This adds support for Arm's Neoverse V1 CPU to AArch64 binutils.
gas/ChangeLog:
2020-09-24 Alex Coplan <alex.coplan@arm.com>
* config/tc-aarch64.c (aarch64_cpu_option_table): Add Neoverse V1.
* doc/c-aarch64.texi: Document Neoverse V1 support.
This commit:
commit afdcafe891
Date: Thu Sep 17 14:30:28 2020 +0800
CSKY: Add objdump option -M abi-names.
cases the build of GDB (configured with --enabled-targets=all) to fail
with this error:
libtool: compile: gcc -DHAVE_CONFIG_H -I. -I../../src/opcodes -I. -I../../src/opcodes -I../bfd -I../../src/opcodes/../include -I../../src/opcodes/../bfd -W -Wall -Wstrict-prototypes -Wmissing-prototypes -Wshadow -Wstack-usage=262144 -Werror -O0 -g3 -D_GLIBCXX_DEBUG=1 -MT csky-dis.lo -MD -MP -MF .deps/csky-dis.Tpo -c ../../src/opcodes/csky-dis.c -o csky-dis.o
../../src/opcodes/csky-dis.c: In function 'csky_output_operand':
../../src/opcodes/csky-dis.c:849:7: error: this 'if' clause does not guard... [-Werror=misleading-indentation]
849 | if (IS_CSKY_V1 (mach_flag))
| ^~
../../src/opcodes/csky-dis.c:851:2: note: ...this statement, but the latter is misleadingly indented as if it were guarded by the 'if'
851 | strcat (str, buf);
| ^~~~~~
this commit adds { ... } around the if body to resolve this issue.
opcodes/ChangeLog:
* csky-dis.c (csky_output_operand): Enclose body of if in curly
braces.
This adds support for Arm's Neoverse N2 CPU to AArch32 binutils. The
Neoverse N2 CPU builds AArch32 at EL0 and therefore needs support in the
AArch32 assembler.
gas/ChangeLog:
2020-09-24 Alex Coplan <alex.coplan@arm.com>
* config/tc-arm.c (arm_cpus): Add Neoverse N2.
* doc/c-arm.texi: Document -mcpu=neoverse-n2.
This stub doesn't have the r2 store at the beginning.
* powerpc.cc (Target_powerpc::Relocate::relocate): Don't skip
first insn of __tls_get_addr_opt stub.
This adds missing support for a power10 version of the __tls_get_addr
call stub implementing DT_PPC64_OPT PPC64_OPT_TLS. Without this,
power10 code using __tls_get_addr fails miserably at runtime unless
the --no-tls-get-addr-optimize option is given.
PR 26656
* elf64-ppc.c (plt_stub_size): Add "odd" param. Use it with
size_power10_offset rather than calculating from start of stub.
Add size for notoc tls_get_addr_opt stub.
(plt_stub_pad): Add "odd" param, pass to plt_stub_size.
(build_tls_get_addr_head, build_tls_get_addr_tail): New functions.
(build_tls_get_addr_stub): Delete.
(ppc_build_one_stub): Use a temp for htab->params->stub_bfd.
Emit notoc tls_get_addr_opt stub. Move eh_frame code to
suit. Adjust code to use bfd_tls_get_addr_head/tail in place
of build_tls_get_addr_stub.
(ppc_size_one_stub): Size notoc tls_get_addr_opt stub.
Adjust plt_stub_size and plt_stub_pad calls. Correct "odd"
when padding stub. Size eh_frame for notoc stub too.
Correct lr_restore value.
(ppc64_elf_relocate_section): Don't skip over first insn of
notoc tls_get_addr_opt stub.
Some of the powerpc64 code editing functions are better run after
dynamic symbols have stabilised in order to make proper decisions
based on SYMBOL_REFERENCES_LOCAL. The dynamic symbols are processed
early in bfd_elf_size_dynamic_sections, before the backend
always_size_sections function is called.
