_function_funcdef_no); ASM_OUTPUT_LABEL (asm_out_file, label); } /* Output a marker (i.e. a label) for the absolute end of the generated code for a function definition. This gets called *after* the epilogue code has been generated. */ static void dwarfout_end_epilogue () { char label[MAX_ARTIFICIAL_LABEL_BYTES]; /* Output a label to mark the endpoint of the code generated for this function. */ sprintf (label, FUNC_END_LABEL_FMT, current_function_funcdef_no); ASM_OUTPUT_LABEL (asm_out_file, label); } static void shuffle_filename_entry (new_zeroth) filename_entry *new_zeroth; { filename_entry temp_entry; filename_entry *limit_p; filename_entry *move_p; if (new_zeroth == &filename_table[0]) return; temp_entry = *new_zeroth; /* Shift entries up in the table to make room at [0]. */ limit_p = &filename_table[0]; for (move_p = new_zeroth; move_p > limit_p; move_p--) *move_p = *(move_p-1); /* Install the found entry at [0]. */ filename_table[0] = temp_entry; } /* Create a new (string) entry for the .debug_sfnames section. */ static void generate_new_sfname_entry () { char label[MAX_ARTIFICIAL_LABEL_BYTES]; fputc ('\n', asm_out_file); ASM_OUTPUT_PUSH_SECTION (asm_out_file, DEBUG_SFNAMES_SECTION); sprintf (label, SFNAMES_ENTRY_LABEL_FMT, filename_table[0].number); ASM_OUTPUT_LABEL (asm_out_file, label); ASM_OUTPUT_DWARF_STRING_NEWLINE (asm_out_file, filename_table[0].name ? filename_table[0].name : ""); ASM_OUTPUT_POP_SECTION (asm_out_file); } /* Lookup a filename (in the list of filenames that we know about here in dwarfout.c) and return its "index". The index of each (known) filename is just a unique number which is associated with only that one filename. We need such numbers for the sake of generating labels (in the .debug_sfnames section) and references to those unique labels (in the .debug_srcinfo and .debug_macinfo sections). If the filename given as an argument is not found in our current list, add it to the list and assign it the next available unique index number. Whatever we do (i.e. whether we find a pre-existing filename or add a new one), we shuffle the filename found (or added) up to the zeroth entry of our list of filenames (which is always searched linearly). We do this so as to optimize the most common case for these filename lookups within dwarfout.c. The most common case by far is the case where we call lookup_filename to lookup the very same filename that we did a lookup on the last time we called lookup_filename. We make sure that this common case is fast because such cases will constitute 99.9% of the lookups we ever do (in practice). If we add a new filename entry to our table, we go ahead and generate the corresponding entry in the .debug_sfnames section right away. Doing so allows us to avoid tickling an assembler bug (present in some m68k assemblers) which yields assembly-time errors in cases where the difference of two label addresses is taken and where the two labels are in a section *other* than the one where the difference is being calculated, and where at least one of the two symbol references is a forward reference. (This bug could be tickled by our .debug_srcinfo entries if we don't output their corresponding .debug_sfnames entries before