Tom Tromey [Thu, 17 Sep 2020 20:11:38 +0000 (14:11 -0600)]
Change management of tdesc_arch_data
While working on something else, I noticed that tdesc_data_cleanup
took a void* parameter. Looking more into this, I found that
tdesc_use_registers expected a transfer of ownership.
I think it's better to express this sort of thing via the type system,
when possible. This patch changes tdesc_data_alloc to return a unique
pointer, changes tdesc_use_registers to accept an rvalue reference,
and then adapts all the users.
Note that a deleter structure is introduced to avoid having to move
tdesc_arch_data to the header file.
Hannes Domani [Thu, 17 Sep 2020 17:10:16 +0000 (19:10 +0200)]
Fix ctrl-c when debugging WOW64 processes
DebugBreakProcess starts a new thread in the target process with the
entry point DbgUiRemoteBreakin, where an int3 triggers a breakpoint
exception for gdb.
But this uses DbgUiRemoteBreakin of the 64bit ntdll.dll even for
WOW64 processes.
It stops in 64bit code, Wow64GetThreadContext reports a wrong pc without
the int3, and gdb lets the target process continue.
So this uses DbgUiRemoteBreakin of the 32bit ntdll.dll as the thread
entry point for WOW64 processes instead.
gdb/ChangeLog:
2020-09-17 Hannes Domani <ssbssa@yahoo.de>
* windows-nat.c (ctrl_c_handler): Use 32bit DbgUiRemoteBreakin
for WOW64 processes.
Tom Tromey [Thu, 17 Sep 2020 17:47:50 +0000 (11:47 -0600)]
Use htab_up in linespec.c
This changes linespec.c to use htab_up rather than explicit calls to
htab_delete. Note that a use still exists in this file, because
linespec_state hasn't been converted to have a real destructor.
gdb/ChangeLog
2020-09-17 Tom Tromey <tom@tromey.com>
Andrew Burgess [Wed, 16 Sep 2020 09:12:39 +0000 (10:12 +0100)]
opcodes/csky: return the default disassembler when there is no bfd
The disassembler function should return a valid disassembler function
even when there is no BFD present. This is implied (I believe) by the
comment in dis-asm.h which says the BFD may be NULL. Further, it
makes sense when considering that the disassembler is used in GDB, and
GDB may connect to a target and perform debugging even without a BFD
being supplied.
This commit makes the csky_get_disassembler function return the
default disassembler configuration when no bfd is supplied, this is
the same default configuration as is used when a BFD is supplied, but
the BFD has no attributes section.
Before the change configuring GDB with --enable-targets=all and
running the tests gdb.base/all-architectures-2.exp results in many
errors, but after this change there are no failures.
opcodes/ChangeLog:
* csky-dis.c (csky_get_disassembler): Don't return NULL when there
is no BFD.
Simon Marchi [Wed, 16 Sep 2020 20:43:39 +0000 (16:43 -0400)]
gdb: update instance_flags field name in gdb-gdb.py.in
Commit 314ad88df63c ("Use type_instance_flags more throughout") changed
the name of field type::instance_flags to type::m_instance_flags. It
however missed changing it in the gdb-gdb.py.in file, which results in
this when trying to use the pretty-printer:
(top-gdb) p *val.type
Traceback (most recent call last):
File "/home/smarchi/build/binutils-gdb/gdb/gdb-gdb.py", line 116, in to_string
% TypeFlagsPrinter(self.val['instance_flags']))
File "/home/smarchi/build/binutils-gdb/gdb/gdb-gdb.py", line 76, in __str__
flag_list = [flag.short_name for flag in TYPE_FLAGS
File "/home/smarchi/build/binutils-gdb/gdb/gdb-gdb.py", line 77, in <listcomp>
if self.val & flag.value]
gdb.error: Argument to arithmetic operation not a number or boolean.
$7 =
This patch fixes it.
gdb/ChangeLog:
* gdb-gdb.py.in (class StructTypePrettyPrinter) <to_string>:
Change instance_flags to m_instance_flags.
Tom Tromey [Wed, 16 Sep 2020 15:49:36 +0000 (09:49 -0600)]
Match demangled name in "skip"
PR gdb/26598 notes that, before commit bcfe6157ca28 ("Use the linkage
name if it exists"), the "skip" command would match the demangled name
of a symbol, but now only matches the linkage name.
This patch fixes this regression. I looked at all calls to
function_name_is_marked_for_skip, and only one used the linkage name.
2020-09-16 Tom Tromey <tromey@adacore.com>
PR gdb/26598:
* infrun.c (fill_in_stop_func): Use find_pc_partial_function_sym.
gdb/testsuite/ChangeLog
2020-09-16 Tom Tromey <tromey@adacore.com>
PR gdb/26598:
* gdb.base/skipcxx.exp: New file.
* gdb.base/skipcxx.cc: New file.
John Baldwin [Wed, 16 Sep 2020 18:40:05 +0000 (11:40 -0700)]
Assume FreeBSD kernels always report exec events.
FreeBSD kernels have reported exec events via the PL_FLAG_EXEC flag
since 8.2 release. The most recent release that did not support this
flag is 7.4 released in November of 2011.
Note that the FreeBSD native target already assumed that PL_FLAG_SCE
and PL_FLAG_SCX were always supported on systems supporting
PT_LWPINFO, but those flags were added at the same time as
PL_FLAG_EXEC. Building the native target on a system without
PL_FLAG_EXEC would have failed to build before this change already as
a result.
John Baldwin [Wed, 16 Sep 2020 18:40:05 +0000 (11:40 -0700)]
Require kinfo_get_file and kinfo_get_vmmap for FreeBSD hosts.
FreeBSD systems have provided these functions in libutil since 7.1
release. The most recent release without support is 6.4 released in
November of 2008.
This also requires libutil-freebsd on GNU/kFreeBSD systems. I assume
that those systems have supported kinfo_get_file and kinfo_get_vmmap
over a similar timeframe.
gdb/ChangeLog:
* configure.ac: Remove check for kinfo_getvmmap().
* configure, config.in: Regenerate.
* fbsd-nat.c (fbsd_read_mapping): Remove
(fbsd_nat_target::find_memory_regions): Remove the procfs version.
(fbsd_nat_target::info_proc): Assume kinfo_getfile() and
kinfo_get_vmmap() are always present.
John Baldwin [Wed, 16 Sep 2020 18:40:05 +0000 (11:40 -0700)]
Assume FreeBSD hosts include support for fetching signal information.
The current layout of siginfo_t and support for fetching it has been
included in FreeBSD kernels since 7.0 release. The most recent
release without support is 6.4 released in November of 2008.
gdb/ChangeLog:
* fbsd-nat.c: Always include support for
TARGET_OBJECT_SIGNAL_INFO.
John Baldwin [Wed, 16 Sep 2020 18:40:05 +0000 (11:40 -0700)]
Assume KERN_PROC_PATHNAME is present on FreeBSD hosts.
FreeBSD kernels have included this sysctl since 6.0 release. The most
recent release without support is 5.5 which was released in May of
2006.
Both the sysctl and the /proc/<pid>/exe file call the same function in
the kernel to obtain the path (vn_fullpath). If the sysctl fails, the
readlink call will also fail, so there is no need to keep the readlink
fallback.
gdb/ChangeLog:
* fbsd-nat.c (fbsd_nat_target::pid_to_exec_file): Always use
sysctl and remove procfs fallback.
John Baldwin [Wed, 16 Sep 2020 18:40:04 +0000 (11:40 -0700)]
Assume that PT_LWPINFO is always defined on FreeBSD hosts.
FreeBSD kernels have included support for this since 5.0 release.
The most recent release without support is 4.11 which was released
in January of 2005.
gdb/ChangeLog:
* fbsd-nat.c: Assume PT_LWPINFO is always defined.
* fbsd-nat.h: Likewise.
