Joel Brobecker [Mon, 2 Feb 2015 03:20:58 +0000 (07:20 +0400)]
[Ada] pspace_data->sym_cache is always NULL
The Ada symbol cache has been designed to have one instance of that
of that cache per program space, and for each instance to be created
on-demand. ada_get_symbol_cache is the function responsible for both
lookup and creation on demand.
Unfortunately, ada_get_symbol_cache forgot to store the reference
to newly created caches, thus causing it to:
- Leak old caches;
- Allocate a new cache each time the cache is being searched or
a new entry is to be inserted.
This patch fixes the issue by avoiding the use of the local variable,
which indirectly allowed the bug to happen. We manipulate the reference
in the program-space data instead.
gdb/ChangeLog:
PR gdb/17854:
* ada-lang.c (ada_get_symbol_cache): Set pspace_data->sym_cache
when allocating a new one.
Eli Zaretskii [Sat, 31 Jan 2015 08:47:14 +0000 (10:47 +0200)]
Make sure TABs are expanded in TUI windows on MS-Windows.
gdb/
2015-01-31 Eli Zaretskii <eliz@gnu.org>
* tui/tui-io.c (tui_expand_tabs): New function.
(tui_puts, tui_redisplay_readline): Expand TABs into the
appropriate number of spaces.
* tui/tui-regs.c: Include tui-io.h.
(tui_register_format): Call tui_expand_tabs to expand TABs into
the appropriate number of spaces.
* tui/tui-io.h: Add prototype for tui_expand_tabs.
Wei-cheng Wang [Sat, 17 Jan 2015 06:30:33 +0000 (14:30 +0800)]
Reverse debugging for PowerPC.
gdb/ChangeLog
Backport from mainline:
* configure.tgt (powerpc*-*-linux): Add linux-record.o to
gdb_target_obs.
(ppc_linux_record_tdep, ppc64_linux_record_tdep): New for linux syscall
record.
(ppc_canonicalize_syscall, ppc_linux_syscall_record,
ppc_linux_record_signal, ppc_init_linux_record_tdep): New functions.
(ppc_linux_init_abi): Set process_record, process_record_signal.
* ppc-tdep.h (struct gdbarch_tdep): Add ppc_syscall_record and
ppc_linux_record_tdep to gdbarch_tdep.
(ppc_process_record): New declaration.
* rs6000-tdep.c (ppc_record_vsr, ppc_process_record_op4,
ppc_process_record_op19, ppc_process_record_op31,
ppc_process_record_op59, ppc_process_record_op60,
ppc_process_record_op63, ppc_process_record): New functions.
gdb/testsuite/ChangeLog
Backport from mainline:
* lib/gdb.exp (supports_process_record): Return true for
powerpc*-*-linux*.
(supports_reverse): Likewise.
Wei-cheng Wang [Sat, 17 Jan 2015 06:29:16 +0000 (14:29 +0800)]
Epilogue unwinder for PowerPC.
gdb/ChangeLog
Backport from mainline:
* rs6000-tdep.c (rs6000_in_function_epilogue_p): Rename to
rs6000_in_function_epilogue_frame_p and add an argument
for frame_info.
(rs6000_epilogue_frame_cache, rs6000_epilogue_frame_this_id,
rs6000_epilogue_frame_prev_register, rs6000_epilogue_frame_sniffer):
New functions.
(rs6000_epilogue_frame_unwind): New.
(rs6000_gdbarch_init): Append epilogue unwinder.
Jan Kratochvil [Sun, 25 Jan 2015 07:57:01 +0000 (08:57 +0100)]
Fix 100x slowdown regression on DWZ files
Since Fedora started to use DWZ DWARF compressor:
http://fedoraproject.org/wiki/Features/DwarfCompressor
GDB has slowed down a lot. To make it clear - DWZ is DWARF structure
rearrangement, "compressor" does not mean any zlib style data compression.
This patch reduces LibreOffice backtrace from 5 minutes to 3 seconds (100x)
and it also reduces memory consumption 20x.
[ benchmark is at the bottom of this mail ]
One can see all DW_TAG_partial_unit have DW_AT_stmt_list 0x0 which causes
repeated decoding of that .debug_line unit on each DW_TAG_imported_unit.
This was OK before as each DW_TAG_compile_unit has its own .debug_line unit.
But since the introduction of DW_TAG_partial_unit by DWZ one should cache
read-in DW_AT_stmt_list .debug_line units.
Fortunately one does not need to cache whole
struct linetable *symtab->linetable
and other data from .debug_line mapping PC<->lines
------------------------------------------------------------------------------
Line Number Statements:
Extended opcode 2: set Address to 0x45c880
Advance Line by 25 to 26
Copy
------------------------------------------------------------------------------
as the only part of .debug_line which GDB needs for DW_TAG_partial_unit is:
------------------------------------------------------------------------------
The Directory Table:
../../gdb
/usr/include/bits
[...]
The File Name Table:
Entry Dir Time Size Name
1 1 0 0 gdb.c
2 2 0 0 string3.h
[...]
------------------------------------------------------------------------------
specifically referenced in GDB for DW_AT_decl_file at a single place:
------------------------------------------------------------------------------
fe = &cu->line_header->file_names[file_index - 1];
SYMBOL_SYMTAB (sym) = fe->symtab;
------------------------------------------------------------------------------
This is because for some reason DW_TAG_partial_unit never contains PC-related
DWARF information. I do not know exactly why, the compression ratio is a bit
lower due to it but thanksfully currently it is that way:
dwz.c:
------------------------------------------------------------------------------
/* These attributes reference code, prevent moving
DIEs with them. */
case DW_AT_low_pc:
case DW_AT_high_pc:
case DW_AT_entry_pc:
case DW_AT_ranges:
die->die_ck_state = CK_BAD;
+
/* State of checksum computation. Not computed yet, computed and
suitable for moving into partial units, currently being computed
and finally determined unsuitable for moving into partial units. */
enum { CK_UNKNOWN, CK_KNOWN, CK_BEING_COMPUTED, CK_BAD } die_ck_state : 2;
------------------------------------------------------------------------------
I have also verified also real-world Fedora debuginfo files really comply with
that assumption with dwgrep
https://github.com/pmachata/dwgrep
using:
------------------------------------------------------------------------------
dwgrep -e 'entry ?DW_TAG_partial_unit child* ( ?DW_AT_low_pc , ?DW_AT_high_pc , ?DW_AT_ranges )' /usr/lib/debug/**
------------------------------------------------------------------------------
BTW I think GDB already does not support the whole DW_TAG_imported_unit and
DW_TAG_partial_unit usage possibilities as specified by the DWARF standard.
I think GDB would not work if DW_TAG_imported_unit was used in some inner
level and not at the CU level (readelf -wi level <1>) - this is how DWZ is
using DW_TAG_imported_unit. Therefore I do not think further assumptions
about DW_TAG_imported_unit and DW_TAG_partial_unit usage by DWZ are a problem
for GDB.
One could save the whole .debug_line decoded PC<->lines mapping (and not just
the DW_AT_decl_file table) but:
* there are some problematic corner cases so one could do it incorrectly
* there are no real world data to really test such patch extension
* such extension could be done perfectly incrementally on top of this patch
benchmark - on Fedora 20 x86_64 and FSF GDB HEAD:
echo -e 'thread apply all bt\nset confirm no\nq'|./gdb -p `pidof soffice.bin` -ex 'set pagination off' -ex 'maintenance set per-command
space' -ex 'maintenance set per-command symtab' -ex 'maintenance set per-command time'
FSF GDB HEAD ("thread apply all bt"):
Command execution time: 333.693000 (cpu), 335.587539 (wall)
---sec
Space used: 1736404992 (+1477189632 for this command)
----MB
vs.