them.) */ static unsigned lookup_filename (file_name) const char *file_name; { filename_entry *search_p; filename_entry *limit_p = &filename_table[ft_entries]; for (search_p = filename_table; search_p < limit_p; search_p++) if (!strcmp (file_name, search_p->name)) { /* When we get here, we have found the filename that we were looking for in the filename_table. Now we want to make sure that it gets moved to the zero'th entry in the table (if it is not already there) so that subsequent attempts to find the same filename will find it as quickly as possible. */ shuffle_filename_entry (search_p); return filename_table[0].number; } /* We come here whenever we have a new filename which is not registered in the current table. Here we add it to the table. */ /* Prepare to add a new table entry by making sure there is enough space in the table to do so. If not, expand the current table. */ if (ft_entries == ft_entries_allocated) { ft_entries_allocated += FT_ENTRIES_INCREMENT; filename_table = (filename_entry *) xrealloc (filename_table, ft_entries_allocated * sizeof (filename_entry)); } /* Initially, add the new entry at the end of the filename table. */ filename_table[ft_entries].number = ft_entries; filename_table[ft_entries].name = xstrdup (file_name); /* Shuffle the new entry into filename_table[0]. */ shuffle_filename_entry (&filename_table[ft_entries]); if (debug_info_level >= DINFO_LEVEL_NORMAL) generate_new_sfname_entry (); ft_entries++; return filename_table[0].number; } static void generate_srcinfo_entry (line_entry_num, files_entry_num) unsigned line_entry_num; unsigned files_entry_num; { char label[MAX_ARTIFICIAL_LABEL_BYTES]; fputc ('\n', asm_out_file); ASM_OUTPUT_PUSH_SECTION (asm_out_file, DEBUG_SRCINFO_SECTION); sprintf (label, LINE_ENTRY_LABEL_FMT, line_entry_num); ASM_OUTPUT_DWARF_DELTA4 (asm_out_file, label, LINE_BEGIN_LABEL); sprintf (label, SFNAMES_ENTRY_LABEL_FMT, files_entry_num); ASM_OUTPUT_DWARF_DELTA4 (asm_out_file, label, SFNAMES_BEGIN_LABEL); ASM_OUTPUT_POP_SECTION (asm_out_file); } static void dwarfout_source_line (line, filename) unsigned int line; const char *filename; { if (debug_info_level >= DINFO_LEVEL_NORMAL /* We can't emit line number info for functions in separate sections, because the assembler can't subtract labels in different sections. */ && DECL_SECTION_NAME (current_function_decl) == NULL_TREE) { char label[MAX_ARTIFICIAL_LABEL_BYTES]; static unsigned last_line_entry_num = 0; static unsigned prev_file_entry_num = (unsigned) -1; unsigned this_file_entry_num; function_section (current_function_decl); sprintf (label, LINE_CODE_LABEL_FMT, ++last_line_entry_num); ASM_OUTPUT_LABEL (asm_out_file, label); fputc ('\n', asm_out_file); if (use_gnu_debug_info_extensions) this_file_entry_num = lookup_filename (filename); else this_file_entry_num = (unsigned) -1; ASM_OUTPUT_PUSH_SECTION (asm_out_file, LINE_SECTION); if (this_file_entry_num != prev_file_entry_num) { char line_entry_label[MAX_ARTIFICIAL_LABEL_BYTES]; sprintf (line_entry_label, LINE_ENTRY_LABEL_FMT, last_line_entry_num); ASM_OUTPUT_LABEL (asm_out_file, line_entry_label); } { const char *tail = strrchr (filename, '/'); if (tail != NULL) filename = tail; } dw2_asm_output_data (4, line, "%s:%u", filename, line); ASM_OUTPUT_DWARF_DATA2 (asm_out_file, 0xffff); ASM_OUTPUT_DWARF_DELTA4 (asm_out_file, label, TEXT_BEGIN_LABEL); ASM_OUTPUT_POP_SECTION (asm_out_file); if (this_file_entry_num != prev_file_entry_num) generate_srcinfo_entry (last_line_entry_num, this_file_entry_num); prev_file_entry_num = this_file_entry_num; } } /* Generate an entry in the .debug_macinfo section. */ static void generate_macinfo_entry (type, offset, string) unsigned int type; rtx offset; const char *string; { if (! use_gnu_debug_info_extensions) return; fputc ('\n', asm_out_file); ASM_OUTPUT_PUSH_SECTION (asm_out_file, DEBUG_MACINFO_SECTION); assemble_integer (gen_rtx_PLUS (SImode, GEN_INT (type << 24), offset), 4, BITS_PER_UNIT, 1); ASM_OUTPUT_DWARF_STRING_NEWLINE (asm_out_file, string); ASM_OUTPUT_POP_SECTION (asm_out_file); } /* Wrapper for toplev.c callback to check debug info level. */ static void dwarfout_start_source_file_check (line, filename) unsigned int line; const char *filename; { if (debug_info_level == DINFO_LEVEL_VERBOSE) dwarfout_start_source_file (line, filename); } static void dwarfout_start_source_file (line, filename) unsigned int line ATTRIBUTE_UNUSED; const char *filename; { char label[MAX_ARTIFICIAL_LABEL_BYTES]; const char *label1, *label2; sprintf (label, SFNAMES_ENTRY_LABEL_FMT, lookup_filename (filename)); label1 = (*label == '*') + label; label2 = (*SFNAMES_BEGIN_LABEL == '*') + SFNAMES_BEGIN_LABEL; generate_macinfo_entry (MACINFO_start, gen_rtx_MINUS (Pmode, gen_rtx_SYMBOL_REF (Pmode, label1), gen_rtx_SYMBOL_REF (Pmode, label2)), ""); } /* Wrapper for toplev.c callback to check debug info level. */ static void dwarfout_end_source_file_check (lineno) unsigned lineno; { if (debug_info_level == DINFO_LEVEL_VERBOSE) dwarfout_end_source_file (lineno); } static void dwarfout_end_source_file (lineno) unsigned lineno; { generate_macinfo_entry (MACINFO_resume, GEN_INT (lineno), ""); } /* Called from check_newline in c-parse.y. The `buffer' parameter contains the tail part of the directive line, i.e. the part which is past the initial whitespace, #, whitespace, directive-name, whitespace part. */ static void dwarfout_define (lineno, buffer) unsigned lineno; const char *buffer; { static int initialized = 0; if (!initialized) { dwarfout_start_source_file (0, primary_filename); initialized = 1; } generate_macinfo_entry (MACINFO_define, GEN_INT (lineno), buffer); } /* Called from check_newline in c-parse.y. The `buffer' parameter contains the tail part of the directive line, i.e. the part which is past the initial whitespace, #, whitespace, directive-name, whitespace part. */ static void dwarfout_undef (lineno, buffer) unsigned lineno; const char *buffer; { generate_macinfo_entry (MACINFO_undef, GEN_INT (lineno), buffer); } /* Set up for Dwarf output at the start of compilation. */ static void dwarfout_init (main_input_filename) const char *main_input_filename; { warning ("support for the DWARF1 debugging format is deprecated"); /* Remember the name of the primary input file. */ primary_filename = main_input_filename; /* Allocate the initial hunk of the pending_sibling_stack. */ pending_sibling_stack = (unsigned *) xmalloc (PENDING_SIBLINGS_INCREMENT * sizeof (unsigned)); pending_siblings_allocated = PENDING_SIBLINGS_INCREMENT; pending_siblings = 1; /* Allocate the initial hunk of the filename_table. */ filename_table = (filename_entry *) xmalloc (FT_ENTRIES_INCREMENT * sizeof (filename_entry)); ft_entries_allocated = FT_ENTRIES_INCREMENT; ft_entries = 0; /* Allocate the initial hunk of the pending_types_list. */ pending_types_list = (tree *) xmalloc (PENDING_TYPES_INCREMENT * sizeof (tree)); pending_types_allocated = PENDING_TYPES_INCREMENT; pending_types = 0; /* Create an artificial RECORD_TYPE node which we can use in our hack to get the DIEs representing types of formal parameters to come out only *after* the DIEs for the formal parameters themselves. */ fake_containing_scope = make_node (RECORD_TYPE); /* Output a starting label for the .text section. */ fputc ('\n', asm_out_file); ASM_OUTPUT_PUSH_SECTION (asm_out_file, TEXT_SECTION_NAME); ASM_OUTPUT_LABEL (asm_out_file, TEXT_BEGIN_LABEL); ASM_OUTPUT_POP_SECTION (asm_out_file); /* Output a starting label for the .data section. */ fputc ('\n', asm_out_file); ASM_OUTPUT_PUSH_SECTION (asm_out_file, DATA_SECTION_NAME); ASM_OUTPUT_LABEL (asm_out_file, DATA_BEGIN_LABEL); ASM_OUTPUT_POP_SECTION (asm_out_file); #if 0 /* GNU C doesn't currently use .data1. */ /* Output a starting label for the .data1 section. */ fputc ('\n', asm_out_file); ASM_OUTPUT_PUSH_SECTION (asm_out_file, DATA1_SECTION_NAME); ASM_OUTPUT_LABEL (asm_out_file, DATA1_BEGIN_LABEL); ASM_OUTPUT_POP_SECTION (asm_out_file); #endif /* Output a starting label for the .rodata section. */ fputc ('\n', asm_out_file); ASM_OUTPUT_PUSH_SECTION (asm_out_file, RODATA_SECTION_NAME); ASM_OUTPUT_LABEL (asm_out_file, RODATA_BEGIN_LABEL); ASM_OUTPUT_POP_SECTION (asm_out_file); #if 0 /* GNU C doesn't currently use .rodata1. */ /* Output a starting label for the .rodata1 section. */ fputc ('\n', asm_out_file); ASM_OUTPUT_PUSH_SECTION (asm_out_file, RODATA1_SECTION_NAME); ASM_OUTPUT_LABEL (asm_out_file, RODATA1_BEGIN_LABEL); ASM_OUTPUT_POP_SECTION (asm_out_file); #endif /* Output a starting label for the .bss section. */ fputc ('\n', asm_out_file); ASM_OUTPUT_PUSH_SECTION (asm_out_file, BSS_SECTION_NAME); ASM_OUTPUT_LABEL (asm_out_file, BSS_BEGIN_LABEL); ASM_OUTPUT_POP_SECTION (asm_out_file); if (debug_info_level >= DINFO_LEVEL_NORMAL) { if (use_gnu_debug_info_extensions) { /* Output a starting label and an initial (compilation directory) entry for the .debug_sfnames section. The starting label will be referenced by the initial entry in the .debug_srcinfo section. */ fputc ('\n', asm_out_file); ASM_OUTPUT_PUSH_SECTION (asm_out_file, DEBUG_SFNAMES_SECTION); ASM_OUTPUT_LABEL (asm_out_file, SFNAMES_BEGIN_LABEL); { const char *pwd = getpwd (); char *dirname; if (!pwd) fatal_io_error ("can't get current directory"); dirname = concat (pwd, "/", NULL); ASM_OUTPUT_DWARF_STRING_NEWLINE (asm_out_file, dirname); free (dirname); } ASM_OUTPUT_POP_SECTION (asm_out_file); } if (debug_info_level >= DINFO_LEVEL_VERBOSE && use_gnu_debug_info_extensions) { /* Output a starting label for the .debug_macinfo section. This label will be referenced by the AT_mac_info attribute in the TAG_compile_unit DIE. */ fputc ('\n', asm_out_file); ASM_OUTPUT_PUSH_SECTION (asm_out_file, DEBUG_MACINFO_SECTION); ASM_OUTPUT_LABEL (asm_out_file, MACINFO_BEGIN_LABEL); ASM_OUTPUT_POP_SECTION (asm_out_file); } /* Generate the initial entry for the .line section. */ fputc ('\n', asm_out_file); ASM_OUTPUT_PUSH_SECTION (asm_out_file, LINE_SECTION); ASM_OUTPUT_LABEL (asm_out_file, LINE_BEGIN_LABEL); ASM_OUTPUT_DWARF_DELTA4 (asm_out_file, LINE_END_LABEL, LINE_BEGIN_LABEL); ASM_OUTPUT_DWARF_ADDR (asm_out_file, TEXT_BEGIN_LABEL); ASM_OUTPUT_POP_SECTION (asm_out_file); if (use_gnu_debug_info_extensions) { /* Generate the initial entry for the .debug_srcinfo section. */ fputc ('\n', asm_out_file); ASM_OUTPUT_PUSH_SECTION (asm_out_file, DEBUG_SRCINFO_SECTION); ASM_OUTPUT_LABEL (asm_out_file, SRCINFO_BEGIN_LABEL); ASM_OUTPUT_DWARF_ADDR (asm_out_file, LINE_BEGIN_LABEL); ASM_OUTPUT_DWARF_ADDR (asm_out_file, SFNAMES_BEGIN_LABEL); ASM_OUTPUT_DWARF_ADDR (asm_out_file, TEXT_BEGIN_LABEL); ASM_OUTPUT_DWARF_ADDR (asm_out_file, TEXT_END_LABEL); #ifdef DWARF_TIMESTAMPS ASM_OUTPUT_DWARF_DATA4 (asm_out_file, time (NULL)); #else ASM_OUTPUT_DWARF_DATA4 (asm_out_file, -1); #endif ASM_OUTPUT_POP_SECTION (asm_out_file); } /* Generate the initial entry for the .debug_pubnames section. */ fputc ('\n', asm_out_file); ASM_OUTPUT_PUSH_SECTION (asm_out_file, DEBUG_PUBNAMES_SECTION); ASM_OUTPUT_DWARF_ADDR (asm_out_file, DEBUG_BEGIN_LABEL); ASM_OUTPUT_POP_SECTION (asm_out_file); /* Generate the initial entry for the .debug_aranges section. */ fputc ('\n', asm_out_file); ASM_OUTPUT_PUSH_SECTION (asm_out_file, DEBUG_ARANGES_SECTION); ASM_OUTPUT_DWARF_DELTA4 (asm_out_file, DEBUG_ARANGES_END_LABEL, DEBUG_ARANGES_BEGIN_LABEL); ASM_OUTPUT_LABEL (asm_out_file, DEBUG_ARANGES_BEGIN_LABEL); ASM_OUTPUT_DWARF_DATA1 (asm_out_file, 1); ASM_OUTPUT_DWARF_ADDR (asm_out_file, DEBUG_BEGIN_LABEL); ASM_OUTPUT_POP_SECTION (asm_out_file); } /* Setup first DIE number == 1. */ NEXT_DIE_NUM = next_unused_dienum++; /* Generate the initial DIE for the .debug section. Note that the (string) value given in the AT_name attribute of the TAG_compile_unit DIE will (typically) be a relative pathname and that this pathname should be taken as being relative to the directory from which the compiler was invoked when the given (base) source file was compiled. */ fputc ('\n', asm_out_file); ASM_OUTPUT_PUSH_SECTION (asm_out_file, DEBUG_SECTION); ASM_OUTPUT_LABEL (asm_out_file, DEBUG_BEGIN_LABEL); output_die (output_compile_unit_die, (PTR) main_input_filename); ASM_OUTPUT_POP_SECTION (asm_out_file); fputc ('\n', asm_out_file); } /* Output stuff that dwarf requires at the end of every file. */ static void dwarfout_finish (main_input_filename) const char *main_input_filename ATTRIBUTE_UNUSED; { char label[MAX_ARTIFICIAL_LABEL_BYTES]; fputc ('\n', asm_out_file); ASM_OUTPUT_PUSH_SECTION (asm_out_file, DEBUG_SECTION); retry_incomplete_types (); fputc ('\n', asm_out_file); /* Mark the end of the chain of siblings which represent all file-scope declarations in this compilation unit. */ /* The (null) DIE which represents the terminator for the (sibling linked) list of file-scope items is *special*. Normally, we would just call end_sibling_chain at this point in order to output a word