The units of the value returned by "frag_now_fix()" and "size" do not
match. "frag_now_fix()" returns bytes (can be 8, 16 or 32 bit), while
"size" is octets (exactly 8 bit).
* dwarf2dbg.c (dwarf2_emit_insn): Fix calculation of line info offset.
Derive the pointer to the DTV array from the %r2 register on 32-bit
powerpc and %r13 on 64-bit powerpc.
gdb/ChangeLog:
* ppc-fbsd-tdep.c (ppcfbsd_get_thread_local_address): New.
(ppcfbsd_init_abi): Install gdbarch
"fetch_tls_load_module_address" and "get_thread_local_address"
methods.
Derive the pointer to the DTV array from the tp register.
gdb/ChangeLog:
* riscv-fbsd-tdep.c (riscv_fbsd_get_thread_local_address): New.
(riscv_fbsd_init_abi): Install gdbarch
"fetch_tls_load_module_address" and "get_thread_local_address"
methods.
Derive the pointer to the DTV array from the gs_base register. As
with FreeBSD/amd64, gs_base is currently only available via the native
target.
gdb/ChangeLog:
* i386-fbsd-tdep.c (i386fbsd_get_thread_local_address): New.
(i386fbsd_init_abi): Install gdbarch
"fetch_tls_load_module_address" and "get_thread_local_address"
methods.
Use the fs_base register to fetch the address of a thread's tcb and
calculate the address of the DTV array. This value is then passed to
fbsd_get_thread_local_address to compute the final variable address.
Note that fs_base is currently only available via the native target as
core dumps on FreeBSD do not store the value of fs_base.
gdb/ChangeLog:
* amd64-fbsd-tdep.c (amd64fbsd_get_thread_local_address): New.
(amd64fbsd_init_abi): Install gdbarch
"fetch_tls_load_module_address" and "get_thread_local_address"
methods.
The fbsd_get_thread_local_address function accepts the base address of
a thread's DTV array and the base address of an object file's link map
and uses this to compute a TLS variable's address. FreeBSD
architectures use an architecture-specific method to determine the
address of the DTV array pointer and call this helper function to
perform the rest of the address calculation.
* fbsd-tdep.c (fbsd_pspace_data_handle): New variable.
(struct fbsd_pspace_data): New type.
(get_fbsd_pspace_data, fbsd_pspace_data_cleanup)
(fbsd_read_integer_by_name, fbsd_fetch_rtld_offsets)
(fbsd_get_tls_index, fbsd_get_thread_local_address): New function.
(_initialize_fbsd_tdep): Initialize 'fbsd_pspace_data_handle'.
* fbsd-tdep.c (fbsd_get_thread_local_address): New prototype.
lookup_struct_elt is a new function which returns a tuple of
information about a component of a structure or union. The returned
tuple contains a pointer to the struct field object for the component
as well as a bit offset of that field within the structure. If the
field names a field in an anonymous substructure, the offset is the
"global" offset relative to the original structure type. If noerr is
set, then the returned tuple will set the field pointer to NULL to
indicate a missing component rather than throwing an error.
lookup_struct_elt_type is now reimplemented in terms of this new
function. It simply returns the type of the returned field.
gdb/ChangeLog:
* gdbtypes.c (lookup_struct_elt): New function.
(lookup_struct_elt_type): Reimplement via lookup_struct_elt.
* gdbtypes.h (struct struct_elt): New type.
(lookup_struct_elt): New prototype.
Update the comment above the function to reflect the code removal and
document the existing behavior.
gdb/ChangeLog:
* gdbtypes.c (lookup_struct_elt_type): Update comment and
remove disabled code block.
Permit TLS variable addresses to be resolved purely by an ABI rather
than requiring a target method. This doesn't try the target method if
the ABI function is present (even if the ABI function fails) to
simplify error handling.
gdb/ChangeLog:
* gdbarch.sh (get_thread_local_address): New method.
* gdbarch.h, gdbarch.c: Regenerate.
* target.c (target_translate_tls_address): Use
gdbarch_get_thread_local_address if present instead of
target::get_thread_local_address.
There isn't an 'objfile' parameter, instead 'load_module_addr' is used
to indicate the executable or shared library. Also, the function
throws errors rather than returning error values.
gdb/ChangeLog:
* target.h (target::get_thread_local_address): Update comment.
