In addition to the existing relocs we need two more to mark all
instructions in the call sequence, PLTCALL on the call itself (plus
the toc restore insn for ppc64), and PLTSEQ on others. All
relocations in a particular sequence have the same symbol.
Example ppc64 ELFv2 assembly:
.reloc .,R_PPC64_PLTSEQ,puts
std 2,24(1)
addis 12,2,puts@plt@ha # .reloc .,R_PPC64_PLT16_HA,puts
ld 12,puts@plt@l(12) # .reloc .,R_PPC64_PLT16_LO_DS,puts
.reloc .,R_PPC64_PLTSEQ,puts
mtctr 12
.reloc .,R_PPC64_PLTCALL,puts
bctrl
ld 2,24(1)
Example ppc32 -fPIC assembly:
addis 12,30,puts+32768@plt@ha # .reloc .,R_PPC_PLT16_HA,puts+0x8000
lwz 12,12,puts+32768@plt@l # .reloc .,R_PPC_PLT16_LO,puts+0x8000
.reloc .,R_PPC_PLTSEQ,puts+32768
mtctr 12
.reloc .,R_PPC_PLTCALL,puts+32768
bctrl
Marking sequences like this allows the linker to convert them to nops
and a direct call if the target symbol turns out to be local.
When the call is __tls_get_addr, each relocation shown above is paired
with an R_PPC*_TLSLD or R_PPC*_TLSGD reloc to additionally mark the
sequence for possible TLS optimization. The TLSLD or TLSGD relocs are
emitted first.
include/
* elf/ppc.h (R_PPC_PLTSEQ, R_PPC_PLTCALL): Define.
* elf/ppc64.h (R_PPC64_PLTSEQ, R_PPC64_PLTCALL): Define.
bfd/
* elf32-ppc.c (ppc_elf_howto_raw): Add PLTSEQ and PLTCALL howtos.
(is_plt_seq_reloc): New function.
(ppc_elf_check_relocs): Handle PLTSEQ and PLTCALL relocs.
(ppc_elf_tls_optimize): Handle inline plt call sequence.
(ppc_elf_relax_section): Handle PLTCALL reloc.
(ppc_elf_relocate_section): Nop out inline plt call sequence when
resolving locally.
* elf64-ppc.c (ppc64_elf_howto_raw): Add R_PPC64_PLTSEQ and
R_PPC64_PLTCALL entries. Comment R_PPC64_TOCSAVE.
(has_tls_get_addr_call): Correct comment.
(is_branch_reloc): Add PLTCALL.
(is_plt_seq_reloc): New function.
(ppc64_elf_check_relocs): Handle PLT16_LO_DS reloc. Set
has_tls_reloc for R_PPC64_TLSGD and R_PPC64_TLSLD. Create plt
entry for R_PPC64_PLTCALL.
(ppc64_elf_tls_optimize): Handle inline plt call sequence.
(ppc_type_of_stub): Handle PLTCALL reloc.
(toc_adjusting_stub_needed): Likewise.
(ppc64_elf_relocate_section): Set "can_plt_call" for PLTCALL
reloc insn. Nop out inline plt call sequence when resolving
locally. Handle __tls_get_addr inline plt call optimization.
elfcpp/
* powerpc.h (R_POWERPC_PLTSEQ, R_POWERPC_PLTCALL): Define.
gold/
* powerpc.cc (Target_powerpc::Track_tls::maybe_skip_tls_get_addr_call):
Handle inline plt sequence relocs.
(Stub_table::Plt_stub_key::Plt_stub_key): Likewise.
(Target_powerpc::Scan::reloc_needs_plt_for_ifunc): Likewise.
(Target_powerpc::Relocate::relocate): Likewise.
Necessary if gcc is to use PLT16 relocs to implement -mlongcall, and
there isn't a good technical reason why local symbols should be
excluded from PLT16 support. Non-ifunc local symbol PLT entries go in
a separate section to other PLT entries. In a fixed position
executable they won't need to be relocated, and in a PIE or shared
library I chose to not implement lazy relocation.
bfd/
* elf64-ppc.c (LOCAL_PLT_ENTRY_SIZE): Define.
(struct ppc_stub_hash_entry): Add symtype field.
(PLT_KEEP): Define.
(struct ppc_link_hash_table): Add pltlocal and relpltlocal.
(create_linkage_sections): Create pltlocal and relpltlocal.
(ppc64_elf_check_relocs): Allow PLT relocs on local symbols.
Set PLT_KEEP.
(ppc64_elf_adjust_dynamic_symbol): Keep PLT entries for inline calls.
(allocate_dynrelocs): Allocate pltlocal and relpltlocal.
(ppc64_elf_size_dynamic_sections): Size pltlocal and relpltlocal.
Keep PLT entries for inline calls against locals.
(ppc_build_one_stub): Use pltlocal as appropriate.
(ppc_size_one_stub): Likewise.
(ppc64_elf_size_stubs): Set symtype.
(build_global_entry_stubs_and_plt): Init pltlocal and write
relpltlocal for globals.
(write_plt_relocs_for_local_syms): Likewise for local syms.
(ppc64_elf_relocate_section): Support PLT for local syms.
* elf32-ppc.c (PLT_KEEP): Define.
(struct ppc_elf_link_hash_table): Add pltlocal and relpltlocal.
(ppc_elf_create_glink): Create pltlocal and relpltlocal.
(ppc_elf_check_relocs): Allow PLT relocs on local symbols.
Set PLT_KEEP. Adjust update_local_sym_info call.
(ppc_elf_adjust_dynamic_symbol): Keep PLT entries for inline calls.
(allocate_dynrelocs): Allocate pltlocal and relpltlocal.
(ppc_elf_size_dynamic_sections): Size pltlocal and relpltlocal.
(ppc_elf_relocate_section): Support PLT16 relocs for local syms.
