The test expect the runto_main to stop at the first line of the function.
Depending on the optimization level, gdb may stop in the prolog or after
the prolog at the first line. To ensure the test stops at the first line
of main, have it explicitly stop at a break point on the first line of the
function.
On PowerPC, the test passes when compiled with no optimization but fails
with all levels of optimization due to gdb stopping in the prolog.
The "add accessors for field (and call site) location" patch caused a
gdb crash when running the internal AdaCore testsuite. This turned
out to be a latent bug in ada-lang.c.
The immediate cause of the bug is that find_struct_field
unconditionally uses TYPE_FIELD_BITPOS. This causes an assert for a
dynamic type.
This patch fixes the problem by doing two things. First, it changes
find_struct_field to use a dummy value for the field offset in the
situation where the offset is not actually needed by the caller. This
works because the offset isn't used in any other way -- only as a
result.
Second, this patch assures that calls to find_struct_field use a
resolved type when the offset is needed. For
value_tag_from_contents_and_address, this is done by resolving the
type explicitly. In ada_value_struct_elt, this is done by passing
nullptr for the out parameters when they are not needed (the second
call in this function already uses a resolved type).
Note that, while we believe the parent field probably can't occur at a
variable offset, the patch still updates this code path, just in case.
I've updated an existing test case to reproduce the crash.
I'm checking this in.
ldelf.c: In function 'ldelf_after_open':
ldelf.c:1049:43: warning: the comparison will always evaluate as 'true' for the address of 'elf_header' will never be NULL [-Waddress]
1049 | && elf_tdata (abfd)->elf_header != NULL
| ^~
In file included from ldelf.c:37:
../bfd/elf-bfd.h:1957:21: note: 'elf_header' declared here
1957 | Elf_Internal_Ehdr elf_header[1]; /* Actual data, but ref like ptr */
* ldelf.c (ldelf_after_open): Remove useless elf_header test.
Mainline gcc:
readelf.c: In function 'find_section':
readelf.c:349:8: error: the comparison will always evaluate as 'true' for the pointer operand in 'filedata->section_headers + (sizetype)((long unsigned int)i * 80)' must not be NULL [-Werror=address]
349 | ((X) != NULL \
| ^~
readelf.c:761:9: note: in expansion of macro 'SECTION_NAME_VALID'
761 | if (SECTION_NAME_VALID (filedata->section_headers + i)
| ^~~~~~~~~~~~~~~~~~
This will likely be fixed in gcc, but inline functions are nicer than
macros.
* readelf.c (SECTION_NAME, SECTION_NAME_VALID),
(SECTION_NAME_PRINT, VALID_SYMBOL_NAME, VALID_DYNAMIC_NAME),
(GET_DYNAMIC_NAME): Delete. Replace with..
(section_name, section_name_valid, section_name_print),
(valid_symbol_name, valid_dynamic_name, get_dynamic_name): ..these
new inline functions. Update use throughout file.
My previous PR27625 patch had a problem or two. For one, the error
"__tls_get_addr call lacks marker reloc" on processing some calls
before hitting a call without markers typically isn't seen. Instead a
gold assertion fails. Either way it would be a hard error, which
triggers on a file contained in libphobos.a when running the gcc
testsuite. A warning isn't even appropriate since the call involved
is one built by hand without any of the arg setup relocations that
might result in linker optimisation.
So this patch reverts most of commit 0af4fcc25dd5, instead entirely
ignoring the problem of mis-optimising old-style __tls_get_addr calls
without marker relocs. We can't handle them gracefully without
another pass over relocations before decisions are made about GOT
entries in Scan::global or Scan::local. That seems too costly, just
to link object files from 2009. What's more, there doesn't seem to be
any way to allow the libphobos explicit __tls_get_addr call, but not
old TLS sequences without marker relocs. Examining instructions
before the __tls_get_addr call is out of the question: program flow
might reach the call via a branch. Putting an R_PPC64_TLSGD marker
with zero sym on the call might be a solution, but current linkers
will then merrily optimise away the call!
PR gold/27625
* powerpc.cc (Powerpc_relobj): Delete no_tls_marker_, tls_marker_,
and tls_opt_error_ variables and accessors. Remove all uses.
When using Bison 3.8, we get this error:
../../gdb/c-exp.y:3455:1: error: 'void c_print_token(FILE*, int, YYSTYPE)' defined but not used [-Werror=unused-function]
That's because bison 3.8 removed YYPRINT support:
https://savannah.gnu.org/forum/forum.php?forum_id=10047
Accordingly, this patch only defines that function for Bison < 3.8.
