When it takes more than one iteration to read the BTS trace, the trace from the
previous iteration is leaked. Fix it.
2014-01-16 Markus Metzger <markus.t.metzger@intel.com>
* common/linux-btrace.c (linux_read_btrace): Free trace from
previous iteration.
For testing multi-line test output, gdb.btrace tests used the following
pattern:
gdb_test "..." "
...\r
..."
Change this to:
gdb_test "..." [join [list \
"..." \
"..."] "\r\n"]
Also extract repeated tests into a test function and shorten or remove
test messages.
2014-01-16 Markus Metzger <markus.t.metzger@intel.com>
testsuite/
* gdb.btrace/function_call_history.exp: Update
* gdb.btrace/instruction_history.exp: Update.
The error message for starting recording twice changed.
Update the expected text to fix resulting regressions.
2014-01-16 Markus Metzger <markus.t.metzger@intel.com>
* gdb.btrace/enable.exp: Update expected text.
This test currently fails on ARM:
(gdb) PASS: gdb.dwarf2/dw2-dos-drive.exp: set breakpoint pending off
break 'z:file.c':func
Cannot access memory at address 0x0
The error is GDB trying to read the prologue at the breakpoint's
address, and failing:
38 throw_error() exceptions.c:444 0x0016728c
37 memory_error() corefile.c:204 0x001d1fcc
36 read_memory() corefile.c:223 0x001d201a
35 read_memory_unsigned_integer() corefile.c:312 0x001d2166
34 arm_skip_prologue() arm-tdep.c:1452 0x00054270
static CORE_ADDR
arm_skip_prologue (struct gdbarch *gdbarch, CORE_ADDR pc)
{
...
for (skip_pc = pc; skip_pc < limit_pc; skip_pc += 4)
{
inst = read_memory_unsigned_integer (skip_pc, 4, byte_order_for_code);
The test doesn't execute the compiled object's code, so GDB will try
to read memory from the binary's sections. Instructions on ARM are
4-byte wide, and thus ARM's prologue scanner reads in 4-byte chunks.
As the section 'func' is put at is only 1 byte long, and no other
section is allocated contiguously:
...
Sections:
Idx Name Size VMA LMA File off Algn
0 .text 00000001 00000000 00000000 00000034 2**0
CONTENTS, ALLOC, LOAD, READONLY, CODE
...
... the exec target fails the read the 4 bytes.
Fix this by increasing the function's size.
gdb/testsuite/ChangeLog:
2014-01-16 Omair Javaid <Omair.Javaid@linaro.org>
* gdb.dwarf2/dw2-dos-drive.S: Increase text section size to 4
bytes.
This adds the "main"-related data into the per-BFD. This is needed
because once symbol sharing across objfiles is complete, computing the
main name as a side effect of symbol reading will no longer work --
the symbols simply won't be re-read.
After this change, set_main_name is only used by the main_name
machinery itself, so this patch makes it static.
2014-01-15 Tom Tromey <tromey@redhat.com>
* dbxread.c (process_one_symbol): Use set_objfile_main_name.
* dwarf2read.c (read_partial_die): Use set_objfile_main_name.
* objfiles.c (get_objfile_bfd_data): Initialize language_of_main.
(set_objfile_main_name): New function.
* objfiles.h (struct objfile_per_bfd_storage) <name_of_main,
language_of_main>: New fields.
(set_objfile_main_name): Declare.
* symtab.c (find_main_name): Loop over objfiles to find the main
name and language.
(set_main_name): Now static.
(get_main_info): Add comment.
* symtab.h (set_main_name): Don't declare.
This moves the "main" name and language into an object attached to the
current progspace. This prevents problems if there are multiple
inferiors tha have different ideas of "main" -- which matters at least
for unwinding, see frame.c:inside_main_func.
2014-01-15 Tom Tromey <tromey@redhat.com>
* symtab.c (main_progspace_key): New global.
(struct main_info): New.
(name_of_main, language_of_main): Remove.
(get_main_info, main_info_cleanup): New function.
(set_main_name, main_name, main_language): Use get_main_info.
(_initialize_symtab): Initialize main_progspace_key.
This makes the global language_of_main static. Now it can be set only
via a new argument to set_main_name.
2014-01-15 Tom Tromey <tromey@redhat.com>
* dbxread.c (process_one_symbol): Update.
* dwarf2read.c (read_partial_die): Update.
* symfile.c (set_initial_language): Call main_language.
* symtab.c (language_of_main): Now static.
(set_main_name): Add 'lang' parameter.
(find_main_name): Update.
(main_language): New function.
(symtab_observer_executable_changed): Update.
* symtab.h (set_main_name): Update.
(language_of_main): Remove.
(main_language): Declare.
This moves the entry point information into the per-BFD object and
arranges not to recompute it when it has already been computed.
2014-01-15 Tom Tromey <tromey@redhat.com>
* symfile.c (init_entry_point_info): Use new "initialized" field.
Update.
* objfiles.h (struct entry_point) <initialized>: New field.
(struct objfile_per_bfd_storage) <ei>: New field, moved from...
(struct objfile) <ei>: ...here. Remove.
* objfiles.c (entry_point_address_query): Update.
This changes the entry point to be unrelocated in the objfile, and
instead applies the relocation when it is used.
2014-01-15 Tom Tromey <tromey@redhat.com>
* objfiles.c (entry_point_address_query): Relocate entry point
address.
(objfile_relocate1): Do not relocate entry point address.
* objfiles.h (struct entry_info) <entry_point>: Update comment.
<the_bfd_section_index>: New field.
* symfile.c (init_entry_point_info): Find the entry point's
section.
This is just a minor cleanup in advance of some other changes, that
modifies solib-frv.c to use entry_point_address_query. I don't have a
good way to test this but I think it is obviously correct.
2014-01-15 Tom Tromey <tromey@redhat.com>
* solib-frv.c (enable_break): Use entry_point_address_query.
This patch adds a news entry about improvements in record-replay and
reverse debugging support for arm*-linux* targets.
gdb/ChangeLog:
2014-01-15 Omair Javaid <omair.javaid@linaro.org>
* NEWS: Add note on improved process record-replay on
arm*-linux* targets.
gdb/ChangeLog:
2014-01-15 Omair Javaid <omair.javaid@linaro.org>
* arm-tdep.c (enum arm_record_result): New enum.