If a TLS variable is provided by a minisym from a separate debug file,
the separate debug file is passed to
gdbarch_fetch_tls_load_module_address. However, the object files
stored in the shared object list are the original object files, not
the separate debug object files. In this case,
svr4_fetch_objfile_link_map was failing to find the link map entry
since the debug object file is not in its internal list, only the
original object file.
gdb/ChangeLog:
* solib-svr4.c (svr4_fetch_objfile_link_map): Look for
objfile->separate_debug_objfile_backlink if not NULL.
The i386 BSD native target uses the same ptrace operations
(PT_[GS]ET[FG]SBASE) as the amd64 BSD native target to fetch and store
the registers.
The amd64 BSD native now uses 'tdep->fsbase_regnum' instead of
hardcoding AMD64_FSBASE_REGNUM and AMD64_GSBASE_REGNUM to support
32-bit targets. In addition, the store operations explicitly zero the
new register value before fetching it from the register cache to
ensure 32-bit values are zero-extended.
gdb/ChangeLog:
* amd64-bsd-nat.c (amd64bsd_fetch_inferior_registers): Use
tdep->fsbase_regnum instead of constants for fs_base and gs_base.
(amd64bsd_store_inferior_registers): Likewise.
* amd64-fbsd-nat.c (amd64_fbsd_nat_target::read_description):
Enable segment base registers.
* i386-bsd-nat.c (i386bsd_fetch_inferior_registers): Use
PT_GETFSBASE and PT_GETGSBASE.
(i386bsd_store_inferior_registers): Use PT_SETFSBASE and
PT_SETGSBASE.
* i386-fbsd-nat.c (i386_fbsd_nat_target::read_description): Enable
segment base registers.
* i386-fbsd-tdep.c (i386fbsd_core_read_description): Likewise.
As on amd64, these registers hold the base address of the fs and gs
segments, respectively. For i386 these two registers are 32 bits.
gdb/ChangeLog:
* amd64-fbsd-nat.c (amd64_fbsd_nat_target::read_description):
Update calls to i386_target_description to add 'segments'
parameter.
* amd64-tdep.c (amd64_init_abi): Set tdep->fsbase_regnum. Don't
add segment base registers.
* arch/i386.c (i386_create_target_description): Add 'segments'
parameter to enable segment base registers.
* arch/i386.h (i386_create_target_description): Likewise.
* features/i386/32bit-segments.xml: New file.
* features/i386/32bit-segments.c: Generate.
* i386-fbsd-nat.c (i386_fbsd_nat_target::read_description): Update
call to i386_target_description to add 'segments' parameter.
* i386-fbsd-tdep.c (i386fbsd_core_read_description): Likewise.
* i386-go32-tdep.c (i386_go32_init_abi): Likewise.
* i386-linux-tdep.c (i386_linux_read_description): Likewise.
* i386-tdep.c (i386_validate_tdesc_p): Add segment base registers
if feature is present.
(i386_gdbarch_init): Pass I386_NUM_REGS to set_gdbarch_num_regs.
Add 'segments' parameter to call to i386_target_description.
(i386_target_description): Add 'segments' parameter to enable
segment base registers.
(_initialize_i386_tdep) [GDB_SELF_TEST]: Add 'segments' parameter
to call to i386_target_description.
* i386-tdep.h (struct gdbarch_tdep): Add 'fsbase_regnum'.
(enum i386_regnum): Add I386_FSBASE_REGNUM and I386_GSBASE_REGNUM.
Define I386_NUM_REGS.
(i386_target_description): Add 'segments' parameter to enable
segment base registers.
gdb/gdbserver/ChangeLog:
* linux-x86-tdesc.c (i386_linux_read_description): Update call to
i386_create_target_description for 'segments' parameter.
* lynx-i386-low.c (lynx_i386_arch_setup): Likewise.
* nto-x86-low.c (nto_x86_arch_setup): Likewise.
* win32-i386-low.c (i386_arch_setup): Likewise.
Changes in DejaGnu 1.6.2 mean that our testsuite will no longer run.
This is because of some confusion over how the gdb.exp file is
handled.
The gdb.exp file is really the tool init file, which is loaded from
within the DejaGnu core, and it should not be loaded directly from any
other file in the testsuite.
DejaGnu tries to prevent the same library being loaded twice by
remembering the names of library files as they are loaded. Until
recently loading the tool init file in DejaGnu was very similar to
loading a library file, as a result, loading the gdb.exp tool init
file simply recorded 'gdb.exp' as having been loaded, future attempts
to load 'gdb.exp' as a library would then be ignored (as the file was
marked as already loaded).