Change-Id: I3cbf2f317630bb72810b00f2d9b2c4b99fa812ad
The test-case gdb.gdb/python-interrupts.exp:
- runs to captured_command_loop
- sets a breakpoint at set_active_ext_lang
- calls a python command
- verifies the command triggers the breakpoint
- sends a signal and verifies the result
The test-case is fragile, because (f.i. with -flto) it cannot be guaranteed
that captured_command_loop and set_active_ext_lang are available for setting
breakpoints.
Reimplement the test-case as unittest, using:
- execute_command_to_string to capture the output
- try/catch to catch the "Error while executing Python code" exception
- a new hook selftests::hook_set_active_ext_lang to raise the signal
Tested on x86_64-linux.
This changes gdb to check the index that is passed to type::field.
This caught one bug in the Ada code when running the test suite
(actually I found the bug first, then realized that the check would
have helped), so this patch fixes that as well.
Regression tested on x86-64 Fedora 34.
The Rust lex selftest fails on our Windows build. I tracked this down
to a use of UTF-32 as a parameter to convert_between_encodings. Here,
iconv_open succeeds, but the actual conversion of a tab character
fails with EILSEQ. I suspect that "UTF-32" is being interpreted as
big-endian, as changing the call to use "UTF-32LE" makes it work.
This patch implements this fix.
The format_pieces selftest currently fails on Windows hosts.
The selftest doesn't handle the "%ll" -> "%I64" rewrite that the
formatter may perform, but also gdbsupport was missing a configure
check for PRINTF_HAS_LONG_LONG. This patch fixes both issues.
A customer-reported problem led us to a bug in dynamic type
resolution. resolve_dynamic_struct will recursively call
resolve_dynamic_type_internal, passing it the sub-object for the
particular field being resolved. While it offsets the address here,
it does not also offset the "valaddr" -- the array of bytes describing
the memory.
This patch fixes the bug, by offsetting both. A test case is included
that can be used to reproduce the bug.
Now that there is a register_test variant that accepts std::function,
it seems to me that the 'selftest' struct and accompanying code is
obsolete -- simply always using std::function is simpler. This patch
implements this idea.
GDB doesn't support loading debug files using build-id from remote
target filesystems.
This is the case when gdbserver attached to a process and a gdb target
remote occurs over tcp.
With this change we make build-id lookups possible:
(gdb) show debug-file-directory
The directory where separate debug symbols are searched for is "/usr/local/lib/debug".
(gdb) set debug-file-directory /usr/lib/debug
(gdb) show sysroot
The current system root is "target:".
(gdb) target extended-remote :46615
Remote debugging using :46615
warning: Can not parse XML target description; XML support was disabled at compile time
Reading /usr/sbin/mariadbd from remote target...
warning: File transfers from remote targets can be slow. Use "set sysroot" to access files locally instead.
Reading /usr/sbin/mariadbd from remote target...
Reading symbols from target:/usr/sbin/mariadbd...
Reading /usr/lib/debug/.build-id/6e/0a874dca5a7ff831396ddc0785d939a192efe3.debug from remote target...
Reading /usr/lib/debug/.build-id/6e/0a874dca5a7ff831396ddc0785d939a192efe3.debug from remote target...
Reading symbols from target:/usr/lib/debug/.build-id/6e/0a874dca5a7ff831396ddc0785d939a192efe3.debug...
Reading /lib/x86_64-linux-gnu/libpcre2-8.so.0 from remote target...
...
Before this change, the lookups would have been (GNU gdb (GDB) Fedora 10.2-3.fc34):
(gdb) target extended-remote :46615
Remote debugging using :46615
Reading /usr/sbin/mariadbd from remote target...
warning: File transfers from remote targets can be slow. Use "set sysroot" to access files locally instead.
Reading /usr/sbin/mariadbd from remote target...
Reading symbols from target:/usr/sbin/mariadbd...
Reading /usr/sbin/0a874dca5a7ff831396ddc0785d939a192efe3.debug from remote target...
Reading /usr/sbin/.debug/0a874dca5a7ff831396ddc0785d939a192efe3.debug from remote target...
Reading /usr/lib/debug//usr/sbin/0a874dca5a7ff831396ddc0785d939a192efe3.debug from remote target...
Reading /usr/lib/debug/usr/sbin//0a874dca5a7ff831396ddc0785d939a192efe3.debug from remote target...
