In ELF, if a data symbol is defined in a shared library and used by
the main program, it will be subject to a "copy relocation". In this
scenario, the main program has a copy of the symbol in question, and a
relocation that tells ld.so to copy the data from the shared library.
Then the symbol in the main program is used to satisfy all references.
This patch changes gdb to handle this scenario. Data symbols coming
from ELF shared libraries get a special flag that indicates that the
symbol's address may be subject to copy relocation.
I looked briefly into handling copy relocations by looking at the
actual relocations in the main program, but this seemed difficult to
do with BFD.
Note that no caching is done here. Perhaps this could be changed if
need be; I wanted to avoid possible problems with either objfile
lifetimes and changes, or conflicts with the long-term (vapor-ware)
objfile splitting project.
gdb/ChangeLog
2019-10-02 Tom Tromey <tromey@adacore.com>
* symmisc.c (dump_msymbols): Don't use MSYMBOL_VALUE_ADDRESS.
* ada-lang.c (lesseq_defined_than): Handle
LOC_STATIC.
* dwarf2read.c (dwarf2_per_objfile): Add can_copy
parameter.
(dwarf2_has_info): Likewise.
(new_symbol): Set maybe_copied on symbol when
appropriate.
* dwarf2read.h (dwarf2_per_objfile): Add can_copy
parameter.
<can_copy>: New member.
* elfread.c (record_minimal_symbol): Set maybe_copied
on symbol when appropriate.
(elf_symfile_read): Update call to dwarf2_has_info.
* minsyms.c (lookup_minimal_symbol_linkage): New
function.
* minsyms.h (lookup_minimal_symbol_linkage): Declare.
* symtab.c (get_symbol_address, get_msymbol_address):
New functions.
* symtab.h (get_symbol_address, get_msymbol_address):
Declare.
(SYMBOL_VALUE_ADDRESS, MSYMBOL_VALUE_ADDRESS): Handle
maybe_copied.
(struct symbol, struct minimal_symbol) <maybe_copied>:
New member.
This changes SYMBOL_VALUE_ADDRESS to be an rvalue. The symbol readers
generally assign using this, so this also introduces
SET_SYMBOL_VALUE_ADDRESS and updates the readers. Making this change
is useful in a subsequent patch, which redefined SYMBOL_VALUE_ADDRESS.
gdb/ChangeLog
2019-10-02 Tom Tromey <tromey@adacore.com>
* coffread.c (process_coff_symbol): Update.
* dwarf2read.c (var_decode_location, new_symbol): Update.
* mdebugread.c (parse_symbol): Update.
* objfiles.c (relocate_one_symbol): Update.
* stabsread.c (define_symbol, fix_common_block)
(scan_file_globals): Update.
* symtab.h (SYMBOL_VALUE_ADDRESS): Expand to an rvalue.
(SET_SYMBOL_VALUE_ADDRESS): New macro.
* xcoffread.c (process_xcoff_symbol): Update.
This parameter is really a boolean, so change the type accordingly
and update the callers.
This is for symbol_set_names, add_psymbol_to_bcache, and
add_psymbol_to_list.
minimal_symbol_reader::record_full was already passing a bool
to symbol_set_names.
gdb/ChangeLog:
2019-09-11 Christian Biesinger <cbiesinger@google.com>
* dbxread.c (read_dbx_symtab): Update.
* dwarf2read.c (load_partial_dies): Update.
* mdebugread.c (parse_partial_symbols): Update.
(handle_psymbol_enumerators): Update.
* psympriv.h (add_psymbol_to_list): Change type of copy_names to bool.
* psymtab.c (add_psymbol_to_bcache): Likewise.
(add_psymbol_to_list): Likewise.
* symtab.c (symbol_set_names): Likewise.
* symtab.h (symbol_set_names): Likewise.
* xcoffread.c (scan_xcoff_symtab): Update.
This introduces a new helper function,
iterate_over_symbols_terminated, and changes psym_map_matching_symbols
to use it. A subsequent patch will introduce a new user of this
function in the DWARF reader.
gdb/ChangeLog
2019-09-10 Tom Tromey <tromey@adacore.com>
* psymtab.c (map_block): Remove.
(psym_map_matching_symbols): Use iterate_over_symbols_terminated.
* symtab.c (iterate_over_symbols_terminated): New function.
* symtab.c (iterate_over_symbols_terminated): Declare.
This changes iterate_over_symbols to return a bool. This allows it to
be reused in another context in a subsequent patch.
gdb/ChangeLog
2019-09-10 Tom Tromey <tromey@adacore.com>
* ada-lang.c (ada_iterate_over_symbols): Return bool.
* language.h (struct language_defn) <la_iterate_over_symbols>:
Return bool.
* symtab.c (iterate_over_symbols): Return bool.
* symtab.h (iterate_over_symbols): Return bool.
The 'info variables', its alias 'whereis', and 'info functions' all
include non-debug symbols in the output by default. The list of
non-debug symbols can sometimes be quite long, resulting in the
debug symbol based results being scrolled off the screen.
This commit adds a '-n' flag to all of the commands listed above that
excludes the non-debug symbols from the results, leaving just the
debug symbol based results.
gdb/ChangeLog:
* cli/cli-utils.c (info_print_options_defs): Delete.
(make_info_print_options_def_group): Delete.
(extract_info_print_options): Delete.
(info_print_command_completer): Delete.
(info_print_args_help): Add extra parameter, and optionally
include text about -n flag.
* cli/cli-utils.h (struct info_print_options): Delete.
(extract_info_print_options): Delete declaration.
(info_print_command_completer): Delete declaration.
(info_print_args_help): Add extra parameter, extend header
comment.
* python/python.c (gdbpy_rbreak): Pass additional parameter to
search_symbols.
* stack.c (struct info_print_options): New type.
(info_print_options_defs): New file scoped variable.
(make_info_print_options_def_group): New static function.
(info_print_command_completer): New static function.
(info_locals_command): Update to use new local functions.
(info_args_command): Likewise.
(_initialize_stack): Add extra parameter to calls to
info_print_args_help.
* symtab.c (search_symbols): Add extra parameter, use this to
possibly excluse non-debug symbols.
(symtab_symbol_info): Add extra parameter, which is passed on to
search_symbols.
(struct info_print_options): New type.
(info_print_options_defs): New file scoped variable.
(make_info_print_options_def_group): New static function.
(info_print_command_completer): New static function.
(info_variables_command): Update to use local functions, and pass
extra parameter through to symtab_symbol_info.
(info_functions_command): Likewise.
(info_types_command): Pass additional argument through to
symtab_symbol_info.
(rbreak_command): Pass extra argument to search_symbols.
(_initialize_symtab): Add extra arguments for calls to
info_print_args_help, and update help text for 'info variables',
'whereis', and 'info functions' commands.
* symtab.h (search_symbols): Add extra argument to declaration.
* NEWS: Mention new flags.
gdb/doc/ChangeLog:
* gdb.texinfo (Symbols): Add information about the -n flag to
"info variables" and "info functions".
gdb/testsuite/ChangeLog:
* gdb.base/info-fun.exp: Extend to test the -n flag for 'info
functions'. Reindent as needed.
* gdb.base/info-var-f1.c: New file.
* gdb.base/info-var-f2.c: New file.
* gdb.base/info-var.exp: New file.
* gdb.base/info-var.h: New file.
Currently, gdb stores the number of lines and an array of file offsets
for the start of each line in struct symtab. This patch moves this
information to the source cache. This has two benefits.
First, it allows gdb to read a source file less frequently.
Currently, a source file may be read multiple times: once when
computing the file offsets, once when highlighting, and then pieces
may be read again while printing source lines. With this change, the
file is read once for its source text and file offsets; and then
perhaps read again if it is evicted from the cache.
Second, if multiple symtabs cover the same source file, then this will
share the file offsets between them. I'm not sure whether this
happens in practice.
gdb/ChangeLog
2019-08-06 Tom Tromey <tromey@adacore.com>
* annotate.c (annotate_source_line): Use g_source_cache.
* source-cache.c (source_cache::get_plain_source_lines): Change
parameters. Populate m_offset_cache.
(source_cache::ensure): New method.
(source_cache::get_line_charpos): New method.
(extract_lines): Move lower. Change parameters.
(source_cache::get_source_lines): Move lower.
* source-cache.h (class source_cache): Update comment.
<get_line_charpos>: New method.
<get_source_lines>: Update comment.
<clear>: Clear m_offset_cache.
<get_plain_source_lines>: Change parameters.
<ensure>: New method
<m_offset_cache>: New member.
* source.c (forget_cached_source_info_for_objfile): Update.
(info_source_command): Use g_source_cache.
(find_source_lines, open_source_file_with_line_charpos): Remove.
(print_source_lines_base, search_command_helper): Use g_source_cache.
* source.h (open_source_file_with_line_charpos): Don't declare.
* symtab.h (struct symtab) <nlines, line_charpos>: Remove.
* tui/tui-source.c (tui_source_window::do_scroll_vertical):
Use g_source_cache.
This has no behavior change in itself, but allows a future patch
to add a function to the Python API to look up symbols in the
static block.
gdb/ChangeLog:
2019-07-24 Christian Biesinger <cbiesinger@google.com>
* compile/compile-object-load.c (compile_object_load): Pass GLOBAL_SCOPE.
* solib-spu.c (spu_lookup_lib_symbol): Pass GLOBAL_SCOPE.
* solib-svr4.c (elf_lookup_lib_symbol): Pass GLOBAL_SCOPE.
* symtab.c (lookup_global_symbol_from_objfile): Add a scope parameter.
* symtab.h (lookup_global_symbol_from_objfile): Likewise.
This patch constifies the return type of main_name. There is a
comment indicating that this wasn't possible at some point in the
past, but whatever the barrier was, it is gone now.
Tested by rebuilding.
gdb/ChangeLog
2019-07-18 Tom Tromey <tromey@adacore.com>
* symtab.c (main_name): Constify return type.
* symfile.c (set_initial_language): Update.
* symtab.h (main_name): Constify return type.
This changes struct minimal_symbol to inherit from general_symbol_info
and updates various macros to cope.
Because MSYMBOL_SET_LANGUAGE and MSYMBOL_SET_NAMES were only used from
a single spot, this patch removes them in favor of simply inlining
their definitions. I consider this to be somewhat cleaner, not least
because the "phony polymorphism" provided by such macros is not useful
in practice.
gdb/ChangeLog
2019-03-15 Tom Tromey <tom@tromey.com>
* symtab.h (struct minimal_symbol): Derive from
general_symbol_info.
(MSYMBOL_VALUE, MSYMBOL_VALUE_RAW_ADDRESS)
(MSYMBOL_VALUE_ADDRESS, MSYMBOL_VALUE_BYTES)
(MSYMBOL_BLOCK_VALUE, MSYMBOL_VALUE_CHAIN, MSYMBOL_LANGUAGE)
(MSYMBOL_SECTION, MSYMBOL_OBJ_SECTION, MSYMBOL_NATURAL_NAME)
(MSYMBOL_LINKAGE_NAME, MSYMBOL_DEMANGLED_NAME)
(MSYMBOL_SEARCH_NAME): Update.
(MSYMBOL_SET_LANGUAGE, MSYMBOL_SET_NAMES): Remove.
* solib.c (gdb_bfd_lookup_symbol_from_symtab): Don't use memset.
* minsyms.c (minimal_symbol_reader::record_full): Update.
Note that print_msymbol_info does not (yet?) print data msymbol
using variable_name_style, as otherwise 'info variables'
would show the non debugging symbols in variable name style,
but 'real' variables would be not styled.
2019-02-12 Philippe Waroquiers <philippe.waroquiers@skynet.be>
* symtab.h (struct minimal_symbol data_p): New const method.
(struct minimal_symbol text_p): Likewise.