(write_global_sym_plt): Init pltlocal and write relpltlocal.
(ppc_finish_symbols): Likewise for locals.
ld/
* emulparams/elf32ppc.sh (OTHER_RELRO_SECTIONS_2): Add .branch_lt.
(OTHER_GOT_RELOC_SECTIONS): Add .rela.branch_lt.
* testsuite/ld-powerpc/elfv2so.d: Update for symbol/stub reordering.
* testsuite/ld-powerpc/relbrlt.d: Likewise.
* testsuite/ld-powerpc/relbrlt.s: Likewise.
* testsuite/ld-powerpc/tlsso.r: Likewise.
* testsuite/ld-powerpc/tlstocso.r: Likewise.
gold/
* powerpc.cc (Target_powerpc::lplt_): New variable.
(Target_powerpc::lplt_section): Associated accessor.
(Target_powerpc::plt_off): Handle local non-ifunc symbols.
(Target_powerpc::make_lplt_section): New function.
(Target_powerpc::make_local_plt_entry): New function.
(Powerpc_relobj::do_relocate_sections): Write out lplt.
(Output_data_plt_powerpc::first_plt_entry_offset): Zero for lplt.
(Output_data_plt_powerpc::add_local_entry): New function.
(Output_data_plt_powerpc::do_write): Ignore lplt.
(Target_powerpc::make_iplt_section): Make lplt first.
(Target_powerpc::make_brlt_section): Make .branch_lt relro.
(Target_powerpc::Scan::local): Handle PLT16 relocs.
The current scheme where we output PLT relocs for global symbols in
finish_dynamic_symbol, and PLT relocs for local symbols when
outputting stubs does not work if PLT entries are to be used for
inline PLT sequences against non-dynamic globals or local symbols.
bfd/
* elf64-ppc.c (ppc_build_one_stub): Move output of PLT relocs
for local symbols to..
(write_plt_relocs_for_local_syms): ..here. New function.
(ppc64_elf_finish_dynamic_symbol): Move output of PLT relocs for
global symbols to..
(build_global_entry_stubs_and_plt): ..here. Rename from
build_global_entry_stubs.
(ppc64_elf_build_stubs): Always call build_global_entry_stubs_and_plt.
Call write_plt_relocs_for_local_syms.
* elf32-ppc.c (get_sym_h): New function.
(ppc_elf_relax_section): Use get_sym_h.
(ppc_elf_relocate_section): Move output of PLT relocs and glink
stubs for local symbols to..
(ppc_finish_symbols): ..here. New function.
(ppc_elf_finish_dynamic_symbol): Move output of PLT relocs for
global syms to..
(write_global_sym_plt): ..here. New function.
* elf32-ppc.h (ppc_elf_modify_segment_map): Delete attribute.
(ppc_finish_symbols): Declare.
ld/
* ppc32elf.em (ppc_finish): Call ppc_finish_symbols.
The PowerPC64 ELFv2 ABI and the PowerPC SysV ABI support a number of
relocations that can be used to create and access a PLT entry.
However, the relocs are not well defined. The PLT16 family of relocs
talk about "the section offset or address of the procedure linkage
table entry". It's plain that we do need a relative address when PIC
as otherwise we'd have dynamic text relocations, but "section offset"
doesn't specify which section. The most obvious one, ".plt", isn't
that useful because there is no readily available way of addressing
the start of the ".plt" section. Much more useful would be "the
GOT/TOC-pointer relative offset of the procedure linkage table entry",
and I suppose you could argue that is a "section offset" of sorts.
For PowerPC64 it is better to use the same TOC-pointer relative
addressing even when non-PIC, since ".plt" may be located outside the
range of a 32-bit address. However, for ppc32 we do want an absolute
address when non-PIC as a GOT pointer may not be set up. Also, for
ppc32 PIC we have a similar situation to R_PPC_PLTREL24 in that the
GOT pointer is set to a location in the .got2 section and we need to
specify the .got2 offset in the PLT16 reloc addend.
This patch supports PLT16 relocations using these semantics. This is
not an ABI change for ppc32 since the relocations were not previously
supported by GNU ld, but is for ppc64 where some of the PLT16 relocs
were supported. I'm not particularly concerned since the old ppc64
PLT16 reloc semantics made them almost completely useless.
bfd/
* elf32-ppc.c (ppc_elf_check_relocs): Handle PLT16 relocs.
(ppc_elf_relocate_section): Likewise.
* elf64-ppc.c (ppc64_elf_check_relocs): Handle PLT16_LO_DS.
(ppc64_elf_relocate_section): Likewise. Correct PLT16
resolution to plt entry relative to toc pointer.
gold/
* powerpc.cc (Target_powerpc::plt_off): New functions.
(is_plt16_reloc): New function.
(Stub_table::plt_off): Use Target_powerpc::plt_off.
(Stub_table::plt_call_size): Use plt_off.
(Stub_table::do_write): Likewise.
(Target_powerpc::Scan::get_reference_flags): Return RELATIVE_REF
for PLT16 relocations.
(Target_powerpc::Scan::reloc_needs_plt_for_ifunc): Return true
for PLT16 relocations.
(Target_powerpc::Scan::global): Make a PLT entry for PLT16 relocations.
(Target_powerpc::Relocate::relocate): Support PLT16 relocations.
(Powerpc_scan_relocatable_reloc::global_strategy): Return RELOC_SPECIAL
for ppc32 plt16 relocs.
It is possible to construct indirect calls to __tls_get_addr in
assembly that confuse TLS optimization. (PowerPC gcc doesn't support
such calls, ignoring -mlongcall for __tls_get_addr.) This patch fixes
the problem by requiring a TLSLD or TLSGD marker reloc before any insn
in an indirect call to __tls_get_addr will be optimized. They also
need additional marker relocs defined in a later patch, so don't
expect the optimization to work just yet. The point here is to
prevent mis-optimization of indirect calls without any marker relocs.