One function, ppc64_elf_tls_setup must run before
bfd_elf_size_dynamic_sections because it changes dynamic symbols.
ppc64_elf_edit_opd and ppc64_elf_inline_plt can run early or late, I
think. ppc64_elf_tls_optimize and ppc64_elf_edit_toc are better run
later.
So this patch arranges to call some edit functions later via
always_size_sections.
PR 26655
bfd/
* elf64-ppc.c (ppc64_elf_func_desc_adjust): Rename to..
(ppc64_elf_edit): Call params->edit.
(ppc64_elf_tls_setup): Don't call _bfd_elf_tls_setup. Return a
bfd_boolean.
* elf64-ppc.h (struct ppc64_elf_params): Add "edit".
(ppc64_elf_tls_setup): Update declaration.
ld/
* emultempl/ppc64elf.em (params): Add ppc_edit.
(ppc_before_allocation): Split off some edit functions to..
(ppc_edit): ..this, new function.
Add option parser for disassembler, and refine the codes of
parse register operand and disassemble register operand.
While strengthen the operands legality check of some instructions.
Co-Authored-By: Lifang Xia <lifang_xia@c-sky.com>
gas/
* config/tc-csky.c (parse_type_ctrlreg): Use function
csky_get_control_regno to operand.
(csky_get_reg_val): Likewise.
(is_reg_sp_with_bracket): Use function csky_get_reg_val
to parse operand.
(is_reg_sp): Refine.
(is_oimm_within_range): Fix, report error when operand
is not constant.
(parse_type_cpreg): Refine.
(parse_type_cpcreg): Refine.
(get_operand_value): Add handle of OPRND_TYPE_IMM5b_LS.
(md_assemble): Fix no error reporting somtimes when
operands number are not fit.
(csky_addc64): Refine.
(csky_subc64): Refine.
(csky_or64): Refine.
(v1_work_fpu_fo): Refine.
(v1_work_fpu_read): Refine.
(v1_work_fpu_writed): Refine.
(v1_work_fpu_readd): Refine.
(v2_work_addc): New function, strengthen the operands legality
check of addc.
* gas/testsuite/gas/csky/all.d : Use register number format when
disassemble register name by default.
* gas/testsuite/gas/csky/cskyv2_all.d : Likewise.
* gas/testsuite/gas/csky/trust.d: Likewise.
* gas/testsuite/gas/csky/cskyv2_ck860.d : Fix.
* gas/testsuite/gas/csky/trust.s : Fix.
opcodes/
* csky-dis.c (using_abi): New.
(parse_csky_dis_options): New function.
(get_gr_name): New function.
(get_cr_name): New function.
(csky_output_operand): Use get_gr_name and get_cr_name to
disassemble and add handle of OPRND_TYPE_IMM5b_LS.
(print_insn_csky): Parse disassembler options.
* opcodes/csky-opc.h (OPRND_TYPE_IMM5b_LS): New enum.
(GENARAL_REG_BANK): Define.
(REG_SUPPORT_ALL): Define.
(REG_SUPPORT_ALL): New.
(ASH): Define.
(REG_SUPPORT_A): Define.
(REG_SUPPORT_B): Define.
(REG_SUPPORT_C): Define.
(REG_SUPPORT_D): Define.
(REG_SUPPORT_E): Define.
(csky_abiv1_general_regs): New.
(csky_abiv1_control_regs): New.
(csky_abiv2_general_regs): New.
(csky_abiv2_control_regs): New.
(get_register_name): New function.
(get_register_number): New function.
(csky_get_general_reg_name): New function.
(csky_get_general_regno): New function.
(csky_get_control_reg_name): New function.
(csky_get_control_regno): New function.
(csky_v2_opcodes): Prefer two oprerans format for bclri and
bseti, strengthen the operands legality check of addc, zext
and sext.
