Tom Tromey [Sat, 22 Jun 2019 18:07:13 +0000 (12:07 -0600)]
Remove "data_content" and "data_content_count" from TUI data window
The TUI has some stub code for adding data other than registers to the
data window. However, it doesn't do anything, and apparently never
has. This removes the dead code.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-data.c (tui_data_item_window): Update.
* tui/tui-windata.h (tui_check_data_values): Don't declare.
* tui/tui-windata.c (tui_display_all_data)
(tui_display_data_from_line): Update.
(tui_check_data_values): Remove.
* tui/tui-regs.c (tui_show_register_group)
(tui_display_reg_element_at_line): Update.
* tui/tui-hooks.c (tui_register_changed)
(tui_refresh_frame_and_register_information): Call
tui_check_register_values.
* tui/tui-data.h (struct tui_data_window) <data_content,
data_content_count, data_type>: Remove.
(enum tui_data_type): Remove.
Tom Tromey [Sat, 22 Jun 2019 17:55:33 +0000 (11:55 -0600)]
Turn tui_first_data_item_displayed into a method
tui_first_data_item_displayed is only called from tui_data_window
methods, so turn it into a method as well.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-windata.h (tui_first_data_item_displayed): Don't
declare.
* tui/tui-windata.c (tui_data_window::first_data_item_displayed):
Rename from tui_first_data_item_displayed. Update.
(tui_data_window::refresh_all)
(tui_data_window::do_scroll_vertical): Update.
* tui/tui-data.h (struct tui_data_window)
<first_data_item_displayed>: Declare new method.
Tom Tromey [Sat, 22 Jun 2019 17:45:58 +0000 (11:45 -0600)]
Remove tui_init_generic_part
tui_init_generic_part has a single caller, so simply inline it there.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-data.h (tui_init_generic_part): Don't declare.
* tui/tui-data.c (tui_init_generic_part): Remove, moving
contents...
(tui_initialize_static_data): ...here.
Tom Tromey [Sat, 22 Jun 2019 06:20:39 +0000 (00:20 -0600)]
Separate out data item window
This introduces a new subclass of tui_gen_win_info for the data item
windows, letting us remove another element from tui_which_element.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-regs.c (tui_show_registers, tui_show_register_group)
(tui_display_registers_from, tui_check_register_values): Update.
(tui_display_register): Remove win_info parameter; update.
(tui_get_register): Change type of parameters.
* tui/tui-data.h (struct tui_data_element): Remove.
(union tui_which_element) <data>: Remove.
<data_window>: Change type.
(struct tui_data_item_window): New.
* tui/tui-data.c (init_content_element): Remove DATA_ITEM_WIN
case. Add assert.
(~tui_data_item_window): New destructor.
(free_content_elements): Remove DATA_ITEM_WIN case.
Tom Tromey [Sat, 22 Jun 2019 06:00:11 +0000 (00:00 -0600)]
Remove two unused enum constants from tui_win_type
This removes a couple of unused constants from enum tui_win_type.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui.h (enum tui_win_type) <MAX_WINDOWS, UNDEFINED_WIN>:
Remove.
Tom Tromey [Fri, 21 Jun 2019 12:16:06 +0000 (06:16 -0600)]
Remove command from tui_which_element
union tui_which_element has a "command" member, but it is never used.
This removes it.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-data.h (struct tui_command_element): Remove.
(union tui_which_element) <command>: Remove.
* tui/tui-data.c (init_content_element): Remove CMD_WIN case. Add
assert.
(free_content_elements): Remove CMD_WIN case.
Tom Tromey [Fri, 21 Jun 2019 12:12:48 +0000 (06:12 -0600)]
Remove layout_def::split
The "split" field in struct layout_def is never used, so this patch
removes it.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-layout.c (tui_set_layout): Update.
* tui/tui-data.h (struct tui_layout_def) <split>: Remove.
* tui/tui-data.c (layout_def): Update.
Tom Tromey [Fri, 21 Jun 2019 02:16:07 +0000 (20:16 -0600)]
Separate out locator window
This introduces a new subclass of tui_gen_win_info for the locator,
letting us remove another element from union tui_which_element.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-wingeneral.c (tui_refresh_all): Update.
* tui/tui-win.c (tui_resize_all, tui_adjust_win_heights)
(tui_source_window_base::set_new_height): Update.
* tui/tui-stack.c (tui_make_status_line): Change parameter type.
Update.
(tui_set_locator_fullname, tui_set_locator_info)
(tui_show_frame_info): Update.
* tui/tui-source.c (tui_set_source_content)
(tui_source_is_displayed): Update.
* tui/tui-layout.c (show_source_disasm_command, show_data)
(show_source_or_disasm_and_command): Update.
* tui/tui-disasm.c (tui_set_disassem_content)
(tui_get_begin_asm_address): Update.
* tui/tui-data.h (struct tui_locator_element): Remove.
(union tui_which_element) <locator>: Remove.
(struct tui_locator_window): New.
(tui_locator_win_info_ptr): Change return type.
* tui/tui-data.c (_locator): Change type.
(tui_locator_win_info_ptr): Change return type.
(init_content_element): Remove LOCATOR_WIN case. Add assert.
(tui_alloc_content): Add assert.
Tom Tromey [Fri, 21 Jun 2019 01:59:06 +0000 (19:59 -0600)]
Separate out execution-info window
This pulls the EXEC_INFO_WIN case out into its own subclass of
tui_gen_win_info. This lets us remove an element from
union tui_which_element.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-winsource.c
(tui_exec_info_window::maybe_allocate_content): New method.
(tui_set_exec_info_content, tui_show_exec_info_content): Update.
* tui/tui-layout.c (init_and_make_win): Add EXEC_INFO_WIN case.
(make_source_or_disasm_window): Add cast.
* tui/tui-data.h (union tui_which_element) <simple_string>:
Remove.
(struct tui_source_info): New.
(struct tui_source_window_base) <execution_info>: Change type.
* tui/tui-data.c (init_content_element): Remove EXEC_INFO_WIN
case, and add assert.
(tui_alloc_content): Add assert.
Tom Tromey [Thu, 20 Jun 2019 21:31:00 +0000 (15:31 -0600)]
Remove tui_alloc_win_info
There is only a single caller of tui_alloc_win_info, and we're going
to add more "new" cases to that caller, so remove tui_alloc_win_info
and inline it into init_and_make_win.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-data.h (tui_alloc_win_info): Don't declare.
* tui/tui-layout.c (init_and_make_win): Use "new" directly.
* tui/tui-data.c (tui_alloc_win_info): Remove.
Tom Tromey [Thu, 20 Jun 2019 21:22:53 +0000 (15:22 -0600)]
Don't check window type in tui_set_win_focus_to
This changes tui_set_win_focus_to so that it no longer checks the
window type. Instead, now tui_unhighlight_win also checks whether the
window can be highlighted.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-win.c (tui_set_win_focus_to): Don't check window type.
* tui/tui-wingeneral.c (tui_unhighlight_win): Check
can_highlight.
Tom Tromey [Tue, 18 Jun 2019 03:11:56 +0000 (21:11 -0600)]
Introduce tui_win_info::make_visible_with_new_height
This changes make_visible_with_new_height to be a method on
tui_win_info, letting us remove a spot that checks the window type.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-win.c (tui_source_window_base::update_tab_width): Call
make_visible_with_new_height method.
(tui_win_info::make_visible_with_new_height): New method.
(tui_source_window_base::do_make_visible_with_new_height)
(tui_data_window::do_make_visible_with_new_height)
(tui_cmd_window::do_make_visible_with_new_height): New methods.
(make_visible_with_new_height): Remove.
(tui_resize_all, tui_adjust_win_heights): Use
make_visible_with_new_height method.
* tui/tui-data.h (struct tui_win_info)
<do_make_visible_with_new_height, make_visible_with_new_height>:
New methods.
(struct tui_source_window_base, struct tui_data_window)
(struct tui_cmd_window) <do_make_visible_with_new_height>: New
methods.
Tom Tromey [Tue, 18 Jun 2019 00:48:00 +0000 (18:48 -0600)]
Introduce tui_win_info::update_tab_width
This introduces a new tui_win_info::update_tab_width method, and
changes the TUI to call it. This fixes another spot that was checking
the window type.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-win.c (tui_source_window_base::update_tab_width): New
method.
(update_tab_width): Call update_tab_width method.
* tui/tui-data.h (struct tui_win_info)
(struct tui_source_window_base) <update_tab_width>: New methods.
Tom Tromey [Mon, 17 Jun 2019 23:15:23 +0000 (17:15 -0600)]
Introduce enum tui_box
This adds a new enum to the TUI, replacing two #define constants,
providing better type safety.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-wingeneral.h (tui_make_window): Change type of "box_it"
parameter.
* tui/tui-wingeneral.c (tui_make_window): Change type of "box_it"
parameter.
(tui_gen_win_info::make_visible): Update.
* tui/tui-layout.c (init_and_make_win): Change type of "box_it"
parameter.
* tui/tui-data.h (enum tui_box): New enum.
(BOX_WINDOW, DONT_BOX_WINDOW): Remove defines.