(arm_record_unsupported_insn): New function.
(arm_record_coproc_data_proc): Removed.
(thumb2_record_ld_st_multiple): New function.
(thumb2_record_ld_st_dual_ex_tbb): New function.
(thumb2_record_data_proc_sreg_mimm): New function.
(thumb2_record_ps_dest_generic): New function.
(thumb2_record_branch_misc_cntrl): New function.
(thumb2_record_str_single_data): New function.
(thumb2_record_ld_mem_hints): New function.
(thumb2_record_ld_word): New function.
(thumb2_record_lmul_lmla_div): New function.
(thumb2_record_decode_insn_handler): New function.
(decode_insn): Add thumb32 instruction handlers.
This patch fixes thumb push instruction recording by replacing base
register from pc to sp.
gdb/ChangeLog:
2014-01-15 Omair Javaid <omair.javaid@linaro.org>
* arm-tdep.c (thumb_record_misc): Update to use sp as base
register for push instruction recording.
This patch corrects the register numbers and removes multiple loops in
recording procedure of instructions involving multiple registers.
gdb/ChangeLog:
2014-01-15 Omair Javaid <omair.javaid@linaro.org>
* arm-tdep.c (thumb_record_misc): Update to correct logical
error while recording ldm, ldmia and pop instructions.
The recent constification of to_detach missed updating the forward
declaration of go32_detach, breaking the build:
../../src/gdb/go32-nat.c:387:1: error: conflicting types for 'go32_detach'
../../src/gdb/go32-nat.c:240:13: note: previous declaration of 'go32_detach' was here
go32_detach is actually defined before it's ever used, making the
forward declaration is unnecessary. So we can just remove it instead
of updating it. While at it, remove all others in the same situation.
Tested by building a djgpp gdb.
gdb/
2014-01-15 Pedro Alves <palves@redhat.com>
* go32-nat.c (go32_open, go32_close, go32_attach, go32_detach)
(go32_resume, go32_fetch_registers, store_register)
(go32_store_registers, go32_prepare_to_store)
(go32_xfer_memory, go32_files_info, go32_kill_inferior)
(go32_create_inferior, go32_can_run, go32_terminal_init)
(go32_terminal_inferior, go32_terminal_ours): Delete forward
declarations.
This introduces async_callback_ftype. This is needed for
make-target-delegates to work properly, as it doesn't implement a real
parser. I think it's also an ok cleanup in its own right.
2014-01-15 Tom Tromey <tromey@redhat.com>
* target.h (async_callback_ftype): New typedef.
(struct target_ops) <to_async>: Use it.
This patch fixes the grammar, and tries to do it in a way that makes
the logic behind the current implementation a little clearer.
gdb/doc/ChangeLog:
(from Yuanhui Zhang <asmwarrior@gmail.com>)
(from Joel Brobecker <brobecker@adacore.com>)
* gdb.texinfo (GDB/MI Output Syntax): Add some missing "nl"
markers. Remove one that was misplaced.
This last patch removes "partial" from the names of
expand_partial_symbol_names and map_partial_symbol_filenames.
It also renames expand_partial_symbol_names to match the
struct quick_symbol_functions "method" that it wraps:
expand_symtabs_matching.
This patch also adds two parameters to expand_symtabs_matching
so that it can fully wrap the underlying quick_symbol_functions method.
This makes it usable in more places.
I thought of having a cover function that still had the same
signature as the old expand_partial_symbol_names function,
but I couldn't think of a good name, and it wasn't clear it was
worth it anyway.
* symfile.h (expand_symtabs_matching): Renamed from
expand_partial_symbol_names. Update prototype.
(map_symbol_filenames): Renamed from map_partial_symbol_filenames.
* symfile.c (expand_symtabs_matching): Renamed from
expand_partial_symbol_names. New args file_matcher, kind.
Rename arg fun to symbol_matcher.
(map_symbol_filenames): Renamed from map_partial_symbol_filenames.
* ada-lang.c (ada_complete_symbol_matcher): Renamed from
ada_expand_partial_symbol_name.
(ada_make_symbol_completion_list): Update to call
expand_symtabs_matching.
(ada_add_global_exceptions): Call expand_symtabs_matching.
* mi/mi-cmd-file.c (mi_cmd_file_list_exec_source_files): Update to
call map_symbol_filenames.
* symtab.c (sources_info): Update to call map_symbol_filenames.
(search_symbols): Call expand_symtabs_matching.
(symbol_completion_matcher): Renamed from expand_partial_symbol_name.
(default_make_symbol_completion_list_break_on): Update to call
expand_symtabs_matching.
(make_source_files_completion_list): Update to call
map_symbol_filenames.
This patch adds two typedefs:
expand_symtabs_file_matcher_ftype
expand_symtabs_symbol_matcher_ftype
It also renames the NAME_MATCHER argument in expand_symtabs_matching.
The function is named expand_symtabs_matching and it takes a name_matcher
argument. Name of what? The symtab? A symbol?
I made it SYMBOL_MATCHER to make it clearer.
* symfile.h (expand_symtabs_file_matcher_ftype): New typedef.
(expand_symtabs_symbol_matcher_ftype): New typedef.
(quick_symbol_functions.expand_symtabs_matching): Update to use.
expand_symtabs_file_matcher_ftype, expand_symtabs_symbol_matcher_ftype.
* symfile.c (expand_partial_symbol_names): Update to use
expand_symtabs_symbol_matcher_ftype.
* dwarf2read.c (dw2_expand_symtabs_matching): Update to use
expand_symtabs_file_matcher_ftype, expand_symtabs_symbol_matcher_ftype.
Arg name_matcher renamed to symbol_matcher.
* psymtab.c (recursively_search_psymtabs): Update to use
expand_symtabs_symbol_matcher_ftype. Arg name_matcher renamed to
sym_matcher.
(expand_symtabs_matching_via_partial): Update to use
expand_symtabs_file_matcher_ftype, expand_symtabs_symbol_matcher_ftype.
Arg name_matcher renamed to symbol_matcher.
This is the first of a set of three patches to cleanup psymtab.c a bit.
Basically, these two functions do not belong in psymtab.c:
expand_partial_symbol_names, map_partial_symbol_filenames,
and "partial" does not belong in the function name.
This first patch moves them to a better location.
The second patch adds some typedefs for function parameters to
quick_symbol_functions.expand_symtabs_matching.