PR symtab/25470 points out that the Zig programming language allows
integers of various bit sizes (including zero), not just sizes that
are a multiple of 8.
This is supported in DWARF by applying both a byte size and a
DW_AT_bit_size.
This patch adds support for this feature to integer and boolean types.
Other base types are not handled -- for floating-point types, this
didn't seem to make sense, and for character types I didn't see much
need. (These can be added later if desired.)
I've also added support for DW_AT_data_bit_offset at the same time. I
don't know whether the Zig compiler requires this, but it was
described in the same section in the DWARF standard and was easy to
add.
A new test case is supplied, using the DWARF assembler.
gdb/ChangeLog
2020-09-23 Tom Tromey <tom@tromey.com>
PR symtab/25470:
* value.c (unpack_long, pack_long, pack_unsigned_long): Handle bit
offset and bit size.
* printcmd.c (print_scalar_formatted): Handle zero-length
integer.
(print_scalar_formatted): Use bit_size_differs_p.
* gdbtypes.h (enum type_specific_kind) <TYPE_SPECIFIC_INT>: New
constant.
(union type_specific): <int_stuff>: New member.
(struct type) <bit_size_differs_p, bit_size, bit_offset>: New
methods.
* gdbtypes.c (init_integer_type, init_boolean_type): Initialize
TYPE_SPECIFIC_FIELD.
(recursive_dump_type, copy_type_recursive): Update.
* dwarf2/read.c (read_base_type): Handle DW_AT_bit_size and
DW_AT_data_bit_offset.
gdb/testsuite/ChangeLog
2020-09-23 Tom Tromey <tom@tromey.com>
* gdb.dwarf2/intbits.exp: New file.
* gdb.dwarf2/intbits.c: New file.
This changes gdb/compile to use gdb_argv directly, rather than
manually managing the arrays itself. A few new helpers are added to
gdb_argv.
gdb/ChangeLog
2020-09-23 Tom Tromey <tom@tromey.com>
* utils.h (class gdb_argv): Add move operators.
<append>: New methods.
* compile/compile.c (build_argc_argv): Remove.
(compile_args_argc): Remove.
(compile_args_argv): Change type.
(set_compile_args): Simplify.
(append_args): Remove.
(filter_args): Remove argcp parameter.
(get_args): Return gdb_argv. Simplify.
(compile_to_object): Update.
This simplifies compile_module cleanup by removing the need to
explicitly free anything. struct setup_sections_data is also cleaned
up a bit.
gdb/ChangeLog
2020-09-23 Tom Tromey <tom@tromey.com>
* compile/compile-object-run.c (do_module_cleanup)
<~do_module_cleanup> :Remove.
(do_module_cleanup): Update.
* compile/compile-object-load.h (struct munmap_list): Add move
assignment operator.
<source_file>: Now a std::string.
<munmap_list>: Rename. No longer a pointer.
* compile/compile-object-load.c (struct setup_sections_data): Add
constructor.
<setup_one_section>: Declare.
<munmap_list>: Move earlier.
<m_bfd>: New member.
<m_last_size, m_last_section_first, m_last_prot,
m_last_max_alignment>: Rename, add initializers where needed.
(setup_sections_data::setup_one_section): Rename from
setup_sections. Update.
(compile_object_load): Update. Don't use bfd_map_over_sections.
This changes the do_module_cleanup structure to simply hold on to the
module itself. This lets us remove most members from
do_module_cleanup.
gdb/ChangeLog
2020-09-23 Tom Tromey <tom@tromey.com>
* compile/compile-object-run.c (struct do_module_cleanup): Add
parameters to constructor. Update destructor.
<source_file, scope, scope_data, out_value_type, out_value_addr,
munmap_list_head, objfile_name_string>: Remove.
<module>: New member.
(do_module_cleanup): Update.
(compile_object_run): Update.