Tom Tromey [Mon, 17 Jun 2019 21:45:14 +0000 (15:45 -0600)]
Remove some TUI static allocations
The TUI statically allocates the "execution_info" for the source and
disassembly windows. However, there's no deep reason to do this, and
this approach makes it harder to allow multiple such windows.
This patch removes the static data and changes the code to simply
allocate these windows as needed. This required pushing some code
into the tui_gen_win_info destructor, but that seems like a good idea
anyhow.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-layout.c (make_source_or_disasm_window): Always use
init_and_make_win for EXEC_INFO_WIN.
* tui/tui-data.h (struct tui_gen_win_info) <~tui_gen_win_info>: No
longer inline.
(struct tui_win_info) <~tui_win_info>: Inline.
(tui_source_exec_info_win_ptr, tui_disassem_exec_info_win_ptr):
Don't declare.
* tui/tui-data.c (source_win, disasm_win): Remove globals.
(tui_source_exec_info_win_ptr, tui_disassem_exec_info_win_ptr):
Remove.
(tui_initialize_static_data): Update.
(~tui_gen_win_info): Handle more cleanup here.
(~tui_source_window_base): Delete "execution_info".
(~tui_win_info): Move code to ~tui_gen_win_info; remove.
Tom Tromey [Mon, 17 Jun 2019 20:08:44 +0000 (14:08 -0600)]
Introduce tui_gen_win_info::reset method
This introduces the tui_gen_win_info::reset method and changes various
places to use it. This led to the realization that the can_highlight
member only needs to be set during construction, so this patch makes
that change as well. Finally, init_and_make_win is drastically
simplified.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-layout.c (make_command_window): Don't set
can_highlight.
(show_source_disasm_command): Call the reset method.
(show_data): Don't set can_highlight. Call the reset method.
(tui_gen_win_info::reset): Rename from init_gen_win_info
(init_and_make_win): Simplify. Return tui_gen_win_info.
(show_source_or_disasm_and_command): Call the reset method.
* tui/tui-data.h (struct tui_gen_win_info) <reset>: New method.
(struct tui_cmd_window): Set can_highlight.
Tom Tromey [Mon, 17 Jun 2019 19:19:15 +0000 (13:19 -0600)]
Move make_visible method to tui_gen_win_info
This moves the make_visible method from tui_win_info to its base
class, tui_gen_win_info. This allows the removal of another window
type check.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-wingeneral.c (tui_gen_win_info::make_visible): Rename
from make_visible.
(tui_make_visible, tui_make_invisible): Rewrite.
(tui_win_info::make_visible): Remove.
(tui_source_window_base::make_visible): Update.
* tui/tui-data.h (struct tui_gen_win_info) <make_visible>: New
method. Moved from...
(struct tui_win_info) <make_visible>: ...here.
Tom Tromey [Mon, 17 Jun 2019 18:56:39 +0000 (12:56 -0600)]
Remove tui_scroll_direction enum
The tui_scroll_direction enum is not really needed, because it's
simple to adapt the various scrolling methods to use the sign of their
argument as the direction in which to scroll.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-winsource.c
(tui_source_window_base::do_scroll_horizontal): Remove direction
parameter.
* tui/tui-windata.c (tui_data_window::do_scroll_vertical): Remove
direction parameter.
* tui/tui-win.c (tui_win_info::forward_scroll)
(tui_win_info::backward_scroll, tui_win_info::left_scroll)
(tui_win_info::right_scroll): Update.
* tui/tui-source.c (tui_source_window::do_scroll_vertical): Remove
direction parameter.
* tui/tui-disasm.c (tui_disasm_window::do_scroll_vertical): Remove
direction parameter.
* tui/tui-data.h (enum tui_scroll_direction): Remove.
(struct tui_win_info) <do_scroll_vertical, do_scroll_horizontal>:
Remove direction parameter.
(struct tui_source_window_base, struct tui_source_window)
(struct tui_disasm_window, struct tui_data_window)
(struct tui_cmd_window): Update.
Tom Tromey [Mon, 17 Jun 2019 18:51:29 +0000 (12:51 -0600)]
Change more TUI functions to take a tui_source_window_base
Several more TUI functions should only be called with a source or
disassembly window. This patch changes these functions so that this
can be caught at compile time.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-winsource.h (tui_set_exec_info_content)
(tui_show_exec_info_content, tui_erase_exec_info_content)
(tui_clear_exec_info_content, tui_update_exec_info): Change
argument to tui_source_window_base.
* tui/tui-winsource.c (tui_set_exec_info_content)
(tui_show_exec_info_content, tui_erase_exec_info_content)
(tui_clear_exec_info_content, tui_update_exec_info): Change
argument to tui_source_window_base.
Tom Tromey [Sun, 16 Jun 2019 23:51:34 +0000 (17:51 -0600)]
Change tui_set_exec_info_content to return void
tui_set_exec_info_content can't return an error, so change it to
return void instead.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-winsource.h (tui_set_exec_info_content): Return void.
* tui/tui-winsource.c (tui_set_exec_info_content): Return void.
Tom Tromey [Sun, 16 Jun 2019 23:51:01 +0000 (17:51 -0600)]
Remove NULL check from tui_set_exec_info_content
tui_set_exec_info_content ensures that the window's "execution_info"
is allocated, so the NULL check here is redundant and can be removed.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-winsource.c (tui_set_exec_info_content): Remove NULL
check.
Tom Tromey [Sun, 16 Jun 2019 23:49:41 +0000 (17:49 -0600)]
Change tui_alloc_source_buffer return type to void
tui_alloc_source_buffer can't actually fail, so change its return type
to void and update the callers.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-winsource.h (tui_alloc_source_buffer): Change return
type to void.
* tui/tui-winsource.c (tui_alloc_source_buffer): Change return
type to void.
* tui/tui-source.c (tui_set_source_content): Update.
* tui/tui-disasm.c (tui_set_disassem_content): Update.
Tom Tromey [Sun, 16 Jun 2019 22:35:31 +0000 (16:35 -0600)]
Introduce tui_gen_win_info::name method
This removes tui_win_name and replaces it with a method on
tui_gen_win_info, removing another spot that switched on window type.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-win.c (window_name_completer, tui_set_focus)
(tui_all_windows_info): Use name method.
* tui/tui-data.h (struct tui_gen_win_info)
(struct tui_source_window, struct tui_disasm_window)
(struct tui_data_window, struct tui_cmd_window) <name>: New
method.
(tui_win_name): Don't declare.
* tui/tui-data.c (tui_partial_win_by_name): Use name method.
(tui_win_name): Remove.
Tom Tromey [Sun, 16 Jun 2019 22:31:37 +0000 (16:31 -0600)]
Change tui_update_source_window for better type safety
tui_update_source_window can only be called for source and disassembly
windows, so change the argument type to enforce this at compile time.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-winsource.h (tui_update_source_window)
(tui_update_source_window_as_is): Change parameter type.
* tui/tui-winsource.c (tui_update_source_window): Change win_info
to be a tui_source_window_base.
(tui_update_source_window_as_is): Likewise.
* tui/tui-win.c (make_visible_with_new_height): Update.
Tom Tromey [Sun, 16 Jun 2019 22:28:12 +0000 (16:28 -0600)]
Introduce refresh_window method
This replaces tui_refresh_win with a new refresh_window method. This
removes another spot that was checking the window's type.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-winsource.c (tui_erase_source_content)
(tui_show_source_content, tui_show_exec_info_content)
(tui_erase_exec_info_content): Use refresh_window method.
* tui/tui-wingeneral.h (tui_refresh_win): Don't declare.
* tui/tui-wingeneral.c (tui_gen_win_info::refresh_window): Rename
from tui_refresh_win.
(tui_data_window::refresh_window): New method.
(tui_win_info::refresh, tui_source_window_base::refresh)
(tui_refresh_all): Use refresh_window method.
* tui/tui-stack.c (tui_show_locator_content): Call refresh_window
method.
* tui/tui-regs.c (tui_display_register): Call refresh_window
method.
* tui/tui-layout.c (show_source_disasm_command)
(show_source_or_disasm_and_command): Call refresh_window method.
* tui/tui-data.h (struct tui_gen_win_info)
(struct tui_data_window, struct tui_cmd_window) <refresh_window>:
New method.
Tom Tromey [Sun, 16 Jun 2019 22:03:54 +0000 (16:03 -0600)]
Derive tui_win_info from tui_gen_win_info
This changes tui_win_info to derive from tui_gen_win_info, rather than
having a tui_gen_win_info as a member. This removes a layer of member
access from the entire TUI, which is why this patch is so large. This
change will enable further removal of switches based on window type.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui.c (tui_rl_other_window, tui_enable)
(tui_is_window_visible, tui_get_command_dimension): Update.
* tui/tui-winsource.c (tui_update_source_window_as_is)
(tui_clear_source_content, tui_erase_source_content)
(tui_show_source_line, tui_source_window_base::refill)
(tui_source_window_base::do_scroll_horizontal)
(tui_source_window_base::set_is_exec_point_at)
(tui_update_breakpoint_info, tui_set_exec_info_content)
(tui_alloc_source_buffer, tui_line_is_displayed)
(tui_addr_is_displayed): Update.