Reading target:/usr/lib/debug/usr/sbin//0a874dca5a7ff831396ddc0785d939a192efe3.debug from remote target...
Missing separate debuginfo for target:/usr/sbin/mariadbd
Try: dnf --enablerepo='*debug*' install /usr/lib/debug/.build-id/6e/0a874dca5a7ff831396ddc0785d939a192efe3.debug
(No debugging symbols found in target:/usr/sbin/mariadbd)
Observe it didn't look for
/usr/lib/debug/.build-id/6e/0a874dca5a7ff831396ddc0785d939a192efe3.debug
on the remote target (where it is) and expected them to be installed
locally.
As a minor optimization, this also changes the build-id lookup such that
if sysroot is empty, no second lookup of the same location is performed.
Change-Id: I5181696d271c325a25a0805a8defb8ab7f9b3f55
Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=17917
bfd * linker.c (_bfd_generic_link_add_one_symbol): Test for a NULL
name before checking to see if the symbol is __gnu_lto_slim.
* archive.c (_bfd_compute_and_write_armap): Likewise.
binutils
* nm.c (filter_symbols): Test for a NULL name before checking to
see if the symbol is __gnu_lto_slim.
* objcopy.c (filter_symbols): Likewise.
A bug was filed against the incorrect underlying type setting for
an enumeration type, which was caused by a copy and paste error.
This patch fixes the problem by setting it by calling objfile_int_type,
which was originally dwarf2_per_objfile::int_type, with ctf_type_size bits.
Also add error checking on ctf_func_type_info call.
I'd missed the fact that the .debug_rnglists dump doesn't exactly
display the contents of the section. Instead readelf rummages through
.debug_info looking for DW_AT_ranges entries, then displays the
entries in .debug_rnglists pointed at, sorted. A simpler dump of the
actual section contents might be more useful and robust, but it was
likely done that way to detect overlap and holes.
Anyway, the headers in .debug_rnglists besides the first are ignored,
and limiting to the unit length of the first header fails if there is
more than one unit.
PR 28459
* dwarf.c (display_debug_ranges): Don't constrain data to length
in header.
Adjust pr28158.rd for glibc 2.34:
$ readelf -W --dyn-syms tmpdir/pr28158
Symbol table '.dynsym' contains 4 entries:
Num: Value Size Type Bind Vis Ndx Name
0: 0000000000000000 0 NOTYPE LOCAL DEFAULT UND
1: 0000000000000000 0 FUNC GLOBAL DEFAULT UND __libc_start_main@GLIBC_2.34 (2)
2: 0000000000000000 0 NOTYPE WEAK DEFAULT UND __gmon_start__
3: 000000000040401c 4 OBJECT GLOBAL DEFAULT 23 foo@VERS_2.0 (3)
$
vs older glibc:
$ readelf -W --dyn-syms tmpdir/pr28158
Symbol table '.dynsym' contains 4 entries:
Num: Value Size Type Bind Vis Ndx Name
0: 0000000000000000 0 NOTYPE LOCAL DEFAULT UND
1: 0000000000000000 0 FUNC GLOBAL DEFAULT UND __libc_start_main@GLIBC_2.2.5 (3)
2: 0000000000000000 0 NOTYPE WEAK DEFAULT UND __gmon_start__
3: 000000000040401c 4 OBJECT GLOBAL DEFAULT 23 foo@VERS_2.0 (2)
$
* testsuite/ld-elf/pr28158.rd: Adjusted for glibc 2.34.
Add .debug_loc support in the dwarf assembler, and use it in new test-case
gdb.dwarf2/loc-sec-offset.exp (which is based on
gdb.dwarf2/loclists-sec-offset.exp).
Tested on x86_64-linux.
We can't get at section->address() until everything is laid out, so
trying to generalise the offset calculation rather than using a value
of 0x8000 (the old object->toc_base_offset()) was bound to fail.
got->g_o_t() is a little better than a hard-coded 0x8000.
* powerpc.cc (Target_powerpc::Scan::local, global): Don't use
toc_pointer() here.
Split .got into two piece, one with the header and entries for small
model got entries, the other with entries for medium/large model got
entries. The idea is to better support mixed pcrel/non-pcrel code
where non-pcrel small-model .toc entries need to be within 32k of the
toc pointer.
* target.h (Target::tls_offset_for_local): Add got param.
(Target::tls_offset_for_global): Likewise.
(Target::do_tls_offset_for_local, do_tls_offset_for_global): Likewise.
* output.h (Output_data_got::Got_entry::write): Add got param.
