Simon pointed out that save_infcall_suspend_state and
save_infcall_control_state could return unique pointers. This patch
implements this idea.
gdb/ChangeLog
2018-09-17 Tom Tromey <tom@tromey.com>
* infrun.c (save_infcall_suspend_state): Return
infcall_suspend_state_up.
(save_infcall_control_state): Return infcall_control_state_up.
* inferior.h (save_infcall_suspend_state)
(save_infcall_control_state): Declare later. Return unique
pointers.
This removes release_stop_context_cleanup, replacing it with a
stop_context destructor. It also mildly c++-ifies this struct.
gdb/ChangeLog
2018-09-17 Tom Tromey <tom@tromey.com>
* infrun.c (struct stop_context): Declare constructor,
destructor, "changed" method.
(stop_context::stop_context): Rename from save_stop_context.
(stop_context::~stop_context): Rename from
release_stop_context_cleanup.
(normal_stop): Update.
(stop_context::changed): Rename from stop_context_changed. Return
bool.
This removes a couple of cleanups from infrun by introducing a couple
of unique_ptr specializations.
gdb/ChangeLog
2018-09-17 Tom Tromey <tom@tromey.com>
* inferior.h (struct infcall_suspend_state_deleter): New.
(infcall_suspend_state_up): New typedef.
(struct infcall_control_state_deleter): New.
(infcall_control_state_up): New typedef.
(make_cleanup_restore_infcall_suspend_state)
(make_cleanup_restore_infcall_control_state): Don't declare.
* infcall.c (call_function_by_hand_dummy): Update.
* infrun.c (do_restore_infcall_suspend_state_cleanup)
(make_cleanup_restore_infcall_suspend_state): Remove.
(do_restore_infcall_control_state_cleanup)
(make_cleanup_restore_infcall_control_state): Remove.
This removes a cleanup from infrun.c by taking advantage of the
previous patch to introduce a use of unique_xmalloc_ptr.
gdb/ChangeLog
2018-09-17 Tom Tromey <tom@tromey.com>
* infrun.c (struct infcall_suspend_state) <registers>: Now a
unique_ptr.
<siginfo_data>: Now a unique_xmalloc_ptr.
(save_infcall_suspend_state, restore_infcall_suspend_state)
(discard_infcall_suspend_state)
(get_infcall_suspend_state_regcache): Update.
This changes infrun.c to use new and delete for infcall_suspend_state.
This enables the coming cleanups.
gdb/ChangeLog
2018-09-17 Tom Tromey <tom@tromey.com>
* gdbthread.h (struct thread_suspend_state): Add initializers.
(class thread_info) <suspend>: Remove initializer.
* infrun.c (struct infcall_suspend_state): Add initializers.
(save_infcall_suspend_state): Use new.
(discard_infcall_suspend_state): Use delete.
Some Python APIs steal references from their caller, and the refcount
checker supports this via an attribute.
However, in gdb with C++ we have a better idiom available: we can use
std::move on a gdbpy_ref<> instead. This makes the semantics obvious
at the point of call, and is safer at runtime as well, because the
callee's gdbpy_ref<> will be emptied.
This patch changes the reference-stealing code in gdb to use rvalue
references instead.
Tested on x86-64 Fedora 28.
gdb/ChangeLog
2018-09-16 Tom Tromey <tom@tromey.com>
* python/python-internal.h (CPYCHECKER_STEALS_REFERENCE_TO_ARG):
Remove.
* python/py-varobj.c (py_varobj_iter_ctor): Change pyiter to
rvalue reference. Remove CPYCHECKER_STEALS_REFERENCE_TO_ARG.
(py_varobj_iter_new): Likewise.
(py_varobj_get_iterator): Use gdbpy_ref.
An review by Simon of an earlier showed a few spots related to
thread_to_thread_object that could be simplified. This also detected
a latent bug, where thread_to_thread_object was inconsistent about
setting the Python exception before a NULL return.
Tested on x86-64 Fedora 28.
gdb/ChangeLog
2018-09-16 Tom Tromey <tom@tromey.com>
* python/py-threadevent.c (py_get_event_thread): Simplify.
* python/py-inferior.c (infpy_thread_from_thread_handle):
Return immediately after calling thread_to_thread_object. Use
Py_RETURN_NONE.
(thread_to_thread_object): Set the exception on a NULL return.
Tom mentioned this a while ago, as a way to give you a cheap sense of
progression in your build, as all object files will be built
alphabetically (including the directory part). I tried it and I think
it's nice.
gdb/ChangeLog:
* Makefile.in (LIBGDB_OBS): Sort COMMON_OBS.
gdb/gdbserver/ChangeLog:
* Makefile.in (gdbserver$(EXEEXT)): Sort OBS.
(gdbreplay$(EXEEXT)): Sort GDBREPLAY_OBS.
($(IPA_LIB)): Sort IPA_OBJS.
ADD_DEPS is defined nowhere, so I presume it's not useful. If I'm wrong
and this is actually used, there should be a comment explaining where it
comes from.
gdb/gdbserver/ChangeLog:
* Makefile.in: Remove references to $(ADD_DEPS).
ld/testsuite/ld-s12z/
* reloc-ext18-1.d: New file.
* reloc-ext18-1.d: New file.
* reloc-ext18-2.d: New file.
* reloc-ext18.d: New file.