The presense of a marker reloc is tracked by a new bit in the tls_mask
field of ppc_link_hash_entry and the corresponding lgot_masks unsigned
char array for local symbols. Since the field is only 8 bits, we've
run out of space. However, tracking TLS use for variables, and
tracking IFUNC for functions are independent, and bits can be reused.
TLS_TLS is always set for TLS usage, so can be used to select the
meaning of the other bits. This patch does that even for elf32-ppc.c
which hasn't yet run out of space in the field.
* elf64-ppc.c (TLS_TLS, TLS_GD, TLS_LD, TLS_TPREL, TLS_DTPREL,
TLS_TPRELGD, TLS_EXPLICIT): Renumber. Test TLS_TLS throughout
file when other TLS flags are tested in a mask.
(TLS_MARK, NON_GOT): Define.
(PLT_IFUNC): Redefine, and test TLS_TLS throughout file as well.
(update_local_sym_info): Don't create got entry when NON_GOT.
(ppc64_elf_check_relocs): Pass NON_GOT with PLT_IFUNC.
Set TLS_MARK.
(get_tls_mask): Do toc lookup if tls_mask is just TLS_MARK.
(ppc64_elf_relocate_section): Likewise.
(ppc64_elf_tls_optimize): Don't attempt to optimize indirect
__tls_get_addr calls lacking a marker reloc.
* elf32-ppc.c (TLS_TLS, TLS_GD, TLS_LD, TLS_TPREL, TLS_DTPREL,
TLS_TPRELGD): Renumber. Update comment.
(TLS_MARK, NON_GOT): Define.
(PLT_IFUNC): Redefine, and test TLS_TLS throughout file as well.
(update_local_sym_info): Don't create got entry when NON_GOT.
(ppc_elf_check_relocs): Pass NON_GOT with PLT_IFUNC.
Set TLS_MARK.
(ppc_elf_tls_optimize): Don't attempt to optimize indirect
__tls_get_addr calls lacking a marker reloc.
I see some failures in the gdb.mi/mi-stack.exp test. The test runs to
the callee4 function:
int callee4 (void)
{
int A=1;
int B=2;
int C;
int D[3] = {0, 1, 2};
C = A + B;
return 0;
}
and expects to be stopped at the A=1 line. However, when gcc generates
some stack protection code, it will stop at the { instead, as shown by
this disassembly (after I did "break callee4" and "run"):
(gdb) disassemble /s
Dump of assembler code for function callee4:
/home/simark/src/binutils-gdb/gdb/testsuite/gdb.mi/mi-stack.c:
26 {
0x00005555555546ca <+0>: push %rbp
0x00005555555546cb <+1>: mov %rsp,%rbp
0x00005555555546ce <+4>: sub $0x20,%rsp
=> 0x00005555555546d2 <+8>: mov %fs:0x28,%rax
0x00005555555546db <+17>: mov %rax,-0x8(%rbp)
0x00005555555546df <+21>: xor %eax,%eax
27 int A=1; /* callee4 begin */
0x00005555555546e1 <+23>: movl $0x1,-0x20(%rbp)
28 int B=2;
0x00005555555546e8 <+30>: movl $0x2,-0x1c(%rbp)
The rest of the test relies on execution stopping on the A=1, so many things
fail after that. This patch uses mi_continue_to_line instead, to stop at the
A=1 line precisely.
gdb/testsuite/ChangeLog:
* gdb.mi/mi-stack.exp (test_stack_frame_listing): Use
mi_continue_to_line.
* gdb.mi/mi-stack.c (callee4): Add comment.
I put the constructor in tracepoint.c because it needs to read
traceframe_number, and I prefer to do that than to expose
traceframe_number.
gdb/ChangeLog:
* tracepoint.c (struct current_traceframe_cleanup): Remove.
(do_restore_current_traceframe_cleanup): Remove.
(restore_current_traceframe_cleanup_dtor): Remove.
(make_cleanup_restore_current_traceframe): Remove.
(scoped_restore_current_traceframe::scoped_restore_current_traceframe):
New.
* tracepoint.h (struct scoped_restore_current_traceframe): New.
* infrun.c (fetch_inferior_event): Use
scoped_restore_current_traceframe.
This patch removes some usages of get_dwarf2_per_objfile, where we can
get hold of the dwarf2_per_objfile object in a simpler way. For
example, it's simpler (and slightly less work) to pass
dwarf2_per_objfile and get the objfile from it than to pass the objfile
and call get_dwarf2_per_objfile.
Ideally, get_dwarf2_per_objfile should only be used in the entry points
of the dwarf2 code, where we receive an objfile.
gdb/ChangeLog:
* dwarf2read.c (create_cus_from_index_list): Replace objfile arg
with dwarf2_per_objfile.
(create_cus_from_index): Likewise.
(create_signatured_type_table_from_index): Likewise.
(dwarf2_read_index): Likewise.
(dwarf2_initialize_objfile): Likewise.
(dwarf2_fetch_die_loc_sect_off): Get dwarf2_per_objfile from
per_cu rather than get_dwarf2_per_objfile.
Those two functions look like good candidates to become methods of
dwarf2_per_objfile. I did that, and added get_tu as well. When
replacing usages of dw2_get_cutu, I changed some instances to get_cutu
and others to get_cu, when appropriate (when we know we want a CU and
not a TU).
gdb/ChangeLog:
* dwarf2read.h (struct signatured_type): Forward declare.
(struct dwarf2_per_objfile) <get_cutu, get_cu, get_tu>:
New methods.
* dwarf2read.c (dwarf2_per_objfile::get_cutu): Rename from...
(dw2_get_cutu): ...this.
(dwarf2_per_objfile::get_cu): Rename from...
(dw2_get_cu): ...this.