* output.cc (Output_data_got::Got_entry::write): Likewise, pass to
tls_offset_for_local/global calls.
(Output_data_got::do_write): Adjust to suit.
* s390.cc (Target_s390::do_tls_offset_for_local): Likewise.
(Target_s390::do_tls_offset_for_global): Likewise.
* powerpc.cc (enum Got_type): Extend with small types, move from
class Target_powerpc.
(Target_powerpc::biggot_): New.
(Traget_powerpc::do_tls_offset_for_local, do_tls_offset_for_global,
got_size, got_section, got_base_offset): Handle biggot_.
(Target_powerpc::do_define_standard_symbols): Adjust.
(Target_powerpc::make_plt_section, do_finalize_sections): Likewise.
(Output_data_got_powerpc::Output_data_got_powerpc): Only make
64-bit header for small got section.
(Output_data_got_powerpc::g_o_t): Only return a result for small
got section.
(Output_data_got_powerpc::write): Only write small got section
header.
(Target_powerpc::Scan::local, global): Select small/big Got_type
and section to suit reloc.
(Target_powerpc::Relocate::relocate): Similarly.
(Sort_toc_sections): Rewrite.
Code in powerpc.cc is pretending to support a per-object toc pointer
value, but powerpc gold has no real support for multi-toc. This patch
removes the pretense, tidying quite a lot in preparation for a
followup patch. If multi-toc is ever to be supported, don't revert
this patch but start by adding object parameter to toc_pointer() and
an object to Branch_stub_key.
* powerpc.cc (Powerpc_relobj::toc_base_offset): Delete.
(Target_powerpc::toc_pointer): New function. Use throughout.
(Target_powerpc::got_base_offset): New function. Use throughout..
(Output_data_got_powerpc::got_base_offset): ..in place of
this. Delete.
(Output_data_got_powerpc::Output_data_got_powerpc): Init
header_index_ to -1u for 64-bit, and make header here.
(Output_data_got_powerpc::set_final_data_size, reserve_ent): Don't
make 64-bit header here.
(Output_data_got_powerpc::g_o_t): Return toc pointer offset in
section for 64-bit. Use throughout.
(Stub_table): Remove toc_base_off_ from Branch_stub_key, and
object param on add_long_branch_entry and find_long_branch_entry.
Adjust all uses.
When printing a variable x in a subroutine foo:
...
subroutine foo (x)
integer(4) :: x (*)
x(3) = 1
end subroutine foo
...
where x is an array with unknown bounds, we get:
...
$ gdb -q -batch outputs/gdb.fortran/array-no-bounds/array-no-bounds \
-ex "break foo" \
-ex run \
-ex "print x"
Breakpoint 1 at 0x4005cf: file array-no-bounds.f90, line 18.
Breakpoint 1, foo (x=...) at array-no-bounds.f90:18
18 x(3) = 1
$1 = <error reading variable>
...
Improve the error message by printing the details of the error, such that we
have instead:
...
$1 = <error reading variable: failed to get range bounds>
...
This is a change in gdb/valprint.c, and grepping through the sources reveals
that this is a common pattern.
Tested on x86_64-linux.
[gdb] Fix address being recorded in rs6000-tdep.c, ppc_process_record_op31.
The GDB record function was recording the variable addr that was passed in
rather than the calculated effective address (ea) by the
ppc_process_record_op31 function.
If the libopcodes disassembler returns a negative value then this
indicates that the disassembly failed for some reason. In disas.c, in
the function gdb_disassembler::print_insn we can see how this is
handled; when we get a negative value back, we call the memory_error
function, which throws an exception.
The problem here is that the address used in the memory_error call is
gdb_disassembler::m_err_memaddr, which is set in
gdb_disassembler::dis_asm_memory_error, which is called from within
the libopcodes disassembler through the
disassembler_info::memory_error_func callback.
However, for this to work correctly, every time the libopcodes
disassembler returns a negative value, the libopcodes disassembler
must have first called the memory_error_func callback.
My first plan was to make m_err_memaddr a gdb::optional, and assert
that it always had a value prior to calling memory_error, however, a
quick look in opcodes/*-dis.c shows that there _are_ cases where a
negative value is returned without first calling the memory_error_func
callback, for example in arc-dis.c and cris-dis.c.
Now, I think that a good argument can be made that these disassemblers
must therefore be broken, except for the case where we can't read
memory, we should always be able to disassemble the memory contents to
_something_, even if it's just '.word 0x....'. However, I certainly
don't plan to go and fix all of the disassemblers.