* reloc-ext18.s: New file.
* reloc-ext24.d: New file.
* reloc-ext24.s: New file.
* reloc-ext32.d: New file.
* reloc-ext32.s: New file.
* reloc-opr.d: New file.
* reloc-opr.d: New file.
* reloc-opr.s: New file.
* reloc-pc-rel-7-15.d: New file.
* reloc-pc-rel-7-15.s: New file.
CPYCHECKER_RETURNS_BORROWED_REF is not used, and I think should never
be used. This patch removes it.
gdb/ChangeLog
2018-09-16 Tom Tromey <tom@tromey.com>
* python/python-internal.h (CPYCHECKER_RETURNS_BORROWED_REF):
Remove.
This changes thread_to_thread_object to return a new reference and
fixes up all the callers.
gdb/ChangeLog
2018-09-16 Tom Tromey <tom@tromey.com>
* python/python-internal.h (thread_to_thread_object): Change
return type.
* python/py-inferior.c (thread_to_thread_object): Return a new
reference.
(infpy_thread_from_thread_handle): Update.
* python/py-infthread.c (gdbpy_selected_thread): Update.
* python/py-stopevent.c (create_stop_event_object): Update.
* python/py-threadevent.c (py_get_event_thread): Return a new
reference.
(py_get_event_thread): Update.
* python/py-event.h (py_get_event_thread): Change return type.
* python/py-continueevent.c (create_continue_event_object):
Update.
This changes pspace_to_pspace_object to return a new reference and
fixes up all the callers.
gdb/ChangeLog
2018-09-16 Tom Tromey <tom@tromey.com>
* python/py-inferior.c (infpy_get_progspace): Update.
* python/python-internal.h (pspace_to_pspace_object): Change
return type.
* python/py-newobjfileevent.c
(create_clear_objfiles_event_object): Update.
* python/py-xmethods.c (gdbpy_get_matching_xmethod_workers):
Update.
* python/python.c (gdbpy_get_current_progspace): Update.
(gdbpy_progspaces): Update.
* python/py-progspace.c (pspace_to_pspace_object): Return a new
reference.
* python/py-objfile.c (objfpy_get_progspace): Update.
* python/py-prettyprint.c (find_pretty_printer_from_progspace):
Update.
There are a number of global functions in the gdb Python module which
really should be methods on Progspace. This patch adds new methods to
Progspace and then redefines these globals in terms of these new
methods.
This version has been rebased on the related changes that Simon
recently put in.
Built and regtested on x86-64 Fedora 28.
gdb/ChangeLog
2018-09-16 Tom Tromey <tom@tromey.com>
* python/lib/gdb/__init__.py (current_progspace, objfiles)
(solib_name, block_for_pc, find_pc_line): New functions.
(execute_unwinders): Update.
* python/py-block.c (gdbpy_block_for_pc): Remove.
* python/py-inferior.c (infpy_get_progspace): New function.
(inferior_object_getset) <progspace>: Add.
* python/py-progspace.c (pspy_objfiles): Rewrite.
(pspy_solib_name, pspy_block_for_pc)
(pspy_find_pc_line, pspy_is_valid): New functions.
(progspace_object_methods): Add entries for solib_name,
block_for_pc, find_pc_line, is_valid.
* python/python-internal.h (gdbpy_block_for_pc)
(build_objfiles_list): Don't declare.
* python/python.c: Don't include solib.h.
(gdbpy_solib_name, gdbpy_find_pc_line)
(gdbpy_get_current_progspace, build_objfiles_list)
(gdbpy_objfiles): Remove.
(GdbMethods) <current_progspace, objfiles, block_for_pc,
solib_name, find_pc_line>: Remove entries.
gdb/doc/ChangeLog
2018-09-16 Tom Tromey <tom@tromey.com>
* python.texi (Basic Python): Update docs for find_pc_line,
solib_name.
(Progspaces In Python): Update docs for current_progspace.
Document block_for_pc, find_pc_line, is_valid, nsolib_name.
Move method documentation before example.
This changes a couple of places in gdbserver to use the GNU style for
metasyntactic variables.
gdb/gdbserver/ChangeLog
2018-09-16 Tom Tromey <tom@tromey.com>
* remote-utils.c (remote_open): Use GNU style for metasyntactic
variables.
* gdbreplay.c (gdbreplay_usage): Use GNU style for metasyntactic
variables.
I searched for other spots that did not use the GNU style for
metasyntactic syntactic variables. This patch fixes most of the ones
I found in gdb proper. There are a few remaining in MI, but I was
unsure whether those should be touched.
gdb/ChangeLog
2018-09-16 Tom Tromey <tom@tromey.com>
* top.c (new_ui_command): Use GNU style for metasyntactic
variables.
* breakpoint.c (stopat_command): Use GNU style for metasyntactic
variables.
* maint.c (maintenance_translate_address): Remove "<>" around
text.
* interps.c (interpreter_exec_cmd): Use GNU style for
metasyntactic variables.
* nto-procfs.c (nto_procfs_target_info): Use GNU style for
metasyntactic variables.
* tracepoint.c (tfind_range_command): Use GNU style for
metasyntactic variables.
(tfind_outside_command): Likewise.
(_initialize_tracepoint): Likewise.
* remote.c (extended_remote_target::create_inferior): Use GNU
style for metasyntactic variables.