THIS PATCH ("thread apply all bt"):
Command execution time: 2.595000 (cpu), 2.607573 (wall)
-sec
Space used: 340058112 (+85917696 for this command)
--MB
FSF GDB HEAD ("thread apply all bt full"):
Command execution time: 466.751000 (cpu), 468.345837 (wall)
---sec
Space used: 2330132480 (+2070974464 for this command)
----MB
vs.
THIS PATCH ("thread apply all bt full"):
Command execution time: 18.907000 (cpu), 18.964125 (wall)
--sec
Space used: 364462080 (+110325760 for this command)
---MB
gdb/ChangeLog
2015-01-25 Jan Kratochvil <jan.kratochvil@redhat.com>
Fix 100x slowdown regression on DWZ files.
* dwarf2read.c (struct dwarf2_per_objfile): Add line_header_hash.
(struct line_header): Add offset and offset_in_dwz.
(dwarf_decode_lines): Add parameter decode_mapping to the declaration.
(free_line_header_voidp): New declaration.
(line_header_hash, line_header_hash_voidp, line_header_eq_voidp): New
functions.
(dwarf2_build_include_psymtabs): Update dwarf_decode_lines caller.
(handle_DW_AT_stmt_list): Use line_header_hash.
(free_line_header_voidp): New function.
(dwarf_decode_line_header): Initialize offset and offset_in_dwz.
(dwarf_decode_lines): New parameter decode_mapping, use it.
(dwarf2_free_objfile): Free line_header_hash.
Pedro Alves [Fri, 23 Jan 2015 11:21:56 +0000 (11:21 +0000)]
Linux: make target_is_async_p return false when async is off
linux_nat_is_async_p currently always returns true, even when the
target is _not_ async. That confuses
gdb_readline_wrapper/gdb_readline_wrapper_cleanup, which
force-disables target-async while the secondary prompt is active. As
a result, when gdb_readline_wrapper returns, the target is left async,
even through it was sync to begin with.
That can result in weird bugs, like the one the test added by this
commit exposes.
gdb/ChangeLog:
2015-01-23 Pedro Alves <palves@redhat.com>
* linux-nat.c (linux_is_async_p): New macro.
(linux_nat_is_async_p):
(linux_nat_terminal_inferior): Check whether the target can async
instead of whether it is already async.
(linux_nat_terminal_ours): Don't check whether the target is
async.
(linux_async_pipe): Use linux_is_async_p.
gdb/testsuite/ChangeLog:
2015-01-23 Pedro Alves <palves@redhat.com>
* gdb.threads/continue-pending-after-query.c: New file.
* gdb.threads/continue-pending-after-query.exp: New file.
Jan Kratochvil [Fri, 23 Jan 2015 09:52:55 +0000 (10:52 +0100)]
compile: Fix function pointers
TBH while I always comment reasons for each of the compilation options in
reality I tried them all and chose that combination that needs the most simple
compile/compile-object-load.c (ld.so emulation) implementation.
gdb/ChangeLog
2015-01-23 Jan Kratochvil <jan.kratochvil@redhat.com>
* compile/compile.c (_initialize_compile): Use -fPIE for compile_args.
gdb/testsuite/ChangeLog
2015-01-23 Jan Kratochvil <jan.kratochvil@redhat.com>
* gdb.compile/compile.exp (pointer to jit function): New test.
Pedro Alves [Thu, 22 Jan 2015 18:41:33 +0000 (18:41 +0000)]
mingw32: fix windows-termcap/curses check
When GDB is configured with "--without-tui --with-curses" or "--with-tui",
$prefer_curses is set to yes. But, that still doesn't mean that curses
will be used. configure will still search for the curses library, and
continue building without it. That's done here:
curses_found=no
if test x"$prefer_curses" = xyes; then
...
AC_SEARCH_LIBS(waddstr, [ncurses cursesX curses])
if test "$ac_cv_search_waddstr" != no; then
curses_found=yes
fi
fi
So if waddstr is not found, meaning curses is not really
available, even though it'd be preferred, $prefer_curses is
'yes', but $curses_found is 'no'.
So the right fix to tell whether we're linking with curses is
$curses_found=yes.
gdb/ChangeLog:
2015-01-22 Pedro Alves <palves@redhat.com>
* configure.ac [*mingw32*]: Check $curses_found instead of
$prefer_curses.
* configure: Regenerate.
* windows-termcap.c: Remove HAVE_CURSES_H, HAVE_NCURSES_H and
HAVE_NCURSES_NCURSES_H checks.
Eli Zaretskii [Thu, 22 Jan 2015 18:23:12 +0000 (20:23 +0200)]
Fix MinGW TUI build
gdb/
2015-01-22 Eli Zaretskii <eliz@gnu.org>
* gdb/tui/tui.c (tui_enable) [__MINGW32__]: If the call to 'newterm'
fails with the 1st arg NULL, try again with "unknown". Don't test
the "cup" capability: it isn't supported by the Windows port of
ncurses, but the Windows console driver is still capable of
supporting TUI.
Eli Zaretskii [Fri, 16 Jan 2015 16:24:16 +0000 (18:24 +0200)]
Make setting TUI border attributes take effect immediately
gdb/
2015-01-16 Eli Zaretskii <eliz@gnu.org>
* tui/tui-win.c (tui_rehighlight_all, tui_set_var_cmd): New
functions.
(_initialize_tui_win) <border-kind, border-mode>:
<active-border-mode>: Use tui_set_var_cmd as the "set" function.
* tui/tui-win.h: Add prototype for tui_rehighlight_all.
Eli Zaretskii [Fri, 16 Jan 2015 15:54:46 +0000 (17:54 +0200)]
Make the change of tab size in TUI mode effective immediately
gdb/ChangeLog:
2015-01-16 Eli Zaretskii <eliz@gnu.org>
* tui/tui-win.c (tui_set_tab_width_command): Delete and
recreate the source and the disassembly windows, to show the
effect of the changed tab size immediately.
Eli Zaretskii [Fri, 16 Jan 2015 11:31:36 +0000 (13:31 +0200)]
Fix TUI-related documentation.
tui/tui-win.c (tui_scroll_left_command, tui_scroll_right_command):
Doc fix.
doc/gdb.texinfo (TUI Commands): Document the possible
values of NAME argument to 'winheight' command. Explain the
effect of 'tabset' setting better.
Eli Zaretskii [Fri, 16 Jan 2015 11:24:20 +0000 (13:24 +0200)]
Leave more space in TUI mode for thread ID.
gdb/tui/tui-data.h (LINE_PREFIX): Make shorter
(MAX_PID_WIDTH): Enlarge from 14 to 19, to leave enough space for
"Thread NNNNN.XXXX" thread ID notation on Windows.
Joel Brobecker [Thu, 15 Jan 2015 03:14:45 +0000 (07:14 +0400)]
Build failure in sim/rx/gdb-if.c on windows
This should fix a build failure reported on x86_64-mingw32 by Daniel
Calcoen due to conflicting declarations of "open". This patch just
renames the static global in sim/rx/gdb-if.c into "rx_sim_is_open".
sim/rx/ChangeLog:
* gdb-if.c (open): Rename to...
