After building GDB
--with-python=/usr/bin/python3
and for example stripping ./gdb and running:
./gdb -data-directory data-directory/ -iex "add-auto-load-safe-path $PWD/gdb-gdb.gdb" -iex "add-auto-load-safe-path $PWD/gdb-gdb.
py" ./gdb
I get:
Make breakpoint pending on future shared library load? (y or [n]) [answered N; input not from terminal]
File "/home/jkratoch/redhat/gdb-test-python3/gdb/gdb-gdb.py", line 91
print "Warning: Cannot find enum type_flag_value type."
^
SyntaxError: Missing parentheses in call to 'print'
(top-gdb) q
gdb/ChangeLog
2016-02-22 Jan Kratochvil <jan.kratochvil@redhat.com>
* gdb-gdb.py (class TypeFlagsPrinter): Use parentheses for print.
elfNN_aarch64_size_stubs, the caller of aarch64_type_stub has redirected
the final destination of long branch veneer to plt stub if the call
should go through it.
It's redundant to do the same check and redirect again from scratch
inside aarch64_type_stub.
bfd/
* elfnn-aarch64. (aarch64_type_of_stub): Remove redundation calcuation
for destination. Remove useless function parameters.
(elfNN_aarch64_size_stubs): Update parameters for aarch64_type_of_stub.
This patch fixes the various code format issues in arm process record
in arm-tdep.c, such as using tab instead of spaces.
gdb:
2016-02-22 Yao Qi <yao.qi@linaro.org>
* arm-tdep.c: Fix code format issues.
There is no need to set BFD compression bits for each section. We
should just set it once in write_object_file.
* write.c (compress_debug): Move BFD compression bits setting
to ...
(write_object_file): Here.
Increment register number by 16 if RegVRex is set.
* config/tc-i386.c (register_number): Check RegVRex.
* testsuite/gas/i386/x86-64-avx512f.s: Add a test for vgatherqpd
with %zmm19 and %zmm3.
* testsuite/gas/i386/x86-64-avx512f-intel.d: Updated.
* testsuite/gas/i386/x86-64-avx512f.d: Likewise.
* symbols.c (decode_local_label_name): Make type a const char *.
* listing.c (print_source): Make type of p const char *.
(print_line): Make type of string const char *.
(buffer_line): Return const char *.
(title): Make type const char *.
(subtitle): Likewise.
(listing_listing): Make type of p const char *.
* messages.c (as_internal_value_out_of_range): Make type of prefix
const char *.
* stabs.c (s_stab_generic): make type of stab_secname, stabstr_secname
and string const char *.
* read.c (_bfd_rel): Make type of name const char *.
* app.c (out_string): Change type to const char *.
(struct app_save::out_string): Likewise.
gdb/testsuite/ChangeLog:
2016-02-18 Wei-cheng Wang <cole945@gmail.com>
* gdb.trace/tspeed.c (myclock): Return wallclock instead of
user+system time.
(trace_speed_test): Determine the iteration count for a time
between 15..30 seconds.
With Intel Memory Protection Extensions it was introduced the concept of
boundary violation. A boundary violations is presented to the inferior as
a segmentation fault having SIGCODE 3. This patch adds a
handler for a boundary violation extending the information displayed
when a bound violation is presented to the inferior. In the stop mode
case the debugger will also display the kind of violation: "upper" or
"lower", bounds and the address accessed.
On no stop mode the information will still remain unchanged. Additional
information about bound violations are not meaningful in that case user
does not know the line in which violation occurred as well.
When the segmentation fault handler is stop mode the out puts will be
changed as exemplified below.