* tui/tui-wingeneral.c (tui_unhighlight_win, tui_highlight_win)
(tui_check_and_display_highlight_if_needed)
(tui_win_info::make_visible, tui_win_info::refresh)
(tui_refresh_all): Update.
* tui/tui-windata.c (tui_first_data_item_displayed)
(tui_delete_data_content_windows, tui_erase_data_content)
(tui_display_all_data, tui_data_window::refresh_all)
(tui_check_data_values): Update.
* tui/tui-win.c (window_name_completer, tui_update_gdb_sizes)
(tui_set_win_focus_to, tui_win_info::forward_scroll)
(tui_win_info::backward_scroll, tui_refresh_all_win)
(tui_resize_all, tui_set_focus, tui_all_windows_info)
(update_tab_width, tui_set_win_height, tui_adjust_win_heights)
(tui_source_window_base::set_new_height)
(tui_data_window::set_new_height)
(make_invisible_and_set_new_height)
(make_visible_with_new_height, new_height_ok)
(parse_scrolling_args): Update.
* tui/tui-stack.c (tui_show_frame_info): Update.
* tui/tui-source.c (tui_set_source_content)
(tui_set_source_content_nil, tui_source_is_displayed)
(tui_source_window::do_scroll_vertical): Update.
* tui/tui-regs.c (tui_show_registers, tui_show_register_group)
(tui_display_registers_from, tui_display_reg_element_at_line)
(tui_check_register_values, tui_reg_command): Update.
* tui/tui-layout.c (tui_default_win_height)
(show_source_disasm_command, show_data, init_and_make_win)
(show_source_or_disasm_and_command): Update.
* tui/tui-io.c (update_cmdwin_start_line, tui_putc, tui_puts)
(tui_redisplay_readline, tui_mld_flush)
(tui_mld_erase_entire_line, tui_mld_getc, tui_cont_sig)
(tui_getc): Update.
* tui/tui-disasm.c (tui_set_disassem_content)
(tui_disasm_window::do_scroll_vertical): Update.
* tui/tui-data.h (struct tui_gen_win_info) <~tui_gen_win_info>:
Now virtual.
(struct tui_win_info): Derive from tui_gen_win_info.
<~tui_win_info>: Mark as override.
<generic>: Remove member.
* tui/tui-data.c (tui_cmd_window::clear_detail, tui_next_win)
(tui_prev_win, tui_partial_win_by_name, tui_win_info)
(~tui_data_window, ~tui_win_info)
(tui_free_all_source_wins_content): Update.
* tui/tui-command.c (tui_refresh_cmd_win): Update.
Tom Tromey [Sun, 16 Jun 2019 21:56:24 +0000 (15:56 -0600)]
Use new and delete for tui_gen_win_info
This changes tui_gen_win_info to be allocated with new and destroyed
with delete.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-layout.c (init_and_make_win): Use new.
* tui/tui-data.h (struct tui_gen_win_info): Add constructor,
destructor, initializers.
(tui_alloc_generic_win_info): Don't declare.
* tui/tui-data.c (_locator): Add argument to constructor.
(source_win, disasm_win): New globals.
(exec_info): Remove.
(tui_source_exec_info_win_ptr, tui_disassem_exec_info_win_ptr):
Update.
(tui_alloc_generic_win_info): Remove.
(init_content_element): Use new.
(tui_win_info::tui_win_info): Update.
(free_content_elements) <case DATA_WIN>: Use delete.
Tom Tromey [Sun, 16 Jun 2019 21:53:36 +0000 (15:53 -0600)]
Change tui_which_element::data_window to be a pointer
A coming patch will add a constructor to tui_gen_win_info. However,
because the tui_which_element union contains an object of this type,
first something must be done here in order to avoid having a union
with a member that has a constructor. This patch changes this element
to be a pointer instead.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-wingeneral.c (tui_refresh_win): Update.
* tui/tui-windata.c (tui_first_data_item_displayed)
(tui_delete_data_content_windows): Update.
* tui/tui-win.c (tui_data_window::set_new_height): Update.
* tui/tui-regs.c (tui_show_registers, tui_show_register_group)
(tui_display_registers_from, tui_check_register_values): Update.
* tui/tui-data.h (union tui_which_element) <data_window>: Now a
pointer.
* tui/tui-data.c (init_content_element): Update. Allocate the new
window.
(tui_free_data_content): Update.
(free_content_elements) <case DATA_WIN>: Free the window.
Tom Tromey [Sun, 16 Jun 2019 21:28:03 +0000 (15:28 -0600)]
Introduce set_highlight method
This introduces the tui_win_info::set_highlight method, and changes
the highlighting-related code to use bool rather than int.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-wingeneral.c (tui_unhighlight_win, tui_highlight_win):
Update.
* tui/tui-layout.c (make_command_window)
(show_source_disasm_command, show_data, init_and_make_win)
(show_source_or_disasm_and_command): Update.
* tui/tui-data.h (struct tui_win_info) <set_highlight>: New
method.
<can_highight, is_highlighted>: Now bool.
(tui_set_win_highlight): Don't declare.
* tui/tui-data.c (tui_set_win_highlight): Remove.
Tom Tromey [Sun, 16 Jun 2019 21:17:52 +0000 (15:17 -0600)]
Remove redundant check from make_visible
This removes a check of the window type from make_visible. This
function already checks that the window type is not CMD_WIN near the
top, so this condition can never be false.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-wingeneral.c (make_visible): Remove check of window
type.
Tom Tromey [Sun, 16 Jun 2019 21:16:52 +0000 (15:16 -0600)]
Introduce max_height method
This introduces the tui_win_info::max_height method and changes
new_height_ok to use it, rather than checking the window type
directly.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-win.c (tui_win_info::max_height)
(tui_cmd_window::max_height): New methods.
(new_height_ok): Call max_height.
* tui/tui-data.h (struct tui_win_info, struct tui_cmd_window)
<max_height>: New method.
Tom Tromey [Sun, 16 Jun 2019 21:12:25 +0000 (15:12 -0600)]
Introduce set_new_height method
This introduces tui_win_info::set_new_height and implements it in the
subclasses as appropriate. This removes another switch on the window
type.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-win.c (tui_source_window_base::set_new_height)
(tui_data_window::set_new_height): New methods.
(make_invisible_and_set_new_height): Call set_new_height method.
* tui/tui-data.h (struct tui_win_info)
(struct tui_source_window_base, struct tui_data_window)
<set_new_height>: New method.
Tom Tromey [Sun, 16 Jun 2019 20:52:08 +0000 (14:52 -0600)]
Introduce the refresh_all method
This introduces the tui_win_info::refresh_all method and implements it
as needed in subclasses. The name and comment are a bit of a guess on
my part. The main benefit of this patch is removing another switch on
the type of the window.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui.c (tui_rl_other_window): Call the refresh_all method.
* tui/tui-windata.c (tui_data_window::refresh_all): Rename from
tui_refresh_data_win.
* tui/tui-win.c (tui_source_window_base::refresh_all): New
method.
(tui_refresh_all_win): Call the refresh_all method.
(tui_set_focus): Likewise.
* tui/tui-data.h (struct tui_win_info) <refresh_all>: New method.
(struct tui_source_window_base, struct tui_data_window) <refresh>:
Likewise.
Tom Tromey [Sun, 16 Jun 2019 20:43:56 +0000 (14:43 -0600)]
Introduce two TUI source window methods
This adds two methods to the TUI source window. These are just
renamings of existing functions. It also changes the source window
list to have a more precise type, letting the code be more type-safe.
This will be useful again later.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-winsource.h (tui_refill_source_window)
(tui_set_is_exec_point_at): Don't declare.
* tui/tui-winsource.c (tui_update_source_windows_with_addr)
(tui_source_window_base::refill): Rename from
tui_refill_source_window.
(tui_source_window_base::do_scroll_horizontal): Update.
(tui_source_window_base::set_is_exec_point_at): Rename from
tui_set_is_exec_point_at.
(tui_update_all_breakpoint_info): Update.
* tui/tui-stack.c (tui_show_frame_info): Update.
* tui/tui-layout.c (show_data): Add cast.
* tui/tui-hooks.c (tui_redisplay_source): Call refill method.
* tui/tui-data.h (struct tui_source_window_base) <refill,
set_is_exec_point_at>: New methods.
(tui_source_windows, tui_add_to_source_windows): Update types.
(tui_add_to_source_windows): Remove redundant declaration.
* tui/tui-data.c (source_windows): Store tui_source_window_base.
(tui_source_windows): Change return type.
(tui_clear_source_windows_detail): Update.
(tui_add_to_source_windows): Change type of parameter.
(tui_free_all_source_wins_content): Update.
Tom Tromey [Sun, 16 Jun 2019 20:27:28 +0000 (14:27 -0600)]
Introduce the refresh method
This adds tui_win_info::refresh and updates tui_source_window_base to
implement it as well. This lets us simplify tui_refresh_all, removing
a check of the window type.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-wingeneral.c (tui_win_info::refresh)
(tui_source_window_base::refresh): New methods.
(tui_refresh_all): Call the refresh method.
* tui/tui-data.h (struct tui_win_info)
(struct tui_source_window_base) <refresh>: New method.