with the value `4' and that word would act as the terminator for the list of DIEs describing file-scope items. Unfortunately, if we were to simply do that, the label that would follow this DIE in the .debug section (i.e. `..D2') would *not* be properly aligned (as it must be on some machines) to a 4 byte boundary. In order to force the label `..D2' to get aligned to a 4 byte boundary, the trick used is to insert extra (otherwise useless) padding bytes into the (null) DIE that we know must precede the ..D2 label in the .debug section. The amount of padding required can be anywhere between 0 and 3 bytes. The length word at the start of this DIE (i.e. the one with the padding) would normally contain the value 4, but now it will also have to include the padding bytes, so it will instead have some value in the range 4..7. Fortunately, the rules of Dwarf say that any DIE whose length word contains *any* value less than 8 should be treated as a null DIE, so this trick works out nicely. Clever, eh? Don't give me any credit (or blame). I didn't think of this scheme. I just conformed to it. */ output_die (output_padded_null_die, (void *) 0); dienum_pop (); sprintf (label, DIE_BEGIN_LABEL_FMT, NEXT_DIE_NUM); ASM_OUTPUT_LABEL (asm_out_file, label); /* should be ..D2 */ ASM_OUTPUT_POP_SECTION (asm_out_file); /* Output a terminator label for the .text section. */ fputc ('\n', asm_out_file); ASM_OUTPUT_PUSH_SECTION (asm_out_file, TEXT_SECTION_NAME); ASM_OUTPUT_LABEL (asm_out_file, TEXT_END_LABEL); ASM_OUTPUT_POP_SECTION (asm_out_file); /* Output a terminator label for the .data section. */ fputc ('\n', asm_out_file); ASM_OUTPUT_PUSH_SECTION (asm_out_file, DATA_SECTION_NAME); ASM_OUTPUT_LABEL (asm_out_file, DATA_END_LABEL); ASM_OUTPUT_POP_SECTION (asm_out_file); #if 0 /* GNU C doesn't currently use .data1. */ /* Output a terminator label for the .data1 section. */ fputc ('\n', asm_out_file); ASM_OUTPUT_PUSH_SECTION (asm_out_file, DATA1_SECTION_NAME); ASM_OUTPUT_LABEL (asm_out_file, DATA1_END_LABEL); ASM_OUTPUT_POP_SECTION (asm_out_file); #endif /* Output a terminator label for the .rodata section. */ fputc ('\n', asm_out_file); ASM_OUTPUT_PUSH_SECTION (asm_out_file, RODATA_SECTION_NAME); ASM_OUTPUT_LABEL (asm_out_file, RODATA_END_LABEL); ASM_OUTPUT_POP_SECTION (asm_out_file); #if 0 /* GNU C doesn't currently use .rodata1. */ /* Output a terminator label for the .rodata1 section. */ fputc ('\n', asm_out_file); ASM_OUTPUT_PUSH_SECTION (asm_out_file, RODATA1_SECTION_NAME); ASM_OUTPUT_LABEL (asm_out_file, RODATA1_END_LABEL); ASM_OUTPUT_POP_SECTION (asm_out_file); #endif /* Output a terminator label for the .bss section. */ fputc ('\n', asm_out_file); ASM_OUTPUT_PUSH_SECTION (asm_out_file, BSS_SECTION_NAME); ASM_OUTPUT_LABEL (asm_out_file, BSS_END_LABEL); ASM_OUTPUT_POP_SECTION (asm_out_file); if (debug_info_level >= DINFO_LEVEL_NORMAL) { /* Output a terminating entry for the .line section. */ fputc ('\n', asm_out_file); ASM_OUTPUT_PUSH_SECTION (asm_out_file, LINE_SECTION); ASM_OUTPUT_LABEL (asm_out_file, LINE_LAST_ENTRY_LABEL); ASM_OUTPUT_DWARF_DATA4 (asm_out_file, 0); ASM_OUTPUT_DWARF_DATA2 (asm_out_file, 0xffff); ASM_OUTPUT_DWARF_DELTA4 (asm_out_file, TEXT_END_LABEL, TEXT_BEGIN_LABEL); ASM_OUTPUT_LABEL (asm_out_file, LINE_END_LABEL); ASM_OUTPUT_POP_SECTION (asm_out_file); if (use_gnu_debug_info_extensions) { /* Output a terminating entry for the .debug_srcinfo section. */ fputc ('\n', asm_out_file); ASM_OUTPUT_PUSH_SECTION (asm_out_file, DEBUG_SRCINFO_SECTION); ASM_OUTPUT_DWARF_DELTA4 (asm_out_file, LINE_LAST_ENTRY_LABEL, LINE_BEGIN_LABEL); ASM_OUTPUT_DWARF_DATA4 (asm_out_file, -1); ASM_OUTPUT_POP_SECTION (asm_out_file); } if (debug_info_level >= DINFO_LEVEL_VERBOSE) { /* Output terminating entries for the .debug_macinfo section. */ dwarfout_end_source_file (0); fputc ('\n', asm_out_file); ASM_OUTPUT_PUSH_SECTION (asm_out_file, DEBUG_MACINFO_SECTION); ASM_OUTPUT_DWARF_DATA4 (asm_out_file, 0); ASM_OUTPUT_DWARF_STRING_NEWLINE (asm_out_file, ""); ASM_OUTPUT_POP_SECTION (asm_out_file); } /* Generate the terminating entry for the .debug_pubnames section. */ fputc ('\n', asm_out_file); ASM_OUTPUT_PUSH_SECTION (asm_out_file, DEBUG_PUBNAMES_SECTION); ASM_OUTPUT_DWARF_DATA4 (asm_out_file, 0); ASM_OUTPUT_DWARF_STRING_NEWLINE (asm_out_file, ""); ASM_OUTPUT_POP_SECTION (asm_out_file); /* Generate the terminating entries for the .debug_aranges section. Note that we want to do this only *after* we have output the end labels (for the various program sections) which we are going to refer to here. This allows us to work around a bug in the m68k svr4 assembler. That assembler gives bogus assembly-time errors if (within any given section) you try to take the difference of two relocatable symbols, both of which are located within some other section, and if one (or both?) of the symbols involved is being forward-referenced. By generating the .debug_aranges entries at this late point in the assembly output, we skirt the issue simply by avoiding forward-references. */ fputc ('\n', asm_out_file); ASM_OUTPUT_PUSH_SECTION (asm_out_file, DEBUG_ARANGES_SECTION); ASM_OUTPUT_DWARF_ADDR (asm_out_file, TEXT_BEGIN_LABEL); ASM_OUTPUT_DWARF_DELTA4 (asm_out_file, TEXT_END_LABEL, TEXT_BEGIN_LABEL); ASM_OUTPUT_DWARF_ADDR (asm_out_file, DATA_BEGIN_LABEL); ASM_OUTPUT_DWARF_DELTA4 (asm_out_file, DATA_END_LABEL, DATA_BEGIN_LABEL); #if 0 /* GNU C doesn't currently use .data1. */ ASM_OUTPUT_DWARF_ADDR (asm_out_file, DATA1_BEGIN_LABEL); ASM_OUTPUT_DWARF_DELTA4 (asm_out_file, DATA1_END_LABEL, DATA1_BEGIN_LABEL); #endif ASM_OUTPUT_DWARF_ADDR (asm_out_file, RODATA_BEGIN_LABEL); ASM_OUTPUT_DWARF_DELTA4 (asm_out_file, RODATA_END_LABEL, RODATA_BEGIN_LABEL); #if 0 /* GNU C doesn't currently use .rodata1. */ ASM_OUTPUT_DWARF_ADDR (asm_out_file, RODATA1_BEGIN_LABEL); ASM_OUTPUT_DWARF_DELTA4 (asm_out_file, RODATA1_END_LABEL, RODATA1_BEGIN_LABEL); #endif ASM_OUTPUT_DWARF_ADDR (asm_out_file, BSS_BEGIN_LABEL); ASM_OUTPUT_DWARF_DELTA4 (asm_out_file, BSS_END_LABEL, BSS_BEGIN_LABEL); ASM_OUTPUT_DWARF_DATA4 (asm_out_file, 0); ASM_OUTPUT_DWARF_DATA4 (asm_out_file, 0); ASM_OUTPUT_LABEL (asm_out_file, DEBUG_ARANGES_END_LABEL); ASM_OUTPUT_POP_SECTION (asm_out_file); } /* There should not be any pending types left at the end. We need this now because it may not have been checked on the last call to dwarfout_file_scope_decl. */ if (pending_types != 0) abort (); } #endif /* DWARF_DEBUGGING_INFO */