The third patch removes "partial" from the function names
and uses them in more places.
* psymtab.c (expand_partial_symbol_names): Delete, moved to symfile.c.
(map_partial_symbol_filenames): Ditto.
* psymtab.h (expand_partial_symbol_names): Delete, moved to symfile.h.
(map_partial_symbol_filenames): Ditto.
* symfile.c (expand_partial_symbol_names): Moved here from psymtab.c.
(map_partial_symbol_filenames): Ditto.
* symfile.h (expand_partial_symbol_names): Moved here from psymtab.h.
(map_partial_symbol_filenames): Ditto.
* symtab.c: Delete #include "psymtab.h".
RECORD_IS_USED and record_full_open look at current_target.to_stratum
to determine whether a record target is in use. This is wrong because
arch_stratum is greater than record_stratum, so if an arch_stratum
target is pushed, RECORD_IS_USED and record_full_open will miss it.
To fix this, we can use the existing find_record_target instead, which
looks up for a record stratum target across the target stack. Since
that means exporting find_record_target in record.h, RECORD_IS_USED
ends up redundant, so the patch eliminates it.
That exercise then reveals other issues:
- adjust_pc_after_break is gating record_full_... calls based on
RECORD_IS_USED. But, record_full_ calls shouldn't be made when
recording with the record-btrace target. So this adds a new
record_full_is_used predicate to be used in that spot.
- record_full_open says "Process record target already running", even
if the recording target is record-btrace ("process record" is the
original complete name of the record-full target). record_btrace_open
only says "The process is already being recorded." and does not
suggest "record stop", like record-full does. The patch factors out
and merges that error to a new record_preopen function that all record
targets call in their open routine.
Tested on x86_64 Fedora 17.
gdb/
2014-01-14 Pedro Alves <palves@redhat.com>
Tom Tromey <tromey@redhat.com>
* infrun.c (use_displaced_stepping): Use find_record_target
instead of RECORD_IS_USED.
(adjust_pc_after_break): Use record_full_is_used instead of
RECORD_IS_USED.
* record-btrace.c (record_btrace_open): Call record_preopen
instead of checking RECORD_IS_USED.
* record-full.c (record_full_shortname)
(record_full_core_shortname): New globals.
(record_full_is_used): New function.
(find_full_open): Call record_preopen instead of checking
RECORD_IS_USED.
(init_record_full_ops): Set the target's shortname to
record_full_shortname.
(init_record_full_core_ops): Set the target's shortname to
record_full_core_shortname.
* record-full.h (record_full_is_used): Declare.
* record.c (find_record_target): Make extern.
(record_preopen): New function.
* record.h (RECORD_IS_USED): Delete macro.
(find_record_target, record_preopen): Declare functions.
This patch changes the type of 'len' from LONGEST to ULONGEST. 'len'
is the argument of gdbarch methods core_xfer_shared_libraries and
core_xfer_shared_libraries_aix.
gdb:
2014-01-14 Yao Qi <yao@codesourcery.com>
* gdbarch.sh (core_xfer_shared_libraries): Change its argument
'len''s type to ULONGEST.
(core_xfer_shared_libraries_aix): Likewise.
* gdbarch.c, gdbarch.h: Reenerated.
* i386-cygwin-tdep.c (windows_core_xfer_shared_libraries):
Change type of 'len' to ULONGEST.
* rs6000-aix-tdep.c (rs6000_aix_ld_info_to_xml): Likewise.
(rs6000_aix_core_xfer_shared_libraries_aix): Likewise.
Hi,
We can use target_xfer_partial_ftype here for
rs6000_xfer_shared_libraries declaration.
gdb:
2014-01-14 Yao Qi <yao@codesourcery.com>
* rs6000-nat.c (rs6000_xfer_shared_libraries): Use
target_xfer_partial_ftype.
PR python/15464
PR python/16113
* valops.c (value_struct_elt_bitpos): New function
* py-type.c (convert_field): Set 'name' attribute of a gdb.Field
object to 'None' if the field name is an empty string ("").
* python/py-value.c (valpy_getitem): Use 'bitpos' and 'type'
attribute to look for a field when 'name' is 'None'.
(get_field_type): New function
testsuite/
* gdb.python/py-type.c: Enhance test case.
* gdb.python/py-value-cc.cc: Likewise
* gdb.python/py-type.exp: Add new tests.
* gdb.python/py-value-cc.exp: Likewise
deprecated_core_resize_section_table is declared but never defined.
This patch removes the stale declaration.
2014-01-13 Tom Tromey <tromey@redhat.com>
* gdbcore.h (deprecated_core_resize_section_table): Remove.
deprecated_flush_hook is declared in defs.h but never defined.
This patch removes the stale declaration.
2014-01-13 Tom Tromey <tromey@redhat.com>
* defs.h (deprecated_flush_hook): Remove.
gdb/
2014-01-07 Jan Kratochvil <jan.kratochvil@redhat.com>
PR threads/16216
* linux-thread-db.c (try_thread_db_load): Add parameter
check_auto_load_safe. Move here the file_is_auto_load_safe call.
(try_thread_db_load_from_pdir_1): Move it there from here.
(try_thread_db_load_from_sdir): Update caller.
(try_thread_db_load_from_dir): Move it there from here.
The rules for generating the output register files look like:
amd64.c : $(srcdir)/../regformats/i386/amd64.dat $(regdat_sh)
$(SHELL) $(regdat_sh) $(srcdir)/../regformats/i386/amd64.dat amd64.c
According to this rule, if regdat.sh is newer than amd64.c, then
regdat.sh shall be invoked on amd64.dat. According to regdat.sh, if the
script determines that the output file amd64.c has not changed, then it
will not overwrite the existing output file. This means that a
subsequent invocation of make will trigger the above rule again as
regdat.sh will be perpetually newer than amd64.c.
This then shows up in the make output like so:
/bin/bash ./../regformats/regdat.sh ./../regformats/i386/amd64-linux.dat amd64-linux.c
amd64-linux.c unchanged.
/bin/bash ./../regformats/regdat.sh ./../regformats/i386/amd64-avx-linux.dat amd64-avx-linux.c
amd64-avx-linux.c unchanged.
...
To fix this pathological behavior, it suffices to have regdat.sh
unconditionally rewrite the output register file.