Tom de Vries [Wed, 16 Sep 2020 16:04:37 +0000 (18:04 +0200)]
[gdb/testsuite] Fix gdb.dlang/watch-loc.exp on ppc64
On ppc64, we run into:
...
/tmp/ccKXURmw.s: Assembler messages:^M
/tmp/ccKXURmw.s:99: Error: can't resolve \
`_Dmain_end' {.text section} - `_Dmain' {.opd section}^M
compiler exited with status 1
...
The unresolvable entity comes from the .debug_aranges section hardcoded in
watch-loc.c:
...
" .4byte _Dmain \n" // Address
" .4byte _Dmain_end - _Dmain \n" // Length
...
where _Dmain is the function label for function _Dmain, and _Dmain_end is a
label after function _Dmain:
...
int _Dmain (void)
{
asm ("_Dmain_label: .globl _Dmain_label");
return 0;
}
asm ("_Dmain_end: .globl _Dmain_end");
...
The problem is that the two labels are not in the same section, because _Dmain
ends up in the .opd section, which "contains the official procedure
descriptors. A pointer to a function shall reference a procedure descriptor in
this section."
Fix this by using function_range instead to get the length.
Also, fix the address part to not use _Dmain, but instead use the function
start as returned by function_range.
Tested on x86_64-linux.
gdb/testsuite/ChangeLog:
2020-09-16 Tom de Vries <tdevries@suse.de>
PR testsuite/26617
* gdb.dlang/watch-loc.c: Only add .debug_aranges if
WITH_DEBUG_SECTIONS is defined.
(xstr, str): New macro.
(.debug_aranges): Use DMAIN_START_0/1/2 and DMAIN_LENGTH
to get start and size of _Dmain.
* gdb.dlang/watch-loc.exp: Pass DMAIN_START_0/1/2 and DMAIN_LENGTH
using additional_flags.
gdb/breakpoint: make a copy of the "commands" command's argument
When GDB reads commands from the input, its internal buffer is re-used
for each line. This is usually just fine because commands are
executed in order; by the time we read the next line, we are already
done with the current line. However, a problematic case is breakpoint
commands that are input from a script. The header (e.g. commands 1 2)
is overwritten with the next line before the breakpoint numbers are
processed completely.
For example, suppose we have the following script:
break main
break main
commands 1 2
print 100123
end
and source this script:
(gdb) source script.gdb
Breakpoint 1 at 0x1245: file main.cpp, line 27.
Breakpoint 2 at 0x1245: file main.cpp, line 27.
No breakpoint number 123.
Note the "No breakpoint number 123." error message. This happens
because GDB first reads "commands 1 2" into its internal buffer
buffer -> "commands 1 2"
and then starts parsing the breakpoint numbers. After parsing the first
token, the "next token" pointer is as below:
buffer -> "commands 1 2"
next-token -----------^
So, if we continue parsing, we would tokenize "2" correctly. However,
before parsing the next number, GDB reads the commands to attach them
to breakpoint 1. Reading the commands causes the buffer to be
overwritten:
buffer -> " print 100123"
next-token -----------^
So, the next time we parse the breakpoint number, we read "123".
To fix, simply create a copy of the arguments of the header.
gdb/ChangeLog:
2020-09-16 Tankut Baris Aktemur <tankut.baris.aktemur@intel.com>
* breakpoint.c (commands_command_1): Make a copy of the 'arg'
argument.
gdb/testsuite/ChangeLog:
2020-09-16 Tankut Baris Aktemur <tankut.baris.aktemur@intel.com>
* gdb.base/bp-cmds-sourced-script.c: New file.
* gdb.base/bp-cmds-sourced-script.exp: New test.
* gdb.base/bp-cmds-sourced-script.gdb: New file.
H.J. Lu [Wed, 16 Sep 2020 14:11:16 +0000 (07:11 -0700)]
elf/x86-64: Adjust relocation for PE/x86-64 inputs
PE linker calls _bfd_relocate_contents to resolve relocation, instead of
bfd_perform_relocation. But ELF linker calls bfd_perform_relocation, not
_bfd_relocate_contents. When linking PE/x86-64 inputs to generate ELF
output, we need to adjust PE/x86-64 relocations in bfd_perform_relocation.
Enable PE/x86-64 in bfd together with PEI/x86-64. Update run_ld_link_tests
to handle bzip2 binary inputs.
bfd/
PR ld/26583
* config.bfd (targ64_selvecs, targ_selvecs): Add x86_64_pe_vec
to x86_64_pei_vec.
* reloc.c: Include "coff/internal.h".
(bfd_perform_relocation): Adjust relocation for PE/x86-64 inputs.
Tom de Vries [Wed, 16 Sep 2020 12:53:26 +0000 (14:53 +0200)]
[gdb/testsuite] Detect gdb prompt after monitor exit
With this gdbserver-support.exp patch:
...
@@ -451,8 +451,10 @@ proc gdbserver_exit { is_mi } {
# We use expect rather than gdb_expect because
# we want to suppress printing exception messages, otherwise,
# remote_expect, invoked by gdb_expect, prints the exceptions.
+ set have_prompt 0
expect {
-i "$gdb_spawn_id" -re "$gdb_prompt $" {
+ set have_prompt 1
exp_continue
}
-i "$server_spawn_id" eof {
@@ -463,6 +465,7 @@ proc gdbserver_exit { is_mi } {
warning "Timed out waiting for EOF in server after $monitor_exit"
}
}
+ gdb_assert {$have_prompt}
}
}
close_gdbserver
...
and with this in parallel:
...
$ stress -c 5
...
we run into this and similar FAILs:
...
FAIL: gdb.multi/multi-target.exp: continue: non-stop=on: $have_prompt
...
In more detail:
...
(gdb) PASS: gdb.multi/multi-target.exp: continue: non-stop=on: inferior 5
Remote debugging from host ::1, port 40712^M
Process build/gdb/testsuite/outputs/gdb.multi/multi-target/multi-target \
created; pid = 11098^M
monitor exit^M
Killing process(es): 11098^M
FAIL: gdb.multi/multi-target.exp: continue: non-stop=on: $have_prompt
spawn build/gdb/testsuite/../../gdb/gdb -nw -nx -data-directory \
build/gdb/testsuite/../data-directory^M
...
After issuing a "monitor exit" command, we should always get a prompt back, so
check for that.
Tested on x86_64-linux.
gdb/testsuite/ChangeLog:
2020-09-16 Tom de Vries <tdevries@suse.de>
* lib/gdbserver-support.exp (gdbserver_exit): Make sure we
get the gdb prompt after issuing "monitor exit".
Tom de Vries [Wed, 16 Sep 2020 11:27:32 +0000 (13:27 +0200)]
[gdb/testsuite] Catch condition evaluation errors in gdb_assert
When running test-case gdb.base/watchpoint-stops-at-right-insn.exp, we may run
into a tcl error, which can be reproduced reliably using this trigger patch:
...
+ set hw_watch_pc ""
gdb_assert {$sw_watch_pc == $hw_watch_pc} \
"hw watchpoint stops at right instruction"
...
such that we have:
...
ERROR: tcl error sourcing watchpoint-stops-at-right-insn.exp.
ERROR: missing operand at _@_
in expression "0x4004b7 == _@_"
(parsing expression "0x4004b7 == ")
invoked from within
"expr $sw_watch_pc == $hw_watch_pc"
("uplevel" body line 1)
invoked from within
"uplevel 1 expr $condition"
(procedure "gdb_assert" line 6)
invoked from within
"gdb_assert {$sw_watch_pc == $hw_watch_pc} \
"hw watchpoint stops at right instruction""
...
A similar problem was fixed in commit 5f0e2eb79e "GDB/testsuite: Fix a
catastrophic step-over-no-symbols.exp failure", by making the assert condition
more robust:
...