The usual output of a segfault is:
Program received signal SIGSEGV, Segmentation fault
0x0000000000400d7c in upper (p=0x603010, a=0x603030, b=0x603050,
c=0x603070, d=0x603090, len=7) at i386-mpx-sigsegv.c:68
68 value = *(p + len);
In case it is a bound violation it will be presented as:
Program received signal SIGSEGV, Segmentation fault
Upper bound violation while accessing address 0x7fffffffc3b3
Bounds: [lower = 0x7fffffffc390, upper = 0x7fffffffc3a3]
0x0000000000400d7c in upper (p=0x603010, a=0x603030, b=0x603050,
c=0x603070, d=0x603090, len=7) at i386-mpx-sigsegv.c:68
68 value = *(p + len);
In mi mode the output of a segfault is:
*stopped,reason="signal-received",signal-name="SIGSEGV",
signal-meaning="Segmentation fault", frame={addr="0x0000000000400d7c",
func="upper",args=[{name="p", value="0x603010"},{name="a",value="0x603030"}
,{name="b",value="0x603050"}, {name="c",value="0x603070"},
{name="d",value="0x603090"},{name="len",value="7"}],
file="i386-mpx-sigsegv.c",fullname="i386-mpx-sigsegv.c",line="68"},
thread-id="1",stopped-threads="all",core="6"
in the case of a bound violation:
*stopped,reason="signal-received",signal-name="SIGSEGV",
signal-meaning="Segmentation fault",
sigcode-meaning="Upper bound violation",
lower-bound="0x603010",upper-bound="0x603023",bound-access="0x60302f",
frame={addr="0x0000000000400d7c",func="upper",args=[{name="p",
value="0x603010"},{name="a",value="0x603030"},{name="b",value="0x603050"},
{name="c",value="0x603070"},{name="d",value="0x603090"},
{name="len",value="7"}],file="i386-mpx-sigsegv.c",
fullname="i386-mpx-sigsegv.c",line="68"},thread-id="1",
stopped-threads="all",core="6"
2016-02-18 Walfred Tedeschi <walfred.tedeschi@intel.com>
gdb/ChangeLog:
* NEWS: Add entry for bound violation.
* amd64-linux-tdep.c (amd64_linux_init_abi_common):
Add handler for segmentation fault.
* gdbarch.sh (handle_segmentation_fault): New.
* gdbarch.c: Regenerate.
* gdbarch.h: Regenerate.
* i386-linux-tdep.c (i386_linux_handle_segmentation_fault): New.
(SIG_CODE_BONDARY_FAULT): New define.
(i386_linux_init_abi): Use i386_mpx_bound_violation_handler.
* i386-linux-tdep.h (i386_linux_handle_segmentation_fault) New.
* i386-tdep.c (i386_mpx_enabled): Add as external.
* i386-tdep.c (i386_mpx_enabled): Add as external.
* infrun.c (handle_segmentation_fault): New function.
(print_signal_received_reason): Use handle_segmentation_fault.
gdb/testsuite/ChangeLog:
* gdb.arch/i386-mpx-sigsegv.c: New file.
* gdb.arch/i386-mpx-sigsegv.exp: New file.
* gdb.arch/i386-mpx-simple_segv.c: New file.
* gdb.arch/i386-mpx-simple_segv.exp: New file.
gdb/doc/ChangeLog:
* gdb.texinfo (Signals): Add bound violation display hints for
a SIGSEGV.
When we're looking at a tracefile trace frame where registers are not
available, and the tracepoint has only one location, we supply
the location's address as the PC register. However, this only works
if PC is not a pseudo register, and individual architectures may want
to guess more registers. Add a gdbarch hook that will handle that.
gdb/ChangeLog:
* arch-utils.c (default_guess_tracepoint_registers): New function.
* arch-utils.h (default_guess_tracepoint_registers): New prototype.
* gdbarch.c: Regenerate.
* gdbarch.h: Regenerate.
* gdbarch.sh: Add guess_tracepoint_registers hook.
* tracefile.c (tracefile_fetch_registers): Use the new gdbarch hook.
Since PR ld/19617 tests require share library support, enable them
only for Linux/GNU/Solaris targets.
* testsuite/ld-elf/pr19617a.d: Enable only for *-*-linux*,
*-*-gnu* and *-*-solaris*.
* testsuite/ld-elf/pr19617b.d: Likewise.
* testsuite/ld-elf/pr19617c.d: Likewise.