(dwarf2_per_objfile::get_tu): New.
(create_addrmap_from_index): Adjust.
(create_addrmap_from_aranges): Adjust.
(dw2_find_last_source_symtab): Adjust.
(dw2_map_symtabs_matching_filename): Adjust.
(dw2_symtab_iter_next): Adjust.
(dw2_print_stats): Adjust.
(dw2_expand_all_symtabs): Adjust.
(dw2_expand_symtabs_with_fullname): Adjust.
(dw2_expand_marked_cus): Adjust.
(dw_expand_symtabs_matching_file_matcher): Adjust.
(dw2_map_symbol_filenames): Adjust.
(dw2_debug_names_iterator::next): Adjust.
(dwarf2_initialize_objfile): Adjust.
(set_partial_user): Adjust.
(dwarf2_build_psymtabs_hard): Adjust.
Most of them are obvious. The ones in dwarf2_record_block_ranges are
less obvious, because it is a bit suspicious to have that many
variables unused. But after inspection, it seems like it dates from
commit 5f46c5a548 ("Code cleanup: Split dwarf2_ranges_read to a
callback"), where dwarf2_record_block_ranges was made to use
dwarf2_ranges_process, which contains the same functionality.
gdb/ChangeLog:
* dwarf2read.c (create_signatured_type_table_from_debug_names):
Remove unused variables.
(dw2_map_symtabs_matching_filename): Likewise.
(dwarf2_record_block_ranges): Likewise.
(dwarf2_read_addr_index): Likewise.
(follow_die_offset): Likewise.
Using this simple test:
static void
break_here ()
{
}
int
main (int argc, char *argv[])
{
fork ();
break_here();
return 0;
}
compiled as a PIE:
$ gcc test.c -g3 -O0 -o test -pie
and running this:
$ ./gdb -nx -q --data-directory=data-directory ./test -ex "b break_here" -ex "set detach-on-fork off" -ex r
gives:
Warning:
Cannot insert breakpoint 1.
Cannot access memory at address 0x64a
Note that GDB might get stopped by SIGTTOU because of this issue:
https://sourceware.org/bugzilla/show_bug.cgi?id=23020
In that case, just use "fg" to continue.
This issue happens only with position-independent executables. Adding
the main objfile for the new inferior (the fork child) causes GDB to try
to reset the breakpoints. However, that new objfile has not been
relocated yet. So the breakpoint on "break_here" resolves to an
unrelocated address, from which we are trying to read/write to set a
breakpoint. Passing SYMFILE_DEFER_BP_RESET avoids that problem. The
executable is relocated just after, in the follow_fork_inferior
function.
The buildbot seems happy with this patch. I don't think it's necessary
to add a new test. Just changing this made many tests go from FAIL to
PASS on my machine, where gcc produces PIE executables by default. If
anything, I think we would need to add a board file that produces
position-independent executables, so that we can run all the tests with
PIE, even on machines where that is not the default.
gdb/ChangeLog:
* progspace.c (clone_program_space): Pass SYMFILE_DEFER_BP_RESET
to symbol_file_add_main.
Enabling "set debug lin-lwp 1" with the MI interpreter doesn't work.
When the sigchld_handler function wants to print a debug output
("sigchld\n"), it uses ui_file_write_async_safe. This ends up in the
default implementation of ui_file::write_async_safe, which aborts GDB.
This patch implements the write_async_safe method for mi_console_file.
The "normal" MI output is line buffered, which means the output
accumulates in m_buffer until a \n is written, at which point it's
flushed in m_raw. The implementation of write_async_safe provided by
this patch bypasses this buffer and writes directly to m_raw. There are
two reasons for this:
(1) Appending to m_buffer (therefore to an std::string) is probably not
async-safe, as it may allocate memory.
(2) We may have a partial output already in m_buffer, so that would lead
to some nested MI output, not so great.
There is probably still a chance to have bad MI output, if
sigchld_handler is invoked in the middle of mi_console_file's flush, and
the line being flushed is only partially sent to m_raw. The solution
would probably be to block signals during flushing. Since this is only
used for debug output, I don't know if it's worth the effort to do that.
To implement write_async_safe, I needed to use the fputstrn_unfiltered,
which does the necessary escaping (e.g. replace \n with \\n). I started
by adding printchar's callback parameters to fputstrn_unfiltered, to be
able to pass async-safe versions of them. It's not easy to provide an
async-safe version of do_fprintf, but it turns out that we can easily
replace printchar's callbacks with a single do_fputc quite easily. The
async-safe version of do_fputc simply calls the underlying ui_file's
write_async_safe method.
gdb/ChangeLog:
PR mi/22299
* mi/mi-console.c (do_fputc_async_safe): New.
(mi_console_file::write_async_safe): New.
(mi_console_file::flush): Adjust calls to fputstrn_unfiltered.
* mi/mi-console.h (class mi_console_file) <write_async_safe>:
New.
* ui-file.c (ui_file::putstrn): Adjust call to
fputstrn_unfiltered.
* utils.c (printchar): Replace do_fputs and do_fprintf
parameters by do_fputc.
(fputstr_filtered): Adjust call to printchar.
(fputstr_unfiltered): Likewise.
(fputstrn_filtered): Likewise.
(fputstrn_unfiltered): Add do_fputc parameter, pass to
printchar.
* utils.h (do_fputc_ftype): New typedef.
(fputstrn_unfiltered): Add do_fputc parameter.
I noticed that regformats/i386/i386-avx.dat did not get re-generated
when doing "make" in the features directory. I think it's a leftover
from commit
f5a29eb0a6 ("Clean up x86 non-linux GDBserver target descriptions")
I build-tested gdbserver with amd64 and i386.
gdb/ChangeLog:
* regformats/i386/i386-avx.dat: Remove.