(rx_sim_is_open): This. Replace all uses of "open" by uses of
"rx_sim_is_open" throughout.
This pulls in some missing prototypes and also adds corresponding entries
to the ChangeLog file. Please note that for one function, strverscmp(),
the ChangeLog entry was already there, but the actual prototype wasn't.
These ChangeLog entries are added:
2014-10-28 Yury Gribov <y.gribov@samsung.com>
* libiberty.h (strtol, strtoul, strtoll, strtoull): New prototypes.
Pedro Alves [Wed, 14 Jan 2015 12:38:47 +0000 (12:38 +0000)]
PR17525 - breakpoint commands not executed when program run from -x script
Executing a gdb script that runs the inferior (from the command line
with -x), and has it hit breakpoints with breakpoint commands that
themselves run the target, is currently broken on async targets
(Linux, remote).
While we're executing a command list or a script, we force the
interpreter to be sync, which results in some functions nesting an
event loop and waiting for the target to stop, instead of returning
immediately and having the top level event loop handle the stop.
The issue with this bug is simply that bpstat_do_actions misses
checking whether the interpreter is sync. When we get here, in the
case of executing a script (or, when the interpreter is sync), the
program has already advanced to the next breakpoint, through
maybe_wait_sync_command_done. We need to process its breakpoints
immediately, just like with a sync target.
Tested on x86_64 Fedora 20.
gdb/
2015-01-14 Pedro Alves <palves@redhat.com>
PR gdb/17525
* breakpoint.c: Include "interps.h".
(bpstat_do_actions_1): Also check whether the interpreter is
async.
gdb/testsuite/
2015-01-14 Pedro Alves <palves@redhat.com>
Joel Brobecker <brobecker@adacore.com>
PR gdb/17525
* gdb.base/bp-cmds-execution-x-script.c: New file.
* gdb.base/bp-cmds-execution-x-script.exp: New file.
* gdb.base/bp-cmds-execution-x-script.gdb: New file.
Pedro Alves [Wed, 14 Jan 2015 12:38:47 +0000 (12:38 +0000)]
PR cli/17828: -batch -ex r breaks terminal
Commit d3d4baed (PR python/17372 - Python hangs when displaying
help()) had the side effect of causing 'gdb -batch' to leave the
terminal in the wrong state if the program was run. E.g,.
$ echo 'main(){*(int*)0=0;}' | gcc -x c -; ./gdb/gdb -batch -ex r ./a.out
Program received signal SIGSEGV, Segmentation fault.
0x00000000004004ff in main ()
$
If you start typing the next command, seemingly nothing happens - GDB
left the terminal with echo disabled.
The issue is that that "r" ends up in fetch_inferior_event, which
calls reinstall_readline_callback_handler_cleanup, which causes
readline to prep the terminal (raw, echo disabled). But "-batch"
causes GDB to exit before the top level event loop is first started,
and then nothing de-preps the terminal.
The reinstall_readline_callback_handler_cleanup function's intro
comment mentions:
"Need to do this as we go back to the event loop, ready to process
further input."
but the implementation forgets the case of when the interpreter is
sync, which indicates we won't return to the event loop yet, or as in
the case of -batch, we have not started it yet.
The fix is to not install the readline callback in that case.
For the test, in this case, checking that command echo still works is
sufficient. Comparing stty output before/after running GDB is even
better. Because stty may not be available, the test tries both ways.
In any case, since expect's spawn (what we use to start gdb) creates a
new pseudo tty, another expect spawn or tcl exec after GDB exits would
not see the wrong terminal settings. So instead, the test spawns a
shell and runs stty and GDB in it.
Tested on x86_64 Fedora 20.
gdb/
2015-01-14 Pedro Alves <palves@redhat.com>
PR cli/17828
* infrun.c (reinstall_readline_callback_handler_cleanup): Don't
reinstall if the interpreter is sync.
gdb/testsuite/
2015-01-14 Pedro Alves <palves@redhat.com>
PR cli/17828
* gdb.base/batch-preserve-term-settings.c: New file.
* gdb.base/batch-preserve-term-settings.exp: New file.
Joel Brobecker [Tue, 13 Jan 2015 14:59:32 +0000 (18:59 +0400)]
gdb/testsuite: Make clean mostlyclean should not delete *.py.
A sanity-check in my release scripts caught something: After having
created the tarballs, I verify that no checked-in file disappeared
in the process, and lo and behod, it found that the following file
got wiped:
- gdb/testsuite/dg-extract-results.py:
And it's not part of the tarball either.
I don't understand while we delete all *.py files in gdb/testsuite,
since I don't see a rule that expected to create one. A run of the
testsuite also doesn't seem to be creating .py files there.
I traced this to the following commit, which unfortunately provided
no explanation. Perhaps we used to run some tests in the gdb/testsuite
directory and caused files to be left behind there. Perhaps we still
do today?
In the meantime, Executive Decision: In order to allow me to create
tarballs without losing files, I removed it. It's easy to put something
back if we find out why it might still be needed.
gdb/testsuite/ChangeLog:
* Makefile.in (clean mostlyclean): Do not delete *.py.
Tested on x86_64-linux by running the src-release.sh script again,
and this time, dg-extract-results.py no longer gets wiped.
% gdb
(gdb) set lang ada
(gdb) python print gdb.lookup_type('character')
Traceback (most recent call last):
File "<string>", line 1, in <module>
gdb.error: No type named character.
Error while executing Python code.
This is because the language_lookup_primitive_type_as_symbol call
was moved to the la_lookup_symbol_nonlocal hook. A couple of
implementations have been upated accordingly, but the Ada version
has not. This patch fixes this omission.
gdb/ChangeLog:
* ada-lang.c (ada_lookup_symbol_nonlocal): If name not found
in static block, then try searching for primitive types.
Patrick Palka [Mon, 12 Jan 2015 22:51:33 +0000 (17:51 -0500)]
Append to input history file instead of overwriting it
This patch makes readline append new history lines to the GDB history
file on exit instead of overwriting the entire history file on exit.
This change allows us to run multiple simultaneous GDB sessions without
having each session overwrite the added history of each other session on
exit.
Care must be taken to ensure that the history file doesn't get corrupted
when multiple GDB processes are trying to simultaneously append to and
then truncate it. Safety is achieved in such a situation by using an
intermediate local history file to mutually exclude multiple processes
from simultaneously performing write operations on the global history
file.
gdb/ChangeLog:
* top.h (gdb_add_history): Declare.
* top.c (command_count): New variable.
(gdb_add_history): New function.
(gdb_safe_append_history): New static function.
(quit_force): Call it.
(command_line_input): Use gdb_add_history instead of
add_history.
* event-top.c (command_line_handler): Likewise.
James Clarke [Mon, 12 Jan 2015 17:13:54 +0000 (21:13 +0400)]
[darwin/gdb] Use <setjmp.h> instead of <machine/setjmp.h>
The `machine/setjmp.h' header is no longer present on OS X 10.10, and is
non-standard. Instead, `darwin-nat.c' should be using the standard
`setjmp.h' header.
gdb/ChangeLog:
2015-01-12 James Clarke <jrtc27@jrtc27.com> (tiny patch)
PR gdb/17046
* darwin-nat.c: Replace <machine/setjmp.h> #include by
<setjmp.h> #include.
Pedro Alves [Mon, 12 Jan 2015 17:10:06 +0000 (17:10 +0000)]
gdb.python/py-prompt.exp: restore GDBFLAGS
The previous change to py-prompt.exp made it return without restoring
GDBFLAGS, resulting in breaking the following tests:
$ make check RUNTESTFLAGS="--target_board=native-gdbserver --directory=gdb.python"
...