This patch series add fork support in target remote,
[PATCH v2 0/3] Target remote mode fork and exec support
https://sourceware.org/ml/gdb-patches/2015-12/msg00144.html
so GDB can be informed about the child, and adjust child correctly in
displaced stepping. The PR server/13796 was fixed by this patch
series actually. Test results on buildbot show this KFAIL->KPASS
change https://sourceware.org/ml/gdb-testers/2015-q4/msg10128.html
gdb/testsuite:
2016-02-18 Yao Qi <yao.qi@linaro.org>
* gdb.base/disp-step-syscall.exp (disp_step_cross_syscall):
Don't call setup_kfail.
Proc do_test in forking-threads-plus-breakpoint.exp has an argument
cond_bp_target, but the test doesn't use it to set
"breakpoint condition-evaluation", which is an oversight in the test.
This patch fixes it by setting "breakpoint condition-evaluation" per
$cond_bp_target.
gdb/testsuite:
2016-02-18 Yao Qi <yao.qi@linaro.org>
* gdb.threads/forking-threads-plus-breakpoint.exp (do_test):
Set "set breakpoint condition-evaluation" per $cond_bp_target.
In embedded environments, including boot loaders, the non-PIC executable
needs to export its symbols to modules loaded in the future. We should
always create dynamic sections for -E/--dynamic-list.
bfd/
PR ld/19617
* elflink.c (elf_link_add_object_symbols): Always create dynamic
sections for -E/--dynamic-list.
ld/
PR ld/19617
* testsuite/ld-elf/pr19617.s: New file.
* testsuite/ld-elf/pr19617a.d: Likewise.
* testsuite/ld-elf/pr19617b.d: Likewise.
* testsuite/ld-elf/pr19617c.d: Likewise.
* Makefile.am (CXX_FOR_TARGET): Check for the presence of an
in-tree xg++ executable after checking for the presence of an
in-tree g++ executable.
* Makefile.in: Regenerate.
gas * read.c (finish_bundle): Avoid recording a negative alignment.
(do_align): Use unsigned values for n, len and max. Only create
a frag if the alignment requirement is greater than the minimum
byte alignment. Avoid recording a negative alignment.
(s_align): Use unsigned values where appropriate.
(bss_alloc): Use an unsigned value for the alignment.
(sizeof_sleb128): Add a comment noting that we encode one octet
per byte, regardless of the value of OCTETS_PER_BYTE_POWER.
(emit_leb129_expr): Abort if the emitted encoding was longer than
expected.
* read.h (output_leb128): Update prototype.
(sizeof_leb128): Update prototype.
(bss_alloc): Update prototype.
* write.c (record_alignment): Use an unsigned value for the
alignment. Do not record alignments less than the minimum
alignment for a byte.
* write.h (record_alignment): Update prototype.
Despite the documentation and the comment in xtensa_move_literals, in
the presence of --text-section-literals and --auto-litpools literals are
moved from the separate literal sections into .init and .fini, because
the check in the xtensa_move_literals is incorrect.
This moving was broken with introduction of auto litpools: some literals
now may be lost. This happens because literal frags emitted from .init
and .fini are not closed when new .literal_position marks new literal
pool. Then frag_align(2, 0, 0) changes type of the last literal frag to
rs_align. rs_align frags are skipped in the xtensa_move_literals. As a
result fixups against such literals are not moved out of .init.literal/
.fini.literal sections producing the following assembler error:
test.S: Warning: fixes not all moved from .init.literal
test.S: Internal error!
Fix check for .init.literal/.fini.literal in the xtensa_move_literals
and don't let it move literals from there in the presence of
--text-section-literals or --auto-litpools.
2016-02-17 Max Filippov <jcmvbkbc@gmail.com>
gas/
* config/tc-xtensa.c (xtensa_move_literals): Fix check for
.init.literal/.fini.literal section name.
* testsuite/gas/xtensa/all.exp: Add init-fini-literals to the
list of xtensa tests.
* testsuite/gas/xtensa/init-fini-literals.d: New file:
init-fini-literals test result patterns.