Tom Tromey [Sun, 16 Jun 2019 20:24:42 +0000 (14:24 -0600)]
Use bool for visibility
This changes the visibility-related functions and data members in the
TUI to use bool rather than int.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui.h (tui_is_window_visible): Return bool.
* tui/tui.c (tui_is_window_visible): Return bool.
* tui/tui-wingeneral.c (tui_make_window, make_visible)
(tui_make_visible, tui_make_invisible)
(tui_win_info::make_visible)
(tui_source_window_base::make_visible, make_all_visible)
(tui_make_all_visible, tui_make_all_invisible): Update.
* tui/tui-windata.c (tui_delete_data_content_windows): Update.
* tui/tui-data.h (struct tui_gen_win_info) <is_visible>: Now
bool.
(struct tui_win_info, struct tui_source_window_base)
(struct tui_cmd_window) <make_visible>: Change parameter to bool.
* tui/tui-data.c (tui_init_generic_part): Update.
Tom Tromey [Sun, 16 Jun 2019 20:22:21 +0000 (14:22 -0600)]
Introduce make_visible method
This introduceds the make_visible to tui_win_info and overrides it in
subclasses as appropriate. This allows the removal of the
tui_win_is_source_type, as it is no longer used.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-wingeneral.c (tui_win_info::make_visible)
(tui_source_window_base::make_visible): New methods.
(make_all_visible): Make method call.
* tui/tui-data.h (struct tui_win_info) <make_visible>: New method.
(struct tui_source_window_base, struct tui_cmd_window): Override
make_visible.
(tui_win_is_source_type): Don't declare.
* tui/tui-data.c (tui_win_is_source_type): Remove.
Tom Tromey [Sun, 16 Jun 2019 20:14:10 +0000 (14:14 -0600)]
Remove an unneeded NULL check
show_source_or_disasm_and_command will either create or reset the
source window, so the final NULL check is not necessary. This patch
removes it.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-layout.c (show_source_or_disasm_and_command): Remove
NULL check.
Tom Tromey [Sun, 16 Jun 2019 20:06:18 +0000 (14:06 -0600)]
Inline constructors and initializers
This inlines the constructors and initializers for tui_cmd_window and
tui_data_window. This makes the code a bit simpler.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-data.h (struct tui_data_window, struct tui_cmd_window):
Inline constructor. Add initializers for members.
* tui/tui-data.c (tui_data_window, tui_cmd_window): Remove
constructors; now inline in class.
Tom Tromey [Sun, 16 Jun 2019 20:04:11 +0000 (14:04 -0600)]
Change tui_data_window::display_regs to bool
This changes tui_data_window::display_regs to bool and updates the
uses.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-regs.c (tui_show_registers): Update.
* tui/tui-data.h (struct tui_data_window) <display_regs>: Now
bool.
* tui/tui-data.c (tui_data_window::clear_detail)
(tui_data_window): Update.
Tom Tromey [Sun, 16 Jun 2019 20:03:04 +0000 (14:03 -0600)]
Remove struct tui_data_info
Like the previous two patches, this removes struct tui_data_info in
favor of inlining its contents into tui_data_window. This was the
last member of the tui_win_info detail union, so that is removed.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-windata.c (tui_display_all_data)
(tui_display_data_from_line, tui_display_data_from)
(tui_check_data_values, tui_data_window::do_scroll_vertical):
Update.
* tui/tui-regs.c (tui_last_regs_line_no)
(tui_line_from_reg_element_no, tui_first_reg_element_no_inline)
(tui_show_registers, tui_show_register_group)
(tui_display_registers_from, tui_display_reg_element_at_line)
(tui_display_registers_from_line, tui_check_register_values)
(tui_reg_next, tui_reg_prev): Update.
* tui/tui-layout.c (tui_set_layout, show_data): Update.
* tui/tui-data.h (struct tui_data_info): Remove. Move contents to
tui_data_window.
(struct tui_win_info) <detail>: Remove. Add new fields from
tui_data_info.
(TUI_DATA_WIN): Add cast.
* tui/tui-data.c (tui_data_window::clear_detail, tui_data_window)
(~tui_data_window): Simplify.
Tom Tromey [Sun, 16 Jun 2019 19:55:51 +0000 (13:55 -0600)]
Remove struct tui_command_info
Like the previous patch, this removes tui_command_info in favor of
putting it elements directly into tui_cmd_window.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-layout.c (show_source_disasm_command)
(show_source_or_disasm_and_command): Update.
* tui/tui-io.c (update_cmdwin_start_line)
(tui_redisplay_readline): Update.
* tui/tui-data.h (struct tui_command_info): Remove.
(struct tui_win_info) <detail>: Remove command_info member.
(struct tui_data_window) <start_line>: New member, from
tui_command_info.
(TUI_CMD_WIN): Add casts.
Tom Tromey [Sun, 16 Jun 2019 19:46:18 +0000 (13:46 -0600)]
Remove struct tui_source_info
The tui_source_info struct is used as a member of the "detail" union
in tui_win_info, and this member of the union is only used by source
and disassembly windows. This patch removes tui_source_info and moves
its members directly to tui_source_window_base. This simplifies the
code by removing a layer of references from many places. In a few
spots, a new cast was needed, but most of these will be removed by the
end of the series.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-winsource.c (tui_update_source_window)
(tui_refill_source_window)
(tui_source_window_base::do_scroll_horizontal)
(tui_update_breakpoint_info, tui_set_exec_info_content)
(tui_show_exec_info_content, tui_erase_exec_info_content)
(tui_clear_exec_info_content): Update.
* tui/tui-wingeneral.c (make_all_visible, tui_refresh_all):
Update.
* tui/tui-win.c (make_invisible_and_set_new_height)
(make_visible_with_new_height): Update.
* tui/tui-source.c (tui_set_source_content)
(tui_show_symtab_source): Update.
* tui/tui-layout.c (extract_display_start_addr)
(show_source_disasm_command, show_data)
(make_source_or_disasm_window)
(show_source_or_disasm_and_command): Update.
* tui/tui-disasm.c (tui_set_disassem_content): Simplify.
(tui_disasm_window::do_scroll_vertical): Remove shadowing
"gdbarch".
* tui/tui-data.h (struct tui_source_info): Remove. Move contents
to tui_source_window_base.
(struct tui_win_info) <detail>: Remove source_info member.
(struct tui_source_window_base) <has_locator>: Inline.
Move contents from tui_source_info; rename has_locator member to
m_has_locator.
(TUI_SRC_WIN, TUI_DISASM_WIN): Add casts.
* tui/tui-data.c (tui_source_window_base::has_locator): Move to
header file.
(tui_source_window_base::clear_detail, ~tui_source_window_base):
Simplify.
(tui_free_all_source_wins_content): Cast to
tui_source_window_base.
Tom Tromey [Sun, 16 Jun 2019 17:37:49 +0000 (11:37 -0600)]
Introduce has_locator method
This changes tui_win_has_locator to be a method on tui_win_info, and
changes the locator code to use bool rather than int.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-win.c (make_invisible_and_set_new_height)
(make_visible_with_new_height): Call has_locator method.
* tui/tui-layout.c (show_source_disasm_command, show_data)
(show_source_or_disasm_and_command): Update for bool change.
* tui/tui-data.h (struct tui_source_info) <has_locator>: Now bool.
(tui_win_info) <has_locator>: New method.
(struct tui_source_window_base) <has_locator>: New method.
(tui_win_has_locator): Don't declare.
* tui/tui-data.c (tui_source_window_base::has_locator): Rename
from tui_win_has_locator.
(tui_source_window_base): Use false, not FALSE.
Tom Tromey [Sun, 16 Jun 2019 17:29:45 +0000 (11:29 -0600)]
Remove tui_clear_win_detail
An earlier patch changed the context of the sole call to
tui_clear_win_detail to make it clear that this can never be called
with a NULL window pointer. So, remove tui_clear_win_detail in favor
of calling the method directly.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-data.h (tui_clear_win_detail): Don't declare.
* tui/tui-data.c (tui_clear_source_windows_detail): Call the
clear_detail method directly.
(tui_clear_win_detail): Remove.
Tom Tromey [Sun, 16 Jun 2019 22:31:56 +0000 (16:31 -0600)]
Don't use TUI_DISASM_WIN in tui_disasm_window method
The previous patch made it clear that the diassembly window scrolling
method was written to assume there could only ever be a single
disassembly window. This changes that spot to use "this" rather than
the TUI_DISASM_WIN global.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-disasm.c (tui_disasm_window::do_scroll_vertical): Use
"this", not TUI_DISASM_WIN.
Tom Tromey [Sun, 16 Jun 2019 17:22:38 +0000 (11:22 -0600)]
Introduce methods for scrolling
This changes the TUI to use virtual methods on the various window
types for scrolling. Window-specific functions for this purpose are
renamed to be methods, and the generic tui_scroll function is removed
as it is no longer called.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-winsource.h (tui_horizontal_source_scroll): Don't
declare.
* tui/tui-winsource.c
(tui_source_window_base::do_scroll_horizontal): Rename from
tui_horizontal_source_scroll.
* tui/tui-windata.h (tui_vertical_data_scroll): Don't declare.
* tui/tui-windata.c (tui_data_window::do_scroll_vertical): Rename
from tui_vertical_data_scroll.
* tui/tui-win.h (tui_scroll): Don't declare.