- gdb_assert {$before_addr != $after_addr} "advanced"
+ gdb_assert {{[string is integer -strict $before_addr] \
+ && [string is integer -strict $after_addr] \
+ && $before_addr != $after_addr}} "advanced"
...
Fix this instead in gdb_assert, by catching errors while evaluating the assert
condition.
Tested on x86_64-linux.
gdb/testsuite/ChangeLog:
2020-09-16 Tom de Vries <tdevries@suse.de>
PR testsuite/26624
* lib/gdb.exp (gdb_assert): Catch errors in condition evaluation.
Andrew Burgess [Tue, 4 Aug 2020 16:07:59 +0000 (17:07 +0100)]
gdb: Convert language_data::la_op_print_tab to a method
Convert language_data::la_op_print_tab member variable to a virtual
method language_defn::opcode_print_table. I changed the name in order
to make it clearer (I hope) what the method does.
There should be no user visible changes after this commit.
gdb/ChangeLog:
* ada-lang.c (ada_language_data): Remove la_op_print_tab
initializer.
(ada_language::opcode_print_table): New member function.
* c-lang.c (c_language_data): Remove la_op_print_tab initializer.
(c_language::opcode_print_table): New member function.
(cplus_language_data): Remove la_op_print_tab initializer.
(cplus_language::opcode_print_table): New member function.
(asm_language_data): Remove la_op_print_tab initializer.
(asm_language::opcode_print_table): New member function.
(minimal_language_data): Remove la_op_print_tab initializer.
(minimal_language::opcode_print_table): New member function.
* d-lang.c (d_language_data): Remove la_op_print_tab initializer.
(d_language::opcode_print_table): New member function.
* expprint.c (print_subexp_standard): Update call to
opcode_print_table.
(op_string): Likewise.
* f-lang.c (f_language_data): Remove la_op_print_tab initializer.
(f_language::opcode_print_table): New member function.
* go-lang.c (go_language_data): Remove la_op_print_tab
initializer.
(go_language::opcode_print_table): New member function.
* language.c (unknown_language_data): Remove la_op_print_tab
initializer.
(unknown_language::opcode_print_table): New member function.
(auto_language_data): Remove la_op_print_tab initializer.
(auto_language::opcode_print_table): New member function.
* language.h (language_data): Remove la_op_print_tab field.
(language_defn::opcode_print_table): Declare new member function.
* m2-lang.c (m2_language_data): Remove la_op_print_tab
initializer.
(m2_language::opcode_print_table): New member function.
* objc-lang.c (objc_language_data): Remove la_op_print_tab
initializer.
(objc_language::opcode_print_table): New member function.
* opencl-lang.c (opencl_language_data): Remove la_op_print_tab
initializer.
(opencl_language::opcode_print_table): New member function.
* p-lang.c (pascal_language_data): Remove la_op_print_tab
initializer.
(pascal_language::opcode_print_table): New member function.
* rust-lang.c (rust_language_data): Remove la_op_print_tab
initializer.
(rust_language::opcode_print_table): New member function.
Andrew Burgess [Tue, 4 Aug 2020 15:31:56 +0000 (16:31 +0100)]
gdb: Convert language_data::la_exp_desc to a method
Convert language_data::la_exp_desc member variable to a virtual
method language_defn::expression_ops. The change of names brings this
method more into line with the existing varobj_ops method, that also
returns a table of function pointers.
There should be no user visible changes after this commit.
gdb/ChangeLog:
* ada-lang.c (ada_language_data): Remove la_exp_desc initializer.
(ada_language::expression_ops): New member function.
* c-lang.c (c_language_data): Remove la_exp_desc initializer.
(c_language::expression_ops): New member function.
(cplus_language_data): Remove la_exp_desc initializer.
(cplus_language::expression_ops): New member function.
(asm_language_data): Remove la_exp_desc initializer.
(asm_language::expression_ops): New member function.
(minimal_language_data): Remove la_exp_desc initializer.
(minimal_language::expression_ops): New member function.
* d-lang.c (d_language_data): Remove la_exp_desc initializer.
(d_language::expression_ops): New member function.
* eval.c (evaluate_subexp): Update call to expression_ops.
* expprint.c (print_subexp): Likewise.
(op_name): Likewise.
(dump_subexp_body): Likewise.
* f-lang.c (f_language_data): Remove la_exp_desc initializer.
(f_language::expression_ops): New member function.
* go-lang.c (go_language_data): Remove la_exp_desc initializer.
(go_language::expression_ops): New member function.
* language.c (language_defn::expression_ops): New function.
(unknown_language_data): Remove la_exp_desc initializer.
(auto_language_data): Likewise.
* language.h (language_data): Remove la_exp_desc field.
(language_defn::expression_ops): Declare new member function.
* m2-lang.c (m2_language_data): Remove la_exp_desc initializer.
(m2_language::expression_ops): New member function.
* objc-lang.c (objc_language_data): Remove la_exp_desc
initializer.
* opencl-lang.c (opencl_language_data): Remove la_exp_desc
initializer.
(opencl_language::expression_ops): New member function.
* p-lang.c (pascal_language_data): Remove la_exp_desc initializer.
* parse.c (operator_length): Update call to expression_ops.
(exp_iterate): Likewise.
* rust-lang.c (rust_language_data): Remove la_exp_desc
initializer.
(ruse_language::expression_ops): New member function.
Andrew Burgess [Sun, 12 Jul 2020 14:09:52 +0000 (15:09 +0100)]
gdb: Convert language_data::la_case_sensitivity to a method
Convert language_data::la_case_sensitivity member variable to a virtual
method language_defn::case_sensitivity.
This is mostly straight forward. The only slight problem is that I
ended up deleting this comment from ada-lang.c:
/* Yes, Ada is case-insensitive, but that's not quite what this
means. */
However, as the comment (which has existed since Ada support was first
added to GDB) doesn't explain _why_ Ada sets case sensitivity to on
despite being a generally case insensitive language, this doesn't
really help me much.
If I understood _why_ the setting doesn't quite mean what it seems to
mean (at least as far as Ada is concerned) then I would extend the
comment on language_defn::case_sensitivity (in language.h) to include
the detail, and note how this impacts Ada. But as it stands I've just
deleted the comment for now.
There should be no user visible changes after this commit.
gdb/ChangeLog:
* ada-lang.c (ada_language_data): Remove la_case_sensitivity
initializer.
* c-lang.c (c_language_data): Likewise.
(cplus_language_data): Likewise.
(asm_language_data): Likewise.
(minimal_language_data): Likewise.
* d-lang.c (d_language_data): Likewise.
* f-lang.c (f_language_data): Likewise.
(f_language::case_sensitivity): New member function.
* go-lang.c (go_language_data): Remove la_case_sensitivity
initializer.
* language.c (enum case_mode): Moved here from language.h.
(case_mode): Make static.
(show_case_command): Update for case_sensitivity being a method.
(set_case_command): Likewise.
(set_range_case): Likewise.
(unknown_language_data): Remove la_case_sensitivity initializer.
(auto_language_data): Likewise.
* language.h (case_mode): Delete, move enum declaration to
language.c.
(language_data): Delete la_case_sensitivity field.
(language_defn::case_sensitivity): New member function.
* m2-lang.c (m2_language_data): Remove la_case_sensitivity
initializer.
* objc-lang.c (objc_language_data): Likewise.
* opencl-lang.c (opencl_language_data): Likewise.
* p-lang.c (pascal_language_data): Likewise.
* rust-lang.c (rust_language_data): Likewise.
Andrew Burgess [Sun, 12 Jul 2020 09:15:32 +0000 (10:15 +0100)]
gdb: Convert language_data::la_range_check to a method
Convert language_data::la_range_check member variable to a virtual
method language_defn::range_checking_on_by_default.
Where the previous member variable was of type 'enum range_check', the
new member function returns a boolean that selects between range
checking being on or off. This removes the possibility of a language
having its default be the third enum state, range_check_warn, which
no language currently used.