* sparc64-tdep.c (adi_examine_command): Use GNU style for
metasyntactic variables.
(adi_assign_command): Likewise.
gdb/testsuite/ChangeLog
2018-09-16 Tom Tromey <tom@tromey.com>
* gdb.base/new-ui.exp (do_execution_tests): Update.
* gdb.base/dbx.exp (test_breakpoints): Update.
I typed this:
(gdb) help set disassembler-options
Set the disassembler options.
Usage: set disassembler-options OPTION [,OPTION]...
See: 'show disassembler-options' for valid option values.
... so I tried what it said and got:
(gdb) show disassembler-options
The current disassembler options are ''
This surprised me a little, so this patch adds some text to explain
the situation when an architecture does not have disassembler options.
While there I noticed one more spot where gdb was not using the GNU
style for metasyntactic variables. This patch fixes this as well.
gdb/ChangeLog
2018-09-16 Tom Tromey <tom@tromey.com>
* disasm.c (show_disassembler_options_sfunc): Use GNU style for
metasyntactic variables. Print message if no disassembler options
are available.
I noticed that get_inferior_args should return const char *, because
it is just returning a reference to something owned by the inferior.
I'm checking this in.
gdb/ChangeLog
2018-09-15 Tom Tromey <tom@tromey.com>
* infcmd.c (get_inferior_args): Return const char *.
* inferior.h (get_inferior_args): Return type now const.
* linux-tdep.c (linux_fill_prpsinfo): Update.
* procfs.c (procfs_target::make_corefile_notes): Update.
AVX "VMOVQ xmm1, xmm2/m64" and "VMOVQ xmm1/m64, xmm2" can only be
encoded with VEX.128. Set Vex=1 on VEX.128 only vmovq and update
assembler tests.
gas/
PR gas/23665
* testsuite/gas/i386/avx-scalar-intel.d: Updated.
* testsuite/gas/i386/avx-scalar.d: Likewise.
* testsuite/gas/i386/x86-64-avx-scalar-intel.d: Likewise.
* testsuite/gas/i386/x86-64-avx-scalar.d: Likewise.
opcodes/
PR gas/23665
* i386-dis.c (vex_len_table): Update VEX_LEN_0F7E_P_1 and
VEX_LEN_0FD6_P_2 entries.
* i386-opc.tbl: Set Vex=1 on VEX.128 only vmovq.
* i386-tbl.h: Regenerated.
This merges the three versions of run_dump_test. Improved warning
handling versus the old gas version shows up a number of tests that
need their disassembly updating:
arm-linuxeabi +FAIL: ARM v1 instructions
arm-linuxeabi +FAIL: Accepted v8-a with ARMv8.1 AdvSIMD.
arm-linuxeabi +FAIL: bl local instructions for v4t.
arm-linuxeabi +FAIL: UDF
and some that now fail due to detecting assembly warnings:
h8300-elf +FAIL: binutils-all/strip-13
h8300-elf +FAIL: binutils-all/strip-14
h8300-elf +FAIL: binutils-all/strip-15
I've generally kept the union of all run_dump_test features, except
that the ld target aliases "cfi" and "shared" have disappeared, as has
the binutils substitution of $srcdir as $scrdir/$subdir.
binutils/
* testsuite/binutils-all/add-symbol.d: Add "section_subst: no".
* testsuite/binutils-all/elfedit.exp: Don't set tempfile or copyfile.
* testsuite/binutils-all/symbols-1.d,
* testsuite/binutils-all/symbols-2.d,
* testsuite/binutils-all/symbols-3.d,
* testsuite/binutils-all/symbols-4.d: Quote '*' and '!'.
* testsuite/binutils-all/add-empty-section.d,
* testsuite/binutils-all/add-section.d,
* testsuite/binutils-all/elfedit.exp,
* testsuite/binutils-all/note-1.d,
* testsuite/binutils-all/pr23633.d: Add $subdir after $srcdir.
* testsuite/config/default.exp (AS, ASFLAGS): Define.
* testsuite/config/hppa.sed: Handle all common symbols.
* testsuite/lib/binutils-common.exp (run_dump_test): New proc,
merged from three other versions.
(slurp_options, file_contents, set_file_contents): Likewise.
(big_or_little_endian, get_standard_section_names): Likewise.
* testsuite/lib/utils-lib.exp (run_dump_test): Delete.
(slurp_options, proc file_contents): Delete.
(get_standard_section_names): Delete.
gas/
* testsuite/config/default.exp: Make tmpdir.
* testsuite/lib/gas-defs.exp (run_dump_test): Delete.
(get_standard_section_names, slurp_options): Delete.
ld/
* testsuite/ld-d10v/reloc-007.d,
* testsuite/ld-d10v/reloc-008.d,
* testsuite/ld-d10v/reloc-015.d,
* testsuite/ld-d10v/reloc-016.d: Remove "error:".
* testsuite/ld-elf/eh-frame-hdr.d,
* testsuite/ld-elf/eh5.d,
* testsuite/ld-gc/personality.d: Replace "cfi" and "shared" in
target list with appropriate proc.
* testsuite/ld-elf/frame.exp: Use check_shared_lib_support rather
than "istarget shared".
* testsuite/lib/ld-lib.exp (proc big_or_little_endian): Delete.