* tui/tui-win.c (tui_win_info::forward_scroll)
(tui_win_info::backward_scroll, tui_win_info::left_scroll)
(tui_win_info::right_scroll): Rename and update.
(tui_scroll_forward_command, tui_scroll_backward_command)
(tui_scroll_left_command, tui_scroll_right_command): Update.
(tui_scroll): Remove.
* tui/tui-source.h: Don't declare tui_vertical_source_scroll.
* tui/tui-source.c (tui_source_window::do_scroll_vertical): Rename
from tui_vertical_source_scroll.
* tui/tui-disasm.h (tui_vertical_disassem_scroll): Don't declare.
* tui/tui-disasm.c (tui_disasm_window::do_scroll_vertical): Rename
from tui_vertical_disassem_scroll.
* tui/tui-data.h (struct tui_win_info) <do_scroll_vertical,
do_scroll_horizontal>: New methods.
<forward_scroll, backward_scroll, left_scroll, right_scroll>:
Likewise.
(struct tui_source_window_base): Add do_scroll_horizontal.
(struct tui_source_window, struct tui_disasm_window): Add
do_scroll_vertical.
(struct tui_data_window, struct tui_cmd_window): Add
do_scroll_horizontal and do_scroll_vertical.
* tui/tui-command.c (tui_dispatch_ctrl_char): Use method calls.
Tom Tromey [Sun, 16 Jun 2019 17:08:43 +0000 (11:08 -0600)]
Create tui_disasm_window
This introduces the new tui_disasm_window class, which represents a
disassembly window. It shares a lot of behavior with the source
window, so a new tui_source_window_base class is also created.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-data.h (struct tui_source_window_base): New struct.
(struct tui_source_window): Derive from tui_source_window_base.
(struct tui_disasm_window): New struct.
* tui/tui-data.c (tui_source_window_base::clear_detail): Rename
from tui_source_window::clear_detail.
(tui_source_window_base): Rename from tui_source_window.
(~tui_source_window_base): Rename from ~tui_source_window.
(tui_alloc_win_info): Create a tui_disasm_window.
Tom Tromey [Sun, 16 Jun 2019 16:52:27 +0000 (10:52 -0600)]
Split the tui_win_info destructor
This patch adds destructors to tui_source_window and tui_data_window,
and splits ~tui_win_info as appropriate.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-data.h (struct tui_source_window)
(struct tui_data_window): Declare destructors.
* tui/tui-data.c (~tui_source_window, ~tui_data_window): New
destructors.
(tui_win_info): Simplify.
Tom Tromey [Sun, 16 Jun 2019 16:49:45 +0000 (10:49 -0600)]
Remove tui_list
This removes the tui_list type in favor of a std::vector.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-winsource.c (tui_display_main)
(tui_update_source_windows_with_addr)
(tui_update_all_breakpoint_info): Update.
* tui/tui-win.c (tui_resize_all, tui_adjust_win_heights)
(new_height_ok, parse_scrolling_args): Update.
* tui/tui-stack.c (tui_show_frame_info): Update.
* tui/tui-data.h (struct tui_list): Remove.
(tui_source_windows): Return a reference to a std::vector.
* tui/tui-data.c (source_windows): Now a std::vector.
(tui_source_windows): Change return type.
(tui_clear_source_windows): Rewrite.
(tui_clear_source_windows_detail, tui_add_to_source_windows)
(tui_free_all_source_wins_content): Rewrite.
Tom Tromey [Sun, 16 Jun 2019 16:42:09 +0000 (10:42 -0600)]
Introduce tui_win_info::clear_detail method
This introduces the clear_detail method and updates the various
subclasses of tui_win_info to implement it. A subsequent patch will
remove tui_clear_win_detail, but that isn't done for now because at
this point it isn't readily obvious that the NULL check is not needed.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-data.h (struct tui_win_info, struct tui_source_window)
(struct tui_data_window, struct tui_cmd_window): Declare
clear_detail method.
* tui/tui-data.c (tui_source_window::clear_detail)
(tui_cmd_window::clear_detail, tui_data_window::clear_detail): New
methods.
(tui_clear_win_detail): Simplify.
Tom Tromey [Sun, 16 Jun 2019 16:37:25 +0000 (10:37 -0600)]
Simplify source and disassembly window creation
Similar to the previous patch, neither make_source_window nor
make_disasm_window could be called when *win_info_ptr was non-NULL.
This patch simplifies the functions by removing the parameter and
having them return their results directly.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-layout.c (make_source_window, make_disasm_window)
(make_source_or_disasm_window): Remove win_info_ptr parameter.
Return the new window.
(show_source_disasm_command, show_data)
(show_source_or_disasm_and_command): Update.
Tom Tromey [Sun, 16 Jun 2019 16:32:35 +0000 (10:32 -0600)]
Simplify command window creation
make_command_window is never called when *win_info_ptr is non-NULL, so
this patch simplifies the function by removing the parameter and
having it return its result directly. This in turn makes it more
obvious that a NULL check in show_source_disasm_command can be
removed.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-layout.c (make_command_window): Remove win_info_ptr
parameter. Return the new window.
(show_source_disasm_command): Update and remove NULL check.
(show_source_or_disasm_and_command): Update.
Tom Tromey [Sun, 16 Jun 2019 16:23:10 +0000 (10:23 -0600)]
Remove an unnecessary NULL check from the TUI
In init_and_make_win, opaque_win_info can't be NULL after a new window
is allocated. This patch removes an unnecessary NULL check.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-layout.c (init_and_make_win): Remove NULL check.
Tom Tromey [Sun, 16 Jun 2019 16:13:13 +0000 (10:13 -0600)]
Create subclasses for different window types
This changes the TUI so that each different major window type has its
own subclass.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-data.h (struct tui_win_info): Make constructor
protected. Make destructor virtual. Add initializers.
(tui_source_window, tui_data_window, tui_cmd_window): New
classes.
* tui/tui-data.c (tui_win_info): Rename from init_win_info. Now a
constructor. Add "type" parameter.
(tui_source_window, tui_data_window, tui_cmd_window): New
constructors.
(tui_alloc_win_info): Instantiate the appropriate subclass.
Tom Tromey [Sun, 16 Jun 2019 15:43:21 +0000 (09:43 -0600)]
Add destructor to tui_win_info
This changes tui_free_window into a destructor for tui_free_window and
then updates the users.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-win.c (tui_resize_all): Use delete.
* tui/tui-data.h (struct tui_win_info) <~tui_win_info>: Declare
destructor.
(tui_free_window): Don't declare.
* tui/tui-data.c (~tui_win_info): Rename from tui_free_window.
Update.
Tom Tromey [Sun, 16 Jun 2019 15:41:06 +0000 (09:41 -0600)]
Use new and delete for TUI windows
This changes tui_win_info to use new and delete, rather than XNEW and
xfree.
gdb/ChangeLog
2019-06-25 Tom Tromey <tom@tromey.com>
* tui/tui-data.h (struct tui_win_info): Add constructor.
* tui/tui-data.c (tui_alloc_win_info): Use new.
(tui_free_window): Use delete.
Jan Beulich [Tue, 25 Jun 2019 10:02:23 +0000 (12:02 +0200)]
ld/plugins: avoid shadowing a C library symbol
With my (oldish) gcc/glibc combination I'm seeing
.../ld/plugin.c: In function \91get_lto_kind\92:
.../ld/plugin.c:664: error: declaration of \91index\92 shadows a global declaration
/usr/include/string.h:487: error: shadowed declaration is here
.../ld/plugin.c: In function \91get_lto_resolution\92:
.../ld/plugin.c:685: error: declaration of \91index\92 shadows a global declaration
/usr/include/string.h:487: error: shadowed declaration is here
.../ld/plugin.c: In function \91get_lto_visibility\92:
.../ld/plugin.c:711: error: declaration of \91index\92 shadows a global declaration
/usr/include/string.h:487: error: shadowed declaration is here
Jan Beulich [Tue, 25 Jun 2019 10:01:50 +0000 (12:01 +0200)]
bfd/elf-properties: avoid shadowing a C library symbol
With my (oldish) gcc/glibc combination I'm seeing
.../bfd/elf-properties.c: In function \91elf_find_and_remove_property\92:
.../bfd/elf-properties.c:244: error: declaration of \91remove\92 shadows a global declaration
/usr/include/stdio.h:157: error: shadowed declaration is here
Jan Beulich [Tue, 25 Jun 2019 09:09:22 +0000 (11:09 +0200)]
readelf: avoid shadowing a libiberty symbol
With my (oldish) gcc I'm seeing
.../binutils/readelf.c: In function \91dump_ctf_indent_lines\92:
.../binutils/readelf.c:13851: error: declaration of \91spaces\92 shadows a global declaration
.../binutils/../include/libiberty.h:253: error: shadowed declaration is here
Jan Beulich [Tue, 25 Jun 2019 09:08:53 +0000 (11:08 +0200)]
objdump: avoid shadowing a libiberty symbol
With my (oldish) gcc I'm seeing
.../binutils/objdump.c: In function \91dump_ctf_indent_lines\92:
.../binutils/objdump.c:3210: error: declaration of \91spaces\92 shadows a global declaration
.../binutils/../include/libiberty.h:253: error: shadowed declaration is here
Jan Beulich [Tue, 25 Jun 2019 07:41:33 +0000 (09:41 +0200)]
x86: correct / adjust debug printing
For quite some time we've been using combinations of bits for
specifying various registers in operands and templates. I think it was
Alan who had indicated that likely the debug printing would need
adjustment as a result. Here we go.