DejaGnu has now changed so that it supports having both a tool init
file and a library with the same name, something that was not possible
before. What this means however is that when the core loads the
'gdb.exp' tool init file it no longer marks the library 'gdb.exp' as
having been loaded. When we then execute 'load_lib gdb.exp' we then
try to reload the 'gdb.exp' file.
Unfortunately our gdb.exp file can only be loaded once. It use of
'rename cd builtin_cd' means that a second attempt to load this file
will fail.
This was discussed on the DejaGnu list here:
http://lists.gnu.org/archive/html/dejagnu/2019-03/msg00000.html
and the suggested advice is that, unless we have some real requirement
to load the tool init file twice, we should remove calls to 'load_lib
gdb.exp' and rely on DejaGnu to load the file for us, which is what
this patch does.
I've tested with native X86-64/GNU Linux and see no regressions.
gdb/testsuite/ChangeLog:
* config/default.exp: Remove 'load_lib gdb.exp'.
* config/monitor.exp: Likewise.
* config/sid.exp: Likewise.
* config/sim.exp: Likewise.
* config/slite.exp: Likewise.
* config/unix.exp: Likewise.
* gdb.base/default.exp: Remove unhelpful comment.
This changes magic_null_ptid, not_sent_ptid, and any_thread_ptid to be
"const". This is a minor improvement that makes it so these can't be
accidentally modified.
Tested by rebuilding. I'm checking this in.
gdb/ChangeLog
2019-03-12 Tom Tromey <tromey@adacore.com>
* remote.c (magic_null_ptid, not_sent_ptid, any_thread_ptid): Now
const. Add initializers.
(_initialize_remote): Don't initialize ptid globals.
$ make test-cp-name-parser
...
CXX test-cp-name-parser.o
src/gdb/cp-name-parser.y: In function ‘int gdb::main(int, char**)’:
src/gdb/cp-name-parser.y:2137:6: error: unused variable ‘len’ [-Werror=unused-variable]
int len;
^~~
cc1plus: all warnings being treated as errors
gdb/ChangeLog:
2019-03-12 Pedro Alves <palves@redhat.com>
* cp-name-parser.y (main): Remove unused 'len' variable.
This makes null_ptid and minus_one_ptid "const". I think this is an
improvement because it means they can't be accidentally modified.
2019-03-12 Tom Tromey <tromey@adacore.com>
* common/ptid.c (null_ptid, minus_one_ptid): Now const.
* common/ptid.h (null_ptid, minus_one_ptid): Now const.
Compressed debug sections can have uncompressed sizes that exceed the
original file size, so we can't use bfd_get_file_size. objdump also
used bfd_get_file_size to limit reloc section size, but I believe the
underlying bug causing the PR22508 out of bounds buffer access was
that we had an integer overflow when calculating the reloc buffer
size. I've fixed that instead in most of the backends, som and
vms-alpha being the exceptions. SOM and vmd-alpha have rather more
serious bugs in their slurp_relocs routines that would need fixing
first if we want to fuss about making them safe against fuzzed object
files.
The patch also fixes a number of other potential overflows by using
the bfd_alloc2/malloc2/zalloc2 memory allocation functions.
bfd/
* coffcode.h (buy_and_read): Delete unnecessary forward decl. Add
nmemb parameter. Use bfd_alloc2.
(coff_slurp_line_table): Use bfd_alloc2. Update buy_and_read calls.
Delete assertion.
(coff_slurp_symbol_table): Use bfd_alloc2 and bfd_zalloc2.
(coff_slurp_reloc_table): Use bfd_alloc2. Update buy_and_read calls.
* coffgen.c (coff_get_reloc_upper_bound): Ensure size calculation
doesn't overflow.
* elf.c (bfd_section_from_shdr): Use bfd_zalloc2. Style fix.
(assign_section_numbers): Style fix.
(swap_out_syms): Use bfd_malloc2.
(_bfd_elf_get_reloc_upper_bound): Ensure size calculation doesn't
overflow.
(_bfd_elf_make_empty_symbol): Style fix.
(elfobj_grok_stapsdt_note_1): Formatting.
* elfcode.h (elf_object_p): Use bfd_alloc2.
(elf_write_relocs, elf_write_shdrs_and_ehdr): Likewise.
(elf_slurp_symbol_table): Use bfd_zalloc2.
(elf_slurp_reloc_table): Use bfd_alloc2.
(_bfd_elf_bfd_from_remote_memory): Use bfd_malloc2.
* elf64-sparc (elf64_sparc_get_reloc_upper_bound): Ensure
size calculation doesn't overflow.
(elf64_sparc_get_dynamic_reloc_upper_bound): Likewise.