Running src/gdb/testsuite/gdb.python/py-prompt.exp ...
Running src/gdb/testsuite/gdb.python/py-section-script.exp ...
ERROR: (timeout) GDB never initialized after 10 seconds.
ERROR: no fileid for gdbuild
ERROR: Couldn't send python print ('test') to GDB.
ERROR: no fileid for gdbuild
ERROR: Couldn't send python print (sys.version_info[0]) to GDB.
ERROR: no fileid for gdbuild
ERROR: Couldn't send python print (sys.version_info[1]) to GDB.
ERROR: no fileid for gdbuild
ERROR: no fileid for gdbuild
...
gdb/testsuite/
2015-01-12 Pedro Alves <palves@redhat.com>
* gdb.python/py-prompt.exp: When the board can't spawn for attach,
restore GDBFLAGS before returning.
Nick Clifton [Mon, 12 Jan 2015 16:08:41 +0000 (16:08 +0000)]
More fixes for memory access errors when running readelf on fuzzed binaries.
PR binutils/17531
* dwarf.c (process_debug_info): Check for abbrev_base being larger
than the section size.
(process_cu_tu_index): Use xcalloc2 to allocate the CU and TU
arrays.
(xcalloc2): New function. Like xcalloc, but checks for overflow.
* dwarf.h (xcalloc2): Prototype.
Jan Beulich [Mon, 12 Jan 2015 14:24:20 +0000 (15:24 +0100)]
gas: allow labeling of CFI instructions
When runtime patching code (like e.g. done by the Linux kernel) there
may be cases where the set of stack frame alterations differs between
unpatched and patched code. Consequently the corresponding unwind data
needs patching too. Locating the right places within an FDE, however,
is rather cumbersome without a way to insert labels in the resulting
section. Hence this patch introduces a new directive, .cfi_label. Note
that with the way CFI data gets emitted currently (at the end of the
assembly process) this can't support local FB- and dollar-labels.
gas/
2015-01-12 Jan Beulich <jbeulich@suse.com>
* gas/dw2gencfi.c (cfi_add_label, dot_cfi_label): New.
(cfi_pseudo_table): Add "cfi_label".
(output_cfi_insn): Handle CFI_label.
(select_cie_for_fde): Als terminate CIE when encountering
CFI_label.
* dw2gencfi.h (cfi_add_label): Declare.
(struct cfi_insn_data): New member "sym_name".
(CFI_label): New.
* read.c (read_symbol_name): Drop "static".
* read.h (read_symbol_name): Declare.
gas/testsuite/
2015-01-12 Jan Beulich <jbeulich@suse.com>
gas/cfi/cfi-label.d, gas/cfi/cfi-label.s: New.
gas/cfi/cfi.exp: Run new tests.
Alan Modra [Mon, 12 Jan 2015 09:54:12 +0000 (20:24 +1030)]
Fix build without makeinfo from release binutils tar
PR 17817
* Makefile.am (aoutx.stamp): cp -p $srcdir/aoutx.texi to keep
timestamps so that makeinfo need not be installed.
(archive.stamp, archures.stamp, bfdt.stamp, cache.stamp,
coffcode.stamp, core.stamp, elf.stamp, elfcode.stamp, mmo.stamp,
format.stamp, libbfd.stamp, bfdio.stamp, bfdwin.stamp,
opncls.stamp, reloc.stamp, section.stamp, syms.stamp, targets.stamp,
init.stamp, hash.stamp, linker.stamp): Similarly.
(bfdver.texi): Use test rather than [ ] in commands.
* Makefile.in: Regenerate.
Jan Kratochvil [Mon, 12 Jan 2015 10:02:46 +0000 (11:02 +0100)]
[testsuite patch] Fix new FAIL: py-frame.exp: test Frame.read_register(rip)
for x86_64 -m32 run one gets:
+FAIL: gdb.python/py-frame.exp: test Frame.read_register(rip)
I do not have x32 OS here but the %rip test should PASS there I think.
On Sun, 11 Jan 2015 14:58:06 +0100, Yao Qi wrote:
With your patch applied, this test is skipped on 'x86_64 -m32'. I
prefer to increasing the test coverage, so how about extending the test
for 'x86_64 -m32'? I mean test Frame.read_register(eip)...
gdb/testsuite/ChangeLog
2015-01-12 Jan Kratochvil <jan.kratochvil@redhat.com>
* gdb.python/py-frame.exp (test Frame.read_register(rip)): Use
is_amd64_regs_target and is_x86_like_target.
Doug Evans [Sun, 11 Jan 2015 23:45:43 +0000 (15:45 -0800)]
Require numeric attributes to specify the form.
gdb/testsuite/ChangeLog:
* lib/dwarf.exp (Dwarf): Flag an error if a numeric attribute value
is given without an explicit form.
* gdb.dwarf2/arr-subrange.exp: Specify forms for all numeric
attributes.
* gdb.dwarf/corrupt.exp: Ditto.
* gdb.dwarf2/enum-type.exp: Ditto.
* gdb.trace/entry-values.exp: Ditto.
* gdb.trace/unavailable-dwarf-piece.exp: Ditto.
Doug Evans [Sun, 11 Jan 2015 23:16:26 +0000 (15:16 -0800)]
Temporarily revert symbol lookup cache.
clear_symtab_users calls breakpoint_re_set before
observer_notify_new_objfile(NULL), and thus symbol lookup
done during breakpoint_re_set will see a stale cache.
Presumably we just need to move the call to observer_notify_new_objfile(NULL)
to before breakpoint_re_set, but need to check for other such issues,
and 7.9 is scheduled to branch tomorrow.
* symtab.c (eq_symbol_entry): Use SYMBOL_SEARCH_NAME and
symbol_matches_domain for symbol comparisons.
* symtab.c (symbol_cache_mark_found): Improve function comment.
Rename parameter objfile to objfile_context.
(symbol_cache_mark_not_found): Improve function comment.
Add symbol lookup cache.
* NEWS: Document new options and commands.
* symtab.c (symbol_cache_key): New static global.
(DEFAULT_SYMBOL_CACHE_SIZE, MAX_SYMBOL_CACHE_SIZE): New macros.
(SYMBOL_LOOKUP_FAILED): New macro.
(symbol_cache_slot_state): New enum.
(block_symbol_cache): New struct.
(symbol_cache): New struct.
(new_symbol_cache_size, symbol_cache_size): New static globals.
(hash_symbol_entry, eq_symbol_entry): New functions.
(symbol_cache_byte_size, resize_symbol_cache): New functions.
(make_symbol_cache, free_symbol_cache): New functions.
(get_symbol_cache, symbol_cache_cleanup): New function.
(set_symbol_cache_size, set_symbol_cache_size_handler): New functions.
(symbol_cache_lookup, symbol_cache_clear_slot): New function.
(symbol_cache_mark_found, symbol_cache_mark_not_found): New functions.
(symbol_cache_flush, symbol_cache_dump): New functions.
(maintenance_print_symbol_cache): New function.
(maintenance_flush_symbol_cache): New function.
(symbol_cache_stats): New function.
(maintenance_print_symbol_cache_statistics): New function.
(symtab_new_objfile_observer): New function.
(symtab_free_objfile_observer): New function.
(lookup_static_symbol, lookup_global_symbol): Use symbol cache.