On my machine, which has a regdat.sh file that is newer than some of the
input register files, this change speeds up every invocation of make
under gdb/ by about 5 seconds.
Tries to compile each header in isolation, thus ensuring headers are
self-contained.
Defaults to checking all $HFILES_NO_SRCDIR headers.
Do:
make check-headers CHECK_HEADERS="header.h list.h"
to check specific headers.
gdb/
2014-01-13 Pedro Alves <palves@redhat.com>
* Makefile.in (CHECK_HEADERS): New variable.
(check-headers:): New rule.
This removes deprecated_set_hook. Insight was the last user of this
hook, but I recently checked in a patch to have it use the
command_param_changed observer instead.
2014-01-13 Tom Tromey <tromey@redhat.com>
* cli/cli-setshow.c (do_set_command): Update.
* defs.h (deprecated_set_hook): Remove.
* top.c (deprecated_set_hook): Remove.
This PC guessing can't work when the PC is a pseudo-register.
Pseudo-register values don't end up stored in the regcache, they're
always recomputed. And, it's actually wrong to try to write a
pseudo-register with regcache_raw_supply. Skip it and add a comment.
gdb/
2014-01-13 Pedro Alves <palves@redhat.com>
* tracepoint.c (tfile_fetch_registers): Don't infer the PC from
the tracepoint if the PC is a pseudo-register.
This removes XCALLOC and replaces it either with XCNEWVEC, or, if the
number of elements being requested was 1, with XCNEW.
2014-01-13 Tom Tromey <tromey@redhat.com>
* defs.h (XCALLOC): Remove.
* bcache.c (bcache_xmalloc): Use XCNEW, not XCALLOC.
(print_bcache_statistics): Use XCNEWVEC, not XCALLOC.
* dwarf2loc.c (allocate_piece_closure): Likewise.
* elfread.c (elf_symfile_segments): Likewise.
(elf_symfile_segments): Likewise.
* gdbtypes.c (copy_type_recursive): Likewise.
* i386-tdep.c (i386_gdbarch_init): Use XCNEW, not XCALLOC.
* jit.c (jit_frame_sniffer): Use XCNEWVEC, not XCALLOC.
* minsyms.c (prim_record_minimal_symbol_full): Use XCNEW, not
XCALLOC.
* mt-tdep.c (mt_gdbarch_init): Likewise.
* opencl-lang.c (allocate_lval_closure): Use XCNEWVEC, not
XCALLOC.
* psymtab.c (psymbol_compare): Use XCNEW, not XCALLOC.
* regcache.c (regcache_xmalloc_1): Use XCNEWVEC, not XCALLOC.
* registry.c (registry_alloc_data): Likewise.
* rs6000-tdep.c (rs6000_gdbarch_init): Use XCNEW, not XCALLOC.
* s390-linux-tdep.c (s390_gdbarch_init): Likewise.
* serial.c (serial_fdopen_ops): Likewise.
* solib-aix.c (solib_aix_get_section_offsets): Use XCNEWVEC, not
XCALLOC.
* spu-tdep.c (spu_gdbarch_init): Use XCNEW, not XCALLOC.
* symfile.c (default_symfile_segments): Use XCNEW and XCNEWVEC,
not XCALLOC.
This patch reorganizes code on gdb/i386-tdep.c's SystemTap SDT probe
support functions. Before it, the code to parse special operands on x86
lived in a single, big function. This patch creates 2 new functions
that makes the code more organized and removes a few indentation levels
(which is always good IMO).
I haven't modified anything logical in the functions, i.e., there's still
one latent bug on i386_stap_parse_special_token_triplet now. I will soon
post a patch to fix this, and to also improve the readability of the two
new functions.
2014-01-12 Sergio Durigan Junior <sergiodj@redhat.com>
* i386-tdep.c (i386_stap_parse_special_token_triplet): New
function, with code from i386_stap_parse_special_token.
(i386_stap_parse_special_token_three_arg_disp): Likewise.
(i386_stap_parse_special_token): Move code to the two functions
above; simplify it.
gdb/doc/
2014-01-10 Pedro Alves <palves@redhat.com>
* gdb.texinfo (Your Program's Environment) <set environment>:
Mention the shell, and point at 'set exec-wrapper'.
S390, the dw2-dir-file-name test case fails in the first
gdb_continue_to_breakpoint. Indeed, the breakpoint is now placed into
the alignment gap *before* the actual function.
This happens because the test case declares the respective "*_start"
symbol as a "loose" label before the function definition, and the
compiler inserts the alignment between that label and the function
itself.
The "*_start" symbols were only necessary because FUNC made the
function static. The fix makes the functions extern instead, thus
making the "*_start" labels unnecessary.
testsuite/
2014-01-10 Andreas Arnez <arnez@linux.vnet.ibm.com>
Pedro Alves <palves@redhat.com>
* gdb.dwarf2/dw2-dir-file-name.c (FUNC): Remove "*_start" symbol.
Make "name" extern.
* gdb.dwarf2/dw2-dir-file-name.exp (out_cu, out_line): Replace
references to ${name}_start by references to ${name}.
A live target is required for `-info-os' to work in non-native
configurations.
(gdb)
Expecting: ^(-info-os[
]+)?(.*\^done,OSDataTable=.*[
]+[(]gdb[)]
[ ]*)
-info-os
^error,msg="Don't know how to get OS data. Try \"help target\"."
(gdb)
FAIL: gdb.mi/mi-info-os.exp: -info-os
If GDB does have a native configuration included, but we're testing
remote, it'll be worse, as if we're not connected yet, -info-os will
run against the default run target, and pass, falsely giving the
impression the remote bits were exercised.
gdb/testsuite/
2014-01-09 Maciej W. Rozycki <macro@codesourcery.com>
Pedro Alves <palves@redhat.com>
* gdb.mi/mi-info-os.exp: Connect to the target with
mi_gdb_target_load.
Although we can tell upfront whether a remote target supports target
side commands, we can only tell whether the target supports that in
combination with a given breakpoint kind (software, hardware,
watchpoints, etc.) when we go and try to insert such a breakpoint kind
the first time. It's not desirable to make remote_insert_breakpoint
simply return -1 in this case, because if the breakpoint was set in a
shared library, insert_bp_location will assume that the breakpoint
insertion failed because the library wasn't mapped in.
insert_bp_location already handles errors/exceptions thrown from the
target_insert_xxx methods, exactly so the backend can tell the user
the detailed reason the insertion of hw breakpoints failed. But, in
the case of software breakpoints, it discards the detailed error
message.