* symtab.c (output_source_filename): Use file name style
to print file name.
(print_symbol_info): Likewise.
(print_msymbol_info): Use address style to print addresses.
Use function name style to print executable text symbols.
(expand_symtab_containing_pc): Use data_p.
(find_pc_sect_compunit_symtab): Likewise.
This changes symbol_set_names to take an objfile_per_bfd_storage
argument, and updates the users. It also changes PSYMBOL_SET_NAMES to
take this argument directly; I feel this clarifies the storage
location of objects created in psymtab.c.
gdb/ChangeLog
2019-01-10 Tom Tromey <tom@tromey.com>
* symtab.h (SYMBOL_SET_NAMES): Update.
(symbol_set_names): Update.
(MSYMBOL_SET_NAMES): Update.
* symtab.c (symbol_set_names): Change argument to be an
objfile_per_bfd_storage.
* psymtab.c (add_psymbol_to_bcache): Update.
* psympriv.h (PSYMBOL_SET_NAMES): Take per_bfd argument.
This removes ALL_COMPUNIT_FILETABS, replacing its uses with ranged for
loops.
Because this is still used in the ALL_OBJFILE_FILETABS macro, in some
places a declaration had to be removed or renamed to avoid shadowing.
gdb/ChangeLog
2019-01-09 Tom Tromey <tom@tromey.com>
* symtab.h (ALL_COMPUNIT_FILETABS): Remove.
(compunit_filetabs): New.
* symtab.c (iterate_over_some_symtabs, find_pc_sect_line): Use
compunit_filetabs.
(info_sources_command, make_source_files_completion_list): Remove
declaration.
* symmisc.c (print_objfile_statistics, dump_objfile)
(maintenance_print_symbols): Remove declaration.
(maintenance_info_symtabs): Use compunit_filetabs.
(maintenance_info_line_tables): Likewise.
* source.c (select_source_symtab): Change local variable name.
(forget_cached_source_info_for_objfile): Remove declaration.
* objfiles.h (ALL_OBJFILE_FILETABS): Use compunit_filetabs.
* objfiles.c (objfile_relocate1): Remove declaration.
* mi/mi-cmd-file.c (mi_cmd_file_list_exec_source_files): Remove
declaration.
* maint.c (count_symtabs_and_blocks): Use compunit_filetabs.
* coffread.c (coff_symtab_read): Remove declaration.
* buildsym.c (buildsym_compunit::end_symtab_with_blockvector): Use
compunit_filetabs.
Declarations for functions in source.c are split between source.h and
symtab.h. This commit moves the small number that are in symtab.h
into source.h. There's just one file that needs to add an include of
source.h in order to build.
I've moved the function header comments from source.c to source.h
inline with the recommended GDB style.
gdb/ChangeLog:
* source.c (select_source_symtab): Move header comment to
declaration in source.h.
(forget_cached_source_info_for_objfile): Likewise.
(forget_cached_source_info): Likewise.
(identify_source_line): Likewise.
* source.h (identify_source_line): Move declaration from symtab.h
and add comment from source.c
(print_source_lines): Likewise.
(forget_cached_source_info_for_objfile): Likewise.
(forget_cached_source_info): Likewise.
(select_source_symtab): Likewise.
(enum print_source_lines_flag): Move definition from symtab.h.
* symtab.h (identify_source_line): Move declaration to source.h.
(print_source_lines): Likewise.
(forget_cached_source_info_for_objfile): Likewise.
(forget_cached_source_info): Likewise.
(select_source_symtab): Likewise.
(enum print_source_lines_flag): Move definition to source.h.
* tui/tui-hooks.c: Add 'source.h' include.
This commit applies all changes made after running the gdb/copyright.py
script.
Note that one file was flagged by the script, due to an invalid
copyright header
(gdb/unittests/basic_string_view/element_access/char/empty.cc).
As the file was copied from GCC's libstdc++-v3 testsuite, this commit
leaves this file untouched for the time being; a patch to fix the header
was sent to gcc-patches first.
gdb/ChangeLog:
Update copyright year range in all GDB files.
Add [-q] [-t TYPEREGEXP] [NAMEREGEXP] args to info [args|functions|locals|variables]
Main changes are:
* stack.c: Add two regexp preg and treg to print_variable_and_value_data
and used them inside do_print_variable_and_value to filter the
variables to print.
* symtab.h: Add a new function bool treg_matches_sym_type_name, that
factorises type matching logic.
* symtab.c: Add type/name matching logic to 'info functions|variables'.
* stack.c : Add type/name matching logic to 'info args|locals'.
gdb/ChangeLog
2018-10-27 Philippe Waroquiers <philippe.waroquiers@skynet.be>
* stack.c (print_variable_and_value_data): Add preg and treg.
(print_frame_local_vars): Add quiet, regexp and t_regexp arguments,
and update callers.
(print_frame_arg_vars): Likewise.
(prepare_reg): New function.
(info_locals_command): Extract info print args and use them.
(info_args_command): Likewise.
(_initialize_stack): Modify on-line help.
* symtab.c (treg_matches_sym_type_name): New function.
(search_symbols): New arg t_regexp.
(symtab_symbol_info): New args quiet, regexp, t_regexp.
(info_variables_command): Extract info print args and use them.
(info_functions_command): Likewise.
(info_types_command): Update call to symtab_symbol_info.
(_initialize_symtab): Modify on-line help.
* symtab.h (treg_matches_sym_type_name): New function.
(search_symbols): New t_regexp arg.
This removes a couple of DEF_VECs from symtab.h, replacing them with
std::vector at the points of use.
gdb/ChangeLog
2018-10-26 Tom Tromey <tom@tromey.com>
* dwarf2read.c (recursively_compute_inclusions): Use std::vector.
(compute_compunit_symtab_includes): Update.
* symtab.h: (symtab_ptr): Remove typedef. Don't define a VEC.
(compunit_symtab_ptr): Likewise.
This patch adds a new symbol searching API based on linespec.c's parser
implementation. This allows users to find "all* matching symbols instead
of the first found match (a la lookup_symbol).
gdb/ChangeLog:
* linespec.c (collect_info::add_symbol): Make virtual.
(struct symbol_searcher_collect_info): New struct.
(symbol_searcher::find_all_symbols): New method.
* symtab.h (class symbol_searcher): New class.
This patch changes the linespec.c APIs to use block_symbol instead of just
a symbol. lookup_symbol et al already return block_symbol's.
gdb/ChangeLog:
* linespec.c (struct linespec) <function_symbols, label_symbols>:
Change to vector of block_symbol. Update all users.
(struct collect_info) <symbols>: Likewise.
(collect_info::add_symbol): Take block_symbol as argument.
Update all callers.
(decode_compound_collector) <m_symbols>: Change type to vector
of block_symbol. Update all users.
(decode_compound_collector::operator ()): Change parameter type
to block_symbol.
(find_method, find_function_symbols, find_linespec_symbols)
(find_label_symbols_in_block, find_label_symbols): Change symbol
vectors to block_symbol vectors.
* symtab.h (symbol_found_callback_ftype): Change parameter type to
block_symbol.
An earlier version of this patch used the returned block in conjunction
with BLOCK_ENTRY_PC to set stop_func_start in fill_in_stop_func() in
infrun.c. While I think this was the correct thing to do, changes
to find_inferior_partial_function could potentially end up with
stop_func_end < stop_func_start, which is definitely wrong. For
this case, we want to set both stop_func_start and stop_func_end
to the start and end of the range containing the function's entry
pc.
I think that this functionality will be useful in many other places
too - it probably ought to be used in all of the various prologue
analyzers in GDB.
The change to infrun.c was simple: the call to
find_pc_partial_function was replaced with a call to
find_function_entry_range_from_pc. The difference between these two
functions is that find_pc_partial_entry_function will (potentially)
return the start and end address corresponding to the range in which
PC is found, but find_function_entry_range_from_pc will (again,
potentially) return the start and end address of the range containing
the entry pc. find_pc_partial_function has the property that
*ADDRESS <= PC < *ENDADDR. This condition does not necessarily hold
for the outputs of find_function_entry_range_from_pc.
It should be noted that for functions which contain only a single
range, the outputs of find_pc_partial_function and
find_function_entry_range_from_pc are identical.
I think it might happen that find_function_entry_range_from_pc will come
to be used in place of many of the calls to find_pc_partial_function
within GDB. Care must be taken in making this change, however, since
some of this code depends on the *ADDRESS <= PC < *ENDADDR property.
Finally, a note regarding the name: I had initially chosen a different
name with a find_pc_partial_ prefix, but Simon suggested the current
name citing the goal of eventually making naming consistent using
the form find_X_from_Y. In this case X is "function_entry_range" and
Y is "pc". Both the name and rationale made sense to me, so that's
how it came to be.
gdb/ChangeLog:
* infrun.c (fill_in_stop_func): Use find_function_entry_range_from_pc
in place of find_pc_partial_function.
* blockframe.c (find_function_entry_range_from_pc): New function.
* symtab.h (find_function_entry_range_from_pc): Declare and document.
This change adds an optional output parameter BLOCK to
find_pc_partial_function. If BLOCK is non-null, then *BLOCK will be
set to the address of the block corresponding to the function symbol
if such a symbol was found during lookup. Otherwise it's set to the
NULL value. Callers may wish to use the block information to
determine whether the block contains any non-contiguous ranges. The
caller may also iterate over or examine those ranges.
When I first started looking at the broken stepping behavior associated
with functions w/ non-contiguous ranges, I found that I could "fix"
the problem by disabling the find_pc_partial_function cache. It would
sometimes happen that the PC passed in would be between the low and
high cache values, but would be in some other function that happens to
be placed in between the ranges for the cached function. This caused
incorrect values to be returned.
So dealing with this cache turns out to be very important for fixing
this problem. I explored three different ways of dealing with the
cache.
My first approach was to clear the cache when a block was encountered
with more than one range. This would cause the non-cache pathway to
be executed on the next call to find_pc_partial_function.
Another approach, which I suspect is slightly faster, checks to see
whether the PC is within one of the ranges associated with the cached
block. If so, then the cached values can be used. It falls back to
the original behavior if there is no cached block.
The current approach, suggested by Simon Marchi, is to restrict the
low/high pc values recorded for the cache to the beginning and end of
the range containing the PC value under consideration. This allows us
to retain the simple (and fast) test for determining whether the
memoized (cached) values apply to the PC passed to
find_pc_partial_function.
Another choice that had to be made regards setting *ADDRESS and
*ENDADDR. There are three possibilities which might make sense:
1) *ADDRESS and *ENDADDR represent the lowest and highest address
of the function.
2) *ADDRESS and *ENDADDR are set to the start and end address of
the range containing the entry pc.
3) *ADDRESS and *ENDADDR are set to the start and end address of
the range in which PC is found.
An earlier version of this patch implemented option #1. I found out
that it's not very useful though and, in fact, returns results that
are incorrect when used in the context of determining the start and
end of the function for doing prologue analysis. While debugging a
function in which the entry pc was in the second range (of a function
containing two non-contiguous ranges), I noticed that
amd64_skip_prologue called find_pc_partial_function - the returned
start address was set to the beginning of the first range. This is
incorrect for this function. What was also interesting was that this
first invocation of find_pc_partial_function correctly set the cache
for the PC on which it had been invoked, but a slightly later call
from skip_prologue_using_sal could not use this cached value because
it was now being used to lookup the very lowest address of the
function - which is in a range not containing the entry pc.
Option #2 is attractive as it would provide a desirable result
when used in the context of prologue analysis. However, many callers,
including some which do prologue analysis want the condition
*ADDRESS <= PC < *ENDADDR to hold. This will not be the case when
find_pc_partial_function is called on a PC that's in a non-entry-pc
range. A later patch to this series adds
find_function_entry_range_from_pc as a wrapper of
find_pc_partial_function.