This all makes sense I think, the language's true/false provides the
default when the global 'set check range' is set to 'auto'.
There should be no user visible changes after this commit.
gdb/ChangeLog:
* ada-lang.c (ada_language_data): Remove la_range_check
initializer.
* c-lang.c (c_language_data): Likewise.
(cplus_language_data): Likewise.
(asm_language_data): Likewise.
(minimal_language_data): Likewise.
* d-lang.c (d_language_data): Likewise.
* f-lang.c (f_language_data): Likewise.
(f_language::range_checking_on_by_default): New member function.
* go-lang.c (go_language_data): Remove la_range_check initializer.
* language.c (enum range_mode): Moved here from language.h.
(range_mode): Made static.
(show_range_command): Update to use
range_checking_on_by_default.
(set_range_command): Likewise.
(set_range_case): Likewise.
(unknown_language_data): Remove la_range_check initializer.
(auto_language_data): Likewise.
* language.h (range_mode): Delete. Enum definition moved to
language.c.
(language_data): Remove la_range_check field.
(language_defn::range_checking_on_by_default): New member
function.
* m2-lang.c (m2_language_data): Remove la_range_check initializer.
(m2_language::range_checking_on_by_default): New member function.
* objc-lang.c (objc_language_data): Remove la_range_check
initializer.
* opencl-lang.c (opencl_language_data): Likewise.
* p-lang.c (pascal_language_data): Likewise.
(pascal_language::range_checking_on_by_default): New member
function.
* rust-lang.c (rust_language_data): Remove la_range_check
initializer.
(rust_language::range_checking_on_by_default): New member
function.
problem looking up some symbols when they have a linkage name
A special case was left behind for Go, however, this special case was
not really needed anymore, it could be handled by having
store_sym_names_in_linkage_form_p return the true for go, instead of
false.
This commit overrides store_sym_names_in_linkage_form_p for Go, and
then removes the special case. As store_sym_names_in_linkage_form_p
is only called once throughout GDB this should be perfectly safe.
There should be no user visible changes after this commit.
gdb/ChangeLog:
* dwarf2/read.c (dwarf2_physname): Remove special case for
language_go.
* go-lang.c (go_language::store_sym_names_in_linkage_form_p): New
member function.
Andrew Burgess [Sun, 5 Jul 2020 08:29:34 +0000 (09:29 +0100)]
gdb: Convert language_data::string_lower_bound to a method
Convert language_data::string_lower_bound member variable to a virtual
method language_defn::string_lower_bound.
Over all of the languages we currently support there are currently
only two values for the lower bound, 0 or 1. I noticed that in all
cases, if a language has C style arrays then the lower bound is 0,
otherwise the lower bound is 1. So the default for the virtual method
in language.h makes use of this, which means languages don't have to
worry about providing a string_lower_bound method at all.
Except for Modula2. This language is defined to not have C style
arrays, but has a string_lower_bound index of 0, this behaviour is
maintained after this commit by having Modula2 be the only language
that overrides the string_lower_bound method.
There should be no user visible changes after this commit.
Andrew Burgess [Fri, 3 Jul 2020 17:51:34 +0000 (18:51 +0100)]
gdb: Move la_language into the language_defn class
Move the language_data::la_language member variable into the
langage_defn class.
I have not made the la_language member variable a method of
langage_defn simply because of the large number of places that
la_language is referenced throughout GDB. I have made the new member
variable constant though, so this should prevent accidental
assignment.
In the future we might consider converting la_language to a method,
but right now my goal is to remove the langage_data class, so I'm
happy to leave la_language as a constant member variable.
There should be no user visible changes after this commit.
Andrew Burgess [Fri, 3 Jul 2020 17:46:16 +0000 (18:46 +0100)]
gdb: Convert la_filename_extensions to a method
Convert language_data::la_filename_extensions member variable to a
virtual method language_defn::filename_extensions.
The new method returns a vector of filename extensions, which means
that where previously we needed a NULL marker on the end of the list,
we can now discard this.
All of the old arrays that contained the extensions now become static
data within each languages filename_extensions method.
I've updated the single use of the filename_extensions method to make
use of this method returning a vector. And, just in case anyone
accidentally adds a NULL marked into a languages extensions list, I've
added a new assert in add_filename_language (symtab.c) to catch this.
gdb/ChangeLog:
* ada-lang.c (ada_extensions): Delete, moved into
ada_language::filename_extensions.
(ada_language_data): Remove la_filename_extensions initializer.
(ada_language::filename_extensions): New member function.
* c-lang.c (c_extensions): Delete, moved into
c_language::filename_extensions.
(c_language_data): Remove la_filename_extensions initializer.
(c_language::filename_extensions): New member function.
(cplus_extensions): Delete, moved into
cplus_language::filename_extensions.
(cplus_language_data): Remove la_filename_extensions initializer.
(cplus_language::filename_extensions): New member function.
(asm_extensions): Delete, moved into
asm_language::filename_extensions.
(asm_language_data): Remove la_filename_extensions initializer.
(asm_language::filename_extensions): New member function.
(minimal_language_data): Remove la_filename_extensions
initializer.
* d-lang.c (d_extensions): Delete, moved into
d_language::filename_extensions.
(d_language_data): Remove la_filename_extensions initializer.
(d_language::filename_extensions): New member function.
* f-lang.c (f_extensions): Delete, moved into
f_language::filename_extensions.
(f_language_data): Remove la_filename_extensions initializer.
(f_language::filename_extensions): New member function.
* go-lang.c (go_language_data): Remove la_filename_extensions
initializer.
* language.c (add_set_language_command): Update now that
filename_extensions returns a vector.
(unknown_language_data): Remove la_filename_extensions
initializer.
(auto_language_data): Likewise.
* language.h (language_data): Remove la_filename_extensions field.
(language_defn::filename_extensions): New member function.
* m2-lang.c (m2_language_data): Remove la_filename_extensions
initializer.
* objc-lang.c (objc_extensions): Delete, moved into
objc_language::filename_extensions.
(objc_language_data): Remove la_filename_extensions initializer.
(objc_language::filename_extensions): New member function.
* opencl-lang.c (opencl_language_data): Remove
la_filename_extensions initializer.
* p-lang.c (pascal_extensions): Delete, moved into
pascal_language::filename_extensions.
(pascal_language_data): Remove la_filename_extensions initializer.
(pascal_language::filename_extensions): New member function.
* rust-lang.c (rust_extensions): Delete, moved into
rust_language::filename_extensions.
(rust_language_data): Remove la_filename_extensions initializer.
(rust_language::filename_extensions): New member function.
* symfile.c (add_filename_language): Add new assert.
Andrew Burgess [Tue, 23 Jun 2020 19:07:09 +0000 (20:07 +0100)]
gdb: Convert la_name and la_natural_name to methods
Convert the two language_data member variables la_name and
la_natural_name to virtual methods in language_defn struct called name
and natural_name respectively.
The virtual methods in the language_defn base class are pure virtual,
as every language must implement these, and as every language has a
unique name there's no sensible default here.
Given that every language must implement these methods I did wonder
about making this data passed into the base class constructor, but in
the end I went with the virtual method approach. I'm open to changing
this approach if people prefer the constructor approach.
During updating the calls to language_defn::name I found in
add_set_language_command a place where we took la_name and then
capitalised the first letter to create a language name that could be
used in the documentation string. I replaced this with a use of
natural_name instead as this seemed a better choice, in most cases
this will make no difference, as for most languages the natural_name
is just the name with the first character in upper case, but for some
languages, for example 'Open-CL' and 'Objective-C' this is not the
case.
In the case of asm_language the name is 'asm', while the natural_name
was previously 'assembly'. I changed the natural name to 'Assembly',
this makes the documentation string case above cleaner, however, this
will change the MI output for -var-info-expression, where the 'lang'
field will change from 'assembly' to 'Assembly'. It is possible this
could be a breaking change if a front-end is relying on the existing
name.
gdb/ChangeLog:
* ada-lang.c (ada_language_data): Remove la_name and
la_natural_name initializers.