So the patch makes insert_bp_location use the error's message for SW
breakpoints too, and, introduces a NOT_SUPPORTED_ERROR error code so
that insert_bp_location doesn't confuse the error for failure due to a
shared library disappearing.
The result is:
(gdb) c
Warning:
Cannot insert breakpoint 2: Target doesn't support breakpoints that have target side commands.
2014-01-09 Pedro Alves <palves@redhat.com>
Hui Zhu <hui@codesourcery.com>
PR gdb/16101
* breakpoint.c (insert_bp_location): Rename hw_bp_err_string to
bp_err_string. Don't mark the location shlib_disabled if the
error thrown wasn't a generic or memory error. Catch errors
thrown while inserting breakpoints in overlayed code. Output
error message of software breakpoints.
* remote.c (remote_insert_breakpoint): If this breakpoint has
target-side commands but this stub doesn't support Z0 packets,
throw NOT_SUPPORTED_ERROR error.
* exceptions.h (enum errors) <NOT_SUPPORTED_ERROR>: New error.
* target.h (target_insert_breakpoint): Extend comment.
(target_insert_hw_breakpoint): Add comment.
... not when a new GDB connection sends the status packet ('?').
Mainly just a cleanup/simplification, as GDB always sends '?' first.
Tested on x86_64 Fedora 17.
2014-01-08 Pedro Alves <palves@redhat.com>
* server.c (handle_status): Don't discard previous queued stop
replies or thread's pending status here.
(main) <disconnection>: Do it here instead.
Currently, when GDB connects in all-stop mode, GDBserver always
responds to the status packet with a GDB_SIGNAL_TRAP, even if the
program is actually stopped for some other signal.
(gdb) tar rem ...
...
(gdb) c
Program received signal SIGUSR1, User defined signal 1.
(gdb) disconnect
(gdb) tar rem ...
(gdb) c
(Or a GDB crash instead of an explicit disconnect.)
This results in the program losing that signal on that last continue,
because gdb will tell the target to resume with no signal (to suppress
the GDB_SIGNAL_TRAP, due to 'handle SISGTRAP nopass'), and that will
actually suppress the real signal the program had stopped for
(SIGUSR1). To fix that, I think we should make GDBserver report the
real signal the thread had stopped for in response to the status
packet:
@item ?
@cindex @samp{?} packet
Indicate the reason the target halted. The reply is the same as for
step and continue.
But, that raises the question -- which thread are we reporting the
status for? Due to how the RSP in all-stop works, we can only report
one status. The status packet's response is a stop reply packet, so
it includes the thread identifier, so it's not a problem packet-wise.
However, GDBserver is currently always reporting the status for first
thread in the thread list, even though that may well not be the thread
that got the signal that caused the program to stop. So the next
logical step would be to report the status for the
last_ptid/last_status thread (the last event reported to gdb), if it's
still around; and if not, fallback to some other thread.
There's an issue on the GDB side with that, though...
GDB currently always adds the thread reported in response to the
status query as the first thread in its list. That means that if we
start with e.g.,
(gdb) info threads
3 Thread 1003 ...
* 2 Thread 1002 ...
1 Thread 1001 ...
And reconnect:
(gdb) disconnect
(gdb) tar rem ...
We end up with:
(gdb) info threads
3 Thread 1003 ...
2 Thread 1001 ...
* 1 Thread 1002 ...
Not a real big issue, but it's reasonably fixable, by having GDB
fetch/sync the thread list before fetching the status/'?', and then
using the status to select the right thread as current on the GDB
side. Holes in the thread numbers are squashed before/after
reconnection (e.g., 2,3,5 becomes 1,2,3), but the order is preserved,
which I think is both good, and good enough.
However (yes, there's more...), the previous GDB that was connected
might have had gdbserver running in non-stop mode, or could have left
gdbserver doing disconnected tracing (which also forces non-stop), and
if the new gdb/connection is in all-stop mode, we can end up with more
than one thread with a signal to report back to gdb. As we can only
report one thread/status (in the all-stop RSP variant; the non-stop
variant doesn't have this issue), we get to do what we do at every
other place we have this situation -- leave events we can't report
right now as pending, so that the next resume picks them up.
Note all this ammounts to a QoI change, within the existing framework.
There's really no RSP change here.
The only user visible change (other than that the signal is program is
stopped at isn't lost / is passed to the program), is in "info
program", that now can show the signal the program stopped for. Of
course, the next resume will respect the pass/nopass setting for the
signal in question. It'd be reasonable to have the initial connection
tell the user the program was stopped with a signal, similar to when
we load a core to debug, but I'm leaving that out for a future change.
I think we'll need to either change how handle_inferior_event & co
handle stop_soon, or maybe bypass them completely (like
fork-child.c:startup_inferior) for that.
Tested on x86_64 Fedora 17.
gdb/gdbserver/
2014-01-08 Pedro Alves <palves@redhat.com>
* gdbthread.h (struct thread_info) <status_pending_p>: New field.
* server.c (visit_actioned_threads, handle_pending_status): New
function.
(handle_v_cont): Factor out parts to ...
(resume): ... this new function. If in all-stop, and a thread
being resumed has a pending status, report it without actually
resuming.
(myresume): Adjust to use the new 'resume' function.
(clear_pending_status_callback, set_pending_status_callback)
(find_status_pending_thread_callback): New functions.
(handle_status): Handle the case of multiple threads having
interesting statuses to report. Report threads' real last signal
instead of always reporting GDB_SIGNAL_TRAP. Look for a thread
with an interesting thread to report the status for, instead of
always reporting the status of the first thread.
gdb/
2014-01-08 Pedro Alves <palves@redhat.com>
* remote.c (remote_add_thread): Add threads silently if starting
up.
(remote_notice_new_inferior): If in all-stop, and starting up,
don't call notice_new_inferior.
(get_current_thread): New function, factored out from ...
(add_current_inferior_and_thread): ... this. Adjust.
(remote_start_remote) <all-stop>: Fetch the thread list. If we
found any thread, then select the remote's current thread as GDB's
current thread too.
gdb/testsuite/
2014-01-08 Pedro Alves <palves@redhat.com>
* gdb.threads/reconnect-signal.c: New file.