Option #3 causes the *ADDRESS <= PC < *ENDADDR property to hold. If
find_pc_partial_function is called with a PC that's within entry pc's
range, then it will correctly return the limits of that range. So, if
the result of a minsym search is passed to find_pc_partial_function
to find the limits, then correct results will be achieved. Returned
limits (for prologue analysis) won't be correct when PC is within some
other (non-entry-pc) range. I don't yet know how big of a problem
this might be; I'm guessing that it won't be a serious problem - if a
compiler generates functions which have non-contiguous ranges, then it
also probably generates DWARF2 CFI which makes a lot of the old
prologue analysis moot.
I've implemented option #3 for this version of the patch. I don't see
any regressions for x86-64. Moreover, I don't expect to see
regressions for other targets either simply because
find_pc_partial_function behaves the same as it did before for the
contiguous address range case. That said, there may be some
adjustments needed if GDB encounters a function requiring prologue
analysis which occupies non-contiguous ranges.
gdb/ChangeLog:
* symtab.h (find_pc_partial_function): Add new parameter `block'.
* blockframe.c (cache_pc_function_block): New static global.
(clear_pc_function_cache): Clear cache_pc_function_block.
(find_pc_partial_function): Move comment to symtab.h. Add
support for non-contiguous blocks.
While experimenting with the previous patch, I noticed this inconsistency
in GDB's output:
(gdb) b 32
Breakpoint 1 at 0x40062f: file inline-break.c, line 32. (1)
(gdb) r
....
Breakpoint 1, func1 (x=1) at inline-break.c:32 (2)
32 return x * 23; /* break here */
(gdb) info breakpoints
Num Type Disp Enb Address What
1 breakpoint keep y 0x40062f in main at inline-break.c:32 (3)
breakpoint already hit 1 time
(gdb)
Notice that when the breakpoint as set, GDB showed "inline-break.c,
line 32" (1), the same line number that was specified in the command.
When we run to the breakpoint, we present the stop at the same line
number, and correctly show "func1" as the function name (2).
But in "info break" output (3), notice that we say "in main", not "in
func1".
The same thing happens if you set a breakpoint by address. I.e.:
(gdb) b *0x40062f
Breakpoint 2 at 0x40062f: file inline-break.c, line 32.
(gdb) info breakpoints
Num Type Disp Enb Address What
2 breakpoint keep y 0x000000000040062f in main at inline-break.c:32
(gdb) r
....
Breakpoint 2, func1 (x=1) at inline-break.c:32
32 return x * 23; /* break here */
The problem is that the breakpoints were set at an inline function,
but when we set such a breakpoint by line number or address, we don't
record the functions symbol in the sal, and as consequence the
breakpoint location does not have an associated symbol either.
Then, in print_breakpoint_location, if the location does not have a
symbol, we call find_pc_sect_function to find one, and this is what
finds "main", because find_pc_sect_function uses
block_linkage_function:
/* Return the symbol for the function which contains a specified
lexical block, described by a struct block BL. The return value
will not be an inlined function; the containing function will be
returned instead. */
struct symbol *
block_linkage_function (const struct block *bl)
To fix this, this commit adds an alternative to find_pc_sect_function
that uses block_containing_function instead:
/* Return the symbol for the function which contains a specified
block, described by a struct block BL. The return value will be
the closest enclosing function, which might be an inline
function. */
struct symbol *
block_containing_function (const struct block *bl)
(It seems odd to me that block_linkage_function says "the CONTAINING
function will be returned", and then block_containing_function says it
returns "the closest enclosing function". Something seems reversed
here. Still, I've kept the same nomenclature and copied the comments,
so that at least there's consistency. Maybe we should fix that up
somehow.)
Then I wondered, why make print_breakpoint_location look up the symbol
every time it is called, instead of just always storing the symbol
when the location is created, since the location already stores the
symbol in some cases. So to find which cases might be missing setting
the symbol in the sal which is used to create the breakpoint location,
I added an assertion to print_breakpoint_location, and ran the
testsuite. That caught a few places, unsurprisingly:
- setting a breakpoint by line number
- setting a breapoint by address
- ifunc resolving
Those are all fixed by this commit. I decided not to add the
assertion to block_linkage_function and leave the existing "if (sym)"
check in place, because it's plausible that we have symtabs with line
info but no symbols. I.e., that would not be a GDB bug, but
a peculiarity of debug info input.
gdb/ChangeLog:
2018-06-29 Pedro Alves <palves@redhat.com>
* blockframe.c (find_pc_sect_containing_function): New function.
* breakpoint.c (print_breakpoint_location): Don't call
find_pc_sect_function.
* linespec.c (create_sals_line_offset): Record the location's
symbol in the sal.
* linespec.c (convert_address_location_to_sals): Fill in sal's
symbol with find_pc_sect_containing_function.
* symtab.c (find_function_start_sal): Rename to ...
(find_function_start_sal_1): ... this.
(find_function_start_sal): Reimplement as wrapper around
find_function_start_sal_1, and use
find_pc_sect_containing_function to fill in the sal's symbol.
(find_function_start_sal(symbol*, bool)): Adjust.
* symtab.h (find_pc_function, find_pc_sect_function): Adjust
comments.
(find_pc_sect_containing_function): Declare.
gdb/testsuite/ChangeLog:
2018-06-29 Pedro Alves <palves@redhat.com>
* gdb.opt/inline-break.exp (line number, address): Add "info
break" tests.
Running the new tests added later in the series on PPC64 (ELFv1)
revealed that the current ifunc support needs a bit of a design rework
to work properly on PPC64/ELFv1, as most of the new tests fail. The
ifunc support only kind of works today if the ifunc symbol and the
resolver have the same name, as is currently tested by the
gdb.base/gnu-ifunc.exp testcase, which is unlike how ifuncs are
written nowadays.
The crux of the problem is that ifunc symbols are really function
descriptors, not text symbols:
44: 0000000000020060 104 FUNC GLOBAL DEFAULT 18 gnu_ifunc_resolver
54: 0000000000020060 104 GNU_IFUNC GLOBAL DEFAULT 18 gnu_ifunc
But, currently GDB only knows about ifunc symbols that are text
symbols. GDB's support happens to work in practice for PPC64 when the
ifunc and resolver are one and only, like in the current
gdb.base/gnu-ifunc.exp testcase:
15: 0000000000020060 104 GNU_IFUNC GLOBAL DEFAULT 18 gnu_ifunc
because in that case, the synthetic ".gnu_ifunc" entry point text
symbol that bfd creates from the actual GNU ifunc "gnu_ifunc" function
(descriptor) symbol ends up with the the "is a gnu ifunc" flag set /
copied over:
(gdb) maint print msymbols
...
[ 8] i 0x9c4 .gnu_ifunc section .text <<< mst_text_gnu_ifunc
...
[29] D 0x20060 gnu_ifunc section .opd crtstuff.c <<< mst_data
But, if the resolver gets a distinct symbol/name from the ifunc
symbol, then we end up with this:
(gdb) maint print msymbols
[ 8] T 0x9e4 .gnu_ifunc_resolver section .text <<< mst_text
...
[29] D 0x20060 gnu_ifunc section .opd crtstuff.c <<< mst_data
[30] D 0x20060 gnu_ifunc_resolver section .opd crtstuff.c <<< mst_data
I have a follow up bfd patch that turns that into:
(gdb) maint print msymbols
+ [ 8] i 0x9e4 .gnu_ifunc section .text <<< mst_text_gnu_ifunc
[ 8] T 0x9e4 .gnu_ifunc_resolver section .text <<< mst_text
...
[29] D 0x20060 gnu_ifunc section .opd crtstuff.c
[30] D 0x20060 gnu_ifunc_resolver section .opd crtstuff.c
but that won't help everything. We still need this patch.
Specifically, when we do a symbol lookup by name, like e.g., to call a
function (see c-exp.y hunk), e.g., "p gnu_ifunc()", then we need to
know that the found "gnu_ifunc" minimal symbol is an ifunc in order to
do some special processing. But, on PPC, that lookup by name finds
the function descriptor symbol, which presently is just a mst_data
symbol, while at present, we look for mst_text_gnu_ifunc symbols to
decide whether to do special GNU ifunc processing. In most of those
places, we could try to resolve the function descriptor with
gdbarch_convert_from_func_ptr_addr, and then lookup the minimal symbol
at the resolved PC, see if that finds a minimal symbol of type
mst_text_gnu_ifunc. If so, then we could assume that the original
mst_dadta / function descriptor "gnu_ifunc" symbol was an ifunc. I
tried it, and it mostly works, even if it's not the most efficient.
However, there's one case that can't work with such a design -- it's
that of the user calling the ifunc resolver directly to debug it, like
"p gnu_ifunc_resolver(0)", expecting that to return the function
pointer of the final function (which is exercised by the new tests
added later). In this case, with the not-fully-working solution, we'd
resolve the function descriptor, find that there's an
mst_text_gnu_ifunc symbol for the resolved address, and proceed
calling the function as if we tried to call "gnu_ifunc", the
user-visible GNU ifunc symbol, instead of the resolver. I.e., it'd be
impossible to call the resolver directly as a normal function.
Introducing mst_data_gnu_ifunc eliminates the need for several
gdbarch_convert_from_func_ptr_addr calls, and, fixes the "call
resolver directly" use case mentioned above too. It's the cleanest
approach I could think of.
In sum, we make GNU ifunc function descriptor symbols get a new
"mst_data_gnu_ifunc" minimal symbol type instead of the bare mst_data
type. So when symbol lookup by name finds such a minimal symbol, we
know we found an ifunc symbol, without resolving the entry/text
symbol. If the user calls the the resolver symbol instead, like "p
gnu_ifunc_resolver(0)", then we'll find the regular mst_data symbol
for "gnu_ifunc_resolver", and we'll call the resolver function as just
another regular function.
With this, most of the GNU ifunc tests added by a later patch pass on
PPC64 too. The following bfd patch fixes the remaining issues.
gdb/ChangeLog:
2018-04-26 Pedro Alves <palves@redhat.com>
* breakpoint.c (set_breakpoint_location_function): Handle
mst_data_gnu_ifunc.
* c-exp.y (variable production): Handle mst_data_gnu_ifunc.
* elfread.c (elf_symtab_read): Give data symbols with
BSF_GNU_INDIRECT_FUNCTION set mst_data_gnu_ifunc type.
(elf_rel_plt_read): Update comment.
* linespec.c (convert_linespec_to_sals): Handle
mst_data_gnu_ifunc.
(minsym_found): Handle mst_data_gnu_ifunc.
* minsyms.c (msymbol_is_function, minimal_symbol_reader::record)
(find_solib_trampoline_target): Handle mst_data_gnu_ifunc.
* parse.c (find_minsym_type_and_address): Handle
mst_data_gnu_ifunc.
* symmisc.c (dump_msymbols): Handle mst_data_gnu_ifunc.
* symtab.c (find_gnu_ifunc): Handle mst_data_gnu_ifunc.
* symtab.h (minimal_symbol_type) <mst_text_gnu_ifunc>: Update
comment.
<mst_data_gnu_ifunc>: New enumerator.
I need to make the ifunc resolving code in elfread.c skip the target
function's prologue like minsym_found does. I thought of factoring
that out to a separate function, but turns out there's already a
comment in find_function_start_sal that says that should agree with
minsym_found...
Instead of making sure the code agrees with a comment, factor out the
common code to a separate function and use it from both places.
Note that the current find_function_start_sal does a bit more than
minsym_found's equivalent (the "We always should ..." bit), though
that's probably a latent bug.
gdb/ChangeLog:
2018-04-26 Pedro Alves <palves@redhat.com>
* linespec.c (minsym_found): Use find_function_start_sal CORE_ADDR
overload.