This introduces compile_module_up, a unique pointer for
compile_module, and changes a few spots to use it.
gdb/ChangeLog
2020-09-23 Tom Tromey <tom@tromey.com>
* compile/compile.c (eval_compile_command): Update.
* compile/compile-object-run.h (compile_object_run): Take a
compile_module_up.
* compile/compile-object-run.c (compile_object_run): Take a
compile_module_up.
* compile/compile-object-load.h (struct compile_module): Add
constructor, destructor.
(compile_module_up): New typedef.
(compile_object_load): Return compile_object_up.
* compile/compile-object-load.c (compile_object_load): Return
compile_module_up.
This changes do_module_cleanup to use new and delete. It also removes
the use of the struct hack from this object -- this requires more
allocations for now, but this will be removed in a subsequent patch.
gdb/ChangeLog
2020-09-23 Tom Tromey <tom@tromey.com>
* compile/compile-object-run.c (struct do_module_cleanup): Add
constructor, destructor.
<objfile_name_string>: Don't use struct hack.
(do_module_cleanup): Use delete.
(compile_object_run): Use new.
This changes gdb's compile code to use std::vector in a couple of
places, rather than manual memory management.
gdb/ChangeLog
2020-09-23 Tom Tromey <tom@tromey.com>
* compile/compile-cplus-types.c
(compile_cplus_convert_struct_or_union): Use std::vector.
(compile_cplus_convert_func): Likewise.
* compile/compile-c-types.c (convert_func): Use std::vector.
sparc can fail at inline prologue skipping. Andrew Burgess tracked
this down to sparc32_skip_prologue, which should use
skip_prologue_using_sal rather than its hand-rolled variant.
I don't have a good way to test this with the gdb test suite (is there
a board file for using qemu? That would help), but it fixes a
regression in the internal AdaCore test suite. We've had this patch
internally at AdaCore for a while, but I just now finally got around
to making sure that backing it out reintroduces the problem.
gdb/ChangeLog
2020-09-21 Tom Tromey <tromey@adacore.com>
* sparc-tdep.c (sparc32_skip_prologue): Use
skip_prologue_using_sal.
This patch corrects "size" and "bitsize" in R_RISCV_RVC_* reloc howtos
so that elfnn-riscv.c:perform_relocation doesn't access past the end
of a section. I've also corrected "size" in the R_RISCV_CALL* reloc
howtos since these relocs apply to two consecutive instructions. That
caused fallout in the assembler with complaints about "fixup not
contained within frag" due to tc-riscv.c:append_insn finishing off a
frag after the auipc insn making up a "call" macro. Which is a little
rude since the CALL reloc also relocates the following jalr. Fixed by
changing the frag handling a little.
I've also changed R_RISCV_ALIGN and R_RISCV_TPREL_ADD marker reloc
howtos to look like R_RISCV_NONE, and corrected dst_mask for numerous
relocs, not that it matters very much.
bfd/
PR 26569
* elfxx-riscv.c (howto_table): Correct size and bitsize of
R_RISCV_RVC_BRANCH, R_RISCV_RVC_JUMP, and R_RISCV_RVC_LUI.
Correct size for R_RISCV_TLS_DTPMOD32, R_RISCV_TLS_DTPREL32,
R_RISCV_CALL, and R_RISCV_CALL_PLT. Make R_RISCV_TPREL_ADD and
R_RISCV_ALIGN like R_RISCV_NONE. Correct dst_mask many relocs.
gas/
* config/tc-riscv.c (append_insn): Don't tie off frags at CALL
relocs.
(riscv_call): Tie them off after the jalr.
(md_apply_fix): Zero fx_size of RELAX fixup.
Tests that use a secondary MI channel (i.e., either tests that call
mi_gdb_start with separate-mi-tty, or all tests when
FORCE_SEPARATE_MI_TTY=1 is specified on the make check command line),
don't close GDB correctly.