When a 64-bits (x86-64) gdbarch is created, it is first born as a
32-bits gdbarch in i386_gdbarch_init. The call gdbarch_init_osabi will
call the handler register for the selected (arch, osabi) pair, such as
amd64_linux_init_abi. The various amd64 handlers call amd64_init_abi,
which turns the gdbarch into a 64-bits one.
When selecting the i386:x86-64 architecture with no osabi, no such
handler is ever called, so the gdbarch stays (wrongfully) a 32-bits one.
My first idea was to manually call amd64_init_abi & al in
i386_gdbarch_init when the osabi is GDB_OSABI_NONE. However, this
doesn't work in a build of GDB where i386 is included as a target but
not amd64. My next option (implemented in this patch), is to allow
registering handlers for GDB_OSABI_NONE. I added two such handlers in
amd64-tdep.c, so now it works the same as for the "normal" osabis. It
required re-ordering things in gdbarch_init_osabi to allow running
handlers for GDB_OSABI_NONE.
Without this patch applied (but with the previous one*) :
(gdb) set osabi none
(gdb) set architecture i386:x86-64
The target architecture is assumed to be i386:x86-64
(gdb) p sizeof(void*)
$1 = 4
and now:
(gdb) set osabi none
(gdb) set architecture i386:x86-64
The target architecture is assumed to be i386:x86-64
(gdb) p sizeof(void*)
$1 = 8
* Before the previous patch, which fixed "set osabi none", this bug was
hidden because we didn't actually try to generate a gdbarch for no
osabi, it would always fall back on Linux. Generating the gdbarch for
amd64/linux did work.
gdb/ChangeLog:
PR gdb/22979
* amd64-tdep.c (amd64_none_init_abi): New function.
(amd64_x32_none_init_abi): New function.
(_initialize_amd64_tdep): Register handlers for x86-64 and
x64_32 with GDB_OSABI_NONE.
* osabi.c (gdbarch_init_osabi): Allow running handlers for the
GDB_OSABI_NONE osabi.
gdb/testsuite/ChangeLog:
PR gdb/22979
* gdb.arch/amd64-osabi.exp: New file.
I was looking for a way to reproduce easily PR 22979 by doing this:
(gdb) set architecture i386:x86-64
(gdb) set osabi none
However, I noticed that even though I did "set osabi none", the gdbarch
gdb created was for Linux:
(gdb) set debug arch 1
(gdb) set architecture i386:x86-64
...
(gdb) set osabi none
gdbarch_find_by_info: info.bfd_arch_info i386:x86-64
gdbarch_find_by_info: info.byte_order 1 (little)
gdbarch_find_by_info: info.osabi 4 (GNU/Linux) <--- Wrong?
gdbarch_find_by_info: info.abfd 0x0
gdbarch_find_by_info: info.tdep_info 0x0
gdbarch_find_by_info: Previous architecture 0x1e6fd30 (i386:x86-64)
selected
gdbarch_update_p: Architecture 0x1e6fd30 (i386:x86-64) unchanged
This is because the value GDB_OSABI_UNKNOWN has an unclear role,
sometimes meaning "no osabi" and sometimes "please selected
automatically". Doing "set osabi none" sets the requested osabi to
GDB_OSABI_UNKNOWN, in which case gdbarch_info_fill overrides it with a
value from the target description, or the built-in default osabi. This
means that it's impossible to force GDB not to use an osabi with "set
osabi". Since my GDB's built-in default osabi is Linux, it always falls
back to GDB_OSABI_LINUX.
To fix it, I introduced GDB_OSABI_NONE, which really means "I don't want
any osabi". GDB_OSABI_UNKNOWN can then be used only for "not set yet,
please auto-detect". GDB_OSABI_UNINITIALIZED now seems unnecessary
since it overlaps with GDB_OSABI_UNKNOWN, so I think it can be removed
and gdbarch_info::osabi can be initialized to GDB_OSABI_UNKNOWN.
gdb/ChangeLog:
PR gdb/22980
* defs.h (enum gdb_osabi): Remove GDB_OSABI_UNINITIALIZED, add
GDB_OSABI_NONE.
* arch-utils.c (gdbarch_info_init): Don't set info->osabi.
* osabi.c (gdb_osabi_names): Add "unknown" entry.
gdb/testsuite/ChangeLog:
PR gdb/22980
* gdb.base/osabi.exp: New file.
This patch started by changing target_read_alloc_1 to return a
byte_vector, to avoid manual memory management (in target_read_alloc_1
and in the callers). To communicate failures to the callers, it
actually returns a gdb::optional<gdb::byte_vector>.
Adjusting target_read_stralloc was a bit more tricky, since it wants to
return a buffer of char, and not gdb_byte. Since you can't just cast a
gdb::byte_vector into a gdb::def_vector<char>, I made
target_read_alloc_1 templated, so both versions (that return vectors of
gdb_byte and char) are generated. Since target_read_stralloc now
returns a gdb::char_vector instead of a gdb::unique_xmalloc_ptr<char>, a
few callers need to be adjusted.
gdb/ChangeLog:
* common/byte-vector.h (char_vector): New type.
* target.h (target_read_alloc): Return
gdb::optional<byte_vector>.
(target_read_stralloc): Return gdb::optional<char_vector>.
(target_get_osdata): Return gdb::optional<char_vector>.
* target.c (target_read_alloc_1): Templatize. Replacement
manual memory management with vector.
(target_read_alloc): Change return type, adjust.
(target_read_stralloc): Change return type, adjust.
(target_get_osdata): Change return type, adjust.
* auxv.c (struct auxv_info) <length>: Remove.
<data>: Change type to gdb::optional<byte_vector>.
(auxv_inferior_data_cleanup): Free auxv_info with delete.
(get_auxv_inferior_data): Allocate auxv_info with new, adjust.
(target_auxv_search): Adjust.
(fprint_target_auxv): Adjust.