* symtab.c (find_function_start_sal(CORE_ADDR, obj_section *,bool)):
New, factored out from ...
(find_function_start_sal(symbol *, int)): ... this. Reimplement
and use bool.
* symtab.h (find_function_start_sal(CORE_ADDR, obj_section *,bool)):
New.
(find_function_start_sal(symbol *, int)): Change boolean parameter
type to bool.
Not used anywhere any longer.
If this is ever reinstated, note that this case:
cache_pc_function_is_gnu_ifunc = TYPE_GNU_IFUNC (SYMBOL_TYPE (f));
was incorrect in that regular symbols never have type marked as GNU
ifunc type, only minimal symbols. At some point I had some fix that
checking the matching minsym here. But in the end I ended up just
eliminating need for this function, so that fix was not necessary.
gdb/ChangeLog:
2018-04-26 Pedro Alves <palves@redhat.com>
* blockframe.c (cache_pc_function_is_gnu_ifunc): Delete. Remove
all references.
(find_pc_partial_function_gnu_ifunc): Rename to ...
(find_pc_partial_function): ... this, and remove references to
'is_gnu_ifunc_p'.
(find_pc_partial_function): Delete old implementation.
* symtab.h (find_pc_partial_function_gnu_ifunc): Delete.
This fixes setting breakpoints on ifunc functions by name after the
ifunc has already been resolved.
In that case, if you have debug info for the ifunc resolver, without
the fix, then gdb puts a breakpoint past the prologue of the resolver,
instead of setting a breakpoint at the ifunc target:
break gnu_ifunc
Breakpoint 4 at 0x7ffff7bd36f2: file src/gdb/testsuite/gdb.base/gnu-ifunc-lib.c, line 34.
(gdb) continue
Continuing.
[Inferior 1 (process 13300) exited normally]
(gdb)
above we should have stopped at "final", but didn't because we never
resolved the ifunc to the final location.
If you don't have debug info for the resolver, GDB manages to resolve
the ifunc target, but, it should be setting a breakpoint after the
prologue of the final function, and instead what you get is that GDB
sets a breakpoint on the first address of the target function. With
the gnu-ifunc.exp tests added by a later patch, we get, without the
fix:
(gdb) break gnu_ifunc
Breakpoint 4 at 0x400753
(gdb) continue
Continuing.
Breakpoint 4, final (arg=1) at src/gdb/testsuite/gdb.base/gnu-ifunc-final.c:20
20 {
vs, fixed:
(gdb) break gnu_ifunc
Breakpoint 4 at 0x40075a: file src/gdb/testsuite/gdb.base/gnu-ifunc-final.c, line 21.
(gdb) continue
Continuing.
Breakpoint 4, final (arg=2) at src/gdb/testsuite/gdb.base/gnu-ifunc-final.c:21
21 return arg + 1;
(gdb)
Fix the problems above by moving the ifunc target resolving to
linespec.c, before we skip a function's prologue. We need to save
something in the sal, so that set_breakpoint_location_function knows
that it needs to create a bp_gnu_ifunc_resolver bp_location. Might as
well just save a pointer to the minsym.
gdb/ChangeLog:
2018-04-26 Pedro Alves <palves@redhat.com>
* breakpoint.c (set_breakpoint_location_function): Don't resolve
ifunc targets here. Instead, if we have an ifunc minsym, use its
address/name.
(add_location_to_breakpoint): Store the minsym and the objfile in
the breakpoint location.
* breakpoint.h (bp_location) <msymbol, objfile>: New fields.
* linespec.c (minsym_found): Resolve GNU ifunc targets here.
Record the minsym in the sal.
* symtab.h (symtab_and_line) <msymbol>: New field.
If the GNU ifunc resolver has the same name as the user visible
symbol, and the resolver has debug info, then the DWARF info for the
resolver masks the ifunc minsym. In that scenario, if you try calling
the ifunc from GDB, you call the resolver instead. With the
gnu-ifunc.exp testcase added in a following patch, you'd see:
(gdb) p gnu_ifunc (3)
$1 = (int (*)(int)) 0x400753 <final>
(gdb) FAIL: gdb.base/gnu-ifunc.exp: resolver_attr=0: resolver_debug=1: resolved_debug=0: p gnu_ifunc (3)
^^^^^^^^^^^^^^^^
That is, we called the ifunc resolver manually, which returned a
pointer to the ifunc target function ("final"). The "final" symbol is
the function that GDB should have called automatically,
~~~~~~~~~~~~
int
final (int arg)
{
return arg + 1;
}
~~~~~~~~~
which is what happens if you don't have debug info for the resolver:
(gdb) p gnu_ifunc (3)
$1 = 4
(gdb) PASS: gdb.base/gnu-ifunc.exp: resolver_attr=0: resolver_debug=0: resolved_debug=1: p gnu_ifunc (3)
^^^^^^^^^^^^^^^^
or if the resolver's symbol has a different name from the ifunc (as is
the case with modern uses of ifunc via __attribute__ ifunc, such as
glibc uses):
(gdb) p gnu_ifunc (3)
$1 = 4
(gdb) PASS: gdb.base/gnu-ifunc.exp: resolver_attr=1: resolver_debug=1: resolved_debug=0: p gnu_ifunc (3)
^^^^^^^^^^^^^^^
in which case after this patch, you can still call the resolver
directly if you want:
(gdb) p gnu_ifunc_resolver (3)
$1 = (int (*)(int)) 0x400753 <final>
gdb/ChangeLog:
2018-04-26 Pedro Alves <palves@redhat.com>
* c-exp.y (variable production): Prefer ifunc minsyms over
regular function symbols.
* symtab.c (find_gnu_ifunc): New function.
* minsyms.h (lookup_msym_prefer): New enum.
(lookup_minimal_symbol_by_pc_section): Replace 'want_trampoline'
parameter by a lookup_msym_prefer parameter.
* symtab.h (find_gnu_ifunc): New declaration.
After the previous patch, on Fedora 27 (glibc 2.26), if you try
calling strlen in the inferior, you now get:
(top-gdb) p strlen ("hello")
'__strlen_avx2' has unknown return type; cast the call to its declared return type
This is correct, because __strlen_avx2 is written in assembly.
We can improve on this though -- if the final ifunc resolved/target
function has no debug info, but the ifunc _resolver_ does have debug
info, we can try extracting the final function's type from the type
that the resolver returns. E.g.,:
typedef size_t (*strlen_t) (const char*);
size_t my_strlen (const char *) { /* some implementation */ }
strlen_t strlen_resolver (unsigned long hwcap) { return my_strlen; }
extern size_t strlen (const char *s);
__typeof (strlen) strlen __attribute__ ((ifunc ("strlen_resolver")));
In the strlen example above, the resolver returns strlen_t, which is a
typedef for pointer to a function that returns size_t. "strlen_t" is
the type of both the user-visible "strlen", and of the the target
function that implements it.
This patch teaches GDB to extract that type.
This is done for actual inferior function calls (in infcall.c), and
for ptype (in eval_call). By the time we get to either of these
places, we've already lost the original symbol/minsym, and only have
values and types to work with. Hence the changes to c-exp.y and
evaluate_var_msym_value, to ensure that we propagate the ifunc
minsymbol's info.
The change to make ifunc symbols have no/unknown return type exposes a
latent problem -- gdb.compile/compile-ifunc.exp calls a no-debug-info
function, but we did not warn about it. The test is fixed by this
commit too.
gdb/ChangeLog:
2018-04-26 Pedro Alves <palves@redhat.com>
* blockframe.c (find_gnu_ifunc_target_type): New function.
(find_function_type): New.
* eval.c (evaluate_var_msym_value): For GNU ifunc types, always
return a value with a memory address.
(eval_call): For calls to GNU ifunc functions, try to find the
type of the target function from the type that the resolver
returns.
* gdbtypes.c (objfile_type): Don't install a return type for ifunc
symbols.
* infcall.c (find_function_return_type): Delete.
(find_function_addr): Add 'function_type' parameter. For calls to
GNU ifunc functions, try to find the type of the target function
from the type that the resolver returns, and return it via
FUNCTION_TYPE.
(call_function_by_hand_dummy): Adjust to use the function type
returned by find_function_addr.
(find_function_addr): Add 'function_type' parameter and move
description here.
* symtab.h (find_function_type, find_gnu_ifunc_target_type): New
declarations.
gdb/testsuite/ChangeLog:
2018-04-26 Pedro Alves <palves@redhat.com>
* gdb.compile/compile-ifunc.exp: Also expect "function has unknown
return type" warnings.
A previous patch fixed verbatim matching in the lookup at the minimal
symbol level, but we should also be finding that same symbol through
the partial/full symtab search.
For example, this is what happens if we use "print" instead of
"break":
(gdb) p <MixedCaseFunc>
$1 = {<text variable, no debug info>} 0x4024dc <MixedCaseFunc>
Before the C++ wildmatching series, GDB knows that MixedCaseFunc is a
function without parameters, and the expression above means calling
it. If you try it before having started the inferior, you'd get the
following (expected) error:
(gdb) print <MixedCaseFunc>
You can't do that without a process to debug.
The main idea behind making the name matcher be determined by the
symbol's language is so that C++ (etc.) wildmatching in linespecs
works even if the current language is not C++, as e.g., when you step
through C or assembly code.
Ada's verbatim matching syntax however ("<...>") isn't quite the same.
It is more a property of the current language than of a particular
symbol's language. We want to support this syntax when debugging an
Ada program, but it's reason of existence is to find non-Ada symbols.
This suggests going back to enabling it depending on current language
instead of language of the symbol being matched.
I'm not entirely happy with the "current_language" reference (though I
think that it's harmless). I think we could try storing the current
language in the lookup_name_info object, and then convert a bunch of
functions more to pass around lookup_name_info objects instead of
"const char *" names. I.e., build the lookup_name_info higher up.
I'm not sure about that, I'll have to think more about it. Maybe
something different will be better. Meanwhile, this gets us going.
I've extended the testcase to also exercise a no-debug-info function,
for extra coverage of the minsyms-only paths.
gdb/ChangeLog:
2018-01-10 Pedro Alves <palves@redhat.com>
PR gdb/22670
* dwarf2read.c
(gdb_index_symbol_name_matcher::gdb_index_symbol_name_matcher):
Adjust to use language_get_symbol_name_matcher instead of
language_defn::la_get_symbol_name_matcher.
* language.c (language_get_symbol_name_matcher): If in Ada mode
and the lookup name is a verbatim match, return Ada's matcher.
* language.h (language_get_symbol_name_matcher): Adjust comment.
(ada_lookup_name_info::verbatim_p):: New method.
gdb/testsuite/ChangeLog:
2018-01-10 Pedro Alves <palves@redhat.com>
PR gdb/22670
* gdb.ada/bp_c_mixed_case.exp: Add intro comment. Test printing C
functions too. Test setting breakpoints and printing C functions
with no debug info too.
* gdb.ada/bp_c_mixed_case/qux.c: New file.
At <https://sourceware.org/ml/gdb-patches/2017-12/msg00298.html>, Joel
wrote:
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Consider the following code which first declares a tagged type (the
equivalent of a class in Ada), and then a procedure which takes a
pointer (access) to this type's 'Class.
package Pck is
type Top_T is tagged record
N : Integer := 1;
end record;
procedure Inspect (Obj: access Top_T'Class);
end Pck;
Putting a breakpoint in that procedure and then running to it triggers
an internal error:
(gdb) break inspect
(gdb) continue
Breakpoint 1, pck.inspect (obj=0x63e010
/[...]/gdb/stack.c:621: internal-error: void print_frame_args(symbol*, frame_info*, int, ui_file*): Assertion `nsym != NULL' failed.