E.g., if you run gdb.mi/mi-exec-run.exp in a loop:
while true; do make check TESTS="gdb.mi/mi-exec-run.exp"; done
you can see more than one gdb running at the same time:
$ ps -ef | grep -v grep | grep "gdb/gdb"
pedro 40507 1 7 15:47 ? 00:00:00 /home/pedro/gdb/build/gdb/testsuite/../../gdb/gdb -nw -nx -data-directory /home/pedro/gdb/build/gdb/testsuite/../data-directory
pedro 40562 1 0 15:47 ? 00:00:00 /home/pedro/gdb/build/gdb/testsuite/../../gdb/gdb -nw -nx -data-directory /home/pedro/gdb/build/gdb/testsuite/../data-directory
pedro 40727 1 0 15:47 ? 00:00:00 /home/pedro/gdb/build/gdb/testsuite/../../gdb/gdb -nw -nx -data-directory /home/pedro/gdb/build/gdb/testsuite/../data-directory
pedro 40786 1 0 15:47 ? 00:00:00 /home/pedro/gdb/build/gdb/testsuite/../../gdb/gdb -nw -nx -data-directory /home/pedro/gdb/build/gdb/testsuite/../data-directory
This commit fixes it.
gdb/testsuite/ChangeLog:
* lib/mi-support.exp (mi_uncatched_gdb_exit) Switch to the main
spawn_id before calling remote_close. Close secondary MI channel.
This changes generic_load to avoid bfd_map_over_sections, in favor of
iteration.
gdb/ChangeLog
2020-09-19 Tom Tromey <tom@tromey.com>
* symfile.c (add_section_size_callback): Remove.
(load_one_section): Rename from load_section_callback. Change
parameters.
(generic_load): Use foreach.
This changes build_section_table to avoid bfd_map_over_sections, in
favor of iteration. In this situation it seemed simple to just remove
the helper function entirely.
gdb/ChangeLog
2020-09-19 Tom Tromey <tom@tromey.com>
* exec.c (add_to_section_table): Remove.
(build_section_table): Use foreach.
This changes elf_symfile_read to avoid bfd_map_over_sections, in favor
of iteration.
gdb/ChangeLog
2020-09-19 Tom Tromey <tom@tromey.com>
* elfread.c (elf_locate_sections): Change parameters.
(elf_symfile_read): Use foreach.
This changes restore_command to avoid bfd_map_over_sections, in favor
of iteration. A helper data structure can also be removed by this
patch.
gdb/ChangeLog
2020-09-19 Tom Tromey <tom@tromey.com>
* cli/cli-dump.c (struct callback_data): Remove.
(restore_one_section): Rename from restore_section_callback.
Change parameters.
(restore_binary_file): Change parameters.
(restore_command): Use foreach.
This changes gcore_memory_sections to avoid bfd_map_over_sections, in
favor of iteration.
gdb/ChangeLog
2020-09-19 Tom Tromey <tom@tromey.com>
* gcore.c (make_output_phdrs): Remove 'ignored' parameter.
(gcore_copy_callback): Likewise.
(gcore_memory_sections): Use foreach.
This changes some ELF osabi tag-sniffing functions to avoid
bfd_map_over_sections, in favor of iteration. I could only readily
test the generic one.
gdb/ChangeLog
2020-09-19 Tom Tromey <tom@tromey.com>
* osabi.h (generic_elf_osabi_sniff_abi_tag_sections): Update.
* osabi.c (generic_elf_osabi_sniff_abi_tag_sections): Change
parameters.
(generic_elf_osabi_sniffer): Use foreach.
* mips-sde-tdep.c (mips_sde_elf_osabi_sniffer): Use foreach.
* arm-tdep.c (arm_elf_osabi_sniffer): Use foreach.
This changes some functions in dwarf2/read.c to avoid
bfd_map_over_sections, in favor of iteration.
gdb/ChangeLog
2020-09-19 Tom Tromey <tom@tromey.com>
* dwarf2/read.c (locate_dwz_sections): Change parameters.