(run_dump_test, slurp_options, file_contents): Delete.
(set_file_contents, istarget): Delete.
We have three copies of run_dump_test in the testsuite. This is a first
step towards consolidating them.
A few tests use objcopy to convert to verilog or srec output, then
check that output is as expected. Those tests can just as easily use
objdump, keeping the set of dump programs (addr2line, nm, objdump,
readelf, size) separate from utilities under test. That in turn makes
auto-detecting the dump program possible in more places.
binutils/
* testsuite/binutils-all/group-7a.d,
* testsuite/binutils-all/group-7b.d,
* testsuite/binutils-all/group-7c.d,
* testsuite/binutils-all/symbols-1.d,
* testsuite/binutils-all/symbols-2.d,
* testsuite/binutils-all/symbols-3.d,
* testsuite/binutils-all/symbols-4.d: Remove DUMPPROG.
gas/
* testsuite/gas/mri/char.d: Don't objcopy to srec, objdump instead.
* testsuite/gas/mri/float.d: Likewise.
* testsuite/lib/gas-defs.exp (run_dump_test): Remove support
for objcopy as a dump tool.
ld/
* testsuite/ld-elf/interleave-0.d,
* testsuite/ld-elf/interleave-4.d: Don't objcopy to srec,
objdump instead.
* testsuite/ld-gc/all-debug-sections.d,
* testsuite/ld-scripts/provide-4.d,
* testsuite/ld-scripts/provide-5.d,
* testsuite/ld-scripts/provide-6.d,
* testsuite/ld-scripts/provide-7.d,
* testsuite/ld-scripts/provide-8.d,
* testsuite/ld-scripts/segment-start.d: Remove PROG, specify
nm instead.
* testsuite/lib/ld-lib.exp (run_dump_test): Remove support
for objcopy as a dump tool.
In the Python code, gdb exceptions may not leak into the Python core.
execute_gdb_command was calling bpstat_do_actions outside of a
TRY/CATCH; which seemed risky. I don't have a test case for this, but
if bpstat_do_actions could ever throw, it could crash gdb.
This patch introduces a new scope in order to preserve the current
semantics, so it is looks a bit bigger than it really is.
Tested on x86-64 Fedora 28.
gdb/ChangeLog
2018-09-07 Tom Tromey <tom@tromey.com>
* python/python.c (execute_gdb_command): Call bpstat_do_actions
inside the TRY.
This patch started as an observation from valgrind that GDB appeared
to be loosing track of some memory associated with types. An example
valgrind stack would be:
24 bytes in 1 blocks are possibly lost in loss record 419 of 5,361
at 0x4C2EA1E: calloc (vg_replace_malloc.c:711)
by 0x623D26: xcalloc (common-utils.c:85)
by 0x623D65: xzalloc(unsigned long) (common-utils.c:95)
by 0x72A066: make_function_type(type*, type**) (gdbtypes.c:510)
by 0x72A098: lookup_function_type(type*) (gdbtypes.c:521)
by 0x73635D: gdbtypes_post_init(gdbarch*) (gdbtypes.c:5439)
by 0x727590: gdbarch_data(gdbarch*, gdbarch_data*) (gdbarch.c:5230)
by 0x735B99: builtin_type(gdbarch*) (gdbtypes.c:5313)
by 0x514D95: elf_rel_plt_read(minimal_symbol_reader&, objfile*, bfd_symbol**) (elfread.c:542)
by 0x51662F: elf_read_minimal_symbols(objfile*, int, elfinfo const*) (elfread.c:1121)
by 0x5168A5: elf_symfile_read(objfile*, enum_flags<symfile_add_flag>) (elfread.c:1207)
by 0x8520F5: read_symbols(objfile*, enum_flags<symfile_add_flag>) (symfile.c:794)
When we look in make_function_type we find a call to TYPE_ZALLOC
(inside the INIT_FUNC_SPECIFIC macro). It is this call to TYPE_ZALLOC
that is allocating memory with xcalloc, that is then getting lost.
The problem is tht calling TYPE_ALLOC or TYPE_ZALLOC currently
allocates memory from either the objfile obstack or by using malloc.
The problem with this is that types are allocated either on the
objfile obstack, or on the gdbarch obstack.
As a result, if we discard a type associated with an objfile then
auxiliary data allocated with TYPE_(Z)ALLOC will be correctly
discarded. But, if we were ever to discard a gdbarch then any
auxiliary type data would be leaked. Right now there are very few
places in GDB where a gdbarch is ever discarded, but it shouldn't hurt
to close down these bugs as we spot them.
This commit ensures that auxiliary type data is allocated from the
same obstack as the type itself, which should reduce leaked memory.
The one problem case that I found with this change was in eval.c,
where in one place we allocate a local type structure, and then used
TYPE_ZALLOC to allocate some space for the type. This local type is
neither object file owned, nor gdbarch owned, and so the updated
TYPE_ALLOC code is unable to find an objstack to allocate space on.
My proposed solution for this issue is that the space should be
allocated with a direct call to xzalloc. We could extend TYPE_ALLOC
to check for type->gdbarch being null, and then fall back to a direct
call to xzalloc, however, I think that making this rare case of a
local type require special handling is not a bad thing, this serves to
highlight that clearing up the memory will require special handling
too.