Accumulator handling for GPRs gets changed to match that for FPU regs.
For this to work, OPERAND_TYPE_ACC{32,64} get repurposed, with their
original uses replaced by direct checks of the two bits of interest,
which is cheaper than operand_type_equal() invocations.
For SIMD registers nothing similar appears to be needed, as respective
operands get stripped from the (copy of the) template before pt() is
reached.
The type change on pi() is to silence a compiler diagnostic. Arguably
its other parameter could also be const-qualified.
Jan Beulich [Tue, 25 Jun 2019 07:40:44 +0000 (09:40 +0200)]
x86: document certain command line options as "dangerous"
Errata BT36, BT41, and BT230 mean that gas may, when using one ofthese
options, produce code that causes #UD on (at least) SandyBridge systems.
Jan Beulich [Tue, 25 Jun 2019 07:28:33 +0000 (09:28 +0200)]
x86: drop dqa_mode
I assume this mode was needed when EVEX.W handling wasn't really correct
yet for other than 64-bit mode. It's clearly not needed anymore. Its
elimination also allows dropping the EVEX.W split of VCVT{,U}SI2SS. (For
the record, the dropped mode would have been wrong if used in any table
entry not already guaranteeing EVEX.W=1.)
Jan Beulich [Tue, 25 Jun 2019 07:27:49 +0000 (09:27 +0200)]
x86: don't open code is_any_vex_encoding()
Jan Beulich [Tue, 25 Jun 2019 07:27:05 +0000 (09:27 +0200)]
x86: simplify OP_I64()
The only meaningful difference from OP_I() is the handling of the
VEX.W=1 case in 64-bit mode for bytemode being v_mode. Funnel
everything else into OP_I(), and drop no longer needed local
variables.
Jan Beulich [Tue, 25 Jun 2019 07:25:26 +0000 (09:25 +0200)]
x86: fix (dis)assembly of certain SSE2 insns in 16-bit mode
MOVNTI was wrongly assembled with a 66h prefix. Add IgnoreSize to
address this. It and the scalar to/from integer conversion insns also
were also wrongly using Ev / Gv, leading to 16-bit register names being
printed when 32-bit ones were meant.
Clone the 32-bit SSE2 test to cover both assembler and disassembler.
Jan Beulich [Tue, 25 Jun 2019 07:23:48 +0000 (09:23 +0200)]
x86-64: also optimize ANDQ with immediate fitting in 7 bits
The same reasoning applies here as did/does for immediates fitting in
31 bits.
Jan Beulich [Tue, 25 Jun 2019 07:22:38 +0000 (09:22 +0200)]
x86: add CVT{,T}PS2PI cases to xmmwords test
I've (not so) recently noticed this further pair which should be tested
here.
Alan Modra [Tue, 25 Jun 2019 07:26:16 +0000 (16:56 +0930)]
Fix logical expression in last commit
* config/tc-ppc.c (ppc_handle_align): Add parentheses.
Alan Modra [Tue, 25 Jun 2019 00:42:58 +0000 (10:12 +0930)]
PowerPC nops
This patch corrects ppc rs_align_code handling to choose the alignment
nops based on the machine in force at the alignment directive rather
than the machine at the end of file.
* config/tc-ppc.h (ppc_nop_select): Declare.
(NOP_OPCODE): Define.
* config/tc-ppc.c (ppc_elf_end, ppc_xcoff_end): Zero ppc_cpu.
(ppc_nop_encoding_for_rs_align_code): New enum.
(ppc_nop_select): New function.
(ppc_handle_align): Don't use ppc_cpu here. Get nop type from frag.
* testsuite/gas/ppc/groupnop.d,
* testsuite/gas/ppc/groupnop.s: New test.
* testsuite/gas/ppc/ppc.exp: Run it.
GDB Administrator [Tue, 25 Jun 2019 00:00:35 +0000 (00:00 +0000)]
Automatic date update in version.in
Jim Wilson [Mon, 24 Jun 2019 20:50:10 +0000 (13:50 -0700)]
RISC-V: Enable lui relaxation for CODE and MERGE sections.
2019-06-24 Ilia Diachkov <ilia.diachkov@optimitech.com>
bfd/
* elfnn-riscv.c (_bfd_riscv_relax_lui): Delete early exit when
SEC_MERGE or SEC_CODE flags are set.
(_bfd_riscv_relax_section): New local symtype. Set sym_sec and
symtype consistently. Don't include sec_addr (sym_sec) in symval.
Add check for SEC_INFO_TYPE_MERGE and call _bfd_merged_section_offset.
Add sec_addr (sym_sec) after handling merge sections.
H.J. Lu [Mon, 24 Jun 2019 18:08:40 +0000 (11:08 -0700)]
elf: Remove the property after reporting its removal
commit
d2ef37ebd9f771d06edf1fdea37970f60b242b2d
Author: H.J. Lu <hjl.tools@gmail.com>
Date: Fri Dec 7 08:30:30 2018 -0800
elf: Report property change when merging properties
failed to remove the property after reporting it has been removed. This
patch corrects it.
bfd/
PR ld/24721
* elf-properties.c (elf_merge_gnu_property_list): Remove the
property after reporting property removal.
ld/
PR ld/24721
* testsuite/ld-x86-64/x86-64.exp: Run PR ld/24721 tests.
* testsuite/ld-x86-64/pr24721-x32.d: New file.
* testsuite/ld-x86-64/pr24721.d: Likewise.
* testsuite/ld-x86-64/pr24721.map: Likewise.
* testsuite/ld-x86-64/pr24721a.s: Likewise.
* testsuite/ld-x86-64/pr24721b.s: Likewise.
Tom de Vries [Mon, 24 Jun 2019 16:26:10 +0000 (18:26 +0200)]
[gdb/testsuite] Fix label reference in implptr-64bit.exp
When running gdb.dwarf2/implptr-64bit.exp with board cc-with-dwz-m, we run into:
...
dwz: dwz.c:2363: checksum_die: \
Assertion `\
((!op_multifile && !rd_multifile && !fi_multifile) || cu != die_cu (ref)) \
&& (!op_multifile || cu->cu_chunk == die_cu (ref)->cu_chunk)' failed.
cc-with-tweaks.sh: line 218: 13030 Aborted \
$DWZ -m ${output_file}.dwz "$output_file" ${output_file}.alt > /dev/null
...
In other words, PR dwz/24170.
The trigger for the dwz PR is when intra-CU references are encoded using
section-relative encoding DW_FORM_ref_addr, but could have been encoded using
CU-relative encoding DW_FORM_ref4.
Fix the intra-CU '%' label reference in implptr-64bit.exp.
Tested on x86_64-linux.
gdb/testsuite/ChangeLog:
2019-06-24 Tom de Vries <tdevries@suse.de>
* gdb.dwarf2/implptr-64bit.exp: Fix intra-CU '%' label reference.
Tom de Vries [Mon, 24 Jun 2019 11:29:23 +0000 (13:29 +0200)]
[gdb/testsuite] Fix DW_AT_decl_file in gdb.trace tests
When running gdb.trace/{entry-values.exp,unavailable-dwarf-piece.exp} with
board cc-with-dwz, we run into two failures related to the DW_AT_decl_file
attribute:
- The encoding DW_FOR_sdata is used for DW_AT_decl_file, while the attribute
is required to have a an "unsigned integer constant" value.
- The DW_AT_decl_file attributes refer to a file with index one, while there's
no such file.
Fix this by using DW_FOR_udata and the value 0, meaning "no file specified".
Tested on x86_64-linux with board native-gdbserver.
gdb/testsuite/ChangeLog:
2019-06-24 Tom de Vries <tdevries@suse.de>
* gdb.trace/entry-values.exp: Use DW_FORM_udata instead of
DW_FOR_sdata for DW_AT_decl_file. Use 0 for DW_AT_decl_file.
* gdb.trace/unavailable-dwarf-piece.exp: Same.
Tom de Vries [Mon, 24 Jun 2019 10:20:39 +0000 (12:20 +0200)]
[gdb/testsuite] Fix inter-cu refs in inlined_subroutine-inheritance.exp
When running gdb.dwarf2/inlined_subroutine-inheritance.exp with board
cc-with-dwz, we run into:
...
dwz: inlined_subroutine-inheritance: Couldn't find DIE referenced by \
DW_AT_abstract_origin
...
The problem is that the DW_AT_abstract_origin attributes refer to DIEs in
other CUs, while the references are encoded using the cu-relative encoding
DW_FORM_ref4.
Fix this by forcing the references to use DW_FORM_ref_addr.
Tested on x86_64-linux.