* testsuite/gas/xtensa/init-fini-literals.s: New file:
init-fini-literals test.
exec_file_locate_attach allocates memory for full_exec_path (using
either exec_file_find, source_full_path_of or xstrdup) but this
memory is never freed. This commit adds the necessary cleanup.
gdb/ChangeLog:
* exec.c (exec_file_locate_attach): Add missing cleanup.
This is necessary for upcoming tracepoint support - otherwise, setting
a tracepoint and a breakpoint on the same address will fail, since gdbserver
won't know about gdb's breakpoint.
Tested on s390x-ibm-linux-gnu and s390-ibm-linux-gnu, RHEL 7.2.
gdb/gdbserver/ChangeLog:
* linux-s390-low.c (s390_supports_z_point_type): New function.
(struct linux_target_ops): Wire s390_supports_z_point_type in.
This patch fixes an internal error that occurs in
gdb.threads/forking-threads-plus-breakpoint.exp:
/blah/binutils-gdb/gdb/target.c:2723: internal-error: Can't determine the
current address space of thread Thread 3170.3170
In default_thread_address_space, find_inferior_ptid couldn't find 3170.3170
because it had been overwritten in inferior_appeared, called as follows:
inferior_appeared
remote_add_inferior
remote_notice_new_inferior
remote_update_thread_list
The cause of the problem was the following sequence of events:
* GDB knows only about the main thread
* the first fork event is reported to GDB, saved as pending_event
* qXfer:threads:read gets the threads from the remote.
remove_new_fork_children id's the fork child from the pending event
and removes it from the list reported to GDB. All the rest of the
threads, including the fork parent, are added to the GDB thread list.
* GDB stops all the threads. All the stop events are pushed onto the
stop reply queue behind the pending fork event. The fork waitstatus
is saved in the fork parent thread's pending status field
thread_info.suspend.
* remote_wait_ns calls queued_stop_reply and process_stop_reply to
remove the fork event from the front of the stop reply queue and save
event information in the thread_info structure for the fork parent
thread. Unfortunately, none of the information saved in this way is
the fork-specific information.
* A subsequent qXfer:threads:read packet gets the thread list including
the fork parent and fork child. remove_new_fork_children checks the
thread list to see if there is a fork parent, doesn't find one, checks
the stop reply queue for a pending fork event, doesn't find one, and
allows the fork child thread to be reported to GDB before the fork
event has been handled. remote_update_thread_list calls
remote_notice_new_thread and overwrites the current (main) thread in
inferior_appeared.
So the fork event has been reported out of target_wait but it was left
pending on the infrun side (infrun.c:save_waitstatus). IOW, the fork
event hasn't been processed by handle_inferior_event yet, so it hasn't
made it to tp->pending_follow yet.
The fix is to check thread_info.suspend along with the
thread_info.pending_follow in remote.c:remove_new_fork_children, to
prevent premature reporting of the fork child thread creation.
gdb/ChangeLog:
PR remote/19496
* remote.c (remove_new_fork_children): Check for pending
fork status in thread_info.suspend.
gdb/testsuite/ChangeLog:
PR remote/19496
* gdb.threads/forking-threads-plus-breakpoint.exp (do_test):
Remove kfail for PR remote/19496.
Just like standard_output_file, standard_temp_file should use multiple
directories to make the tests parallel-safe. However,
standard_temp_file is sometimes called in some procedures that are not
test-specific. For example, gdb_init uses it, but is called once before
all test files are ran. Therefore, we can't organize it in a
temp/gdb.subdir/testname layout, like standard_output_file.
Because it's just meant for temporary files that don't really need to be
inspected after the test, we can just put them in a directory based on
the runtest pid. There is always a single exp file being executed by a
particular runtest invocation at any given time, so it should be safe.
gdb/testsuite/ChangeLog:
* lib/gdb.exp (standard_temp_file): Return a path specific to
the runtest invocation.