What's special about this subprogram is that it takes an access to
what we call a 'Class type, and for implementation reasons, the
compiler adds an extra argument named "objL". If you are curious why,
it allows the compiler for perform dynamic accessibility checks that
are mandated by the language.
If we look at the location where we get the internal error (in
stack.c), we find that we are looping over the symbol of each
parameter, and for each parameter, we do:
/* We have to look up the symbol because arguments can have
two entries (one a parameter, one a local) and the one we
want is the local, which lookup_symbol will find for us.
[...]
nsym = lookup_symbol (SYMBOL_LINKAGE_NAME (sym),
b, VAR_DOMAIN, NULL).symbol;
gdb_assert (nsym != NULL);
The lookup_symbol goes through the lookup structure, which means the
symbol's linkage name ("objL") gets transformed into a
lookup_name_info object (in block_lookup_symbol), before it gets fed
to the block symbol dictionary iterators. This, in turn, triggers the
symbol matching by comparing the "lookup" name which, for Ada, means
among other things, lowercasing the given name to "objl". It is this
transformation that causes the lookup find no matches, and therefore
trip this assertion.
Going back to the "offending" call to lookup_symbol in stack.c, what
we are trying to do, here, is do a lookup by linkage name. So, I
think what we mean to be doing is a completely literal symbol lookup,
so maybe not even strcmp_iw, but actually just plain strcmp???
In the past, in practice, you could get that effect by doing a lookup
using the C language. But that doesn't work, because we still end up
somehow using Ada's lookup_name routine which transforms "objL".
So, ideally, as I hinted before, I think what we need is a way to
perform a literal lookup so that searches by linkage names like the
above can be performed.
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
This commit fixes the problem by implementing something similar to
Joel's literal idea, but with some important differences.
I considered adding a symbol_name_match_type::LINKAGE and supporting
searching by linkage name for any language, but the problem with that
is that the dictionaries only work with SYMBOL_SEARCH_NAME, because
that's what is used for hashing. We'd need separate dictionaries for
hashed linkage names.
So with the current symbol tables infrastructure, it's not literal
linkage names that we want to pass down, but instead literal _search_
names (SYMBOL_SEARCH_NAME, etc.).
However, psymbols have no overload/function parameter info in C++, so
a straight strcmp doesn't work properly for C++ name matching.
So what we do is be a little less aggressive then and add a new
symbol_name_match_type::SEARCH_SYMBOL instead that takes as input a
non-user-input search symbol, and then we skip any decoding/demangling
steps and make:
- Ada treat that as a verbatim match,
- other languages treat it as symbol_name_match_type::FULL.
This also fixes the new '"maint check-psymtabs" for Ada' testcase for
me (gdb.ada/maint_with_ada.exp). I've not removed the kfail yet
because Joel still sees that testcase failing with this patch.
That'll be fixed in follow up patches.
gdb/ChangeLog:
2018-01-05 Pedro Alves <palves@redhat.com>
PR gdb/22670
* ada-lang.c (literal_symbol_name_matcher): New function.
(ada_get_symbol_name_matcher): Use it for
symbol_name_match_type::SEARCH_NAME.
* block.c (block_lookup_symbol): New parameter 'match_type'. Pass
it down instead of assuming symbol_name_match_type::FULL.
* block.h (block_lookup_symbol): New parameter 'match_type'.
* c-valprint.c (print_unpacked_pointer): Use
lookup_symbol_search_name instead of lookup_symbol.
* compile/compile-object-load.c (get_out_value_type): Pass down
symbol_name_match_type::SEARCH_NAME.
* cp-namespace.c (cp_basic_lookup_symbol): Pass down
symbol_name_match_type::FULL.
* cp-support.c (cp_get_symbol_name_matcher): Handle
symbol_name_match_type::SEARCH_NAME.
* infrun.c (insert_exception_resume_breakpoint): Use
lookup_symbol_search_name.
* p-valprint.c (pascal_val_print): Use lookup_symbol_search_name.
* psymtab.c (maintenance_check_psymtabs): Use
symbol_name_match_type::SEARCH_NAME and SYMBOL_SEARCH_NAME.
* stack.c (print_frame_args): Use lookup_symbol_search_name and
SYMBOL_SEARCH_NAME.
* symtab.c (lookup_local_symbol): Don't demangle the lookup name
if symbol_name_match_type::SEARCH_NAME.
(lookup_symbol_in_language): Pass down
symbol_name_match_type::FULL.
(lookup_symbol_search_name): New.
(lookup_language_this): Pass down
symbol_name_match_type::SEARCH_NAME.
(lookup_symbol_aux, lookup_local_symbol): New parameter
'match_type'. Pass it down.
* symtab.h (symbol_name_match_type::SEARCH_NAME): New enumerator.
(lookup_symbol_search_name): New declaration.
(lookup_symbol_in_block): New 'match_type' parameter.
gdb/testsuite/ChangeLog:
2018-01-05 Joel Brobecker <brobecker@adacore.com>
PR gdb/22670
* gdb.ada/access_tagged_param.exp: New file.
* gdb.ada/access_tagged_param/foo.adb: New file.
A following patch will add support for wild matching for C++ symbols,
making completing on "b push_ba" on a C++ program complete to
std::vector<...>::push_back, std::string::push_back etc., like:
(gdb) b push_ba[TAB]
std::vector<...>::push_back(....)
std::string<...>::push_back(....)
Currently, we compute the "lowest common denominator" between all
completion candidates (what the input line is adjusted to) as the
common prefix of all matches. That's problematic with wild matching
as above, as then we'd end up with TAB changing the input line to
"b std::", losing the original input, like:
(gdb) b push_ba[TAB]
std::vector<...>::push_back(....)
std::string<...>::push_back(....)
(gdb) b std::
while obviously we'd want it to adjust itself to "b push_back(" instead:
(gdb) b push_ba[TAB]
std::vector<...>::push_back(....)
std::string<...>::push_back(....)
(gdb) b push_back(
This patch adds the core code necessary to support this, though
nothing really makes use of it yet in this patch.
gdb/ChangeLog:
2017-11-29 Pedro Alves <palves@redhat.com>
* ada-lang.c (ada_lookup_name_info::matches): Change type of
parameter from completion_match to completion_match_result.
Adjust.
(do_wild_match, do_full_match, ada_symbol_name_matches): Likewise.
* completer.c (completion_tracker::maybe_add_completion): Add
match_for_lcd parameter and use it.
(completion_tracker::add_completion): Likewise.
* completer.h (class completion_match_for_lcd): New class.
(completion_match_result::match_for_lcd): New field.
(completion_match_result::set_match): New method.
(completion_tracker): Add comments.
(completion_tracker::add_completion): Add match_for_lcd parameter.
(completion_tracker::reset_completion_match_result): Reset
match_for_lcd too.
(completion_tracker::maybe_add_completion): Add match_for_lcd
parameter.
(completion_tracker::m_lowest_common_denominator_unique): Extend
comments.
* cp-support.c (cp_symbol_name_matches_1)
(cp_fq_symbol_name_matches): Change type of parameter from
completion_match to completion_match_result. Adjust.
* language.c (default_symbol_name_matcher): Change type of
parameter from completion_match to completion_match_result.
Adjust.
* language.h (completion_match_for_lcd): Forward declare.
(default_symbol_name_matcher): Change type of parameter from
completion_match to completion_match_result.
* symtab.c (compare_symbol_name): Adjust.
(completion_list_add_name): Pass the match_for_lcd to the tracker.
* symtab.h (ada_lookup_name_info::matches): Change type of
parameter from completion_match to completion_match_result.
(symbol_name_matcher_ftype): Likewise, and update comments.
This patch converts the generic probe interface (gdb/probe.[ch]) to
C++, and also performs some cleanups that were on my TODO list for a
while.
The main changes were the conversion of 'struct probe' to 'class
probe', and 'struct probe_ops' to 'class static_probe_ops'. The
former now contains all the "dynamic", generic methods that act on a
probe + the generic data related to it; the latter encapsulates a
bunch of "static" methods that relate to the probe type, but not to a
specific probe itself.
I've had to do a few renamings (e.g., on 'struct bound_probe' the
field is called 'probe *prob' now, instead of 'struct probe *probe')
because GCC was complaining about naming the field using the same name
as the class. Nothing major, though. Generally speaking, the logic
behind and the design behind the code are the same.
Even though I'm sending a series of patches, they need to be tested
and committed as a single unit, because of inter-dependencies. But it
should be easier to review in separate logical units.
I've regtested this patch on BuildBot, no regressions found.
gdb/ChangeLog:
2017-11-22 Sergio Durigan Junior <sergiodj@redhat.com>
* break-catch-throw.c (fetch_probe_arguments): Use
'probe.prob' instead of 'probe.probe'.
* breakpoint.c (create_longjmp_master_breakpoint): Call
'can_evaluate_arguments' and 'get_relocated_address' methods
from probe.
(create_exception_master_breakpoint): Likewise.
(add_location_to_breakpoint): Use 'sal->prob' instead of
'sal->probe'.
(bkpt_probe_insert_location): Call 'set_semaphore' method from
probe.
(bkpt_probe_remove_location): Likewise, for 'clear_semaphore'.
* elfread.c (elf_get_probes): Use 'static_probe_ops' instead
of 'probe_ops'.
(probe_key_free): Call 'delete' on probe.
(check_exception_resume): Use 'probe.prob' instead of
'probe.probe'.
* location.c (string_to_event_location_basic): Call
'probe_linespec_to_static_ops'.
* probe.c (class any_static_probe_ops): New class.
(any_static_probe_ops any_static_probe_ops): New variable.
(parse_probes_in_pspace): Receive 'static_probe_ops' as
argument. Adjust code to reflect change.
(parse_probes): Use 'static_probe_ops' instead of
'probe_ops'. Adjust code to reflect change.
(find_probes_in_objfile): Call methods to get name and
provider from probe.
(find_probe_by_pc): Use 'result.prob' instead of
'result.probe'. Call 'get_relocated_address' method from
probe.
(collect_probes): Adjust comment and argument list to receive
'static_probe_ops' instead of 'probe_ops'. Adjust code to
reflect change. Call necessary methods from probe.
(compare_probes): Call methods to get name and provider from
probes.
(gen_ui_out_table_header_info): Receive 'static_probe_ops'
instead of 'probe_ops'. Use 'std::vector' instead of VEC,
adjust code accordingly.
(print_ui_out_not_applicables): Likewise.
(info_probes_for_ops): Rename to...
(info_probes_for_spops): ...this. Receive 'static_probe_ops'
as argument instead of 'probe_ops'. Adjust code. Call
necessary methods from probe.
(info_probes_command): Use 'info_probes_for_spops'.
(enable_probes_command): Pass correct argument to
'collect_probes'. Call methods from probe.
(disable_probes_command): Likewise.
(get_probe_address): Move to 'any_static_probe_ops::get_address'.
(get_probe_argument_count): Move to
'any_static_probe_ops::get_argument_count'.
(can_evaluate_probe_arguments): Move to
'any_static_probe_ops::can_evaluate_arguments'.
(evaluate_probe_argument): Move to
'any_static_probe_ops::evaluate_argument'.
(probe_safe_evaluate_at_pc): Use 'probe.prob' instead of
'probe.probe'.
(probe_linespec_to_ops): Rename to...
(probe_linespec_to_static_ops): ...this. Adjust code.
(probe_any_is_linespec): Rename to...
(any_static_probe_ops::is_linespec): ...this.
(probe_any_get_probes): Rename to...
(any_static_probe_ops::get_probes): ...this.
(any_static_probe_ops::type_name): New method.
(any_static_probe_ops::gen_info_probes_table_header): New
method.
(compute_probe_arg): Use 'pc_probe.prob' instead of
'pc_probe.probe'. Call methods from probe.
(compile_probe_arg): Likewise.
(std::vector<const probe_ops *> all_probe_ops): Delete.