Tested with commit
c24bdb023c "Introduce dwarf2_cu::get_builder" reverted,
and verified that the test-case fails in the same way before and after this
patch.
gdb/testsuite/ChangeLog:
2019-06-24 Tom de Vries <tdevries@suse.de>
* gdb.dwarf2/inlined_subroutine-inheritance.exp:
GDB Administrator [Mon, 24 Jun 2019 00:00:31 +0000 (00:00 +0000)]
Automatic date update in version.in
Alan Modra [Sun, 23 Jun 2019 02:58:39 +0000 (12:28 +0930)]
PR24704, Internal error building skiboot for powerpc64-linux-gnu
While the skiboot linker script bears some culpability in this PR,
it's also true that the GOT indirect to GOT relative optimisation for
16-bit offsets isn't safe. At least, it isn't safe to remove the GOT
entry based on distance between the GOT pointer and symbol calculated
from the preliminary layout. So this patch removes that optimisation,
and reduces the range allowed for 32-bit and 34-bit offsets.
PR 24704
bfd/
* elf64-ppc.c (R_PPC64_GOT16_DS): Don't set has_gotrel.
(ppc64_elf_edit_toc): Don't remove R_PPC64_GOT16_DS got entries.
Reduce range of offsets allowed for other GOT relocs.
ld/
* testsuite/ld-powerpc/elfv2exe.d: Update.
* testsuite/ld-powerpc/elfv2so.d: Update.
Alan Modra [Sun, 23 Jun 2019 02:40:02 +0000 (12:10 +0930)]
PR24689 again, string table corruption
Depending on optimisation level and gcc version, git commit
890f750a3b
introduces a false positive warning that i_shdrp may be used
uninitialized.
PR 24689
* elfcode.h (elf_object_p): Warning fix.
GDB Administrator [Sun, 23 Jun 2019 00:00:55 +0000 (00:00 +0000)]
Automatic date update in version.in
Tom Tromey [Sat, 22 Jun 2019 17:50:19 +0000 (11:50 -0600)]
Remove tui_first_data_element_no_in_line
tui_first_data_element_no_in_line is never used. This patch removes
it. Tested by rebuilding, and by grep.
gdb/ChangeLog
2019-06-22 Tom Tromey <tom@tromey.com>
* tui/tui-windata.h (tui_first_data_element_no_in_line): Don't
declare.
* tui/tui-windata.c (tui_first_data_element_no_in_line): Remove.
Tom Tromey [Sun, 16 Jun 2019 16:04:49 +0000 (10:04 -0600)]
Remove two unused functions from the TUI
This removes two unused functions from the TUI. According to
"git grep -G", they have never been used.
gdb/ChangeLog
2019-06-22 Tom Tromey <tom@tromey.com>
* tui/tui-data.h (tui_del_window, tui_del_data_windows): Don't
declare.
* tui/tui-data.c (tui_del_window, tui_del_data_windows): Remove.
Tom de Vries [Sat, 22 Jun 2019 08:59:37 +0000 (10:59 +0200)]
[gdb] Fix s390x -m31 build
When building gdb on s390x with -m31, we run into this Wformat
warning (which Werror turns into an error):
...
gdb/dwarf2read.c: In function \
'void create_addrmap_from_aranges(dwarf2_per_objfile*, \
dwarf2_section_info*)':
gdb/dwarf2read.c:3277:22: error: format '%zu' expects argument of type \
'size_t', but argument 3 has type 'int' [-Werror=format=]
warning (_("Section .debug_aranges in %s entry at offset %zu "
...
The Wformat warning is triggered in this statement:
...
warning (_("Section .debug_aranges in %s entry at offset %zu "
"length %s exceeds section length %s, "
"ignoring .debug_aranges."),
objfile_name (objfile), entry_addr - section->buffer,
plongest (bytes_read + entry_length),
pulongest (section->size));
...
where 'entry_addr - section->buffer' is of type ptrdiff_t and '%zu' prints an
unsigned with the same size as size_t/ssize_t.
On s390x with -m31, we have:
- size_t : unsigned long int (32-bit)
- ptrdiff_t: int (32-bit)
Wformat warns against this because even though long int and int have the same
size, the types are not compatible.
[ The Wformat warning is to similar to what we would get for x86_64 -m32
(where long and int are also the same size) and:
...
int i;
printf ("%ld", i);
... ]
Fix this by using '%s' and plongest instead of '%zu' to print ptrdiff_t.
Build and reg-tested on x86_64.
gdb/ChangeLog:
2019-06-22 Tom de Vries <tdevries@suse.de>
* dwarf2read.c (create_addrmap_from_aranges)
(read_debug_names_from_section): Print ptrdiff_t using '%s' and plongest
instead of '%zu'.
GDB Administrator [Sat, 22 Jun 2019 00:00:20 +0000 (00:00 +0000)]
Automatic date update in version.in
Tom de Vries [Fri, 21 Jun 2019 21:56:18 +0000 (23:56 +0200)]
[gdb/testsuite] Compile index-cache.c with -Wl,--build-id
When testing gdb.base/index-cache.exp using a gcc build without
--enable-linker-build-id we get:
...
FAIL: gdb.base/index-cache.exp: \
test_cache_enabled_miss: at least one file was created
FAIL: gdb.base/index-cache.exp: \
test_cache_enabled_miss: couldn't get executable build id
FAIL: gdb.base/index-cache.exp: \
test_cache_enabled_hit: check index-cache stats
...
With "set debug index-cache on" we find:
...
(gdb) file index-cache
Reading symbols from index-cache...
index cache: objfile index-cache has no build id
...
The problem is that a build-id is required for the index-cache functionality.
Fix this by compiling index-cache.c with -Wl,--build-id.
Tested on x86_64-linux.
gdb/testsuite/ChangeLog:
2019-06-21 Tom de Vries <tdevries@suse.de>
* gdb.base/index-cache.exp: Add additional_flags=-Wl,--build-id.
H.J. Lu [Fri, 21 Jun 2019 20:18:41 +0000 (13:18 -0700)]
i386: Break i386-dis-evex.h into small files
Break i386-dis-evex.h into small files such that each file is included
just once.
* i386-dis-evex.h: Break into ...
* i386-dis-evex-len.h: New file.
* i386-dis-evex-mod.h: Likewise.
* i386-dis-evex-prefix.h: Likewise.
* i386-dis-evex-reg.h: Likewise.
* i386-dis-evex-w.h: Likewise.
* i386-dis.c: Include i386-dis-evex-reg.h, i386-dis-evex-prefix.h,
i386-dis-evex.h, i386-dis-evex-len.h, i386-dis-evex-w.h and
i386-dis-evex-mod.h.
Simon Marchi [Fri, 21 Jun 2019 18:10:15 +0000 (14:10 -0400)]
dwarf2read: Get rid of VEC (dwarf2_section_info_def)
This patch removes uses of VEC (dwarf2_section_info_def) in favor of
std::vector<dwarf2_section_info>. The conversion is relatively
straightforward, no function changes are intended.
gdb/ChangeLog:
* dwarf2read.h (dwarf2_section_info_def): Remove.
(DEF_VEC_O (dwarf2_section_info_def)): Remove.
* dwarf2read.c (struct dwo_sections) <types>: Change type to
std::vector<dwarf2_section_info>.
(struct dwo_file) <~dwo_file>: Remove.
(dwarf2_per_objfile::~dwarf2_per_objfile): Don't manually free
types field.
(dwarf2_per_objfile::locate_sections): Adjust to std::vector.
(dwarf2_read_debug_names): Likewise.
(create_debug_types_hash_table): Change parameter type to
array_view, adjust code accordingly.
(dwarf2_locate_dwo_sections): Adjust to std::vector.
(partial_die_info::fixup): Likewise.
(determine_prefix): Likewise.
* dwarf-index-write.c (write_psymtabs_to_index): Adjust.
Simon Marchi [Fri, 21 Jun 2019 18:10:57 +0000 (14:10 -0400)]
dwarf2read: Make dwo_file::dbfd a gdb_bfd_ref_ptr
This removes the manual call to gdb_bfd_ref in favor of gdb_bfd_ref_ptr.
gdb/ChangeLog:
* dwarf2read.c (struct dwo_file) <dbfd>: Change type to
gdb_bfd_ref_ptr.
<~dwo_file>: Remove call to gdb_bfd_unref.
(open_and_init_dwo_file): Move gdb_bfd_ref_ptr into dbfd field. Call
gdb_bfd_ref_ptr::get.
Simon Marchi [Fri, 21 Jun 2019 18:10:14 +0000 (14:10 -0400)]
dwarf2read: C++ify dwo_file
This patch changes dwo_file to be allocated/deallocated with new/delete,
so that we can start using C++ features in it, and in struct
dwo_sections.
The free_dwo_file function becomes the destructor of struct dwo_file
(and will disappear in upcoming patches, which will use gdb_bfd_ref_ptr
for dbfd and an std::vector for sections.types).
gdb/ChangeLog:
* dwarf2read.h (struct dwarf2_per_objfile) <dwo_files>: Change
type to htab_up.
* dwarf2read.c (struct dwo_file): Initialize fields.
<~dwo_file>: New.
(free_dwo_file): Remove, move content to ~dwo_file.
(struct dwo_file_deleter): Remove.
(dwo_file_up>: Remove custom deleter.
(free_dwo_files): Remove.
(dwarf2_per_objfile::~dwarf2_per_objfile): Don't explicitly free
dwo_files.
(process_skeletonless_type_units): Call unique_ptr::get.