(dwarf2_get_dwz_file): Use foreach.
(dwarf2_locate_dwo_sections): Change parameters.
(open_and_init_dwo_file): Use foreach.
(dwarf2_locate_common_dwp_sections): Change parameters.
(open_and_init_dwp_file): Use foreach.
This changes some functions in symfile.c to avoid
bfd_map_over_sections, in favor of iteration. Some helper types can
also be removed due to this change.
gdb/ChangeLog
2020-09-19 Tom Tromey <tom@tromey.com>
* symfile.h: (find_lowest_section): Don't declare.
* symfile.c (find_lowest_section): Now static. Change
parameters.
(struct place_section_arg): Remove.
(place_section): Change parameters.
(addr_info_make_relative): Use foreach.
(symfile_dummy_outputs): Remove.
(default_symfile_relocate): Use foreach.
This changes build_objfile_section_table to avoid
bfd_map_over_sections, in favor of iteration.
gdb/ChangeLog
2020-09-19 Tom Tromey <tom@tromey.com>
* objfiles.c (add_to_objfile_sections): Rename from
add_to_objfile_sections_full.
(add_to_objfile_sections): Remove.
(build_objfile_section_table): Use foreach.
This changes get_stap_base_address to avoid bfd_map_over_sections, in
favor of iteration.
gdb/ChangeLog
2020-09-19 Tom Tromey <tom@tromey.com>
* stap-probe.c (get_stap_base_address_1): Remove.
(get_stap_base_address): Use foreach.
This changes gdb_bfd_close_or_warn to avoid bfd_map_over_sections, in
favor of iteration.
gdb/ChangeLog
2020-09-19 Tom Tromey <tom@tromey.com>
* gdb_bfd.c (free_one_bfd_section): Remove 'abfd' and 'ignore'
parameters.
(gdb_bfd_close_or_warn): Use foreach.
This changes core_target_open to avoid bfd_map_over_sections, in favor
of iteration.
gdb/ChangeLog
2020-09-19 Tom Tromey <tom@tromey.com>
* corelow.c (add_to_thread_list): Change parameters.
(core_target_open): Use foreach.
This adds a new overload of gdb_bfd_sections, that accepts a
gdb_bfd_ref_ptr. This also fixes the formatting of the existing
function, since I happened to notice it was mildly off.
gdb/ChangeLog
2020-09-19 Tom Tromey <tom@tromey.com>
* gdb_bfd.h (gdb_bfd_sections): New overload. Fix formatting of
existing function.
This commit makes the whitespace usage when printing Fortran arrays
more consistent, and more inline with how we print C arrays.
Currently a 2 dimensional Fotran array is printed like this, I find
the marked whitespace unpleasant:
(( 1, 2, 3) ( 4, 5, 6) )
^ ^ ^
After this commit the same array is printed like this:
((1, 2, 3) (4, 5, 6))
Which seems more inline with how we print C arrays, in the case of C
arrays we don't add extra whitespace before the first element.
gdb/ChangeLog:
* f-valprint.c (f77_print_array_1): Adjust printing of whitespace
for arrays.
gdb/testsuite/ChangeLog:
* gdb.fortran/array-slices.exp: Update expected results.
* gdb.fortran/class-allocatable-array.exp: Likewise.
* gdb.fortran/multi-dim.exp: Likewise.
* gdb.fortran/vla-type.exp: Likewise.
* gdb.mi/mi-vla-fortran.exp: Likewise.
The Fortran specific OP_F77_UNDETERMINED_ARGLIST is currently handled
in the generic expression handling code. There's no reason why this
should be the case, so this commit moves handling of this into Fortran
specific files.
There should be no user visible changes after this commit.
gdb/ChangeLog:
* eval.c: Remove 'f-lang.h' include.
(value_f90_subarray): Moved to f-lang.c.
(eval_call): Renamed to...