(std::vector<const static_probe_ops *> all_static_probe_ops):
New variable.
(_initialize_probe): Use 'all_static_probe_ops' instead of
'all_probe_ops'.
* probe.h (struct info_probe_column) <field_name>: Delete
extraneous newline
(info_probe_column_s): Delete type and VEC.
(struct probe_ops): Delete. Replace with...
(class static_probe_ops): ...this and...
(clas probe): ...this.
(struct bound_probe) <bound_probe>: Delete extraneous
newline. Adjust constructor to receive 'probe' instead of
'struct probe'.
<probe>: Rename to...
<prob>: ...this. Delete extraneous newline.
<objfile>: Delete extraneous newline.
(register_probe_ops): Delete unused prototype.
(info_probes_for_ops): Rename to...
(info_probes_for_spops): ...this. Adjust comment.
(get_probe_address): Move to 'probe::get_address'.
(get_probe_argument_count): Move to
'probe::get_argument_count'.
(can_evaluate_probe_arguments): Move to
'probe::can_evaluate_arguments'.
(evaluate_probe_argument): Move to 'probe::evaluate_argument'.
* solib-svr4.c (struct svr4_info): Adjust comment.
(struct probe_and_action) <probe>: Rename to...
<prob>: ...this.
(register_solib_event_probe): Receive 'probe' instead of
'struct probe' as argument. Use 'prob' instead of 'probe'
when applicable.
(solib_event_probe_action): Call 'get_argument_count' method
from probe. Adjust comment.
(svr4_handle_solib_event): Adjust comment. Call
'evaluate_argument' method from probe.
(svr4_create_probe_breakpoints): Call 'get_relocated_address'
from probe.
(svr4_create_solib_event_breakpoints): Use 'probe' instead of
'struct probe'. Call 'can_evaluate_arguments' from probe.
* symfile.h: Forward declare 'class probe' instead of 'struct
probe'.
* symtab.h: Likewise.
(struct symtab_and_line) <probe>: Rename to...
<prob>: ...this.
* tracepoint.c (start_tracing): Use 'prob' when applicable.
Call probe methods.
(stop_tracing): Likewise.
This changes struct symbol to use an enum to encode the concrete
subclass of a particular symbol. Note that "enum class" doesn't work
properly with bitfields, so a plain enum is used.
2017-11-17 Tom Tromey <tom@tromey.com>
* symtab.h (enum symbol_subclass_kind): New.
(struct symbol) <is_cplus_template_function, is_rust_vtable>:
Remove.
<subclass>: New member.
(SYMBOL_IS_CPLUS_TEMPLATE_FUNCTION): Update.
* rust-lang.c (rust_get_trait_object_pointer): Update.
* dwarf2read.c (read_func_scope): Update.
(read_variable): Update.
This changes template_symbol to derive from symbol, which seems a bit
cleaner; and also more consistent with rust_vtable_symbol.
2017-11-17 Tom Tromey <tom@tromey.com>
* dwarf2read.c (read_func_scope): Update.
* symtab.h (struct template_symbol): Derive from symbol.
<base>: Remove.
In Rust, virtual tables work a bit differently than they do in C++. In
C++, as you know, they are connected to a particular class hierarchy.
Rust, instead, can generate a virtual table for potentially any type --
in fact, one such virtual table for each trait (a trait is similar to an
abstract class or to a Java interface) that a type implements.
Objects that are referenced via a trait can't currently be inspected by
gdb. This patch implements the Rust equivalent of "set print object".
gdb relies heavily on the C++ ABI to decode virtual tables; primarily to
make "set print object" work; but also "info vtbl". However, Rust does
not currently have a specified ABI, so this approach seems unwise to
emulate.
Instead, I've changed the Rust compiler to emit some DWARF that
describes trait objects (previously their internal structure was
opaque), vtables (currently just a size -- but I hope to expand this in
the future), and the concrete type for which a vtable was emitted.
The concrete type is expressed as a DW_AT_containing_type on the
vtable's type. This is a small extension to DWARF.
This patch adds a new entry to quick_symbol_functions to return the
symtab that holds a data address. Previously there was no way in gdb to
look up a full (only minimal) non-text symbol by address. The psymbol
implementation of this method works by lazily filling in a map that is
added to the objfile. This avoids slowing down psymbol reading for a
feature that is likely to not be used too frequently.
I did not update .gdb_index. My thinking here is that the DWARF 5
indices will obsolete .gdb_index soon-ish, meaning that adding a new
feature to them is probably wasted work. If necessary I can update the
DWARF 5 index code when it lands in gdb.
Regression tested on x86-64 Fedora 25.
2017-11-17 Tom Tromey <tom@tromey.com>
* symtab.h (struct symbol) <is_rust_vtable>: New member.
(struct rust_vtable_symbol): New.
(find_symbol_at_address): Declare.
* symtab.c (find_symbol_at_address): New function.
* symfile.h (struct quick_symbol_functions)
<find_compunit_symtab_by_address>: New member.
* symfile-debug.c (debug_qf_find_compunit_symtab_by_address): New
function.
(debug_sym_quick_functions): Link to
debug_qf_find_compunit_symtab_by_address.
* rust-lang.c (rust_get_trait_object_pointer): New function.
(rust_evaluate_subexp) <case UNOP_IND>: New case. Call
rust_get_trait_object_pointer.
* psymtab.c (psym_relocate): Clear psymbol_map.
(psym_fill_psymbol_map, psym_find_compunit_symtab_by_address): New
functions.
(psym_functions): Link to psym_find_compunit_symtab_by_address.
* objfiles.h (struct objfile) <psymbol_map>: New member.
* dwarf2read.c (dwarf2_gdb_index_functions): Update.
(process_die) <DW_TAG_variable>: New case. Call read_variable.
(rust_containing_type, read_variable): New functions.
2017-11-17 Tom Tromey <tom@tromey.com>
* gdb.rust/traits.rs: New file.
* gdb.rust/traits.exp: New file.
Currently "b foo[TAB]" offers data symbols as completion candidates.
This doesn't make sense, since you can't set a breakpoint on data
symbols, only on code symbols.
(gdb) b globa[TAB]
(gdb) b global [ENTER]
Function "global" not defined.
Make breakpoint pending on future shared library load? (y or [n]) n
(gdb) info symbol global
global in section .rodata
So this patch makes linespec completion ignore data symbols.
gdb/ChangeLog:
2017-11-08 Pedro Alves <palves@redhat.com>
* ada-lang.c (ada_make_symbol_completion_list): Use
completion_skip_symbol.
* symtab.c (symbol_is_function_or_method(minimal_symbol*)): New.
(symbol_is_function_or_method(symbol*)): New.
(add_symtab_completions): Add complete_symbol_mode parameter. Use
completion_skip_symbol.
(default_collect_symbol_completion_matches_break_on): Use
completion_skip_symbol. Pass down mode.
(collect_file_symbol_completion_matches): Pass down mode.
* symtab.h (symbol_is_function_or_method): New declarations.
(completion_skip_symbol): New template function.
sym_text_len existed to strip parameters out of the lookup name. Now
that that's handled by the lookup_name_info objects, the
sym_text/sym_text_len parameters are no longer necessary.
gdb/ChangeLog:
2017-11-08 Pedro Alves <palves@redhat.com>
* ada-lang.c (ada_make_symbol_completion_list): Remove text and
text_len locals and don't pass them down.
* symtab.c (completion_list_add_name): Remove
sym_text/sym_text_len parameters and adjust.
(completion_list_add_symbol, completion_list_add_msymbol)
(completion_list_objc_symbol, completion_list_add_fields)
(add_symtab_completions): Likewise.
(default_collect_symbol_completion_matches_break_on)
(collect_file_symbol_completion_matches): Remove sym_text_len
local and don't pass it down.
* symtab.h (completion_list_add_name): Remove
sym_text/sym_text_len parameters.
A few places in the completion code look for a "(" to find a
function's parameter list, in order to strip it, because psymtabs (and
gdb index) don't include parameter info in the symbol names.
See compare_symbol_name and
default_collect_symbol_completion_matches_break_on.
This is too naive. Consider:
ns_overload2_test::([TAB]
We'd want to complete that to:
ns_overload2_test::(anonymous namespace)::struct_overload2_test
Or:
b (anonymous namespace)::[TAB]
That currently completes to:
b (anonymous namespace)
Which is obviously broken. This patch makes that work.
Also, the current compare_symbol_name hack means that while this
works:
"b function([TAB]" -> "b function()"
This does not:
"b function ([TAB]"
This patch fixes that. Whitespace "ignoring" now Just Works, i.e.,
assuming a symbol named "function(int, long)", this:
b function ( int , lon[TAB]
completes to:
b function ( int , long)
To address all of this, this patch builds on top of the rest of the
series, and pushes the responsibility of stripping parameters from a
lookup name to the new lookup_name_info object, where we can apply
per-language rules. Also note that we now only make a version of the
lookup name with parameters stripped out where it's actually required
to do that, in the psymtab and GDB index code.
For C++, the right way to strip parameters is with "cp_remove_params",
which uses a real parser (cp-name-parser.y) to split the name into a
component tree and then discards parameters.
The trouble for completion is that in that case we have an incomplete
name, like "foo::func(int" and thus cp_remove_params throws an error.
This patch sorts that by adding a cp_remove_params_if_any variant of
cp_remove_params that tries removing characters from the end of the
string until cp_remove_params works. So cp_remove_params_if_any
behaves like this:
With a complete name:
"foo::func(int)" => foo::func(int) # cp_remove_params_1 succeeds the first time.
With an incomplete name:
"foo::func(int" => NULL # cp_remove_params fails the first time.
"foo::func(in" => NULL # and again...
"foo::func(i" => NULL # and again...
"foo::func(" => NULL # and again...
"foo::func" => "foo::func" # success!
Note that even if this approach removes significant rightmost
characters, it's still OK, because this parameter stripping is only
necessary for psymtabs and gdb index, where we're determining whether
to expand a symbol table. Say cp_remove_params_if_any returned
"foo::" above for "foo::func(int". That'd cause us to expand more
symtabs than ideal (because we'd expand all symtabs with symbols that
start with "foo::", not just "foo::func"), but then when we actually
look for completion matches, we'd still use the original lookup name,
with parameter information ["foo::func(int"], and thus we'll return no
false positive to the user. Whether the stripping works as intended
and doesn't strip too much is thus covered by a unit test instead of a
testsuite test.
The "if_any" part of the name refers to the fact that while
cp_remove_params returns NULL if the input name has no parameters in
the first place, like:
"foo::func" => NULL # cp_remove_params
cp_remove_params_if_any still returns the function name:
"foo::func" => "foo::func" # cp_remove_params_if_any
gdb/ChangeLog:
2017-11-08 Pedro Alves <palves@redhat.com>
* Makefile.in (SUBDIR_UNITTESTS_SRCS): Add
unittests/lookup_name_info-selftests.c.
(SUBDIR_UNITTESTS_OBS): Add lookup_name_info-selftests.o.
* cp-support.c: Include "selftest.h".
(cp_remove_params_1): Rename from cp_remove_params. Add
'require_param' parameter, and handle it.
(cp_remove_params): Reimplement.
(cp_remove_params_if_any): New.
(selftests::quote): New.
(selftests::check_remove_params): New.
(selftests::test_cp_remove_params): New.
(_initialize_cp_support): Install
selftests::test_cp_remove_params.
* cp-support.h (cp_remove_params_if_any): Declare.
* dwarf2read.c :Include "selftest.h".
(dw2_expand_symtabs_matching_symbol): Use
lookup_name_info::make_ignore_params.