This special case of a local type is interesting as the types owner
field (contained within the main_type) is completely null. While
reflecting on this I looked at how types use the get_type_arch
function. It seems clear that, based on how this is used, it is never
intended that null will be returned from this function. This only
goes to reinforce, how locally alloctaed types, with no owner, are
both special, and need to be handled carefully. To help spot errors
earlier, I added an assert into get_type_arch that the returned arch
is not null.
Inside gdbarch.c I found a few other places where auxiliary type data
was being allocated directly on the heap rather than on the types
obstack. I have fixed these to call TYPE_ALLOC now.
Finally, it is worth noting that as we don't clean up our gdbarch
objects yet, then this will not make much of an impact on the amount
of memory reported as lost at program termination time. Memory
allocated for auxiliary type information is still not freed, however,
it is now on the correct obstack. If we do ever start freeing our
gdbarch structures then the associated type data will be cleaned up
correctly.
Tested on X86-64 GNU/Linux with no regressions.
gdb/ChangeLog:
* eval.c (fake_method::fake_method): Call xzalloc directly for a
type that is neither object file owned, nor gdbarch owned.
* gdbtypes.c (get_type_gdbarch): Add an assert that returned
gdbarch is non-NULL.
(alloc_type_instance): Allocate non-objfile owned types on the
gdbarch obstack.
(copy_type_recursive): Allocate TYPE_FIELDS and TYPE_RANGE_DATA
using TYPE_ALLOC to ensure memory is allocated on the correct
obstack.
* gdbtypes.h (TYPE_ALLOC): Allocate space on either the objfile
obstack, or the gdbarch obstack.
(TYPE_ZALLOC): Rewrite using TYPE_ALLOC.
Define a new procedure, `run_mips_undefweak_test', and use it to iterate
over several scenarios involving undefined weak symbols resolving to
zero, verifying expected regular MIPS, MIPS16 and microMIPS code, GOT
and dynamic symbol table generation, as well as the setting of the
EI_ABIVERSION field in the ELF file header. In particular ensure that
symbol versioning works and that `__gnu_absolute_zero' gets assigned a
version (any will do) even if it has not been listed for exportation in
a linker version script.
ld/
PR ld/21375
* testsuite/ld-mips-elf/pr21375-abi.hd: New test.
* testsuite/ld-mips-elf/pr21375-noabi.hd: New test.
* testsuite/ld-mips-elf/pr21375.dd: New test.
* testsuite/ld-mips-elf/pr21375h.dd: New test.
* testsuite/ld-mips-elf/pr21375p.dd: New test.
* testsuite/ld-mips-elf/pr21375ph.dd: New test.
* testsuite/ld-mips-elf/pr21375s.dd: New test.
* testsuite/ld-mips-elf/pr21375s-n32.dd: New test.
* testsuite/ld-mips-elf/pr21375s-n64.dd: New test.
* testsuite/ld-mips-elf/pr21375sh.dd: New test.
* testsuite/ld-mips-elf/pr21375sh-n32.dd: New test.
* testsuite/ld-mips-elf/pr21375sh-n64.dd: New test.
* testsuite/ld-mips-elf/pr21375shg.dd: New test.
* testsuite/ld-mips-elf/pr21375sx.dd: New test.
* testsuite/ld-mips-elf/pr21375sxh.dd: New test.
* testsuite/ld-mips-elf/pr21375sm16.dd: New test.
* testsuite/ld-mips-elf/pr21375sm16h.dd: New test.
* testsuite/ld-mips-elf/pr21375su.dd: New test.
* testsuite/ld-mips-elf/pr21375su-n32.dd: New test.
* testsuite/ld-mips-elf/pr21375su-n64.dd: New test.
* testsuite/ld-mips-elf/pr21375suh.dd: New test.
* testsuite/ld-mips-elf/pr21375suh-n32.dd: New test.
* testsuite/ld-mips-elf/pr21375suh-n64.dd: New test.
* testsuite/ld-mips-elf/pr21375sux.dd: New test.
* testsuite/ld-mips-elf/pr21375suxh.dd: New test.
* testsuite/ld-mips-elf/pr21375.gd: New test.
* testsuite/ld-mips-elf/pr21375h.gd: New test.
* testsuite/ld-mips-elf/pr21375p.gd: New test.
* testsuite/ld-mips-elf/pr21375ph.gd: New test.
* testsuite/ld-mips-elf/pr21375s.gd: New test.
* testsuite/ld-mips-elf/pr21375s-n32.gd: New test.
* testsuite/ld-mips-elf/pr21375s-n64.gd: New test.
* testsuite/ld-mips-elf/pr21375sh.gd: New test.
* testsuite/ld-mips-elf/pr21375sh-n32.gd: New test.
* testsuite/ld-mips-elf/pr21375sh-n64.gd: New test.
* testsuite/ld-mips-elf/pr21375shg.gd: New test.
* testsuite/ld-mips-elf/pr21375shl.gd: New test.
* testsuite/ld-mips-elf/pr21375shv.gd: New test.
* testsuite/ld-mips-elf/pr21375sx.gd: New test.
* testsuite/ld-mips-elf/pr21375sxh.gd: New test.
* testsuite/ld-mips-elf/pr21375.sd: New test.
* testsuite/ld-mips-elf/pr21375-irix.sd: New test.
* testsuite/ld-mips-elf/pr21375h.sd: New test.