(ada_language::name): New member function.
(ada_language::natural_name): New member function.
* c-lang.c (c_language_data): Remove la_name and
la_natural_name initializers.
(c_language::name): New member function.
(c_language::natural_name): New member function.
(cplus_language_data): Remove la_name and
la_natural_name initializers.
(cplus_language::name): New member function.
(cplus_language::natural_name): New member function.
(asm_language_data): Remove la_name and
la_natural_name initializers.
(asm_language::name): New member function.
(asm_language::natural_name): New member function.
(minimal_language_data): Remove la_name and
la_natural_name initializers.
(minimal_language::name): New member function.
(minimal_language::natural_name): New member function.
* compile/compile.c (compile_to_object): Update call to
lanugage_defn::name.
* d-lang.c (d_language_data): Remove la_name and
la_natural_name initializers.
(d_language::name): New member function.
(d_language::natural_name): New member function.
* expprint.c (print_subexp_standard): Update call to
language_defn::name.
(dump_raw_expression): Likewise
(dump_prefix_expression): Likewise.
* f-lang.c (f_language_data): Remove la_name and
la_natural_name initializers.
(f_language::name): New member function.
(f_language::natural_name): New member function.
* go-lang.c (go_language_data): Remove la_name and
la_natural_name initializers.
(go_language::name): New member function.
(go_language::natural_name): New member function.
* language.c (show_language_command): Update call to
language_defn::name.
(set_language_command): Likewise.
(language_enum): Likewise.
(language_str): Likewise.
(add_set_language_command): Likewise, use
language_defn::natural_name in the doc string.
(unknown_language_data): Remove la_name and
la_natural_name initializers.
(unknown_language::name): New member function.
(unknown_language::natural_name): New member function.
(auto_language_data): Remove la_name and
la_natural_name initializers.
(auto_language::name): New member function.
(auto_language::natural_name): New member function.
(language_lookup_primitive_type_as_symbol): Update call to
language_defn::name.
* language.h (language_data): Remove la_name and la_natural_name
member variables.
(language_defn::name): New member function.
(language_defn::natural_name): New member function.
* m2-lang.c (m2_language_data): Remove la_name and
la_natural_name initializers.
(m2_language::name): New member function.
(m2_language::natural_name): New member function.
* mi/mi-cmd-var.c (mi_cmd_var_info_expression): Update call to
language_defn::natural_name.
* objc-lang.c (objc_language_data): Remove la_name and
la_natural_name initializers.
(objc_language::name): New member function.
(objc_language::natural_name): New member function.
* opencl-lang.c (opencl_language_data): Remove la_name and
la_natural_name initializers.
(opencl_language::name): New member function.
(opencl_language::natural_name): New member function.
* p-lang.c (pascal_language_data): Remove la_name and
la_natural_name initializers.
(pascal_language::name): New member function.
(pascal_language::natural_name): New member function.
* rust-lang.c (rust_language_data): Remove la_name and
la_natural_name initializers.
(rust_language::name): New member function.
(rust_language::natural_name): New member function.
* symtab.c (lookup_language_this): Update call to
language_defn::name.
Tom de Vries [Wed, 16 Sep 2020 08:57:17 +0000 (10:57 +0200)]
[gdb/testsuite] Fix unbalanced braces in gdb.tui/new-layout.exp
On SLE-11 with tcl version 8.5, we run into:
...
UNRESOLVED: gdb.tui/new-layout.exp: testcase aborted due to \
invalid command name: }
ERROR: Couldn't send tui new-layout example src 1} to GDB.
...
Apparently tcl 8.6 is more forgiving about unbalanced braces.
Expose the unbalanced braces on tcl 8.6 by temporarily wrapping the
test-case in:
...
proc do_this { args } {
uplevel 1 {*}$args
}
do_this {
...
}
...
and escape them using a backslash.
Tom Tromey [Wed, 16 Sep 2020 00:44:37 +0000 (18:44 -0600)]
Remove TYPE_CODE_INT case from f_value_print_inner
I looked through the various language value-print functions, to see if
any code could be consolidated. Pretty much all I found was that
f_value_print_inner does not need to handle TYPE_CODE_INT itself, but
can simply dispatch to the generic printer.
gdb/ChangeLog
2020-09-15 Tom Tromey <tom@tromey.com>
Tom Tromey [Wed, 16 Sep 2020 00:44:37 +0000 (18:44 -0600)]
Handle member pointers directly in generic_value_print
TYPE_CODE_MEMBERPTR and TYPE_CODE_METHODPTR are only used for C++, so
it seems to me that the generic value-printing code ought to handle
these cases -- that way, printing these objects will work even when
the current language is not C++. This patch implements this idea.
gdb/ChangeLog
2020-09-15 Tom Tromey <tom@tromey.com>
* rust-lang.c (rust_value_print_inner): Remove TYPE_CODE_MEMBERPTR
and TYPE_CODE_METHODPTR cases.
* c-valprint.c (c_value_print_memberptr): Move to valprint.c.
(c_value_print_inner): Update.
* valprint.c (generic_value_print_memberptr): New function, from
c_value_print_memberptr.
(generic_value_print): Use it. Call cplus_print_method_ptr.
Tom Tromey [Tue, 15 Sep 2020 17:08:56 +0000 (11:08 -0600)]
Don't use PyInt_FromLong
Avoid the use of PyInt_FromLong, preferring gdb_py_object_from_longest
instead. I found found another spot that was incorrectly handling
errors (see gdbpy_create_ptid_object) while writing this patch; it is
fixed here.
gdb/ChangeLog
2020-09-15 Tom Tromey <tromey@adacore.com>
* python/python-internal.h (PyInt_FromLong): Remove define.
* python/py-value.c (convert_value_from_python): Use
gdb_py_object_from_longest.
* python/py-type.c (typy_get_code): Use
gdb_py_object_from_longest.
* python/py-symtab.c (salpy_get_line): Use
gdb_py_object_from_longest.
* python/py-symbol.c (sympy_get_addr_class, sympy_line): Use
gdb_py_object_from_longest.
* python/py-record.c (recpy_gap_reason_code): Use
gdb_py_object_from_longest.
* python/py-record-btrace.c (recpy_bt_insn_size)
(recpy_bt_func_level, btpy_list_count): Use
gdb_py_object_from_longest.
* python/py-infthread.c (gdbpy_create_ptid_object): Use
gdb_py_object_from_longest. Fix error handling.
* python/py-framefilter.c (bootstrap_python_frame_filters): Use
gdb_py_object_from_longest.
* python/py-frame.c (frapy_type, frapy_unwind_stop_reason): Use
gdb_py_object_from_longest.
* python/py-breakpoint.c (bppy_get_type, bppy_get_number)
(bppy_get_thread, bppy_get_task, bppy_get_hit_count)
(bppy_get_ignore_count): Use gdb_py_object_from_longest.
Tom Tromey [Tue, 15 Sep 2020 17:08:56 +0000 (11:08 -0600)]
Don't use PyLong_FromLongLong
This changes gdb to avoid PyLong_FromLongLong, preferring to use
gdb_py_object_from_longest instead.
gdb/ChangeLog
2020-09-15 Tom Tromey <tromey@adacore.com>
* python/py-infevents.c (create_register_changed_event_object):
Use gdb_py_object_from_longest.
* python/py-exitedevent.c (create_exited_event_object): Use
gdb_py_object_from_longest.