* gdb.threads/reconnect-signal.exp: New file.
Now that the GDB 7.7 branch has been created, we can
bump the version number.
gdb/ChangeLog:
GDB 7.7 branch created (79301218fa):
* version.in: Bump version to 7.7.50.DATE-cvs.
This patch fixes these errors below:
../../binutils-gdb/gdb/spu-linux-nat.c: In function ‘spu_symbol_file_add_from_memory’:
../../binutils-gdb/gdb/spu-linux-nat.c:368:3: error: pointer targets in passing argument 2 of ‘spu_proc_xfer_spu’ differ in signedness [-Werror=pointer-sign]
../../binutils-gdb/gdb/spu-linux-nat.c:232:1: note: expected ‘gdb_byte *’ but argument is of type ‘char *’
../../binutils-gdb/gdb/spu-linux-nat.c: In function ‘spu_xfer_partial’:
../../binutils-gdb/gdb/spu-linux-nat.c:598:7: error: pointer targets in passing argument 1 of ‘strtoulst’ differ in signedness [-Werror=pointer-sign]
In file included from ../../binutils-gdb/gdb/defs.h:769:0,
from ../../binutils-gdb/gdb/spu-linux-nat.c:21:
../../binutils-gdb/gdb/utils.h:43:15: note: expected ‘const char *’ but argument is of type ‘gdb_byte *’
gdb:
2014-01-08 Yao Qi <yao@codesourcery.com>
* spu-linux-nat.c (spu_symbol_file_add_from_memory): Change
type of 'id' to gdb_byte. Cast 'id' to 'const char *'.
(spu_xfer_partial): Cast 'buf' to 'const char *'.
This patch fixes the following build error:
../../binutils-gdb/gdb/spu-linux-nat.c:383:5: error: passing argument 2 of ‘symbol_file_add_from_bfd’ makes pointer from integer without a cast [-Werror]
In file included from ../../binutils-gdb/gdb/spu-linux-nat.c:29:0:
../../binutils-gdb/gdb/symfile.h:444:24: note: expected ‘const char *’ but argument is of type ‘int’
../../binutils-gdb/gdb/spu-linux-nat.c:383:5: error: passing argument 3 of ‘symbol_file_add_from_bfd’ makes integer from pointer without a cast [-Werror]
In file included from ../../binutils-gdb/gdb/spu-linux-nat.c:29:0:
../../binutils-gdb/gdb/symfile.h:444:24: note: expected ‘int’ but argument is of type ‘void *’
../../binutils-gdb/gdb/spu-linux-nat.c:383:5: error: passing argument 5 of ‘symbol_file_add_from_bfd’ makes integer from pointer without a cast [-Werror]
In file included from ../../binutils-gdb/gdb/spu-linux-nat.c:29:0:
../../binutils-gdb/gdb/symfile.h:444:24: note: expected ‘int’ but argument is of type ‘void *’
../../binutils-gdb/gdb/spu-linux-nat.c:383:5: error: too few arguments to function ‘symbol_file_add_from_bfd’
Argument 'name' was added to function symbol_file_add_from_bfd by this patch
[patchv4 4/5] Keep objfile original filename
https://sourceware.org/ml/gdb-patches/2013-09/msg00683.html
but caller of symbol_file_add_from_bfd in spu-linux-nat.c wasn't updated.
This patch fixes the build error.
gdb:
2014-01-08 Yao Qi <yao@codesourcery.com>
* spu-linux-nat.c (spu_symbol_file_add_from_memory): Pass
return value of bfd_get_filename to symbol_file_add_from_bfd.
* coff-pe-read.c (struct read_pe_section_data): Add index field.
(add_pe_exported_sym): Use SECTION_DATA->INDEX for call
to prim_record_mininal_symbol_and_info.
(add_pe_forwarded_sym): Use known section number of forwarded symbol
in call to prim_record_minimal_symbol_and_info.
(read_pe_exported_syms): Set index field of section_data.
gdb/ChangeLog:
2014-01-07 Edjunior Barbosa Machado <emachado@linux.vnet.ibm.com>
* source.c (add_path): Fix check for duplicated paths in the previously
included paths.
gdb/testsuite/ChangeLog:
2014-01-07 Jan Kratochvil <jan.kratochvil@redhat.com>
* gdb.base/source-dir.exp: New file.
This patch fixes this build error below:
../../binutils-gdb/gdb/spu-linux-nat.c:616:1: error: no previous prototype for ‘_initialize_spu_nat’ [-Werror=missing-prototypes]
gdb:
2014-01-07 Yao Qi <yao@codesourcery.com>
* spu-linux-nat.c (_initialize_spu_nat): Declare.
When I verify my changes to target.h doesn't break build on aix, I get
the following build error on a clean GDB checkout.
../../binutils-gdb/gdb/aix-thread.c: In function 'pdc_read_regs':
../../binutils-gdb/gdb/aix-thread.c:366:4: error: passing argument 3 of 'ptrace32' makes integer from pointer without a cast [-Werror]
if (!ptrace32 (PTT_READ_GPRS, tid, gprs32, 0, NULL))
^
../../binutils-gdb/gdb/aix-thread.c:263:1: note: expected 'long long int' but argument is of type 'uint32_t *'
ptrace32 (int req, int id, addr_ptr addr, int data, int *buf)
^
../../binutils-gdb/gdb/aix-thread.c:375:42: error: cast from pointer to integer of different size [-Werror=pointer-to-int-cast]
if (!ptrace32 (PTT_READ_FPRS, tid, (addr_ptr) fprs, 0, NULL))
^
../../binutils-gdb/gdb/aix-thread.c:392:39: error: cast from pointer to integer of different size [-Werror=pointer-to-int-cast]
if (!ptrace32 (PTT_READ_SPRS, tid, (addr_ptr) &sprs32, 0, NULL))
GCC uses -maix32 in default, so the 'long long' is 64 bit and address
is 32 bit. Such warnings should go away if -maix64 is used.
In this patch, I cast the parameter to uintptr_t first, and then cast
to addr_ptr.
gdb:
2014-01-07 Yao Qi <yao@codesourcery.com>
Joel Brobecker <brobecker@adacore.com>
* aix-thread.c (pdc_read_regs): Cast parameter to uintptr_t.
(pdc_write_regs): Likewise.