* testsuite/ld-mips-elf/pr21375h-irix.sd: New test.
* testsuite/ld-mips-elf/pr21375p.sd: New test.
* testsuite/ld-mips-elf/pr21375p-irix.sd: New test.
* testsuite/ld-mips-elf/pr21375ph.sd: New test.
* testsuite/ld-mips-elf/pr21375ph-irix.sd: New test.
* testsuite/ld-mips-elf/pr21375s.sd: New test.
* testsuite/ld-mips-elf/pr21375s-irix.sd: New test.
* testsuite/ld-mips-elf/pr21375s-n32.sd: New test.
* testsuite/ld-mips-elf/pr21375s-n32-irix.sd: New test.
* testsuite/ld-mips-elf/pr21375s-n64.sd: New test.
* testsuite/ld-mips-elf/pr21375s-n64-irix.sd: New test.
* testsuite/ld-mips-elf/pr21375sh.sd: New test.
* testsuite/ld-mips-elf/pr21375sh-irix.sd: New test.
* testsuite/ld-mips-elf/pr21375sh-n32.sd: New test.
* testsuite/ld-mips-elf/pr21375sh-n32-irix.sd: New test.
* testsuite/ld-mips-elf/pr21375sh-n64.sd: New test.
* testsuite/ld-mips-elf/pr21375sh-n64-irix.sd: New test.
* testsuite/ld-mips-elf/pr21375shg.sd: New test.
* testsuite/ld-mips-elf/pr21375shg-irix.sd: New test.
* testsuite/ld-mips-elf/pr21375shl.sd: New test.
* testsuite/ld-mips-elf/pr21375shl-irix.sd: New test.
* testsuite/ld-mips-elf/pr21375shv.sd: New test.
* testsuite/ld-mips-elf/pr21375shv-irix.sd: New test.
* testsuite/ld-mips-elf/pr21375sx.sd: New test.
* testsuite/ld-mips-elf/pr21375sx-irix.sd: New test.
* testsuite/ld-mips-elf/pr21375sxh.sd: New test.
* testsuite/ld-mips-elf/pr21375sxh-irix.sd: New test.
* testsuite/ld-mips-elf/pr21375.ld: New test linker script.
* testsuite/ld-mips-elf/pr21375-xgot.ld: New test linker script.
* testsuite/ld-mips-elf/pr21375.ver: New test version script.
* testsuite/ld-mips-elf/pr21375v.ver: New test version script.
* testsuite/ld-mips-elf/pr21375.s: New test source.
* testsuite/ld-mips-elf/pr21375-mips16.s: New test source.
* testsuite/ld-mips-elf/pr21375-n32.s: New test source.
* testsuite/ld-mips-elf/pr21375-n64.s: New test source.
* testsuite/ld-mips-elf/pr21375-xgot.s: New test source.
* testsuite/ld-mips-elf/mips-elf.exp (run_mips_undefweak_test):
New procedure; run the new tests.
We have an issue in the MIPS backend, with the handling of undefined
hidden and internal weak symbols. References to such symbols are
supposed to resolve to 0 according to the ELF gABI[1]:
"Unresolved weak symbols have a zero value."
and the 64-bit MIPS psABI[2]:
"If a symbol with one of these [hidden or internal] attributes has no
definition within the executable/DSO being linked, then it must be
resolved to allocated space if common, resolved to zero if weak, or an
error reported otherwise."
however if a GOT relocation is used, then a local GOT entry is created
and used to satisfy the reference. Such an entry is then (in DSO and
PIE binaries) subject to the usual load-time relocation, which means a
non-zero value will be returned if the base address is non-zero. This
will defeat the usual run-time sequence like:
void a (void) __attribute__ ((visibility ("hidden"), weak));
void
x (void)
{
if (a)
a ();
}
This can be reproduced with this simple code:
$ cat libtest.c
extern int a __attribute__ ((visibility ("hidden"), weak));
int *
x (void)
{
return &a;
}
$ cat test.c
int *x (void);
int
main (void)
{
printf ("a: %p\n", x ());
return 0;
}
$ gcc -shared -fPIC -o libtest.so libtest.c
$ gcc -o test test.c -Wl,-rpath,$(pwd) libtest.so
$ ./test
a: 0x77184000
$
The usual approach targets take is making all the steps required to
assign a GOT entry for the symbol referred, and then leave its contents
at zero with no dynamic relocation attached, therefore ensuring that the
value does not change at load time. However this is not going to work
with the implicitly relocated GOT the MIPS psABI specifies[3]:
"The dynamic linker relocates the global offset table by first adding
the difference between the base where the shared object is loaded and
the value of the dynamic tag DT_MIPS_BASE_ADDRESS to all local global
offset table entries."
and we cannot therefore use the local GOT part.
And we cannot offhand use the global part either, as the symbol would
then have to be exported and possibly wrongly preempt symbols in other
modules involved in the dynamic load, because as per the ELF gABI[1] we
are not allowed to enter a hidden or internal symbol into the dynamic
symbol table (and then use its associated GOT entry):
"A hidden symbol contained in a relocatable object must be either
removed or converted to STB_LOCAL binding by the link-editor when the
relocatable object is included in an executable file or shared object."
and:
"An internal symbol contained in a relocatable object must be either
removed or converted to STB_LOCAL binding by the link-editor when the
relocatable object is included in an executable file or shared object."