Tom Tromey [Tue, 15 Sep 2020 17:08:56 +0000 (11:08 -0600)]
Don't use gdb_py_long_from_ulongest
Remove the gdb_py_long_from_ulongest defines and change the Python
layer to prefer gdb_py_object_from_ulongest. While writing this I
noticed that the error handling in archpy_disassemble was incorrect --
it could call PyDict_SetItemString with a NULL value. This patch also
fixes this bug.
gdb/ChangeLog
2020-09-15 Tom Tromey <tromey@adacore.com>
* python/python-internal.h (gdb_py_long_from_ulongest): Remove
defines.
* python/py-value.c (valpy_long): Use
gdb_py_object_from_ulongest.
* python/py-symtab.c (salpy_get_pc): Use
gdb_py_object_from_ulongest.
(salpy_get_last): Likewise.
* python/py-record-btrace.c (recpy_bt_insn_pc): Use
gdb_py_object_from_ulongest.
* python/py-lazy-string.c (stpy_get_address): Use
gdb_py_object_from_ulongest.
* python/py-frame.c (frapy_pc): Use gdb_py_object_from_ulongest.
* python/py-arch.c (archpy_disassemble): Use
gdb_py_object_from_ulongest and gdb_py_object_from_longest. Fix
error handling.
Tom Tromey [Tue, 15 Sep 2020 15:27:01 +0000 (09:27 -0600)]
Avoid running one Rust test against older LLVM
LLVM 8.0 introduced some changes to let the Rust compiler emit DWARF
variant parts. Before this change, the compiler would emit two types
with the same name, and unfortunately gdb happens to pick the wrong
one. So, this patch disables the test when using an older version of
LLVM.
gdb/testsuite/ChangeLog
2020-09-15 Tom Tromey <tromey@adacore.com>
PR rust/26197:
* lib/rust-support.exp (rust_llvm_version): New proc.
* gdb.rust/simple.exp: Check rust_llvm_version.
Tom Tromey [Tue, 15 Sep 2020 14:38:22 +0000 (08:38 -0600)]
Use arrays rather than pointers for global string constants
My understanding is that it's mildly better to use a static const
array, as opposed to a "const char *", for a global string constant,
when possible. This makes sense to me because the pointer requires a
load from an address, whereas the array is just the address.
So, I searched for these in gdb and gdbserver. This patch fixes the
ones I found.
gdb/ChangeLog
2020-09-15 Tom Tromey <tromey@adacore.com>
* unittests/memory-map-selftests.c (valid_mem_map): Now array.
* ui-style.c (ansi_regex_text): Now array.
* rust-exp.y (number_regex_text): Now array.
* linespec.c (linespec_quote_characters): Now array.
* jit.c (jit_break_name, jit_descriptor_name, reader_init_fn_sym):
Now arrays.
gdbserver/ChangeLog
2020-09-15 Tom Tromey <tromey@adacore.com>
* linux-x86-low.cc (xmltarget_i386_linux_no_xml)
(xmltarget_amd64_linux_no_xml): Now arrays.
Alan Modra [Tue, 15 Sep 2020 11:25:02 +0000 (20:55 +0930)]
PR26610, ARM's "VFPv3 vldr to vmov" gas testcase fail
I removed a few too many parentheses in git commit 7af677524e2. This
patch fixes that problem, rewriting the expression so it won't happen
again. The patch also avoids more UB with shifts of signed values.
PR 26610
* config/tc-arm.c (move_or_literal_pool): Correct extraction of
bignum. Use unsigned "v"
(is_double_a_single): Make "v" and "mantissa" unsigned. Formatting.
(double_to_single): Likewise.
CRIS: fix PR ld/26589, a missing NULL check in fix for PR ld/22269
Not sure why there wasn't a NULL check in the ld/22269 patch
(e01c16a8) at the time, as there was one for the corresponding patch
to elf32-m68k.c (5056ba1d).
Incidentally, I had missed that in 2017, as a prerequisite for the
ld/22269 series, the check_relocs function finally were made "safe"!
(I.e. the number of references and symbol types are final, garbage
collection done, so port-specific accounting can be made sanely.)
Committed.
bfd:
PR ld/26589
* elf32-cris.c (cris_elf_check_relocs): Add missing NULL check
on argument before calling UNDEFWEAK_NO_DYNAMIC_RELOC.
ld:
PR ld/26589
* testsuite/ld-elf/pr26589.d, testsuite/ld-elf/locref3.s: New test.
Michael Mullin [Tue, 15 Sep 2020 00:39:54 +0000 (20:39 -0400)]
gdb: fix compile break when HAVE_LIBEXPAT is not defined
Fixes:
CXX xml-tdesc.o
/home/smarchi/src/binutils-gdb/gdb/xml-tdesc.c: In function const target_desc* file_read_description_xml(const char*):
/home/smarchi/src/binutils-gdb/gdb/xml-tdesc.c:681:60: error: too few arguments to function target_desc* tdesc_parse_xml(const char*, xml_fetch_another, void*)
681 | return tdesc_parse_xml (tdesc_str->data (), fetch_another);
| ^
/home/smarchi/src/binutils-gdb/gdb/xml-tdesc.c:44:1: note: declared here
44 | tdesc_parse_xml (const char *document, xml_fetch_another fetcher,
| ^~~~~~~~~~~~~~~
Commit 8400a90d19c5 ("gdb: change xml_fetch_another a function_view")
removed the `baton` parameter of `tdesc_parse_xml`, but didn't update
the version of the function used when GDB is built with no libexpat
support. Remove the parameter in that definition too.
Pedro Alves [Mon, 14 Sep 2020 20:16:59 +0000 (21:16 +0100)]
Rewrite enum_flags, add unit tests, fix problems
This patch started by adding comprehensive unit tests for enum_flags.
For the testing part, it adds:
- tests of normal expected uses of the API.
- checks that _invalid_ uses of the API would fail to compile. I.e.,
it validates that enum_flags really is a strong type, and that
incorrect mixing of enum types would be caught at compile time. It
pulls that off making use of SFINEA and C++11's decltype/constexpr.
This revealed many holes in the enum_flags API. For example, the f1
assignment below currently incorrectly fails to compile:
enum_flags<flags> f1 = FLAG1;
enum_flags<flags> f2 = FLAG2 | f1;
The unit tests also revealed that this useful use case doesn't work:
It fails to compile because enum_flags<flag> and flag are convertible
to each other.
Turns out that making enum_flags be implicitly convertible to the
backing raw enum type was not a good idea.
If we make it convertible to the underlying type instead, we fix that
ternary operator use case, and, we find cases throughout the codebase
that should be using the enum_flags but were using the raw backing
enum instead. So it's a good change overall.
Also, several operators were missing.
These holes and more are plugged by this patch, by reworking how the
enum_flags operators are implemented, and making use of C++11's
feature of being able to delete methods/functions.
There are cases in gdb/compile/ where we need to call a function in a
C plugin API that expects the raw enum. To address cases like that,
this adds a "raw()" method to enum_flags. This way we can keep using
the safer enum_flags to construct the value, and then be explicit when
we need to get at the raw enum.
This makes most of the enum_flags operators constexpr. Beyond
enabling more compiler optimizations and enabling the new unit tests,
this has other advantages, like making it possible to use operator|
with enum_flags values in switch cases, where only compile-time
constants are allowed:
enum_flags<flags> f = FLAG1 | FLAG2;
switch (f)
{
case FLAG1 | FLAG2:
break;
}
Currently that fails to compile.
It also switches to a different mechanism of enabling the global
operators. The current mechanism isn't namespace friendly, the new
one is.
It also switches to C++11-style SFINAE -- instead of wrapping the
return type in a SFINAE-friently structure, we use an unnamed template
parameter. I.e., this:
Note that the static_assert inside operator~() was converted to a
couple overloads (signed vs unsigned), because static_assert is too
late for SFINAE-based tests, which is important for the CHECK_VALID
unit tests.
Tested with gcc {4.8, 7.1, 9.3} and clang {5.0.2, 10.0.0}.
gdb/ChangeLog:
* Makefile.in (SELFTESTS_SRCS): Add
unittests/enum-flags-selftests.c.