* avr-tdep.c (avr_io_reg_read_command): Adjust.
* linux-tdep.c (linux_spu_make_corefile_notes): Adjust.
(linux_make_corefile_notes): Adjust.
* osdata.c (get_osdata): Adjust.
* remote.c (remote_get_threads_with_qxfer): Adjust.
(remote_memory_map): Adjust.
(remote_traceframe_info): Adjust.
(btrace_read_config): Adjust.
(remote_read_btrace): Adjust.
(remote_pid_to_exec_file): Adjust.
* solib-aix.c (solib_aix_get_library_list): Adjust.
* solib-dsbt.c (decode_loadmap): Don't free buf.
(dsbt_get_initial_loadmaps): Adjust.
* solib-svr4.c (svr4_current_sos_via_xfer_libraries): Adjust.
* solib-target.c (solib_target_current_sos): Adjust.
* tracepoint.c (sdata_make_value): Adjust.
* xml-support.c (xinclude_start_include): Adjust.
(xml_fetch_content_from_file): Adjust.
* xml-support.h (xml_fetch_another): Change return type.
(xml_fetch_content_from_file): Change return type.
* xml-syscall.c (xml_init_syscalls_info): Adjust.
* xml-tdesc.c (file_read_description_xml): Adjust.
(fetch_available_features_from_target): Change return type.
(target_fetch_description_xml): Adjust.
(target_read_description_xml): Adjust.
Relocations referring to discarded sections are now treated as errors
instead of warnings.
Also with this patch, we will now print the section group signature and the
object file with the prevailing definition of that group along with the
name of the symbol that the relocation is referring to. This additional
information should be much more useful to anyone trying to track down
the source of such errors.
To do so, we now map each discarded section to the Kept_section info in
the Layout class, and defer the logic that maps a discarded section to
its counterpart in the kept group. This gives us the information we need
to identify the signature symbol given the discarded section, and the
name of the object file that provided the prevailing (i.e., first)
definition of that group.
gold/
* object.cc (Sized_relobj_file::include_section_group): Store
reference to Kept_section info for discarded comdat sections
regardless of size. Move size checking to map_to_kept_section.
(Sized_relobj_file::include_linkonce_section): Likewise.
(Sized_relobj_file::map_to_kept_section): Add section name parameter.
Insert size checking logic from above functions.
(Sized_relobj_file::find_kept_section_object): New method.
(Sized_relobj_file::get_symbol_name): New method.
* object.h (Sized_relobj_file::map_to_kept_section): Add section_name
parameter. Adjust all callers.
(Sized_relobj_file::find_kept_section_object): New method.
(Sized_relobj_file::get_symbol_name): New method.
(Sized_relobj_file::Kept_comdat_section): Replace object and shndx
fields with sh_size, kept_section, symndx, and is_comdat fields.
(Sized_relobj_file::set_kept_comdat_section): Replace kept_object
and kept_shndx parameters with is_comdat, symndx, sh_size, and
kept_section.
(Sized_relobj_file::get_kept_comdat_section): Likewise.
* target-reloc.h (enum Comdat_behavior): Change CB_WARNING to CB_ERROR.
Adjust all references.
(issue_undefined_symbol_error): New function template.
(relocate_section): Pass section name to map_to_kept_section.
Move discarded section code to new function above.
* aarch64.cc (Target_aarch64::scan_reloc_section_for_stubs): Move
declaration for gsym out one level. Call issue_discarded_error.
* arm.cc (Target_arm::scan_reloc_section_for_stubs): Likewise.
* powerpc.cc (Relocate_comdat_behavior): Change CB_WARNING to CB_ERROR.
This changes value::contents to be a unique_xmalloc_ptr, removing a
small bit of manual memory management.
gdb/ChangeLog
2018-04-06 Tom Tromey <tom@tromey.com>
* value.c (~value): Update.
(struct value) <contents>: Now unique_xmalloc_ptr.
(value_contents_bits_eq, allocate_value_contents)
(value_contents_raw, value_contents_all_raw)
(value_contents_for_printing, value_contents_for_printing_const)
(set_value_enclosing_type): Update.
This changes value::parent to a value_ref_ptr. This removes a bit of
manual reference count management.
gdb/ChangeLog
2018-04-06 Tom Tromey <tom@tromey.com>
* value.c (~value): Update.
(struct value) <parent>: Now a value_ref_ptr.
(value_parent, set_value_parent, value_address, value_copy):
Update.
This adds a constructor and destructor to struct value, and then
changes value.c to use "new" and "delete".
While doing this I noticed a memory leak -- value_decref was not
freeing value::optimized_out. This patch fixes this leak.
gdb/ChangeLog
2018-04-06 Tom Tromey <tom@tromey.com>
* value.c (struct value): Add constructor, destructor, and member
initializers.
(allocate_value_lazy, value_decref): Update.
This patch converts all_values to simply hold a list of references to
values. Now, there's no need to have a value record whether or not it
is released -- there is only a single reference-counting mechanism for
values. So, this also removes value::next, value::released, and
value_next.
gdb/ChangeLog
2018-04-06 Tom Tromey <tom@tromey.com>
* value.c (struct value) <released, next>: Remove.
(all_values): Now a std::vector.
(allocate_value_lazy): Update.
(value_next): Remove.
(value_mark, value_free_to_mark, release_value)
(value_release_to_mark): Update.
This patch changes value_release_to_mark and fetch_subexp_value to
return a std::vector of value references, rather than relying on the
"next" field that is contained in a struct value. This makes it
simpler to reason about the returned values, and also allows for the
removal of free_value_chain.
gdb/ChangeLog
2018-04-06 Tom Tromey <tom@tromey.com>
* value.h (fetch_subexp_value, value_release_to_mark): Update.
(free_value_chain): Remove.
* value.c (free_value_chain): Remove.