* mach-o.c (bfd_mach_o_get_reloc_upper_bound): Likewise.
* pdp11.c (get_reloc_upper_bound): Copy aoutx.h version.
binutils/
* objdump.c (load_specific_debug_section): Don't compare section
size against file size.
(dump_relocs_in_section): Don't compare reloc size against file size.
Print "failed to read relocs" on bfd_get_reloc_upper_bound error.
dwarf2_find_containing_comp_unit has two assignments to "this_cu" in
quick succession, both of which are just:
this_cu = dwarf2_per_objfile->all_comp_units[low];
... with no intervening assignments.
This patch removes the second assignment. I'm checking this in as
obvious. Tested on x86-64 Fedora 29.
gdb/ChangeLog
2019-03-11 Tom Tromey <tromey@adacore.com>
* dwarf2read.c (dwarf2_find_containing_comp_unit): Remove
redundant assignment to "this_cu".
We return INCOMPATIBLE_TYPE_BADNESS for all these type codes, so we might as
well just let them go to the default case.
Incidentally, this patch also makes this false positive error go away when
compiling with gcc (Ubuntu 7.3.0-27ubuntu1~18.04) 7.3.0, default compiler on
Ubuntu 18.04.
CXX gdbtypes.o
/home/smarchi/src/binutils-gdb/gdb/gdbtypes.c: In function ‘rank rank_one_type(type*, type*, value*)’:
/home/smarchi/src/binutils-gdb/gdb/gdbtypes.c:4259:1: error: control reaches end of non-void function [-Werror=return-type]
}
^
gdb/ChangeLog:
* gdbtypes.c (rank_one_type): Remove unnecessary cases from switch.
Without this patch, the help stops after 'e.g.' :
(gdb) apropos \(inferior\|thread\) event
set print inferior-events -- Set printing of inferior events (e.g.
set print thread-events -- Set printing of thread events (such as thread start and exit)
show print inferior-events -- Show printing of inferior events (e.g.
show print thread-events -- Show printing of thread events (such as thread start and exit)
Using the same notation as for the thread evenets (i.e. 'such as') gives:
(gdb) apropos \(inferior\|thread\) event
set print inferior-events -- Set printing of inferior events (such as inferior start and exit)
set print thread-events -- Set printing of thread events (such as thread start and exit)
show print inferior-events -- Show printing of inferior events (such as inferior start and exit)
show print thread-events -- Show printing of thread events (such as thread start and exit)
gdb/ChangeLog:
2019-03-08 Eli Zaretskii <eliz@gnu.org>
PR/24315
* utils.c (can_emit_style_escape) [_WIN32]: Don't disable styling
on MS-Windows if $TERM is not defined.
* cli/cli-style.c: Set cli_styling to 1 in the MinGW build.
* posix-hdep.c (gdb_console_fputs):
* mingw-hdep.c (rgb_to_16colors, gdb_console_fputs): New
functions.
* ui-file.h (gdb_console_fputs): Add prototype.
* ui-file.c (stdio_file::puts): Call gdb_console_fputs, and fall
back to fputs only if the former returns zero.
Not padding string merge section output to its alignment can cause
failures of the S-record tests when input string merge sections are
padded, since the ELF linker output for the single string section
would shrink compared to the SREC linker output. That might result in
following sections having different addresses.
On the other hand, padding string merge section output when input
string merge sections are *not* padded can also cause failures, in
this case due to the ELF linker output for the string section being
larger (due to padding) than the SREC linker output.
It would be better to write a more robust test, but it is also nice
to leave input unchanged when no string merges occur.
PR 24311
* merge.c (merge_strings): Return secinfo. Don't pad section
to alignment here.
(_bfd_merge_sections): Pad section to alignment here, if input
sections contributing to merged output all pad to alignment.
Formatting.
TARGET_WAITKIND_NO_RESUMED doesn't have an associated event thread, so
we shouldn't be referring to inferior_thread() assuming it points to
one.
This was caught on the multi-target branch, where we always switch to
no-thread-selected whenever we start handling an event, exactly to
catch places that incorrectly use "inferior_ptid/inferior_thread()"
without switching to the right event thread / target.
Here, on the branch, we assert in inferior_thread() because
TARGET_WAITKIND_NO_RESUMED doesn't have an associated event thread, so
inferior_ptid is still null_ptid.
gdb/ChangeLog:
2019-03-07 Pedro Alves <palves@redhat.com>
* infrun.c (normal_stop): Also check for
TARGET_WAITKIND_NO_RESUMED before referring to inferior_thread().