(selftests::dw2_expand_symtabs_matching::mock_mapped_index)
(selftests::dw2_expand_symtabs_matching::string_or_null)
(selftests::dw2_expand_symtabs_matching::check_match)
(selftests::dw2_expand_symtabs_matching::test_symbols)
(selftests::dw2_expand_symtabs_matching::run_test): New.
(_initialize_dwarf2_read): Register
selftests::dw2_expand_symtabs_matching::run_test.
* psymtab.c (psym_expand_symtabs_matching): Use
lookup_name_info::make_ignore_params.
* symtab.c (demangle_for_lookup_info::demangle_for_lookup_info):
If the lookup name wants to ignore parameters, strip them.
(compare_symbol_name): Remove sym_text/sym_text_len parameters and
code handling '('.
(completion_list_add_name): Don't pass down sym_text/sym_text_len.
(default_collect_symbol_completion_matches_break_on): Don't try to
strip parameters.
* symtab.h (lookup_name_info::lookup_name_info): Add
'ignore_parameters' parameter.
(lookup_name_info::ignore_parameters)
(lookup_name_info::make_ignore_params): New methods.
(lookup_name_info::m_ignore_parameters): New field.
* unittests/lookup_name_info-selftests.c: New file.
Summary:
- This is preparation for supporting wild name matching on C++ too.
- This is also preparation for TAB-completion fixes.
- Makes symbol name matching (think strcmp_iw) be based on a per-language method.
- Merges completion and non-completion name comparison (think
language_ops::la_get_symbol_name_cmp generalized).
- Avoid re-hashing lookup name multiple times
- Centralizes preparing a name for lookup (Ada name encoding / C++ Demangling),
both completion and non-completion.
- Fixes Ada latent bug with verbatim name matches in expressions
- Makes ada-lang.c use common|symtab.c completion code a bit more.
Ada's wild matching basically means that
"(gdb) break foo"
will find all methods named "foo" in all packages. Translating to
C++, it's roughly the same as saying that "break klass::method" sets
breakpoints on all "klass::method" methods of all classes, no matter
the namespace. A following patch will teach GDB about fullname vs
wild matching for C++ too. This patch is preparatory work to get
there.
Another idea here is to do symbol name matching based on the symbol
language's algorithm. I.e., avoid dependency on current language set.
This allows for example doing
(gdb) b foo::bar< int > (<tab>
and having gdb name match the C++ symbols correctly even if the
current language is C or Assembly (or Rust, or Ada, or ...), which can
easily happen if you step into an Assembly/C runtime library frame.
By encapsulating all the information related to a lookup name in a
class, we can also cache hash computation for a given language in the
lookup name object, to avoid recomputing it over and over.
Similarly, because we don't really know upfront which languages the
lookup name will be matched against, for each language we store the
lookup name transformed into a search name. E.g., for C++, that means
demangling the name. But for Ada, it means encoding the name. This
actually forces us to centralize all the different lookup name
encoding in a central place, resulting in clearer code, IMO. See
e.g., the new ada_lookup_name_info class.
The lookup name -> symbol search name computation is also done only
once per language.
The old language->la_get_symbol_name_cmp / symbol_name_cmp_ftype are
generalized to work with both completion, and normal symbol look up.
At some point early on, I had separate completion vs non-completion
language vector entry points, but a single method ends up being better
IMO for simplifying things -- the more we merge the completion /
non-completion name lookup code paths, the less changes for bugs
causing completion vs normal lookup finding different symbols.
The ada-lex.l change is necessary because when doing
(gdb) p <UpperCase>
then the name that is passed to write_ write_var_or_type ->
ada_lookup_symbol_list misses the "<>", i.e., it's just "UpperCase",
and we end up doing a wild match against "UpperCase" lowercased by
ada_lookup_name_info's constructor. I.e., "uppercase" wouldn't ever
match "UpperCase", and the symbol lookup fails.
This wouldn't cause any regression in the testsuite, but I added a new
test that would pass before the patch and fail after, if it weren't
for that fix.
This is latent bug that happens to go unnoticed because that
particular path was inconsistent with the rest of Ada symbol lookup by
not lowercasing the lookup name.
Ada's symbol_completion_add is deleted, replaced by using common
code's completion_list_add_name. To make the latter work for Ada, we
needed to add a new output parameter, because Ada wants to return back
a custom completion candidates that are not the symbol name.
With this patch, minimal symbol demangled name hashing is made
consistent with regular symbol hashing. I.e., it now goes via the
language vector's search_name_hash method too, as I had suggested in a
previous patch.
dw2_expand_symtabs_matching / .gdb_index symbol names were a
challenge. The problem is that we have no way to telling what is the
language of each symbol name found in the index, until we expand the
corresponding full symbol, which is off course what we're trying to
avoid. Language information is simply not considered in the index
format... Since the symbol name hashing and comparison routines are
per-language, we now have a problem. The patch sorts this out by
matching each name against all languages. This is inneficient, and
indeed slows down completion several times. E.g., with:
$ cat script.cmd
set pagination off
set $count = 0
while $count < 400
complete b string_prin
printf "count = %d\n", $count
set $count = $count + 1
end
$ time gdb --batch -q ./gdb-with-index -ex "source script-string_printf.cmd"
I get, before patch (-O2, x86-64):
real 0m1.773s
user 0m1.737s
sys 0m0.040s
While after patch (-O2, x86-64):
real 0m9.843s
user 0m9.482s
sys 0m0.034s
However, the following patch will optimize this, and will actually
make this use case faster compared to the "before patch" above:
real 0m1.321s
user 0m1.285s
sys 0m0.039s
gdb/ChangeLog:
2017-11-08 Pedro Alves <palves@redhat.com>
* ada-lang.c (ada_encode): Rename to ..
(ada_encode_1): ... this. Add throw_errors parameter and handle
it.
(ada_encode): Reimplement.
(match_name): Delete, folded into full_name.
(resolve_subexp): No longer pass the encoded name to
ada_lookup_symbol_list.
(should_use_wild_match): Delete.
(name_match_type_from_name): New.
(ada_lookup_simple_minsym): Use lookup_name_info and the
language's symbol_name_matcher_ftype.
(add_symbols_from_enclosing_procs, ada_add_local_symbols)
(ada_add_block_renamings): Adjust to use lookup_name_info.
(ada_lookup_name): New.
(add_nonlocal_symbols, ada_add_all_symbols)
(ada_lookup_symbol_list_worker, ada_lookup_symbol_list)
(ada_iterate_over_symbols): Adjust to use lookup_name_info.
(ada_name_for_lookup): Delete.
(ada_lookup_encoded_symbol): Construct a verbatim name.
(wild_match): Reverse sense of return type. Use bool.
(full_match): Reverse sense of return type. Inline bits of old
match_name here.
(ada_add_block_symbols): Adjust to use lookup_name_info.
(symbol_completion_match): Delete, folded into...
(ada_lookup_name_info::matches): ... .this new method.
(symbol_completion_add): Delete.
(ada_collect_symbol_completion_matches): Add name_match_type
parameter. Adjust to use lookup_name_info and
completion_list_add_name.
(get_var_value, ada_add_global_exceptions): Adjust to use
lookup_name_info.
(ada_get_symbol_name_cmp): Delete.
(do_wild_match, do_full_match): New functions.
(ada_lookup_name_info::ada_lookup_name_info): New method.
(ada_symbol_name_matches, ada_get_symbol_name_matcher): New
functions.
(ada_language_defn): Install ada_get_symbol_name_matcher.
* ada-lex.l (processId): If name starts with '<', copy it
verbatim.
* block.c (block_iter_match_step, block_iter_match_first)
(block_iter_match_next, block_lookup_symbol)
(block_lookup_symbol_primary, block_find_symbol): Adjust to use
lookup_name_info.
* block.h (block_iter_match_first, block_iter_match_next)
(ALL_BLOCK_SYMBOLS_WITH_NAME): Adjust to use lookup_name_info.
* c-lang.c (c_language_defn, cplus_language_defn)
(asm_language_defn, minimal_language_defn): Adjust comments to
refer to la_get_symbol_name_matcher.
* completer.c (complete_files_symbols)
(collect_explicit_location_matches, symbol_completer): Pass a
symbol_name_match_type down.
* completer.h (class completion_match, completion_match_result):
New classes.
(completion_tracker::reset_completion_match_result): New method.
(completion_tracker::m_completion_match_result): New field.
* cp-support.c (make_symbol_overload_list_block): Adjust to use
lookup_name_info.
(cp_fq_symbol_name_matches, cp_get_symbol_name_matcher): New
functions.
* cp-support.h (cp_get_symbol_name_matcher): New declaration.
* d-lang.c: Adjust comments to refer to
la_get_symbol_name_matcher.
* dictionary.c (dict_vector) <iter_match_first, iter_match_next>:
Adjust to use lookup_name_info.
(dict_iter_match_first, dict_iter_match_next)
(iter_match_first_hashed, iter_match_next_hashed)
(iter_match_first_linear, iter_match_next_linear): Adjust to work
with a lookup_name_info.
* dictionary.h (dict_iter_match_first, dict_iter_match_next):
Likewise.
* dwarf2read.c (dw2_lookup_symbol): Adjust to use lookup_name_info.
(dw2_map_matching_symbols): Adjust to use symbol_name_match_type.
(gdb_index_symbol_name_matcher): New class.
(dw2_expand_symtabs_matching) Adjust to use lookup_name_info and
gdb_index_symbol_name_matcher. Accept a NULL symbol_matcher.
* f-lang.c (f_collect_symbol_completion_matches): Adjust to work
with a symbol_name_match_type.
(f_language_defn): Adjust comments to refer to
la_get_symbol_name_matcher.
* go-lang.c (go_language_defn): Adjust comments to refer to
la_get_symbol_name_matcher.
* language.c (default_symbol_name_matcher)
(language_get_symbol_name_matcher): New functions.
(unknown_language_defn, auto_language_defn): Adjust comments to
refer to la_get_symbol_name_matcher.
* language.h (symbol_name_cmp_ftype): Delete.
(language_defn) <la_collect_symbol_completion_matches>: Add match
type parameter.
<la_get_symbol_name_cmp>: Delete field.
<la_get_symbol_name_matcher>: New field.
<la_iterate_over_symbols>: Adjust to use lookup_name_info.
(default_symbol_name_matcher, language_get_symbol_name_matcher):
Declare.
* linespec.c (iterate_over_all_matching_symtabs)
(iterate_over_file_blocks): Adjust to use lookup_name_info.
(find_methods): Add language parameter, and use lookup_name_info
and the language's symbol_name_matcher_ftype.
(linespec_complete_function): Adjust.
(lookup_prefix_sym): Use lookup_name_info.
(add_all_symbol_names_from_pspace): Adjust.
(find_superclass_methods): Add language parameter and pass it
down.
(find_method): Pass symbol language down.
(find_linespec_symbols): Don't demangle or Ada encode here.
(search_minsyms_for_name): Add lookup_name_info parameter.
(add_matching_symbols_to_info): Add name_match_type parameter.
Use lookup_name_info.
* m2-lang.c (m2_language_defn): Adjust comments to refer to
la_get_symbol_name_matcher.
* minsyms.c: Include <algorithm>.
(add_minsym_to_demangled_hash_table): Remove table parameter and
add objfile parameter. Use search_name_hash, and add language to
demangled languages vector.
(struct found_minimal_symbols): New struct.
(lookup_minimal_symbol_mangled, lookup_minimal_symbol_demangled):
New functions.
(lookup_minimal_symbol): Adjust to use them. Don't canonicalize
input names here. Use lookup_name_info instead. Lookup up
demangled names once for each language in the demangled names
vector.
(iterate_over_minimal_symbols): Use lookup_name_info. Lookup up
demangled names once for each language in the demangled names
vector.
(build_minimal_symbol_hash_tables): Adjust.
* minsyms.h (iterate_over_minimal_symbols): Adjust to pass down a
lookup_name_info.