* btrace.c (ftrace_update_caller, ftrace_fixup_calle): Use
btrace_function_flags instead of enum btrace_function_flag.
* compile/compile-c-types.c (convert_qualified): Use
enum_flags::raw.
* compile/compile-cplus-symbols.c (convert_one_symbol)
(convert_symbol_bmsym):
* compile/compile-cplus-types.c (compile_cplus_convert_method)
(compile_cplus_convert_struct_or_union_methods)
(compile_cplus_instance::convert_qualified_base):
* go-exp.y (parse_string_or_char): Add cast to int.
* unittests/enum-flags-selftests.c: New file.
* record-btrace.c (btrace_thread_flag_to_str): Change parameter's
type to btrace_thread_flags from btrace_thread_flag.
(record_btrace_cancel_resume, record_btrace_step_thread): Change
local's type to btrace_thread_flags from btrace_thread_flag. Add
cast in DEBUG call.
gdbsupport/ChangeLog:
* enum-flags.h: Include "traits.h".
(DEF_ENUM_FLAGS_TYPE): Declare a function instead of defining a
structure.
(enum_underlying_type): Update comment.
(namespace enum_flags_detail): New. Move struct zero_type here.
(EnumIsUnsigned, EnumIsSigned): New.
(class enum_flags): Make most methods constexpr.
(operator&=, operator|=, operator^=): Take an enum_flags instead
of an enum_type. Make rvalue ref versions deleted.
(operator enum_type()): Delete.
(operator&, operator|, operator^, operator~): Delete, moved out of
class.
(raw()): New method.
(is_enum_flags_enum_type_t): Declare.
(ENUM_FLAGS_GEN_BINOP, ENUM_FLAGS_GEN_COMPOUND_ASSIGN)
(ENUM_FLAGS_GEN_COMP): New. Use them to reimplement global
operators.
(operator~): Now constexpr and reimplemented.
(operator<<, operator>>): New deleted functions.
* valid-expr.h (CHECK_VALID_EXPR_5, CHECK_VALID_EXPR_6): New.
Pedro Alves [Mon, 14 Sep 2020 20:16:57 +0000 (21:16 +0100)]
Rewrite valid-expr.h's internals in terms of the detection idiom (C++17/N4502)
An earlier attempt at doing this had failed (wouldn't work in GCCs
around 4.8, IIRC), but now that I try again, it works. I suspect that
my previous attempt did not use the pre C++14-safe void_t (in
traits.h).
I want to switch to this model because:
- It's the standard detection idiom that folks will learn starting
with C++17.
- In the enum_flags unit tests, I have a static_assert that triggers
a warning (resulting in build error), which GCC does not suppress
because the warning is not being triggered in the SFINAE context.
Switching to the detection idiom fixes that. Alternatively,
switching to the C++03-style expression-validity checking with a
varargs overload would allow addressing that, but I think that
would be going backwards idiomatically speaking.
- While this patch shows a net increase of lines of code, the magic
being added to traits.h can be removed in a few years when we start
requiring C++17.
Pedro Alves [Mon, 14 Sep 2020 20:16:56 +0000 (21:16 +0100)]
Use type_instance_flags more throughout
A later patch in this series will rewrite enum_flags fixing some API
holes. That would cause build failures around code using
type_instance_flags. Or rather, that should be using it, but wasn't.
This patch fixes it by using type_instance_flags throughout instead of
plain integers.
Note that we can't make the seemingly obvious change to struct
type::instance_flags:
Because G++ complains then that 9 bits isn't sufficient for holding
all values of type_instance_flag_value.
So the patch adds an type::instance_flags() method, which takes care
of casting appropriately, and adds a separate type::set_instance_flags
method, following the pattern of the ongoing TYPE_XXX macro
elimination. This converts uses of TYPE_INSTANCE_FLAGS to
type::instance_flags() in the places where the code was already being
touched, but there are still many references to the
TYPE_INSTANCE_FLAGS macro left behind. Those could/should be fully
replaced at some point.
gdb/ChangeLog:
* avr-tdep.c (avr_address_class_type_flags): Return
type_instance_flags.
(avr_address_class_type_flags_to_name): Take a
type_instance_flags.
(avr_address_class_name_to_type_flags): Return bool and take a
type_instance_flags.
* d-lang.c (build_d_types): Use type::set_instance_flags.
* ft32-tdep.c (ft32_address_class_type_flags): Return
type_instance_flags.
(ft32_address_class_type_flags_to_name): Take a
type_instance_flags.
(ft32_address_class_name_to_type_flags): Return bool and take a
type_instance_flags.
(ft32_gdbarch_init): Use type::set_instance_flags.
* eval.c (fake_method::fake_method): Use type::set_instance_flags.
* gdbarch.h, gdbarch.c: Regenerate.
* gdbarch.sh (address_class_type_flags): Use type_instance_flags.
(address_class_name_to_type_flags): Use type_instance_flags and
bool.
* gdbtypes.c (address_space_name_to_int)
(address_space_int_to_name, make_qualified_type): Use
type_instance_flags.
(make_qualified_type): Use type_instance_flags and
type::set_instance_flags.
(make_type_with_address_space, make_cv_type, make_vector_type)
(check_typedef): Use type_instance_flags.
(recursive_dump_type): Cast type_instance_flags to unsigned for
printing.
(copy_type_recursive): Use type::set_instance_flags.
(gdbtypes_post_init): Use type::set_instance_flags.
* gdbtypes.h (struct type) <instance_flags>: Rename to ...
<m_instance_flags>: ... this.
<instance_flags, set_instance_flags>: New methods.
(TYPE_INSTANCE_FLAGS): Use the instance_flags method.
(SET_TYPE_INSTANCE_FLAGS): New.
(address_space_name_to_int, address_space_int_to_name)
(make_type_with_address_space): Pass flags using
type_instance_flags instead of int.
* stabsread.c (cleanup_undefined_types_noname): Use
type::set_instance_flags.
* s390-tdep.c (s390_address_class_type_flags): Return
type_instance_flags.
(s390_address_class_type_flags_to_name): Take a
type_instance_flags.
(s390_address_class_name_to_type_flags): Return bool and take a
type_instance_flags.
* type-stack.c (type_stack::follow_types): Use
type_instance_flags.
* dwarf2/read.c (read_tag_pointer_type): Use type_instance_flags.
Tom Tromey [Mon, 14 Sep 2020 18:09:59 +0000 (12:09 -0600)]
Apply 'const' in more places
Many global arrays in gdb could be marked "const" but are not. This
patch changes some of them. (There may be other arrays that could
benefit from this treatment. I only examined arrays of strings.)
This lets the linker move some symbols to the readonly data section.
For example, previously:
Simon Marchi [Mon, 14 Sep 2020 18:02:30 +0000 (14:02 -0400)]
gdb: turn gdb::bcache's function pointers into virtual methods
The two function pointers optionally passed to gdb::bcache are very good
candidates to be turned into virtual methods, this patch does that in
the most straightforward / unsurprising way.
Simon Marchi [Mon, 14 Sep 2020 15:51:04 +0000 (11:51 -0400)]
gdb: don't use inferior_ptid in linux_nat_wait_1
target_ops::wait implementations should not rely on the value of
inferior_ptid on entry. While looking at another wait-related patch, I
noticed that the code in linux_nat_wait_1, checking for a newly created
process, did just that. This patch fixes it. Note that I didn't see
any bug, this "fix" is simply to make the function respect the
target_ops::wait contract.
Instead of checking inferior_ptid, check for the passed in `ptid`
value.
During startup, linux_nat_wait_1 gets called a few times with the
pid-only ptid, while startup_inferior waits for the expected number of
exec events. For this reason, I needed to add a `find_lwp_pid` call to
ensure that the actions of changing the main thread's ptid, and adding
the initial lwp, were done only once for a given process.