(_initialize_symtab): Init symbol_cache_key. New parameter
maint symbol-cache-size. New maint commands print symbol-cache,
print symbol-cache-statistics, flush-symbol-cache.
Install new_objfile, free_objfile observers.
gdb/doc/ChangeLog:
* gdb.texinfo (Symbols): Document new commands
"maint print symbol-cache", "maint print symbol-cache-statistics",
"maint flush-symbol-cache". Document new option
"maint set symbol-cache-size".
Doug Evans [Sun, 11 Jan 2015 22:06:34 +0000 (14:06 -0800)]
PR gdb/15830
gdb/ChangeLog:
PR gdb/15830
* NEWS: The "maint demangle" command is renamed as "demangle".
* demangle.c: #include cli/cli-utils.h, language.h.
(demangle_command): New function.
(_initialize_demangle): Add new command "demangle".
* maint.c (maintenance_demangle): Stub out.
(_initialize_maint_cmds): Update help text for "maint demangle",
and mark as deprecated.
gdb/doc/ChangeLog:
* gdb.texinfo (Debugging C Plus Plus): Mention "demangle".
(Symbols): Ditto.
(Maintenance Commands): Delete docs for "maint demangle".
gdb/testsuite/ChangeLog:
* gdb.base/maint.exp: Remove references to "maint demangle".
* gdb.cp/demangle.exp: Update. "maint demangle" -> "demangle".
Add tests for explicitly specifying language to demangle.
* gdb.dlang/demangle.exp: Ditto.
H.J. Lu [Sun, 11 Jan 2015 16:04:27 +0000 (08:04 -0800)]
Only discard space for pc-relative relocs symbols
When building PIE, we should only discard space for pc-relative relocs
symbols which turn out to need copy relocs.
bfd/
PR ld/17827
* elf64-x86-64.c (elf_x86_64_allocate_dynrelocs): For PIE,
only discard space for pc-relative relocs symbols which turn
out to need copy relocs.
Doug Evans [Sun, 11 Jan 2015 06:27:10 +0000 (22:27 -0800)]
Add symbol lookup cache.
gdb/ChangeLog:
Add symbol lookup cache.
* NEWS: Document new options and commands.
* symtab.c (symbol_cache_key): New static global.
(DEFAULT_SYMBOL_CACHE_SIZE, MAX_SYMBOL_CACHE_SIZE): New macros.
(SYMBOL_LOOKUP_FAILED): New macro.
(symbol_cache_slot_state): New enum.
(block_symbol_cache): New struct.
(symbol_cache): New struct.
(new_symbol_cache_size, symbol_cache_size): New static globals.
(hash_symbol_entry, eq_symbol_entry): New functions.
(symbol_cache_byte_size, resize_symbol_cache): New functions.
(make_symbol_cache, free_symbol_cache): New functions.
(get_symbol_cache, symbol_cache_cleanup): New function.
(set_symbol_cache_size, set_symbol_cache_size_handler): New functions.
(symbol_cache_lookup, symbol_cache_clear_slot): New function.
(symbol_cache_mark_found, symbol_cache_mark_not_found): New functions.
(symbol_cache_flush, symbol_cache_dump): New functions.
(maintenance_print_symbol_cache): New function.
(maintenance_flush_symbol_cache): New function.
(symbol_cache_stats): New function.
(maintenance_print_symbol_cache_statistics): New function.
(symtab_new_objfile_observer): New function.
(symtab_free_objfile_observer): New function.
(lookup_static_symbol, lookup_global_symbol): Use symbol cache.
(_initialize_symtab): Init symbol_cache_key. New parameter
maint symbol-cache-size. New maint commands print symbol-cache,
print symbol-cache-statistics, flush-symbol-cache.
Install new_objfile, free_objfile observers.
doc/ChangeLog:
* gdb.texinfo (Symbols): Document new commands
"maint print symbol-cache", "maint print symbol-cache-statistics",
"maint flush-symbol-cache". Document new option
"maint set symbol-cache-size".
Joel Brobecker [Sat, 10 Jan 2015 05:50:44 +0000 (09:50 +0400)]
Fix use of wrong struct i387_xsave field in i387_cache_to_xsave
gdb/gdbserver/ChangeLog:
* i387-fp.c (i387_cache_to_xsave): In look over
num_avx512_zmmh_high_registers, replace use of struct i387_xsave
zmmh_low_space field by use of zmmh_high_space.
Tested on x86_64-linux, using boards/native-gdbserver.exp.
Andrew Burgess [Sun, 4 Jan 2015 00:03:16 +0000 (00:03 +0000)]
gas/avr: Prevent incorrect overflow errors for diff fixups.
When fixups are converted to a difference type within md_apply_fix, we
previously left the contents of VALP (the value that was initially
computed within write.c:fixup_segment) unchanged. This is harmless,
except that this value is used within write.c:fixup_segment once we
return from md_apply_fix to perform an overflow check.
In some cases, the value computed in write.c:fixup_segment is so wrong
that an overflow error can be triggered. These errors are incorrect.
This patch avoids the overflow errors by adjusting the value in
write.c:fixup_segment using the VALP pointer in md_apply_fix.
A test for this issue is included.
gas/ChangeLog:
* config/tc-avr.c (md_apply_fix): Update the contents of VALP for
diff fixups.
gas/testsuite/ChangeLog:
* gas/avr/large-debug-line-table.d: New file.
* gas/avr/large-debug-line-table.s: New file.
Cary Coutant [Fri, 9 Jan 2015 23:55:50 +0000 (15:55 -0800)]
Don't align start of segment unless alignment is larger than page size.
This fixes an issue where a page-aligned data section, combined with -z relro,
could lead to a gap between text and data segments larger than a page, and
we would fail to overlap the segments in the file.
gold/
* layout.cc (Layout::set_segment_offsets): Don't align start of segment
unless alignment is larger than page size.
Patrick Palka [Sun, 30 Nov 2014 16:47:16 +0000 (11:47 -0500)]
Don't munge yacc's #line directives
The #line directives within GDB's autogenerated yacc files (e.g.
c-exp.c) are being incorrectly munged, causing these directives to refer
to nonexistent source files, e.g.
The munging happens due to a sed expression added by commit 954d8cae
whose intended purpose[1] was to work around the fact that ylwrap emitted #line
directives without any directory information, e.g.
#line 36 "c-exp.y"
So the sed expression was meant to munge such directives to refer to
absolute paths instead. But the behavior of ylwrap was changed some
years ago[2] to emit absolute paths within #line directives. And when
our local copy of ylwrap was synced by commit e30465112, the sed
expression in question became unnecessary, and indeed harmful.
This patch removes the now-obsolete sed expression. The emitted #line
directives are now correct without it.
Patrick Palka [Fri, 9 Jan 2015 18:27:56 +0000 (13:27 -0500)]
Consolidate the custom TUI query hook with the default query hook
This patch primarily rewrites defaulted_query() to use
gdb_readline_wrapper() to prompt the user for input, like
prompt_for_continue() does. The motivation for this rewrite is to be
able to reuse the default query hook in TUI, obviating the need for a
custom TUI query hook.
However, having TUI use the default query mechanism exposed a couple of
latent bugs in tui_redisplay_readline() related to the handling of
multi-line prompts, in particular GDB's multi-line quit prompt.
The first issue is an off-by-one error in the calculation of the height
of the prompt. The check in question should be col <= prev_col, not c <
prev_col, to properly account for the case when a prompt contains
multiple consecutive newlines. Failing to do so makes TUI have the
wrong idea of the vertical height of the prompt. This patch fixes the
column check.