What I do propose to do then, is make m_err_memaddr a gdb::optional,
but now, instead of always calling memory_error, I add a new path
which just calls error complaining about an unknown error. This new
path is only used if m_err_memaddr doesn't have a value (indicating
that the memory_error_func callback was not called).
To test this I just augmented one of the disassemblers to always
return -1, before this patch I see this:
Dump of assembler code for function main:
0x000101aa <+0>: Cannot access memory at address 0x0
And after this commit I now see:
Dump of assembler code for function main:
0x000101aa <+0>: unknown disassembler error (error = -1)
This doesn't really help much, but that's because there's no way to
report non memory errors out of the disasembler, because, it was not
expected that the disassembler would ever report non memory errors.
When running test-case gdb.fortran/call-no-debug.exp with target board
native-gdbserver, I run into:
...
(gdb) PASS: gdb.fortran/call-no-debug.exp: print string_func_ (&'abcdefg', 3)
call (integer) string_func_ (&'abcdefg', 3)^M
$2 = 0^M
(gdb) FAIL: gdb.fortran/call-no-debug.exp: call (integer) string_func_ (&'abcdefg', 3)
...
The problem is that gdb_test is used to match inferior output.
Fix this by using gdb_test_stdio.
Tested on x86_64-linux.
When running with target board native-gdbserver, we run into a number of FAILs
due to use of the start command (and similar), which is not supported when
use_gdb_stub == 1.
Fix this by:
- requiring use_gdb_stub == 0 for the entire test-case, or
- guarding some tests in the test-case with use_gdb_stub == 0.
Tested on x86_64-linux.
When running test-case gdb.python/python.exp, we have:
...
PASS: gdb.python/python.exp: starti via gdb.execute, not from tty
PASS: gdb.python/python.exp: starti via interactive input
...
The two tests are instances of the same test, with different values for
starti command argument from_tty, so it's strange that the test names are so
different.
This is due to using a gdb_test nested in a gdb_test_multiple, with the inner
one using a different test name than the outer one. [ That could still make
sense if both produced passes, but that's not the case here. ]
Fix this by using $gdb_test_name, such that we have:
...
PASS: gdb.python/python.exp: starti via gdb.execute, not from tty
PASS: gdb.python/python.exp: starti via gdb.execute, from tty
...
Also make this more readable by using variables.
Tested on x86_64-linux.
When running test-case gdb.base/batch-exit-status.exp with target board
native-gdbserver, I run into (added missing double quotes for clarity):
...
builtin_spawn $build/gdb/testsuite/../../gdb/gdb -nw -nx \
-data-directory $build/gdb/testsuite/../data-directory \
-iex "set height 0" -iex "set width 0" \
-ex "set auto-connect-native-target off" \
-iex "set sysroot" -batch ""^M
: No such file or directory.^M
PASS: gdb.base/batch-exit-status.exp: 1x: \
No such file or directory: [lindex $result 2] == 0
FAIL: gdb.base/batch-exit-status.exp: 1x: \
No such file or directory: [lindex $result 3] == $expect_status
...
As in commit a02a90c114c "[gdb/testsuite] Set sysroot earlier in
local-board.exp", the problem is the use of -ex for
"set auto-connect-native-target off", which makes that the last command to
be executed, and consequently determines the return status.
Fix this by using -iex instead.
Tested on x86_64-linux.
When running test-case gdb.arch/i386-mpx.exp with target board
native-gdbserver, I run into:
...
(gdb) PASS: gdb.arch/i386-mpx.exp: verify size for bnd0
Remote debugging from host ::1, port 42328^M
quit^M
A debugging session is active.^M
^M
Inferior 1 [process 19679] will be killed.^M
^M
Quit anyway? (y or n) monitor exit^M
Please answer y or n.^M
A debugging session is active.^M
^M
Inferior 1 [process 19679] will be killed.^M
^M
Quit anyway? (y or n) WARNING: Timed out waiting for EOF in server after monitor exit
...
The problem is that the test-case sends a quit at the end (without verifying
the result of this in any way):
...
send_gdb "quit\n"
...
Fix this by removing the quit.
Tested on x86_64-linux.
The pseudo register handling for ARM uses some hardcoded constants to
determine types and names. In preparation to the upcoming MVE support
patch (that will add another pseudo register), this patch refactors and
reorganizes things in order to simplify handling of future pseudo registers.
We keep track of the first pseudo register number in a group and the number of
pseudo registers in that group.
Right now we only have the S and Q pseudo registers.