This was not needed before, since thread_change_ptid, through the
thread_ptid_changed observer, ends up changing inferior_ptid. So the
second time around, inferior_ptid was not a pid-only ptid.
That find_lwp_pid won't add much overhead, as it will only be called
when the ptid is a pid-only ptid. And AFAIK, that only happens during
inferior startup.
An alternative to that `find_lwp_pid` call might be to make
startup_inferior realize that the main thread has changed ptid, and make
it wait for the new ptid. But that doesn't look easy to do.
Regtested on amd64/Linux.
gdb/ChangeLog:
* linux-nat.c (linux_nat_wait_1): Don't use inferior_ptid when
checking for initial lwp.
Tom Tromey [Mon, 14 Sep 2020 14:30:10 +0000 (08:30 -0600)]
Various m68k fixes for gdb
Recently I tried the m68k port of gdb. It had some issues, which are
fixed in this patch.
* Various types of return values were not being handled properly. In
particular:
* arrays are returned by following the same convention as
structures. This matters in languages like Ada, where an array
can in fact be returned as a value.
* "long double" was not being handled correctly in
m68k_svr4_return_value.
* GCC's m68k back end does not return vector types in registers, so
change gdb to follow.
* GCC's m68k back end doesn't faithfully implement the ABI, and so
some objects with unusual size (not possible in C, but possible in
Ada) are not returned correctly.
* gcc implements an m68k ABI variant that it simply describes as
"embedded". This ABI is similar to the SVR4 ABI, but rather than
returning pointer-typed values in %a0, such values are returned in
%d0. To support this, an ELF osabi sniffer is added.
* Commit 85f7484a ("m68k: tag floating-point ABI used") adds an
attribute that can be used to recognize when hard- or soft-float is
in use. gdb can now read this tag and choose the ABI accordingly.
I was unable to run the gdb test suite with this patch. Instead, I
tested it using qemu and the internal AdaCore test suite.
gdb/ChangeLog
2020-09-14 Tom Tromey <tromey@adacore.com>
* m68k-tdep.c (m68k_extract_return_value): Use
pointer_result_regnum.
(m68k_store_return_value): Likewise.
(m68k_reg_struct_return_p): Handle vectors and arrays.
(m68k_return_value): Handle arrays.
(m68k_svr4_return_value): Fix single-element aggregate handling.
Handle long double. Adjust for embedded ABI.
(m68k_svr4_init_abi): Set pointer_result_regnum.
(m68k_embedded_init_abi): New function.
(m68k_gdbarch_init): Handle Tag_GNU_M68K_ABI_FP.
(m68k_osabi_sniffer): New function.
(_initialize_m68k_tdep): Register osabi sniffer.
* m68k-tdep.h (struct gdbarch_tdep) <pointer_result_regnum>: New
member.
Simon Marchi [Mon, 14 Sep 2020 15:12:55 +0000 (11:12 -0400)]
gdb: change xml_fetch_another a function_view
The xml_fetch_another is currently a plain function pointer type, with a
`void *` baton parameter. To improve type-safety, change this to a
function_view. Any required data is captured by a lambda at the call
site.
gdb/ChangeLog:
* xml-support.h (xml_fetch_another): Change type to be a
function_view.
(xml_process_xincludes): Remove baton parameter.
(xml_fetch_content_from_file): Change baton parameter to
dirname.
* xml-support.c (struct xinclude_parsing_data)
<xinclude_parsing_data>: Remove baton parameter.
<fetcher_baton>: Remove.
(xinclude_start_include): Adjust.
(xml_process_xincludes): Adjust.
(xml_fetch_content_from_file): Replace baton parameter with
dirname.
* xml-syscall.c (syscall_parse_xml): Remove baton parameter.
(xml_init_syscalls_info): Use a lambda.
* xml-tdesc.c (tdesc_parse_xml): Remove baton parameter.
(file_read_description_xml): Use a lambda.
(fetch_available_features_from_target): Change baton parameter
to target_ops.
(target_read_description_xml): Use a lambda.
(target_fetch_description_xml): Use a lambda.
(string_read_description_xml): Update.
Nick Clifton [Mon, 14 Sep 2020 15:14:24 +0000 (16:14 +0100)]
Add a new ".nop" directive to the assembler to allow the creation of no-op instructions in an architeture neutral manner.
* read.c (s_nop): New function. Handles the .nop directive.
(potable): Add entry for "nop".
(s_nops): Code tidy.
* read.h (s_nop): Add prototype.
* config/tc-bpf.h (md_single_noop_insn): Define.
* config/tc-mmix.h (md_single_noop_insn): Define.
* config/tc-or1k.h (md_single_noop_insn): Define.
* config/tc-s12z.c (md_assemble): Preserve the input line pointer,
rather than corrupting it.
* write.c (relax_segment): Update error message regarding
non-absolute values passed to .fill and .nops.
* NEWS: Mention the new directive.
* doc/as.texi: Document the new directive.
* doc/internals.texi: Document the new internal macros used to
implement the new directive.
* testsuite/gas/all/nop.s: New test.
* testsuite/gas/all/nop.d: New test control file.
* testsuite/gas/all/gas.exp: Run the new test.
* testsuite/gas/elf/dwarf-5-nop-for-line-table.s: New test.
* testsuite/gas/elf/dwarf-5-nop-for-line-table.d: New test
control file.
* testsuite/gas/elf/elf.exp: Run the new test.
* testsuite/gas/i386/space1.l: Adjust expected output.
Add the `endianity_is_not_default` and `set_endianity_is_not_default`
methods on `struct type`, in order to remove the
`TYPE_ENDIANITY_NOT_DEFAULT` macro. In this patch, the macro is changed
to use the getter, so all the call sites of the macro that are used as a
setter are changed to use the setter method directly. The next patch
will remove the macro completely.
gdb/ChangeLog:
* gdbtypes.h (struct type) <endianity_is_not_default,
set_endianity_is_not_default>: New methods.
(TYPE_ENDIANITY_NOT_DEFAULT): Use
type::endianity_is_not_default, change all write call sites to
use type::set_endianity_is_not_default.
Add the `is_fixed_instance` and `set_is_fixed_instance` methods on `struct
type`, in order to remove the `TYPE_FIXED_INSTANCE` macro. In this patch, the
macro is changed to use the getter, so all the call sites of the macro that are
used as a setter are changed to use the setter method directly. The next patch
will remove the macro completely.
gdb/ChangeLog:
* gdbtypes.h (struct type) <is_fixed_instance,
set_is_fixed_instance>: New methods.
(TYPE_FIXED_INSTANCE): Use type::is_fixed_instance, change all
write call sites to use type::set_is_fixed_instance.
Add the `is_gnu_ifunc` and `set_is_gnu_ifunc` methods on `struct type`, in
order to remove the `TYPE_GNU_IFUNC` macro. In this patch, the macro is
changed to use the getter, so all the call sites of the macro that are
used as a setter are changed to use the setter method directly. The
next patch will remove the macro completely.
gdb/ChangeLog:
* gdbtypes.h (struct type) <is_gnu_ifunc, set_is_gnu_ifunc>: New methods.
(TYPE_GNU_IFUNC): Use type::is_gnu_ifunc, change all write call sites to
use type::set_is_gnu_ifunc.
Add the `stub_is_supported` and `set_stub_is_supported` methods on `struct type`, in
order to remove the `TYPE_STUB_SUPPORTED` macro. In this patch, the macro is
changed to use the getter, so all the call sites of the macro that are
used as a setter are changed to use the setter method directly. The
next patch will remove the macro completely.
gdb/ChangeLog:
* gdbtypes.h (struct type) <stub_is_supported, set_stub_is_supported>: New methods.
(TYPE_STUB_SUPPORTED): Use type::stub_is_supported, change all write call sites to
use type::set_stub_is_supported.
projects / thirdparty / binutils-gdb.git / log