The second issue is that cur_line does not get updated to reflect the
cursor position if the user's prompt cursor is at the end of the prompt
(i.e. if rl_point == rl_end). cur_line only gets updated if rl_point
lies between 0..rl_end-1 because that is the bounds of the for loop
responsible for updating cur_line. This patch changes the loop's bounds
to 0..rl_end so that cur_line always gets updated.
With these two bug fixes out of the way, the default query mechanism
works well in TUI even with multi-line prompts like GDB's quit prompt.
gdb/ChangeLog:
* utils.c (defaulted_query): Rewrite to use gdb_readline_wrapper
to prompt for input.
* tui/tui-hooks.c (tui_query_hook): Remove.
(tui_install_hooks): Don't set deprecated_query_hook.
* tui/tui-io.c (tui_redisplay_readline): Fix off-by-one error in
height calculation. Always update the command window's cur_line.
Pedro Alves [Mon, 29 Dec 2014 19:41:07 +0000 (19:41 +0000)]
add non-stop test that stresses thread starvation issues
This commit adds a non-stop mode test originally inspired by
signal-while-stepping-over-bp-other-thread.exp, that exposes the
thread starvation issues fixed by the previous patches. It sets a set
of threads stepping in parallel, and has one of them get a signal.
Without the previous fixes, this would fail with timeouts.
gdb/testsuite/
2015-01-09 Pedro Alves <palves@redhat.com>
* gdb.threads/non-stop-fair-events.c: New file.
* gdb.threads/non-stop-fair-events.exp: New file.
Pedro Alves [Mon, 29 Dec 2014 19:41:07 +0000 (19:41 +0000)]
[gdbserver] linux-low.c: better starvation avoidance, handle non-stop mode too
This patch applies the same starvation avoidance improvements of the
previous patch to the Linux gdbserver side.
Without this, the test added by the following commit
(gdb.threads/non-stop-fair-events.exp) always fails with time outs.
gdb/gdbserver/
2015-01-09 Pedro Alves <palves@redhat.com>
* linux-low.c (step_over_bkpt): Move higher up in the file.
(handle_extended_wait): Don't store the stop_pc here.
(get_stop_pc): Adjust comments and rename to ...
(check_stopped_by_breakpoint): ... this. Record whether the LWP
stopped for a software breakpoint or hardware breakpoint.
(thread_still_has_status_pending_p): New function.
(status_pending_p_callback): Use
thread_still_has_status_pending_p. If the event is no longer
interesting, resume the LWP.
(handle_tracepoints): Add assert.
(maybe_move_out_of_jump_pad): Remove cancel_breakpoints call.
(wstatus_maybe_breakpoint): New function.
(cancel_breakpoint): Delete function.
(check_stopped_by_watchpoint): New function, factored out from
linux_low_filter_event.
(lp_status_maybe_breakpoint): Delete function.
(linux_low_filter_event): Remove filter_ptid argument.
Leave thread group exits pending here. Store the LWP's stop PC.
Always leave events pending.
(linux_wait_for_event_filtered): Pull all events out of the
kernel, and leave them all pending.
(count_events_callback, select_event_lwp_callback): Consider all
events.
(cancel_breakpoints_callback, linux_cancel_breakpoints): Delete.
(select_event_lwp): Only give preference to the stepping LWP in
all-stop mode. Adjust comments.
(ignore_event): New function.
(linux_wait_1): Delete 'retry' label. Use ignore_event. Remove
references to cancel_breakpoints. Adjust to renames. Also give
equal priority to all LWPs that have had events in non-stop mode.
If reporting a software breakpoint event, unadjust the LWP's PC.
(linux_wait): If linux_wait_1 returned an ignored event, retry.
(stuck_in_jump_pad_callback, move_out_of_jump_pad_callback):
Adjust.
(linux_resume_one_lwp): Store the LWP's PC. Adjust.
(resume_status_pending_p): Use thread_still_has_status_pending_p.
(linux_stopped_by_watchpoint): Adjust.
(linux_target_ops): Remove reference to linux_cancel_breakpoints.
* linux-low.h (enum lwp_stop_reason): New.
(struct lwp_info) <stop_pc>: Adjust comment.
<stopped_by_watchpoint>: Delete field.
<stop_reason>: New field.
* linux-x86-low.c (x86_linux_prepare_to_resume): Adjust.
* mem-break.c (software_breakpoint_inserted_here)
(hardware_breakpoint_inserted_here): New function.
* mem-break.h (software_breakpoint_inserted_here)
(hardware_breakpoint_inserted_here): Declare.
* target.h (struct target_ops) <cancel_breakpoints>: Remove field.
(cancel_breakpoints): Delete.
* tracepoint.c (clear_installed_tracepoints, stop_tracing)
(upload_fast_traceframes): Remove references to
cancel_breakpoints.
Pedro Alves [Wed, 7 Jan 2015 12:48:32 +0000 (12:48 +0000)]
linux-nat.c: better starvation avoidance, handle non-stop mode too
Running the testsuite with a series that reimplements user-visible
all-stop behavior on top of a target running in non-stop mode revealed
problems related to event starvation avoidance.
For example, I see
gdb.threads/signal-while-stepping-over-bp-other-thread.exp failing.
What happens is that GDB core never gets to see the signal event. It
ends up processing the events for the same threads over an over,
because Linux's waitpid(-1, ...) returns that first task in the task
list that has an event, starving threads on the tail of the task list.
So I wrote a non-stop mode test originally inspired by
signal-while-stepping-over-bp-other-thread.exp, to stress this
independently of all-stop on top of non-stop. Fixing it required the
changes described below. The test will be added in a following
commit.
1) linux-nat.c has code in place that picks an event LWP at random out
of all that have had events. This is because on the kernel side,
"waitpid(-1, ...)" just walks the task list linearly looking for the
first that had an event. But, this code is currently only used in
all-stop mode. So with a multi-threaded program that has multiple
events triggering debug events in parallel, GDB ends up starving some
threads.
To make the event randomization work in non-stop mode too, the patch
makes us pull out all the already pending events on the kernel side,
with waitpid, before deciding which LWP to report to the core.
There's some code in linux_wait that takes care of leaving events
pending if they were for LWPs the caller is not interested in. The
patch moves that to linux_nat_filter_event, so that we only have one
place that leaves events pending. With that in place, conceptually,
the flow is simpler and more normalized:
#1 - walk the LWP list looking for an LWP with a pending event to report.
#2 - if no pending event, pull events out of the kernel, and store
them in the LWP structures as pending.
#3- goto #1.
2) Then, currently the event randomization code only considers SIGTRAP
(or trap-like) events. That means that if e.g., have have multiple
threads stepping in parallel that hit a breakpoint that needs stepping
over, and one gets a signal, the signal may end up never getting
processed, because GDB will always be giving priority to the SIGTRAPs.
The patch fixes this by making the randomization code consider all
kinds of pending events.
3) If multiple threads hit a breakpoint, we report one of those, and
"cancel" the others. Cancelling means decrementing the PC, and
discarding the event. If the next time the LWP is resumed the
breakpoint is still installed, the LWP should hit it again, and we'll
report the hit then. The problem I found is that this delays threads
from advancing too much, with the kernel potentially ending up
scheduling the same threads over and over, and others not advancing.