(allocate_dwo_file_hash_table): Add deleter to created hash
table. Change return type to htab_up.
(lookup_dwo_file_slot): Don't memset dwo_file, call
unique_ptr::get.
(create_dwo_unit_in_dwp_v1): Allocate dwo_file with new.
(create_dwo_unit_in_dwp_v2): Likewise.
(open_and_init_dwo_file): Likewise.
(free_dwo_file_from_slot): Remove.
Simon Marchi [Fri, 21 Jun 2019 18:10:13 +0000 (14:10 -0400)]
dwarf2read: Use bool for dwarf2_section_info fields
Use bool instead of char where applicable in dwarf2_section_info.
No functional changes intended.
gdb/ChangeLog:
* dwarf2read.h (struct dwarf2_section_info) <readin,
is_virtual>: Change type to bool.
* dwarf2read.c (dwarf2_read_section, create_dwp_v2_section): Use
true instead of 1.
Tom de Vries [Fri, 21 Jun 2019 15:03:27 +0000 (17:03 +0200)]
[gdb/testsuite] Mark ptype_union.exp as unsupported for cc-with-gdb-index
When testing gdb with board cc-with-gdb-index, we run into:
...
FAIL: gdb.ada/ptype_union.exp: ptype global
FAIL: gdb.ada/ptype_union.exp: print global
...
The index is not supported for Ada (PR24713), and cc-with-gdb-index does not
add an index for Ada test-cases. However, this test-case compiles C sources,
for which cc-with-gdb-index does add an index. In gdb we load the executable
containing the index and set the language to Ada, resulting in gdb trying to
handle something that is not supported.
Fix the fail by marking this unsupported.
Tested on x86_64-linux.
gdb/testsuite/ChangeLog:
2019-06-21 Tom de Vries <tdevries@suse.de>
PR testsuite/24518
PR ada/24713
* gdb.ada/ptype_union.exp: Mark as unsupported if executable contains
index.
Andreas Schwab [Fri, 21 Jun 2019 12:42:22 +0000 (14:42 +0200)]
Add gnulib to gdb release tarball
* src-release.sh (GDB_SUPPORT_DIRS): Add gnulib.
Gary Benson [Fri, 21 Jun 2019 12:23:59 +0000 (13:23 +0100)]
Fix gnulib/update-gnulib.sh
This commit fixes two paths in update-gnulib.sh that weren't updated
when gnulib was moved to toplevel.
gnulib/ChangeLog:
* update-gnulib.sh: Adjust paths.
Nick Alcock [Wed, 19 Jun 2019 14:56:52 +0000 (15:56 +0100)]
libctf: fix ctf_open endianness problems with raw CTF files
ctf_open (or, rather, ctf_fdopen, which underlies it) has several
endianness problems, even though it was written after the
endian-swapping code was implemented, so should have been endian-aware.
Even though the comment right above the relevant check says that it wil
check for CTF magic in any endianness, it only checks in the native
endianness, so opening raw LE CTF files on BE, or vice-versa, will fail.
It also checks the CTF version by hand, without ever endianness-swapping
the header, so that too will fail, and is entirely redundant because
ctf_simple_open does the job properly in any case. We have a similar
problem in the next if block, which checks for raw CTF archives: we are
checking in the native endianness while we should be doing a le64toh()
on it to check in little-endian form only: so opening CTF archives
created on the local machine will fail if the local machine is
big-endian.
Adding insult to injury, if ctf_simple_open then fails, we go on and try
to turn it into a single-element CTF archive regardless, throwing the
error away. Since this involves dereferencing null pointers it is not
likely to work very well.
libctf/
* ctf-open-bfd.c: Add swap.h and ctf-endian.h.
(ctf_fdopen): Check for endian-swapped raw CTF magic, and
little-endian CTF archive magic. Do not check the CTF version:
ctf_simple_open does that in endian-safe ways. Do not dereference
null pointers on open failure.
Nick Alcock [Wed, 19 Jun 2019 11:34:56 +0000 (12:34 +0100)]
libctf: endianness fixes
Testing of the first code to generate CTF_K_SLICEs on big-endian
revealed a bunch of new problems in this area. Most importantly, the
trick we did earlier to avoid wasting two bytes on padding in the
ctf_slice_t is best avoided: because it leads to the whole file after
that point no longer being naturally aligned, all multibyte accesses
from then on must use memmove() to avoid unaligned access on platforms
where that is fatal. In future, this is planned, but for now we are
still doing direct access in many places, so we must revert to making
ctf_slice_t properly aligned for storage in an array.
Rather than wasting bytes on padding, we boost the size of cts_offset
and cts_bits. This is still a waste of space (we cannot have offsets or
bits in bitfields > 256) but it cannot be avoided for now, and slices
are not so common that this will be a serious problem.
A possibly-worse endianness problem fixed at the same time involves
a codepath used only for foreign-endian, uncompressed CTF files, where
we were not copying the actual CTF data into the buffer, leading to
libctf reading only zeroes (or, possibly, uninitialized garbage).
Finally, when we read in a CTF file, we copy the header and work from
the copy. We were flipping the endianness of the header copy, and of
the body of the file buffer, but not of the header in the file buffer
itself: so if we write the file back out again we end up with an
unreadable frankenfile with header and body of different endiannesses.
Fix by flipping both copies of the header.
include/
* ctf.h (ctf_slice_t): Make cts_offset and cts_bits unsigned
short, so following structures are properly aligned.
libctf/
* ctf-open.c (get_vbytes_common): Return the new slice size.
(ctf_bufopen): Flip the endianness of the CTF-section header copy.
Remember to copy in the CTF data when opening an uncompressed
foreign-endian CTF file. Prune useless variable manipulation.
Nick Alcock [Wed, 19 Jun 2019 11:27:18 +0000 (12:27 +0100)]
libctf: unidentified type kinds on open are a sign of file corruption
If we see a CTF type with a kind we do not recognize in its ctt_info
during opening, we cannot skip it and continue opening the file: if the
type kind is unknown, we do not know how long its vlen is, and we cannot
have skipped past it: so if we continue reading we will almost certainly
read in part of the vlen as if it were a new ctf_type_t.
Avoid this trouble by considering unknown type kinds to be a reason to
return ECTF_CORRUPT, just like everything else that reads in type kinds
does.
libctf/
* ctf-open.c (ctf_types): Fail when unidentified type kinds are
seen.
Nick Alcock [Wed, 19 Jun 2019 11:23:38 +0000 (12:23 +0100)]
libctf: dump header offsets into the debugging output
This is an essential first piece of info needed to debug both libctf
writing and reading problems, and we weren't recording it anywhere!
(This is a short-term fix: fairly soon, we will record all of this in a
form that outlives ctf_bufopen, and then ctf_dump() will be able to dump
it like it can everything else.)
libctf/
* ctf-open.c (ctf_bufopen): Dump header offsets into the debugging
output.
Nick Alcock [Wed, 19 Jun 2019 11:20:47 +0000 (12:20 +0100)]
libctf: drop mmap()-based CTF data allocator
This allocator has the ostensible benefit that it lets us mprotect() the
memory used for CTF storage: but in exchange for this it adds
considerable complexity, since we have to track allocation sizes
ourselves for use at freeing time, note whether the data we are storing
was ctf_data_alloc()ed or not so we know if we can safely mprotect()
it... and while the mprotect()ing has found few bugs, it *has* been the
cause of more than one due to errors in all this tracking leading to us
mprotect()ing bits of the heap and stuff like that.
We are about to start composing CTF buffers from pieces so that we can
do usage-based optimizations on the strtab. This means we need
realloc(), which needs nonportable mremap() and *more* tracking of the
*original* allocation size, and the complexity and bureaucracy of all of
this is just too high for its negligible benefits.
Drop the whole thing and just use malloc() like everyone else. It knows
better than we do when it is safe to use mmap() under the covers,
anyway.
While we're at it, don't leak the entire buffer if ctf_compress_write()
fails to compress it.
libctf/
* ctf-subr.c (_PAGESIZE): Remove.
(ctf_data_alloc): Likewise.
(ctf_data_free): Likewise.
(ctf_data_protect): Likewise.
* ctf-impl.h: Remove declarations.
* ctf-create.c (ctf_update): No longer call ctf_data_protect: use
ctf_free, not ctf_data_free.
(ctf_compress_write): Use ctf_data_alloc, not ctf_alloc. Free
the buffer again on compression error.
* ctf-open.c (ctf_set_base): No longer track the size: call
ctf_free, not ctf_data_free.
(upgrade_types): Likewise. Call ctf_alloc, not ctf_data_alloc.
(ctf_bufopen): Likewise. No longer call ctf_data_protect.
Nick Alcock [Wed, 19 Jun 2019 11:14:16 +0000 (12:14 +0100)]
libctf: handle errors on dynhash insertion better
We were missing several cases where dynhash insertion might fail, likely
due to OOM but possibly for other reasons. Pass the errors on.
libctf/
* ctf-create.c (ctf_dtd_insert): Pass on error returns from
ctf_dynhash_insert.
(ctf_dvd_insert): Likewise.
(ctf_add_generic): Likewise.
(ctf_add_variable): Likewise.
* ctf-impl.h: Adjust declarations.