(evaluate_subexp_do_call): ...this, is no longer static, header
comment moved into header file.
(evaluate_funcall): Update call to eval_call.
(skip_undetermined_arglist): Moved to f-lang.c.
(fortran_value_subarray): Likewise.
(evaluate_subexp_standard): OP_F77_UNDETERMINED_ARGLIST handling
moved to evaluate_subexp_f.
(calc_f77_array_dims): Moved to f-lang.c
* expprint.c (print_subexp_funcall): New function.
(print_subexp_standard): OP_F77_UNDETERMINED_ARGLIST handling
moved to print_subexp_f, OP_FUNCALL uses new function.
(dump_subexp_body_funcall): New function.
(dump_subexp_body_standard): OP_F77_UNDETERMINED_ARGLIST handling
moved to dump_subexp_f, OP_FUNCALL uses new function.
* expression.h (evaluate_subexp_do_call): Declare.
* f-lang.c (value_f90_subarray): Moved from eval.c.
(skip_undetermined_arglist): Likewise.
(calc_f77_array_dims): Likewise.
(fortran_value_subarray): Likewise.
(evaluate_subexp_f): Add OP_F77_UNDETERMINED_ARGLIST support.
(operator_length_f): Likewise.
(print_subexp_f): Likewise.
(dump_subexp_body_f): Likewise.
* fortran-operator.def (OP_F77_UNDETERMINED_ARGLIST): Move
declaration of this operation to here.
* parse.c (operator_length_standard): OP_F77_UNDETERMINED_ARGLIST
support moved to operator_length_f.
* parser-defs.h (dump_subexp_body_funcall): Declare.
(print_subexp_funcall): Declare.
* std-operator.def (OP_F77_UNDETERMINED_ARGLIST): Moved to
fortran-operator.def.
This commit is a refactor of part of the Fortran array and string
handling code.
The current code is split into two blocks, linked, weirdly, with a
goto. After this commit all the code is moved to its own function,
and arrays and strings are now handled using the same code; this will
be useful later when I want to add array stride support where strings
will want to be treated just like arrays, but is a good clean up even
without the array stride work, which is why I'm merging it now.
For now the new function is added as a static within eval.c, even
though the function is Fortran only. A following commit will remove
some of the Fortran specific code from eval.c into one of the Fortran
specific files, including this new function.
There should be no user visible changes after this commit.
gdb/ChangeLog:
* eval.c (fortran_value_subarray): New function, content is taken
from...
(evaluate_subexp_standard): ...here, in two places. Now arrays
and strings both call the new function.
(calc_f77_array_dims): Add header comment, handle strings.
With -m32 -fcf-protection, GCC generates an `endbr32` instruction at the
function entry:
[hjl@gnu-cfl-2 gdb]$ cat /tmp/x.c
int
main(void)
{
return 0;
}
[hjl@gnu-cfl-2 gdb]$ gcc -g -fcf-protection /tmp/x.c -m32
(gdb) b main
Breakpoint 1 at 0x8049176: file /tmp/x.c, line 3.
(gdb) r
Breakpoint 1, main () at /tmp/x.c:3
3 {
(gdb) disass
Dump of assembler code for function main:
=> 0x08049176 <+0>: endbr32
0x0804917a <+4>: push %ebp
0x0804917b <+5>: mov %esp,%ebp
0x0804917d <+7>: mov $0x0,%eax
0x08049182 <+12>: pop %ebp
0x08049183 <+13>: ret
End of assembler dump.
(gdb)
Update i386_analyze_prologue to skip `endbr32`:
(gdb) b main
Breakpoint 1 at 0x804917d: file /tmp/x.c, line 4.
(gdb) r
Breakpoint 1, main () at /tmp/x.c:4
4 return 0;
(gdb) disass
Dump of assembler code for function main:
0x08049176 <+0>: endbr32
0x0804917a <+4>: push %ebp
0x0804917b <+5>: mov %esp,%ebp
=> 0x0804917d <+7>: mov $0x0,%eax
0x08049182 <+12>: pop %ebp
0x08049183 <+13>: ret
End of assembler dump.