So we have to choose something else.
Our choice is further limited by the need for the reference associated
with the GOT relocation to stay within the signed 16-bit limit from the
GOT pointer base register, while being compliant with the ELF gABI and
the MIPS psABI. However as Alan Modra has observed[4] one possibility
is to edit (relax) the code such that the GOT reference is removed
altogether.
Based on these observations then modify MIPS BFD linker backend code to:
1. Interpret code associated with GOT relocations and relax the usual LW
or LD instructions into a corresponding immediate load operation that
places the value of 0 in the intended register, while leaving the GOT
entry allocated and initialized as usually.
2. Leave any other instructions associated with GOT relocations in place
and instead redirect the reference to a global GOT entry associated
with a special `__gnu_absolute_zero' symbol created for this purpose,
whose value is 0, SHN_ABS section marks it absolute, binding is
global and export class protected, ensuring that the locally provided
value is always used at load time, and that the value is not
relocated by the dynamic loader.
3. Adjust any high-part GOT relocation used, typically associated with
a LUI instruction, accordingly, so that run-time consistency is
maintained, either by resolving to the original entry if the
instruction associated with the corresponding low-part GOT relocation
has been relaxed to an immediate load (in which case the value loaded
with LUI will be overwritten), or by also redirecting the reference
to `__gnu_absolute_zero' to complete the GOT access sequence if that
symbol has been used.
4. Add a target `elf_backend_hide_symbol' hook, for the three MIPS ABIs,
which prevents the `__gnu_absolute_zero' symbol from being forced
local, to ensure that the redirection works and the symbol remains
global/protected with existing linker scripts unchanged.
5. Observing the issue with handling SHN_ABS symbols in the GNU dynamic
loader, covered by glibc PR 19818, set the EI_ABIVERSION field in the
ELF file header produced to 4 (ABI_ABSOLUTE) if `__gnu_absolute_zero'
symbol has been produced and the target configured indicates the GNU
operating system, so that broken versions of the GNU dynamic loader
gracefully reject the file in loading rather than going astray. Keep
EI_ABIVERSION at the original value for other operating systems or if
no `__gnu_absolute_zero' symbol has been made.
The name of the special `__gnu_absolute_zero' has no meaning other than
how a human reader can interpret it, as it is ignored in dynamic loading
in the handling of the scenarios concerned. This is because the symbol
resolves locally, and it's only the symbol's attributes that matter so
that the associated GOT entry remains unchanged at load time.
Therefore the name is somewhat arbitrary, observing however the need to
use the name space reserved for the system so that it does not conflict
with a possible user symbol, and hence the leading underscore, and also
the `gnu' infix to denote a GNU feature. Other implementations wishing
to address the problem in a similar way may choose a different name and
have the solution still work, possibly with a mixture of modules used in
a dynamic having symbols of different names provided, which will however
not interact with each other due to the protected export class.
The symbol can be referred explicitly, however the name is an internal
implementation detail rather than a part of the ABI, and therefore no
specific semantics is guaranteed.
One limitation of this change is that if `__gnu_absolute_zero' has been
already defined, then we do not wipe the old definition and all kinds of
odd behavior can result. This is however like with other symbols we
internally define, such as `_GLOBAL_OFFSET_TABLE_' or `__rld_map', and
therefore left as a possible future enhancement.
As an optimization the relaxation of LW and LD instructions to a load of
immediate zero is always made, even SVR4 PIC code for code that will end
up in a regular (non-PIE) executable, because there is a cache advantage
with the avoidance of a load from the GOT, even if it is otherwise
guaranteed to remain zero. It does not reliably happen though, due to a
symbol exportation issue affecting executables, covered by PR ld/21805.
One existing test case needs to be updated, as it triggers relaxation
introduced with this change and consequently linker output does not
match expectations anymore. As we want to keep the original issue
covered with the test case modify it then to use the LWL instruction in
place of LW, and adjust the output expected accordingly.
References:
[1] "System V Application Binary Interface - DRAFT - 19 October 2010",
The SCO Group, Section "Symbol Table",
<http://www.sco.com/developers/gabi/2012-12-31/ch4.symtab.html>
[2] "64-bit ELF Object File Specification, Draft Version 2.5", MIPS
Technologies / Silicon Graphics Computer Systems, Order Number
007-4658-001, Section 2.5 "Symbol Table", p. 22,
<http://techpubs.sgi.com/library/manuals/4000/007-4658-001/pdf/007-4658-001.pdf>
[3] "SYSTEM V APPLICATION BINARY INTERFACE, MIPS RISC Processor
Supplement, 3rd Edition", Section "Global Offset Table", p. 5-10,
<http://www.linux-mips.org/pub/linux/mips/doc/ABI/mipsabi.pdf>
[4] "Undo dynamic symbol state after regular object sym type mismatch",
<https://sourceware.org/ml/binutils/2017-07/msg00265.html>
bfd/
PR ld/21375
* elfxx-mips.h (_bfd_mips_elf_hide_symbol): New prototype.
(_bfd_mips_elf_linker_flags): Update prototype.
* elf32-mips.c (elf_backend_hide_symbol): New macro.
* elf64-mips.c (elf_backend_hide_symbol): Likewise.
* elfn32-mips.c (elf_backend_hide_symbol): Likewise.