In save-trace.exp, we want to test loading of a tracepoint definition
file with a relative path (I am not sure why in fact). We currently use
"savetrace-relative.tr", which ends up directly in testsuite/. If we
use [standard_output_file] on that path, it becomes absolute. I decided
to just replace [pwd] with . (a dot) in the path given by
standard_output_file to make it relative. However, this trick only
works because [pwd] is a prefix of the standard output directory. So I
added a check to verify that precondition.
gdb/testsuite/ChangeLog:
* gdb.trace/save-trace.exp: Change relative path to be in the
standard output directory.
x86_64_mov_to_lea5.o and x86_64_mov_to_lea6.o must be assembled with
-mrelax-relocations=yes since when assembler is configured with
--disable-x86-relax-relocations, R_X86_64_GOTPCRELX relocation won't
be generated. This linker optimization requires R_X86_64_GOTPCRELX
relocation.
This patch also removed the duplicated x86_64_overflow_pc32.o target.
* testsuite/Makefile.am (x86_64_mov_to_lea5.o): Pass
-mrelax-relocations=yes to $(TEST_AS).
(x86_64_mov_to_lea6.o): Likewise.
(x86_64_overflow_pc32.o): Remove duplicated target.
* testsuite/Makefile.in: Regenerated.
I see the following error in testing aarch64 GDB debugging arm
program.
(gdb) PASS: gdb.reverse/readv-reverse.exp: set breakpoint at marker2
continue
Continuing.
=================================================================
==32273==ERROR: AddressSanitizer: attempting free on address which was not malloc()-ed: 0x000000ce4c00 in thread T0
#0 0x2ba5615645c7 in __interceptor_free (/usr/lib/x86_64-linux-gnu/libasan.so.1+0x545c7)^M
#1 0x4be8b5 in VEC_CORE_ADDR_cleanup /home/yao/SourceCode/gnu/gdb/git/gdb/common/gdb_vecs.h:34^M
#2 0x5e6d95 in do_my_cleanups /home/yao/SourceCode/gnu/gdb/git/gdb/common/cleanups.c:154^M
#3 0x64c99a in fetch_inferior_event /home/yao/SourceCode/gnu/gdb/git/gdb/infrun.c:3975^M
#4 0x678437 in inferior_event_handler /home/yao/SourceCode/gnu/gdb/git/gdb/inf-loop.c:44^M
#5 0x5078f6 in remote_async_serial_handler /home/yao/SourceCode/gnu/gdb/git/gdb/remote.c:13223^M
#6 0x4cecfd in run_async_handler_and_reschedule /home/yao/SourceCode/gnu/gdb/git/gdb/ser-base.c:137^M
#7 0x676864 in gdb_wait_for_event /home/yao/SourceCode/gnu/gdb/git/gdb/event-loop.c:834^M
#8 0x676a27 in gdb_do_one_event /home/yao/SourceCode/gnu/gdb/git/gdb/event-loop.c:323^M
#9 0x676aed in start_event_loop /home/yao/SourceCode/gnu/gdb/git/gdb/event-loop.c:347^M
#10 0x6706d2 in captured_command_loop /home/yao/SourceCode/gnu/gdb/git/gdb/main.c:318^M
#11 0x66db8c in catch_errors /home/yao/SourceCode/gnu/gdb/git/gdb/exceptions.c:240^M
#12 0x6716dd in captured_main /home/yao/SourceCode/gnu/gdb/git/gdb/main.c:1157^M
#13 0x66db8c in catch_errors /home/yao/SourceCode/gnu/gdb/git/gdb/exceptions.c:240^M
#14 0x671b7a in gdb_main /home/yao/SourceCode/gnu/gdb/git/gdb/main.c:1165^M
#15 0x467684 in main /home/yao/SourceCode/gnu/gdb/git/gdb/gdb.c:32^M
#16 0x2ba563ed7ec4 in __libc_start_main (/lib/x86_64-linux-gnu/libc.so.6+0x21ec4)^M
#17 0x4676b2 (/scratch/yao/gdb/build-git/aarch64-linux-gnu/gdb/gdb+0x4676b2)
looks we should discard cleanup if function
arm_linux_software_single_step returns early, or create cleanup when
it is needed.
gdb:
2016-02-16 Yao Qi <yao.qi@linaro.org>
* arm-linux-tdep.c (arm_linux_software_single_step): Assign
'old_chain' later.