(value_release_to_mark): Return a std::vector.
* ppc-linux-nat.c (num_memory_accesses): Change "chain" to a
std::vector.
(check_condition): Update.
* eval.c (fetch_subexp_value): Change "val_chain" to a
std::vector.
* breakpoint.c (update_watchpoint): Update.
(can_use_hardware_watchpoint): Change "vals" to a std::vector.
free_all_values is unused, so this removes it.
gdb/ChangeLog
2018-04-06 Tom Tromey <tom@tromey.com>
* value.h (free_all_values): Remove.
* value.c (free_all_values): Remove.
This simplifies the value history implementation by replacing the
current data structure with a std::vector, and by making the value
history simply hold a reference to each value.
gdb/ChangeLog
2018-04-06 Tom Tromey <tom@tromey.com>
* value.c (VALUE_HISTORY_CHUNK, struct value_history_chunk)
(value_history_chain, value_history_count): Remove.
(value_history): New global.
(record_latest_value, access_value_history, show_values)
(preserve_values): Update.
This patch removes some manual reference count manipulation by
changing last_examine_value to be a value_ref_ptr and then updating
the users.
gdb/ChangeLog
2018-04-06 Tom Tromey <tom@tromey.com>
* printcmd.c (last_examine_address): Change type to
value_ref_ptr.
(do_examine, x_command): Update.
Now that value_ref_ptr exists, it is possible to simplify breakpoint
and bpstat memory management by using a value_ref_ptr rather than
manually handling the reference counts.
gdb/ChangeLog
2018-04-06 Tom Tromey <tom@tromey.com>
* value.c (release_value): Update.
* breakpoint.h (struct watchpoint) <val>: Now a value_ref_ptr.
(struct bpstats) <val>: Now a value_ref_ptr.
* breakpoint.c (update_watchpoint, breakpoint_init_inferior)
(~bpstats, bpstats, bpstat_clear_actions, watchpoint_check)
(~watchpoint, print_it_watchpoint, watch_command_1)
(invalidate_bp_value_on_memory_change): Update.
struct value is internally reference counted and so, while it also has
some ownership rules unique to it, it makes sense to use a gdb_ref_ptr
when managing it automatically.
This patch removes the existing unique_ptr specialization in favor of
a reference-counted pointer. It also introduces two other
clarifications:
1. Rename value_free to value_decref, which I think is more in line
with what the function actually does; and
2. Change release_value to return a gdb_ref_ptr. This change allows
us to remove the confusing release_value_or_incref function,
primarily by making it much simpler to reason about the result of
release_value.
gdb/ChangeLog
2018-04-06 Tom Tromey <tom@tromey.com>
* varobj.c (varobj_clear_saved_item)
(update_dynamic_varobj_children, install_new_value, ~varobj):
Update.
* value.h (value_incref): Move declaration earlier.
(value_decref): Rename from value_free.
(struct value_ref_policy): New.
(value_ref_ptr): New typedef.
(struct value_deleter): Remove.
(gdb_value_up): Remove typedef.
(release_value): Change return type.
(release_value_or_incref): Remove.
* value.c (set_value_parent): Update.
(value_incref): Change return type.
(value_decref): Rename from value_free.
(value_free_to_mark, free_all_values, free_value_chain): Update.
(release_value): Return value_ref_ptr.
(release_value_or_incref): Remove.
(record_latest_value, set_internalvar, clear_internalvar):
Update.
* stack.c (info_frame_command): Don't call value_free.
* python/py-value.c (valpy_dealloc, valpy_new)
(value_to_value_object): Update.
* printcmd.c (do_examine): Update.
* opencl-lang.c (lval_func_free_closure): Update.
* mi/mi-main.c (register_changed_p): Don't call value_free.
* mep-tdep.c (mep_frame_prev_register): Don't call value_free.
* m88k-tdep.c (m88k_frame_prev_register): Don't call value_free.
* m68hc11-tdep.c (m68hc11_frame_prev_register): Don't call
value_free.
* guile/scm-value.c (vlscm_free_value_smob)
(vlscm_scm_from_value): Update.
* frame.c (frame_register_unwind, frame_unwind_register_signed)
(frame_unwind_register_unsigned, get_frame_register_bytes)
(put_frame_register_bytes): Don't call value_free.
* findvar.c (address_from_register): Don't call value_free.
* dwarf2read.c (dwarf2_compute_name): Don't call value_free.
* dwarf2loc.c (entry_data_value_free_closure)
(value_of_dwarf_reg_entry, free_pieced_value_closure)
(dwarf2_evaluate_loc_desc_full): Update.
* breakpoint.c (update_watchpoint, breakpoint_init_inferior)
(~bpstats, bpstats, bpstat_clear_actions, watchpoint_check)
(~watchpoint, watch_command_1)
(invalidate_bp_value_on_memory_change): Update.
* alpha-tdep.c (alpha_register_to_value): Don't call value_free.
As shown in PR 23022, building with clang-6 and Python 2 trips on the
fact that the Python 2 headers use the "register" keyword:
/usr/include/python2.7/unicodeobject.h:534:5: error: 'register' storage class specifier is deprecated and incompatible with C++17 [-Werror,-Wdeprecated-register]
register PyObject *obj, /* Object */
^~~~~~~~~
This patch adds -Wno-error=deprecated-register to our flags, so that we can
still see this class of warnings, but they don't cause a build failure.
gdb/ChangeLog:
PR gdb/23022
* warning.m4: Add -Wno-error=deprecated-register.
* configure: Re-generate.
I happened to notice that objdump was not printing "Rust" when showing
the DW_AT_language for a CU:
<10> DW_AT_language : 28 (Unknown: 1c)
This patch adds all the new language constants from DWARF 5 to
binutils/dwarf.c.