So the patch switches away from cancelling the breakpoints, and
instead remembering that the LWP had stopped for a breakpoint. If on
resume the breakpoint is still installed, we report it. If it's no
longer installed, we discard the pending event then. This is actually
how GDBserver used to handle this before d50171e4 (Teach linux
gdbserver to step-over-breakpoints), but with the difference that back
then we'd delay adjusting the PC until resuming, which made it so that
"info threads" could wrongly see threads with unadjusted PCs.
gdb/
2015-01-09 Pedro Alves <palves@redhat.com>
* breakpoint.c (hardware_breakpoint_inserted_here_p): New
function.
* breakpoint.h (hardware_breakpoint_inserted_here_p): New
declaration.
* linux-nat.c (linux_nat_status_is_event): Move higher up in file.
(linux_resume_one_lwp): Store the thread's PC. Adjust to clear
stop_reason.
(check_stopped_by_watchpoint): New function.
(save_sigtrap): Reimplement.
(linux_nat_stopped_by_watchpoint): Adjust.
(linux_nat_lp_status_is_event): Delete.
(stop_wait_callback): Only call save_sigtrap after storing the
pending status.
(status_callback): If the thread had been stopped for a breakpoint
that has since been removed, discard the event and resume the LWP.
(count_events_callback, select_event_lwp_callback): Use
lwp_status_pending_p instead of linux_nat_lp_status_is_event.
(cancel_breakpoint): Rename to ...
(check_stopped_by_breakpoint): ... this. Record whether the LWP
stopped for a software breakpoint or hardware breakpoint.
(select_event_lwp): Only give preference to the stepping LWP in
all-stop mode. Adjust comments.
(stop_and_resume_callback): Remove references to new_pending_p.
(linux_nat_filter_event): Likewise. Leave exit events of the
leader thread pending here. Handle signal short circuiting here.
Only call save_sigtrap after storing the pending waitstatus.
(linux_nat_wait_1): Remove 'retry' label. Remove references to
new_pending. Don't handle leaving events the caller is not
interested in pending here, nor handle signal short-circuiting
here. Also give equal priority to all LWPs that have had events
in non-stop mode. If reporting a software breakpoint event,
unadjust the LWP's PC.
* linux-nat.h (enum lwp_stop_reason): New.
(struct lwp_info) <stop_pc>: New field.
(struct lwp_info) <stopped_by_watchpoint>: Delete field.
(struct lwp_info) <stop_reason>: New field.
* x86-linux-nat.c (x86_linux_prepare_to_resume): Adjust.
Pedro Alves [Mon, 29 Dec 2014 19:41:06 +0000 (19:41 +0000)]
linux-nat.c: always mark execing LWP as resumed
A subsequent patch will make the Linux backend's target_wait method
pull all events out of the kernel (with waitpid) and store them as
pending status in the LWP structure if no pending status was already
available. Then, the backend goes over the pending statuses and pick
one to report to the core.
With that, the existing thread-execl.exp test exposes a bug, like:
(gdb) set scheduler-locking on
(gdb) PASS: gdb.threads/thread-execl.exp: schedlock on: set scheduler-locking on
next
FAIL: gdb.threads/thread-execl.exp: schedlock on: get to main in new image (timeout)
Recall that when the non-leader thread execs, all threads in the
process die, the execing thread changes its pid to the tgid, and then
waitpid returns an exec event to the tgid. If GDB didn't resume the
leader LWP, then GDB sees an event for an LWP that was supposedly
stopped, and thus not marked as resumed. Because the code that picks
a pending event to report to the core ignores not-resumed LWPs:
/* Return non-zero if LP has a wait status pending. */
static int
status_callback (struct lwp_info *lp, void *data)
{
/* Only report a pending wait status if we pretend that this has
indeed been resumed. */
if (!lp->resumed)
return 0;
the event ends up pending forever, thus the timeout.
gdb/
2015-01-09 Pedro Alves <palves@redhat.com>
* linux-nat.c (linux_handle_extended_wait) <PTRACE_EVENT_EXEC>:
Set the LWP's 'resumed' flag.
Pedro Alves [Wed, 7 Jan 2015 19:41:06 +0000 (19:41 +0000)]
linux-nat.c: clean up pending status checking and resuming LWPs
Whenever we resume an LWP, we must clear a few flags and flush the
LWP's register cache. We actually currently flush the register cache
of all LWPs, but that's unnecessary. This patch makes us flush the
register cache of only the LWP that is resumed. Instead of open
coding all that in many places, we use a helper function.
Likewise, we have two fields in the LWP structure where a pending
status may be recorded. Add a helper predicate that checks both and
use it throughout instead of open coding the checks.
gdb/
2015-01-09 Pedro Alves <palves@redhat.com>
* linux-nat.c (linux_resume_one_lwp): New function.
(resume_lwp): Use lwp_status_pending_p and linux_resume_one_lwp.
(linux_nat_resume): Use lwp_status_pending_p and
linux_resume_one_lwp.
(linux_handle_syscall_trap): Use linux_resume_one_lwp.
(linux_handle_extended_wait): Use linux_resume_one_lwp.
(status_callback, running_callback): Use lwp_status_pending_p.
(lwp_status_pending_p): New function.
(stop_and_resume_callback): Use lwp_status_pending_p.
(linux_nat_filter_event): Use linux_resume_one_lwp.
(linux_nat_wait_1): Always use status_callback to look for an LWP
with a pending status. Use linux_resume_one_lwp.
(resume_stopped_resumed_lwps): Use lwp_status_pending_p and
linux_resume_one_lwp.
Pedro Alves [Mon, 29 Dec 2014 19:41:05 +0000 (19:41 +0000)]
cleanup and speed up (software_)breakpoint_inserted_here_p
Factor out common code, and use the more efficient
ALL_BP_LOCATIONS_AT_ADDR.
gdb/
2015-01-09 Pedro Alves <palves@redhat.com>
* breakpoint.c (bp_location_inserted_here_p): New function,
factored out from ...
(breakpoint_inserted_here_p): ... here. Use
ALL_BP_LOCATIONS_AT_ADDR.
(software_breakpoint_inserted_here_p): Use
bp_location_inserted_here_p and ALL_BP_LOCATIONS_AT_ADDR.
Pedro Alves [Mon, 29 Dec 2014 19:41:05 +0000 (19:41 +0000)]
watch_thread_num.exp and targets with fairer event reporting
This patch fixes the watch_thread_num.exp test to work when the target
is better at making event handling be fair among threads.
I wrote patches that make GDB native and GDBserver event handling
fairer between threads. That is, if threads A and B both
simultaneously trigger some debug event, GDB will pick either A or B
at random, rather than always handling the event of A first. There's
code for that in the Linux backends (gdb and gdbserver) already, but
it can be improved, and only works in all-stop mode.
With those fixes in place, I found that the watch_thread_num.exp would
often time out. The problem is that the test only works _because_
event handling isn't as fair as intended. With the fairness fixes,
the test falls victim of PR10116 (gdb drops watchpoints on
multi-threaded apps) quite often.
To expand on the PR10116 reference, consider that stop events are
serialized to GDB core, through target_wait. Say a thread-specific
watchpoint as set on thread A. When the "right" thread and some other
"wrong" thread both trigger a watchpoint simultaneously, the target
may report the "wrong" thread's hit to GDB first (thread B). When
handling that event, GDB notices the watchpoint is for another thread,
and so shouldn't cause a user-visible stop. On resume, GDB saves the
now current value of the watched expression. Afterwards, the "right"
thread (thread A) reports its watchpoint trigger. But the watched
value hasn't changed since GDB last saved it, and so GDB doesn't
report the watchpoint hit to the user.