(gdb)
Tested with
$ make check RUNTESTFLAGS="--target_board='unix{-m32,}' i386-prologue-skip-cf-protection.exp"
on Fedora 32/x86-64.
2020-0X-YY Victor Collod <vcollod@nvidia.com>
gdb/ChangeLog:
PR gdb/26635
* i386-tdep.c (i386_skip_endbr): Add a helper function to skip endbr.
(i386_analyze_prologue): Call i386_skip_endbr.
gdb/testsuite/ChangeLog:
PR gdb/26635
* gdb.arch/amd64-prologue-skip-cf-protection.exp: Make the test
compatible with i386, and move it to...
* gdb.arch/i386-prologue-skip-cf-protection.exp: ... here.
* gdb.arch/amd64-prologue-skip-cf-protection.c: Move to...
* gdb.arch/i386-prologue-skip-cf-protection.c: ... here.
In this commit:
commit 6108fd1823
Date: Thu Sep 17 11:47:50 2020 -0600
Use htab_up in type copying
A use after free bug was introduced. In compile-object-run.c, in the
function compile_object_run, the code used to look like this:
htab_t copied_types;
/* .... snip .... */
/* OBJFILE may disappear while FUNC_TYPE still will be in use. */
copied_types = create_copied_types_hash (objfile);
func_type = copy_type_recursive (objfile, func_type, copied_types);
htab_delete (copied_types);
/* .... snip .... */
call_function_by_hand_dummy (func_val, NULL, args,
do_module_cleanup, data);
The copied_types table exists on the obstack of objfile, but is
deleted once the call to copy_type_recursive has been completed.
After the change the code now looks like this:
/* OBJFILE may disappear while FUNC_TYPE still will be in use. */
htab_up copied_types = create_copied_types_hash (objfile);
func_type = copy_type_recursive (objfile, func_type, copied_types.get ());
/* .... snip .... */
call_function_by_hand_dummy (func_val, NULL, args,
do_module_cleanup, data);
The copied_types is now a unique_ptr and deleted automatically when it
goes out of scope.
The problem however is that objfile, and its included obstack, may be
deleted by the call to do_module_cleanup, which is called by
call_function_by_hand_dummy.
This means that in the new code the objfile, and its obstack, are
deleted before copied_types is deleted, and as copied_types is on the
objfiles obstack, we are now reading undefined memory.
The solution in this commit is to wrap the call to
create_copied_types_hash and copy_type_recursive into a new static
helper function. The htab_up will then be deleted within the new
function's scope, before objfile is deleted.
This resolves some non-deterministic test failures I was seeing in
gdb.compile/*.exp tests.
gdb/ChangeLog:
* compile/compile-object-run.c (create_copied_type_recursive): New
function.
(compile_object_run): Use new function.
Add gas and opcodes support for two xBPF-exclusive ALU operations:
SDIV (signed division) and SMOD (signed modulo), and add tests for
them in gas.
cpu/
* bpf.cpu (insn-op-code-alu): Add SDIV and SMOD.
(define-alu-insn-bin, daib): Take ISAs as an argument.
(define-alu-instructions): Update calls to daib pmacro with
ISAs; add sdiv and smod.
gas/
* testsuite/gas/bpf/alu-xbpf.d: New file.
* testsuite/gas/bpf/alu-xbpf.s: Likewise.
* testsuite/gas/bpf/alu32-xbpf.d: Likewise.
* testsuite/gas/bpf/alu32-xbpf.d: Likewise.
* testuiste/gas/bpf/bpf.exp: Run new tests.
opcodes/
* bpf-desc.c: Regenerate.
* bpf-desc.h: Likewise.
* bpf-opc.c: Likewise.
* bpf-opc.h: Likewise.