2018-04-06 Tom Tromey <tom@tromey.com>
* dwarf.c (read_and_display_attr_value): Add missing DW_LANG
constants from DWARF 5.
R_X86_64_converted_reloc_bit is set in elf_x86_64_convert_load_reloc
which is called from elf_x86_64_check_relocs. Since it is used only
internally by linker, there is no need to mask it out in
elf_x86_64_info_to_howto.
* elf64-x86-64.c (elf_x86_64_info_to_howto): Don't mask out
R_X86_64_converted_reloc_bit.
config/plugins.m4 has
if test "$plugins" = "yes"; then
AC_SEARCH_LIBS([dlopen], [dl])
fi
Plugin uses dlsym, but libasan.so only intercepts dlopen, not dlsym:
[hjl@gnu-tools-1 binutils-text]$ nm -D /lib64/libasan.so.4| grep " dl"
0000000000038580 W dlclose
U dl_iterate_phdr
000000000004dc50 W dlopen
U dlsym
U dlvsym
[hjl@gnu-tools-1 binutils-text]$
Testing dlopen for libdl leads to false negative when -fsanitize=address
is used. It results in link failure:
../bfd/.libs/libbfd.a(plugin.o): undefined reference to symbol 'dlsym@@GLIBC_2.16'
dlsym should be used to check if libdl is needed for plugin.
bfd/
PR gas/22318
* configure: Regenerated.
binutils/
PR gas/22318
* configure: Regenerated.
gas/
PR gas/22318
* configure: Regenerated.
gprof/
PR gas/22318
* configure: Regenerated.
ld/
PR gas/22318
* configure: Regenerated.
Sync with GCC
2018-04-05 H.J. Lu <hongjiu.lu@intel.com>
PR gas/22318
* plugins.m4 (AC_PLUGINS): Use dlsym to check if libdl is needed.
2018-02-14 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
PR target/84148
* cet.m4: Check if target support multi-byte NOPS (SSE).
Fix a typo: `.dc.w' -> `.dc.l' in `strip-13mips64.s', correcting a bug
from commit 2f8ceb3899 ("binutils/testsuite: Support REL and MIPS64
reloc formats with `strip-13'"). For relocation format correctness only
as there is no observable change in test results due to the lack of
connection between the second relocation entry affected and the examined
error message produced.
binutils/
* testsuite/binutils-all/strip-13mips64.s: Use `.dc.l' rather
than `.dc.w' in second relocation.
The record_got_page_entry function records and updates the maximum
number of GOT page entries which may be required by an object. In the
case where an existing GOT page entry was expanded, only the entry
belonging to output GOT would have its page count updated. This leaves
the entry belonging to the object GOT with the num_pages count of 1 it
was originally initialized with. Later on when GOTs are being merged in a
multi-got link, this causes the value of entry->num_pages in
add_got_page_entries to always be 1 and underestimates the number of pages
required for the new entry. This in turn leads to an assertion failure in
get_got_page_offset where we run out of pages.
Fix by obtaining the object's GOT entry unconditionally and not just
the first time it gets created. Now that entry2 is always valid, remove
the useless NULL checks.
gold/
PR gold/22770
* mips.cc (Mips_got_info::record_got_page_entry): Fetch existing
page entries for the object's GOT.
PR 23030
* emulparams/elf64_ia64.sh (OTHER_READONLY_SECTIONS): Make sure
that the .IA_64.unwind_info and .IA_64.unwind sections are not
subject to garbage collection.
linespec.h was inculding vec.h, but doesn't expose any VECs.
So, this include can be removed.
ChangeLog
2018-04-05 Tom Tromey <tom@tromey.com>
* linespec.h: Remove include of "vec.h".
This removes VEC(typep) from linespec.c in favor of std::vector. It
also removes the "typep" typedef. This change allowed the removal of
some cleanups.
I believe the previous cleanup code in find_superclass_methods could
result in a memory leak, so this patch is an improvement in that way
as well.
ChangeLog
2018-04-05 Tom Tromey <tom@tromey.com>
* linespec.c (typep): Remove typedef.
(find_methods, find_superclass_methods): Take a std::vector.
(find_method): Use std::vector.
This changes some spots in linespec.c to take a std::vector. This
patch spilled out to objc-lang.c a bit as well. This change allows
for the removal of some cleanups.
ChangeLog
2018-04-05 Tom Tromey <tom@tromey.com>
* utils.c (compare_strings): Remove.
* utils.h (compare_strings): Remove.
* objc-lang.h (find_imps): Update.
* objc-lang.c (find_methods): Take a std::vector.
(uniquify_strings, find_imps): Likewise.
* linespec.c (find_methods): Take a std::vector.
(decode_objc): Use std::vector.
(add_all_symbol_names_from_pspace, find_superclass_methods): Take
a std::vector.
(find_method, find_function_symbols): Use std::vector.
I wanted to use streq with std::unique in another (upcoming) patch in
this seres, so I changed it to return bool. To my surprise, this lead
to regressions. The cause turned out to be that streq was used as an
htab callback -- by casting it to the correct function type. This
sort of cast is invalid, so this patch adds a variant which is
directly suitable for use by htab. (Note that I did not add an
overload, as I could not get that to work with template deduction in
the other patch.)
ChangeLog
2018-04-05 Tom Tromey <tom@tromey.com>
* completer.c (completion_tracker::completion_tracker): Remove
cast.
(completion_tracker::discard_completions): Likewise.
* breakpoint.c (ambiguous_names_p): Remove cast.
* ada-lang.c (_initialize_ada_language): Remove cast.
* utils.h (streq): Update.
(streq_hash): Add new declaration.
* utils.c (streq): Return bool.
(streq_hash): New function.
Alan Modra
GDB Administrator
Simon Marchi
Simon Marchi
Cary Coutant
Tom Tromey
H.J. Lu
Maciej W. Rozycki
James Cowgill
Nick Clifton