* objc-lang.c (objc_language_defn): Adjust comment to refer to
la_get_symbol_name_matcher.
* objfiles.h: Include <vector>.
(objfile_per_bfd_storage) <demangled_hash_languages>: New field.
* opencl-lang.c (opencl_language_defn): Adjust comment to refer to
la_get_symbol_name_matcher.
* p-lang.c (pascal_language_defn): Adjust comment to refer to
la_get_symbol_name_matcher.
* psymtab.c (psym_lookup_symbol): Use lookup_name_info.
(match_partial_symbol): Use symbol_name_match_type,
lookup_name_info and psymbol_name_matches.
(lookup_partial_symbol): Use lookup_name_info.
(map_block): Use symbol_name_match_type and lookup_name_info.
(psym_map_matching_symbols): Use symbol_name_match_type.
(psymbol_name_matches): New.
(recursively_search_psymtabs): Use lookup_name_info and
psymbol_name_matches. Rename 'kind' parameter to 'domain'.
(psym_expand_symtabs_matching): Use lookup_name_info. Rename
'kind' parameter to 'domain'.
* rust-lang.c (rust_language_defn): Adjust comment to refer to
la_get_symbol_name_matcher.
* symfile-debug.c (debug_qf_map_matching_symbols)
(debug_qf_map_matching_symbols): Use symbol_name_match_type.
(debug_qf_expand_symtabs_matching): Use lookup_name_info.
* symfile.c (expand_symtabs_matching): Use lookup_name_info.
* symfile.h (quick_symbol_functions) <map_matching_symbols>:
Adjust to use symbol_name_match_type.
<expand_symtabs_matching>: Adjust to use lookup_name_info.
(expand_symtabs_matching): Adjust to use lookup_name_info.
* symmisc.c (maintenance_expand_symtabs): Use
lookup_name_info::match_any ().
* symtab.c (symbol_matches_search_name): New.
(eq_symbol_entry): Adjust to use lookup_name_info and the
language's matcher.
(demangle_for_lookup_info::demangle_for_lookup_info): New.
(lookup_name_info::match_any): New.
(iterate_over_symbols, search_symbols): Use lookup_name_info.
(compare_symbol_name): Add language, lookup_name_info and
completion_match_result parameters, and use them.
(completion_list_add_name): Make extern. Add language and
lookup_name_info parameters. Use them.
(completion_list_add_symbol, completion_list_add_msymbol)
(completion_list_objc_symbol): Add lookup_name_info parameters and
adjust. Pass down language.
(completion_list_add_fields): Add lookup_name_info parameters and
adjust. Pass down language.
(add_symtab_completions): Add lookup_name_info parameters and
adjust.
(default_collect_symbol_completion_matches_break_on): Add
name_match_type parameter, and use it. Use lookup_name_info.
(default_collect_symbol_completion_matches)
(collect_symbol_completion_matches): Add name_match_type
parameter, and pass it down.
(collect_symbol_completion_matches_type): Adjust.
(collect_file_symbol_completion_matches): Add name_match_type
parameter, and use lookup_name_info.
* symtab.h: Include <string> and "common/gdb_optional.h".
(enum class symbol_name_match_type): New.
(class ada_lookup_name_info): New.
(struct demangle_for_lookup_info): New.
(class lookup_name_info): New.
(symbol_name_matcher_ftype): New.
(SYMBOL_MATCHES_SEARCH_NAME): Use symbol_matches_search_name.
(symbol_matches_search_name): Declare.
(MSYMBOL_MATCHES_SEARCH_NAME): Delete.
(default_collect_symbol_completion_matches)
(collect_symbol_completion_matches)
(collect_file_symbol_completion_matches): Add name_match_type
parameter.
(iterate_over_symbols): Use lookup_name_info.
(completion_list_add_name): Declare.
* utils.c (enum class strncmp_iw_mode): Moved to utils.h.
(strncmp_iw_with_mode): Now extern.
* utils.h (enum class strncmp_iw_mode): Moved from utils.c.
(strncmp_iw_with_mode): Declare.
gdb/testsuite/ChangeLog:
2017-11-08 Pedro Alves <palves@redhat.com>
* gdb.ada/complete.exp (p <Exported_Capitalized>): New test.
(p Exported_Capitalized): New test.
(p exported_capitalized): New test.
Currently, we have a mess of symbol name hashing/comparison routines.
There's msymbol_hash for mangled names, and dict_hash and
msymbol_hash_iw for demangled names. Then there's strcmp_iw,
strcmp_iw_ordered and Ada's full_match/wild_match, which all have to
agree with the hashing routines. That's why dict_hash is really about
Ada names. From the inconsistency department, minimal symbol hashing
doesn't go via dict_hash, so Ada's wild matching can't ever work with
minimal symbols.
This patch starts fixing this, by doing two things:
#1 - adds a language vector method to let each language decide how to
compute a symbol name hash.
#2 - makes dictionaries know the language of the symbols they hold,
and then use the dictionaries language to decide which hashing
method to use.
For now, this is just scaffolding, since all languages install the
default method. The series will make C++ install its own hashing
method later on, and will add per-language symbol name comparison
routines too.
This patch was originally based on a patch that Keith wrote for the
libcc1/C++ WIP support.
gdb/ChangeLog:
2017-11-08 Keith Seitz <keiths@redhat.com>
Pedro Alves <palves@redhat.com>
* ada-lang.c (ada_language_defn): Install
default_search_name_hash.
* buildsym.c (struct buildsym_compunit): <language>: New field.
(finish_block_internal): Pass language when creating dictionaries.
(start_buildsym_compunit, start_symtab): New language parameters.
Use them.
(restart_symtab): Pass down compilation unit's language.
* buildsym.h (enum language): Forward declare.
(start_symtab): New 'language' parameter.
* c-lang.c (c_language_defn, cplus_language_defn)
(asm_language_defn, minimal_language_defn): Install
default_search_name_hash.
* coffread.c (coff_start_symtab): Adjust.
* d-lang.c (d_language_defn): Install default_search_name_hash.
* dbxread.c (struct symloc): Add 'pst_language' field.
(PST_LANGUAGE): Define.
(start_psymtab, read_ofile_symtab): Use it.
(process_one_symbol): New 'language' parameter. Pass it down.
* dictionary.c (struct dictionary) <language>: New field.
(DICT_LANGUAGE): Define.
(dict_create_hashed, dict_create_hashed_expandable)
(dict_create_linear, dict_create_linear_expandable): New parameter
'language'. Set the dictionary's language.
(iter_match_first_hashed): Adjust to rename.
(insert_symbol_hashed): Assert we don't see mismatching
languages. Adjust to rename.
(dict_hash): Rename to ...
(default_search_name_hash): ... this and make extern.
* dictionary.h (struct language_defn): Forward declare.
(dict_create_hashed): New parameter 'language'.
* dwarf2read.c (dwarf2_start_symtab): Pass down language.
* f-lang.c (f_language_defn): Install default_search_name_hash.
* go-lang.c (go_language_defn): Install default_search_name_hash.
* jit.c (finalize_symtab): Pass compunit's language to dictionary
creation.
* language.c (unknown_language_defn, auto_language_defn):
* language.h (language_defn::la_search_name_hash): New field.
(default_search_name_hash): Declare.
* m2-lang.c (m2_language_defn): Install default_search_name_hash.
* mdebugread.c (new_block): New parameter 'language'.
* mdebugread.c (parse_symbol): Pass symbol language to block
allocation.
(psymtab_to_symtab_1): Pass down language.
(new_symtab): Pass compunit's language to block allocation.
* objc-lang.c (objc_language_defn): Install
default_search_name_hash.
* opencl-lang.c (opencl_language_defn):
* p-lang.c (pascal_language_defn): Install
default_search_name_hash.
* rust-lang.c (rust_language_defn): Install
default_search_name_hash.
* stabsread.h (enum language): Forward declare.
(process_one_symbol): Add 'language' parameter.
* symtab.c (search_name_hash): New function.
* symtab.h (search_name_hash): Declare.
* xcoffread.c (read_xcoff_symtab): Pass language to start_symtab.
This changes search_symbols to return a std::vector, replacing the
previous linked list approach. This allows the removal of some
cleanups, as well as the use of std::sort and std::unique, saving some
code and extra allocations in sort_search_symbols_remove_dups.
Regression tested by the buildbot.
gdb/ChangeLog
2017-10-08 Tom Tromey <tom@tromey.com>
* symtab.c (free_search_symbols, do_free_search_symbols_cleanup)
(make_cleanup_free_search_symbols): Remove.
(search_symbols): Return std::vector.
(symbol_search::compare_search_syms): Now member of
symbol_search. Change arguments.
(sort_search_symbols_remove_dups): Change arguments. Rewrite.
(symtab_symbol_info, rbreak_command): Update.
* symtab.h (struct symbol_search) <next>: Remove.
Add constructors.
(symbol_search::operator<): New function.
(symbol_search::operator==): New function.
(search_symbols): Remove std::vector.
(free_search_symbols, make_cleanup_free_search_symbols): Remove.
(symbol_search::compare_search_syms): Declare.
Currently, with an ambiguous "list first,last", we get:
(gdb) list bar,main
Specified first line 'bar' is ambiguous:
file: "src/gdb/testsuite/gdb.cp/overload.cc", line number: 97
file: "src/gdb/testsuite/gdb.cp/overload.cc", line number: 98
This commit makes gdb's output above a bit clearer by printing the
symbol name as well:
(gdb) list bar,main
Specified first line 'bar' is ambiguous:
file: "src/gdb/testsuite/gdb.cp/overload.cc", line number: 97, symbol: "bar(A)"
file: "src/gdb/testsuite/gdb.cp/overload.cc", line number: 98, symbol: "bar(B)"
And while at it, makes gdb print the symbol name when actually listing
multiple locations too. I.e., before (with "set listsize 2"):
(gdb) list bar
file: "src/gdb/testsuite/gdb.cp/overload.cc", line number: 97
96
97 int bar (A) { return 11; }
file: "src/gdb/testsuite/gdb.cp/overload.cc", line number: 98
97 int bar (A) { return 11; }
98 int bar (B) { return 22; }
After:
(gdb) list bar
file: "src/gdb/testsuite/gdb.cp/overload.cc", line number: 97, symbol: "bar(A)"
96
97 int bar (A) { return 11; }
file: "src/gdb/testsuite/gdb.cp/overload.cc", line number: 98, symbol: "bar(B)"
97 int bar (A) { return 11; }
98 int bar (B) { return 22; }
Currently, the result of decoding a linespec loses information about
the original symbol that was found. All we end up with is an address.
This makes it difficult to find the original symbol again to get at
its print name. Fix that by storing a pointer to the symbol in the
sal. We already store the symtab and obj_section, so it feels like a
natural progression to me. This avoids having to do any extra symbol
lookup too.
gdb/ChangeLog:
2017-09-20 Pedro Alves <palves@redhat.com>
* cli/cli-cmds.c (list_command): Use print_sal_location.
(print_sal_location): New function.
(ambiguous_line_spec): Use print_sal_location.
* linespec.c (symbol_to_sal): Record the symbol in the sal.
* symtab.c (find_function_start_sal): Likewise.
* symtab.h (symtab_and_line::symbol): New field.
gdb/testsuite/ChangeLog:
2017-09-20 Pedro Alves <palves@redhat.com>
* gdb.base/list-ambiguous.exp (test_list_ambiguous_symbol): Expect
symbol names in gdb's output.
* gdb.cp/overload.exp ("list all overloads"): Likewise.
Tom Tromey
Christian Biesinger
Andrew Burgess
Tom Tromey
Philippe Waroquiers
Joel Brobecker
Simon Marchi
Keith Seitz
Kevin Buettner
Pedro Alves
Sergio Durigan Junior
John Baldwin