(fetch_regs_kernel_thread): Likewise.
(store_regs_kernel_thread): Likewise.
Consider the following code:
type Element is abstract tagged null record;
type GADataType is interface;
type Data_Type is new Element and GADataType with record
I : Integer := 42;
end record;
Result1 : Data_Type;
GGG1 : GADataType'Class := GADataType'Class (Result1);
When trying to create a varobj for variable ggg1, GDB currently
returns an object which has no child:
-var-create ggg1 * ggg1
^done,name="ggg1",numchild="0",[...]
This is incorrect, it should return an object which has one child
(field "i"). This is because tagged-type objects are dynamic, and
we need to apply a small transformation in order to get their actual
type. This is already done on the GDB/CLI side in ada-valprint,
and it needs to be done on the ada-varobj side as well.
gdb/ChangeLog:
* ada-varobj.c (ada_varobj_adjust_for_child_access): Convert
tagged type objects to their actual type.
gdb/testsuite/ChangeLog:
* gdb.ada/mi_interface: New testcase.
Consider the following types:
type Time_T is record
Secs : Integer;
end record;
Before : Time_T := (Secs => 1384395743);
In this example, we assume that type Time_T is the number of seconds
since Epoch, and so added a Python pretty-printer, to print this
type in a more human-friendly way. For instance:
(gdb) print before
$1 = Thu Nov 14 02:22:23 2013 (1384395743)
However, we've noticed that things stop working when this type is
embedded inside another record, and we try to print that record.
For instance, with the following declarations:
type Composite is record
Id : Integer;
T : Time_T;
end record;
Afternoon : Composite := (Id => 1, T => (Secs => 1384395865));
(gdb) print afternoon
$2 = (id => 1, t => (secs => 1384395865))
We expected instead:
(gdb) print afternoon
$2 = (id => 1, t => Thu Nov 14 02:24:25 2013 (1384395865))
This patch fixes the problem by making sure that we try to print
each field via a call to val_print, rather than calling ada_val_print
directly. We need to go through val_print, as the val_print
handles all language-independent features such as calling the
pretty-printer, knowing that ada_val_print will get called eventually
if actual Ada-specific printing is required (which should be the
most common scenario).
And because val_print takes the language as parameter, we enhanced
the print_field_values and print_variant_part to also take a language.
As a bonus, this allows us to remove a couple of references to
current_language.
gdb/ChangeLog:
* ada-valprint.c (print_field_values): Add "language" parameter.
Update calls to print_field_values and print_variant_part.
Pass new parameter "language" in call to val_print instead
of "current_language". Replace call to ada_val_print by call
to val_print.
(print_variant_part): Add "language" parameter.
(ada_val_print_struct_union): Update call to print_field_values.
gdb/testsuite/ChangeLog:
* gdb.ada/pp-rec-component.exp, gdb.ada/pp-rec-component.py,
gdb.ada/pp-rec-component/foo.adb, gdb.ada/pp-rec-component/pck.adb,
gdb.ada/pp-rec-component/pck.ads: New files.
ada_print_floating declares a char buffer with a size that we're hoping
to always be large enough to hold any string representation of a float
value. But that's not really necessary, and also forces us to create
a small wrapper (ui_memcpy) to perform the extraction from a temporary
stream into this buffer. This patches fixes both issues by relying on
ui_file_xstrdup. This forces us to make a few adjustments that are
minor in nature, as we now need to defer the cleanup to the end of
the function.
gdb/ChangeLog:
* ada-valprint.c (ui_memcpy): Delete.
(ada_print_floating): Update documentation. Add empty line
between between function documentation and implementation.
Delete variable "buffer". Use ui_file_xstrdup in place of
ui_file_put. Minor adjustments following this change.
This patch creates a new function called "ada_val_print_string"
whose code is directly extracted out of ada_val_print_array.
The extracted code is then replaced by a call to this new function,
followed by a "return". The return avoids the need for an "else"
branch, with the associated block nesting. The latter is not really
terrible in this case, but it seems more readable this way.
gdb/ChangeLog:
* ada-valprint.c (ada_val_print_string): New function,
extracted from ada_val_print_array.
(ada_val_print_array): Replace extracted code by call
to ada_val_print_string followed by a return. Move
"else" branch to the function's top block.
This patch moves ada_val_print_array to group it with the other
ada_val_print_* function which are being called by ada_val_print_1.
Since this function is in the same situation, it is more logical
to move it within that group.
It also rationalizes the function's prototype to match the prototype
of the other ada_val_print_* routines.
gdb/ChangeLog:
* ada-valprint.c (ada_val_print_array): Move implementation
down. Rename parameter "offset" and "val" into "offset_aligned"
and "original_value" respectively. Add parameter "offset".
The logic as currently implemented in this function was a little
difficult to follow, due to the nested of if/else conditions,
but most of the time, the "else" block was very simple. So this
patch re-organizes the code to use fewer levels of nesting by
using return statements, and writing the code as a sequence of
"if something simple, then handle it and return" blocks.
While touching this code, this patch changes the cryptic "???"
printed when trying to print a reference pointing to an undefined
type. This should only ever happen if the debugging information
was corrupted or improperly read. But in case that happens, we now
print "<ref to undefined type>" instead. This is more in line
with how we print other conditions such as optimized out pieces,
or synthetic pointers.
gdb/ChangeLog:
* ada-valprint.c (ada_val_print_ref): Rewrite by mostly
re-organizing the code. Change the "???" message printed
when target type is a TYPE_CODE_UNDEF into
"<ref to undefined type>".
The function print_record is a fairly small and straightforward
function which is only called from one location. So this patch
inlines the code at the point of call.
One small advantage is that the context of use of this patch has
now become such that we can assume that TYPE is not a typedef,
nor an enum. So thhe call to ada_check_typedef is unnecessary,
and this patch removes it.
gdb/ChangeLog:
* ada-valprint.c (print_record): Delete, implementation inlined...
(ada_val_print_struct_union): ... here. Remove call to
ada_check_typedef in inlined implementation.
The idea of this patch is that it's hard to have a global view of
ada_val_print_1 because its body spans over too many lines. Also,
each individual "case" block within the giant "switch" can be hard
to isolate if spanning over multiple pages as well.
gdb/ChangeLog:
* ada-valprint.c (ada_val_print_gnat_array): New function,
extracted from ada_val_print_1;
(ada_val_print_ptr, ada_val_print_num, ada_val_print_enum)
(ada_val_print_flt, ada_val_print_struct_union)
(ada_val_print_ref): Likewise.