The way the test is written, the watchpoint is associated with the
first thread that happens to report an event. It happens that GDB is
processing events much more often for one of the threads, which
usually will be that same first thread.
Hacking the test with "set debug infrun 1", we see exactly that:
However, if the target is better at giving equal priority to all
threads, the PR10116 issue happens often, so it may take quite a while
for the right thread to be the first to report its watchpoint event
just after the memory being watched really changed, resulting in test
time outs.
Here's the number of events handled for each thread on a gdbserver run
with the event fairness patches:
Note how the number of events is much higher. The test routinely
takes over 10 seconds to finish on my machine rather than under a
second as with unpatched gdbserver, when it succeeds, but often it'll
fail with timeouts too.
So to make the test robust, this patch switches the tests to using
"awatch" instead of "watch", as access watchpoints don't care about
the watchpoint's "old value". With this, the test always finishes
quickly, and we can even bump the number of threads concurrently
writting to the shared variable, to have better assurance we're really
testing the case of the "wrong" thread triggering a watchpoint.
Here's the number of events I see for each thread on a run on my
machine, with a gdbserver patched with the event fairness series:
gdb/testsuite/
2015-01-09 Pedro Alves <palves@redhat.com>
* gdb.base/annota1.exp (thread_test): Use srcfile and binfile from
the global scope. Set a breakpoint after all threads are started
rather than stepping over two source lines. Expect the prompt.
* gdb.base/watch_thread_num.c (threads_started_barrier): New
global.
(NUM): Now 15.
(main): Use threads_started_barrier to wait for all threads to
start. Main thread no longer calls thread_function. Exit after
180 seconds.
(loop): New function.
(thread_function): Wait on threads_started_barrier barrier. Call
'loop' at each iteration.
* gdb.base/watch_thread_num.exp: Continue to breakpoint after all
threads have started, instead of hardcoding number of "next"
steps. Use an access watchpoint instead of a write watchpoint.
Pedro Alves [Mon, 29 Dec 2014 19:41:04 +0000 (19:41 +0000)]
gdb.threads/{siginfo-thread.c,watchthreads-reorder.c,ia64-sigill.c} races with GDB
These three test all spawn a few threads and then send a SIGSTOP to
their parent GDB in order to pause it while the new threads set things
up for the test. With a GDB patch that changes the inferior thread's
scheduling a bit, I sometimes see:
FAIL: gdb.threads/siginfo-threads.exp: catch signal 0 (timeout)
...
FAIL: gdb.threads/watchthreads-reorder.exp: reorder1: continue a (timeout)
...
FAIL: gdb.threads/ia64-sigill.exp: continue (timeout)
...
The issue is that the test program stops GDB before it had a chance of
processing the new thread's clone event:
(gdb) PASS: gdb.threads/siginfo-threads.exp: get pid
continue
Continuing.
Stopping GDB PID 21541.
Waiting till the threads initialize their TIDs.
FAIL: gdb.threads/siginfo-threads.exp: catch signal 0 (timeout)
On Linux (at least), new threads start stopped, and the debugger must
resume them. The fix is to make the test program wait for the new
threads to be running before stopping GDB.
gdb/testsuite/
2015-01-09 Pedro Alves <palves@redhat.com>
* gdb.threads/ia64-sigill.c (threads_started_barrier): New global.
(thread_func): Wait on barrier.
(main): Wait for all threads to start before stopping GDB.
* gdb.threads/siginfo-threads.c (threads_started_barrier): New
global.
(thread1_func, thread2_func): Wait on barrier.
(main): Wait for all threads to start before stopping GDB.
* gdb.threads/watchthreads-reorder.c (threads_started_barrier):
New global.
(thread1_func, thread2_func): Wait on barrier.
(main): Wait for all threads to start before stopping GDB.
Pedro Alves [Tue, 16 Dec 2014 16:12:25 +0000 (16:12 +0000)]
Linux: Skip thread_db thread event reporting if PTRACE_EVENT_CLONE is supported
[A test I wrote stumbled on a libthread_db issue related to thread
event breakpoints. See glibc PR17705:
[nptl_db: stale thread create/death events if debugger detaches]
https://sourceware.org/bugzilla/show_bug.cgi?id=17705
This patch avoids that whole issue by making GDB stop using thread
event breakpoints in the first place, which is good for other reasons
as well, anyway.]
Before PTRACE_EVENT_CLONE (Linux 2.6), the only way to learn about new
threads in the inferior (to attach to them) or to learn about thread
exit was to coordinate with the inferior's glibc/runtime, using
libthread_db. That works by putting a breakpoint at a magic address
which is called when a new thread is spawned, or when a thread is
about to exit. When that breakpoint is hit, all threads are stopped,
and then GDB coordinates with libthread_db to read data structures out
of the inferior to learn about what happened. Then the breakpoint is
single-stepped, and then all threads are re-resumed. This isn't very
efficient (stops all threads) and is more fragile (inferior's thread
list in memory may be corrupt; libthread_db bugs, etc.) than ideal.
When the kernel supports PTRACE_EVENT_CLONE (which we already make use
of), there's really no need to use libthread_db's event reporting
mechanism to learn about new LWPs. And if the kernel supports that,
then we learn about LWP exits through regular WIFEXITED wait statuses,
so no need for the death event breakpoint either.
GDBserver has been likewise skipping the thread_db events for a long
while:
https://sourceware.org/ml/gdb-patches/2007-10/msg00547.html
There's one user-visible difference: we'll no longer print about
threads being created and exiting while the program is running, like:
This is exactly the same behavior as when debugging against remote
targets / gdbserver. I actually think that's a good thing (and as
such have listed this in the local/remote parity wiki page a while
ago), as the printing slows down the inferior. It's also a
distraction to keep bothering the user about short-lived threads that
she won't be able to interact with anyway. Instead, the user (and
frontend) will be informed about new threads that currently exist in
the program when the program next stops:
Program received signal SIGINT, Interrupt.
[Switching to Thread 0x7ffff796b700 (LWP 7796)]
clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:81
81 testq %rax,%rax
(gdb) info threads
A couple of tests had assumptions on GDB thread numbers that no longer
hold.
Tested on x86_64 Fedora 20.
gdb/
2014-01-09 Pedro Alves <palves@redhat.com>
Skip enabling event reporting if the kernel supports
PTRACE_EVENT_CLONE.
* linux-thread-db.c: Include "nat/linux-ptrace.h".
(thread_db_use_events): New function.
(try_thread_db_load_1): Check thread_db_use_events before enabling
event reporting.
(update_thread_state): New function.
(attach_thread): Use it. Check thread_db_use_events before
enabling event reporting.
(thread_db_detach): Check thread_db_use_events before disabling
event reporting.
(find_new_threads_callback): Check thread_db_use_events before
enabling event reporting. Update the thread's state if not using
libthread_db events.
gdb/testsuite/
2014-01-09 Pedro Alves <palves@redhat.com>
* gdb.threads/fork-thread-pending.exp: Switch to the main thread
instead of to thread 2.
* gdb.threads/signal-command-multiple-signals-pending.c (main):
Add barrier around each pthread_create call instead of around all
calls.
* gdb.threads/signal-command-multiple-signals-pending.exp (test):
Set a break on thread_function and have the child threads hit it
one at at a time.
projects / thirdparty / binutils-gdb.git / log