* elfxx-mips.c (mips_elf_link_hash_table): Add
`use_absolute_zero' and `gnu_target' members.
(mips_elf_record_global_got_symbol): Call
`_bfd_mips_elf_hide_symbol' rather than
`_bfd_elf_link_hash_hide_symbol'.
(mips_use_local_got_p): Return FALSE if the symbol is absolute.
(mips_elf_obtain_contents): Reorder function.
(mips_elf_nullify_got_load): New function.
(mips_elf_calculate_relocation): Add `contents' parameter.
Nullify GOT loads or if it is not possible, then redirect GOT
relocations to the `__gnu_absolute_zero' symbol, for references
that are supposed to resolve to zero.
(mips_elf_define_absolute_zero): New function.
(_bfd_mips_elf_check_relocs): Prepare for arrangements made in
`mips_elf_calculate_relocation' for references made via the GOT
that are supposed to resolve to zero.
(_bfd_mips_elf_hide_symbol): New function.
(_bfd_mips_elf_linker_flags): Add the `gnu_target' parameter,
set the `gnu_target' member of the MIPS hash table.
(MIPS_LIBC_ABI_ABSOLUTE): New enumeration constant.
(_bfd_mips_post_process_headers): Use it.
ld/
PR ld/21375
* emultempl/mipself.em: Set `gnu_target' according to ${target}.
(mips_create_output_section_statements): Update call to
`_bfd_mips_elf_linker_flags'.
* testsuite/ld-mips-elf/pr21334.s: Use LWL rather than LW.
* testsuite/ld-mips-elf/pr21334.dd: Update accordingly.
Move code used to store the contents of a relocated field in output into
a separate function, `mips_elf_store_contents', complementing existing
`mips_elf_obtain_contents'.
bfd/
* elfxx-mips.c (mips_elf_store_contents): New function...
(mips_elf_perform_relocation): ... factored out from here.
Fix an issue with the SEGMENT_START builtin function where its result is
absolute when taken from the default supplied, and section-relative when
taken from a `-T' command-line override. This is against documentation,
inconsistent and unexpected, and with PIE executables gives an incorrect
result with the `__executable_start' symbol.
Make the result of SEGMENT_START always section-relative then.
ld/
* ldexp.c (fold_binary): Always make the result of SEGMENT_START
section-relative.
* testsuite/ld-scripts/segment-start.d: New test.
* testsuite/ld-scripts/segment-start.ld: New test linker script.
* testsuite/ld-scripts/segment-start.s: New test source.
* testsuite/ld-scripts/script.exp: Run the new test.
Avoid a division by zero and thus a linker crash in SEGMENT_START script
builtin function handling, by not checking the value supplied with a
`-T' command-line override against the maximum page size if that has not
been set.
ld/
* ldexp.c (fold_binary): Check that `config.maxpagesize' is
non-zero before using it as a divisor.
Verify that -mevexwig=1 has no impact on non-WIG EVEX instruction encoding.
PR gas/23642
* testsuite/gas/i386/evex-wig2.d: New file.
* testsuite/gas/i386/evex-wig2.s: Likewise.
* testsuite/gas/i386/x86-64-evex-wig2.d: Likewise.
* testsuite/gas/i386/x86-64-evex-wig2.s: Likewise.
* testsuite/gas/i386/i386.exp: Run evex-wig2 and
x86-64-evex-wig2.
Add VEXWIG, defined as 3, to indicate that the VEX.W/EVEX.W bit is
ignored by such VEX/EVEX instructions, aka WIG instructions. Set
VexW=3 on VEX/EVEX WIG instructions. Update assembler to check
VEXWIG when setting the VEX.W bit.
gas/
PR gas/23642
* config/tc-i386.c (build_vex_prefix): Check VEXWIG when setting
the VEX.W bit.
(build_evex_prefix): Check VEXWIG when setting the EVEX.W bit.
opcodes/
PR gas/23642
* i386-opc.h (VEXWIG): New.
* i386-opc.tbl: Set VexW=3 on VEX/EVEX WIG instructions.
* i386-tbl.h: Regenerated.
Update x86 disassembler to ignore the EVEX.W bit in EVEX vcvt[u]si2s[sd]
instructions in 32-bit mode.
gas/
PR binutils/23655
* testsuite/gas/i386/evex.d: New file.
* testsuite/gas/i386/evex.s: Likewise.
* testsuite/gas/i386/i386.exp: Run evex.
opcodes/
PR binutils/23655
* i386-dis-evex.h (evex_table): Replace Eq with Edqa for
vcvtsi2ss%LQ, vcvtsi2sd%LQ, vcvtusi2ss%LQ and vcvtusi2sd%LQ.
* i386-dis.c (Edqa): New.
(dqa_mode): Likewise.
(intel_operand_size): Handle dqa_mode as m_mode.
(OP_E_register): Handle dqa_mode as dq_mode.
(OP_E_memory): Set shift for dqa_mode based on address_mode.
I noticed that call_function_by_hand_dummy has a block that only
exists to declare a variable, like:
{
int i;
for (i = ...0)
...
}
This patch removes the unnecessary and the extra indentation by moving
the declaration into the "for".
gdb/ChangeLog
2018-09-14 Tom Tromey <tom@tromey.com>
* infcall.c (call_function_by_hand_dummy): Remove unnecessary
block.