(ada_val_print_1): Delete variables i and elttype.
Replace extracted-out code by call to corresponding
new functions.
I am not sure why this function was called in the first place, but
it disrupts the printing flow when in GDB/MI mode, ending the current
console stream output, and starting a new one. It's not clear whether,
with the code as currently written, the problem is actually visible
or only latent. But, it becomes visible when we replace one of the
"return" statements in the "switch" block just above by a "break"
statement (this is something I'd like to do, and what made me realize
the problem). With the gdb_flush call (after having replaced the
"return" statement as explained above), we get:
% gdb -q -i=mi ada_prg
(gdb)
print 1
&"print 1\n"
!! -> ~"$1 = 1"
!! -> ~"\n"
^done
With the gdb_flush call removed, we now get the entire output into
a single stream.
(gdb)
print 1
&"print 1\n"
~"$1 = 1"
~"\n"
^done
gdb/ChangeLog:
* ada-valprint.c (ada_val_print_1): Remove call to gdb_flush.
This is to standardize a little bit how printing is done, and in
particular make sure that everyone goes through val_print when
printing sub-objects. This helps making sure that standard features
handled by val_print get activated when expected.
gdb/ChangeLog:
* ada-valprint.c (ada_val_print_1): Replace calls to
ada_val_print_1 by calls to val_print.
This is to help calling val_print. We would like to be more systematic
in calling val_print when printing, because it allows us to make sure
we take advantage of the standard features such as pretty-printing
which are handled by val_print.
gdb/ChangeLog:
* ada-valprint.c (ada_val_print_1): Add parameter "language".
Update calls to self accordingly. Replace calls to c_val_print
by calls to val_print.
Advance function declarations add to the maintenance cost, since
any update to the function prototype needs to be made twice.
For static functions, this is not necessary, and this patch
reorders the function so as to reduce the use of such advanche
declarations.
gdb/ChangeLog:
* ada-valprint.c (print_record): Delete declaration.
(adjust_type_signedness, ada_val_print_1): Likewise.
(ada_val_print): Move function implementation down.
(print_variant_part, print_field_values, print_record):
Move function implementation up.
Consider the following declarations:
typedef long our_time_t;
our_time_t current_time = 1384395743;
The purpose of this patch is to allow the use of a pretty-printer
for variables of type our_time_t. Normally, pretty-printing sniffers
use the tag name in order to determine which, if any, pretty-printer
should be used. But in the case above, the tag name is not set, since
it does not apply to integral types.
This patch extends the gdb.Type list of attributes to also include
the name of the type, thus allowing the sniffer to match against
that name. With that change, I was able to write a pretty-printer
which displays our variable as follow:
(gdb) print current_time
$1 = Thu Nov 14 02:22:23 2013 (1384395743)
gdb/ChangeLog:
* python/py-type.c (typy_get_name): New function.
(type_object_getset): Add entry for attribute "name".
* NEWS: Add entry mentioning this new attribute.
gdb/doc/ChangeLog:
* gdb.texinfo (Types In Python): Document new attribute Types.name.
gdb/testsuite:
* gdb.python/py-pp-integral.c: New file.
* gdb.python/py-pp-integral.py: New file.
* gdb.python/py-pp-integral.exp: New file.
Tested on x86_64-linux.
This patch removes the if statement and the comments together.
gdb:
2014-01-07 Yao Qi <yao@codesourcery.com>
* gnu-nat.c (set_exceptions_cmd): Remove an empty body 'if'
statement.
This patch fixes the following error.
../../../git/gdb/gnu-nat.c: In function 'info_port_rights':
../../../git/gdb/gnu-nat.c:3083:11: error: passing argument 1 of 'parse_to_comma_and_eval' from incompatible pointer type [-Werror]
In file included from ../../../git/gdb/breakpoint.h:23:0,
from ../../../git/gdb/inferior.h:37,
from ../../../git/gdb/gnu-nat.c:55:
../../../git/gdb/value.h:763:22: note: expected 'const char **' but argument is of type 'char **'
gdb:
2014-01-07 Yao Qi <yao@codesourcery.com>
* gnu-nat.c (info_port_rights): Add qualifier const to
argument args.
This patch fixes the following error:
../../../git/gdb/gnu-nat.c: In function 'trace_me':
../../../git/gdb/gnu-nat.c:2106:8: error: old-style function definition [-Werror=old-style-definition]
gdb:
2014-01-07 Yao Qi <yao@codesourcery.com>
* gnu-nat.c (trace_me): Use 'void' for empty argument list.
inf_tid_to_proc is not defined at all. This patch is to remove its
declaration.
gdb:
2014-01-07 Yao Qi <yao@codesourcery.com>
* gnu-nat.c (inf_tid_to_proc): Remove declaration.
This patch fixes this error below by declaring _initialize_gnu_nat.
../../../git/gdb/gnu-nat.c:3447:1: error: no previous prototype for '_initialize_gnu_nat' [-Werror=missing-prototypes]
gdb:
2014-01-07 Yao Qi <yao@codesourcery.com>
* gnu-nat.c (_initialize_gnu_nat): Declare.
This patch changes the return type of gdbarch_byte_order and
gdbarch_byte_order_for_code, from 'int' to 'enum bfd_endian'.
gdb:
2014-01-07 Yao Qi <yao@codesourcery.com>
* gdbarch.sh (byte_order, byte_order_for_code): Change type to
'enum bfd_endian'.
(struct gdbarch_info) <byte_order>: Change type to
'enum bfd_endian'.
<byte_order_for_code>: Likewise.
* gdbarch.c, gdbarch.h: Regenerated.
This removes the last uses of the obsolete CONST macro from the tree.
I'm checking this in. Tested by rebuilding.
2014-01-06 Tom Tromey <tromey@redhat.com>
* doublest.c (convert_doublest_to_floatformat): Use const, not
CONST.
* somread.c (som_symtab_read): Likewise.
gdb/ChangeLog:
* top.c (print_gdb_version): Set copyright year to 2014.
gdb/gdbserver/ChangeLog:
* gdbserver.c (gdbserver_version): Set copyright year to 2014.
* gdbreplay.c (gdbreplay_version): Likewise.
