Cary Coutant [Sun, 22 Mar 2015 18:59:21 +0000 (11:59 -0700)]
Fix support for i386 TLS GD-to-IE optimization.
There are two cases to support, one with an SIB-form (6-byte) LEA,
the other with a 5-byte LEA and a NOP after the call __tls_get_addr.
Gold did not yet support the second case. This patch adds that
support.
gold/
PR gold/18106
* i386.cc (Target_i386::Relocate::tls_gd_to_ie): Fix support for
non-SIB form of lea, with nop after the call.
Cary Coutant [Sun, 22 Mar 2015 05:30:44 +0000 (22:30 -0700)]
Fix internal error with -z relro when .tbss is last relro section.
When calculating the padding necessary to align the end of the relro
segment to a page boundary, gold erroneously ignores the .tdata section
when checking to see if there are any relro sections (so if .tdata
is the only relro section, we fail to align the segment properly),
and erroneously pads the cumulative size of the segment based on
the alignment of .tbss. If there are no relro sections following .tbss,
it then fails to note the padding needed at the end of .tdata.
This patch fixes both problems. is_first_section_relro() will return
true when it sees a .tdata section, and we do not align the cumulative
size until after checking for the .tbss section.
gold/
PR gold/14217
* output.cc (Output_segment::is_first_section_relro): Don't ignore
.tdata section.
(Output_segment::set_section_addresses): Don't align size of relro
segment for .tbss.
Cary Coutant [Sun, 22 Mar 2015 04:09:46 +0000 (21:09 -0700)]
Fix bug when optimizing string pools of aligned strings.
Tail optimization of string pools (enabled when linker is run with -O2
or greater) should not be done when the section alignment is greater
than the size of the characters in the strings; otherwise, unaligned
strings may result.
gold/
PR gold/18010
* stringpool.cc (Stringpool_template): Don't optimize if section
alignment is greater than sizeof(char).
Cary Coutant [Sun, 22 Mar 2015 02:03:00 +0000 (19:03 -0700)]
PR gold/18048: Fix INCLUDE directive support for gold
This patch fixes INCLUDE directives in script files, so that when
an INCLUDE appears inside a sections block, section commands block,
or memory def block, the contents are parsed in the appropriate
context.
gold/
PR gold/18048
* script-c.h (script_include_directive): Add first_token parameter.
* script.cc (script_include_directive): Add first_token parameter, and
pass it to read_script_file.
* yyscript.y (PARSING_SECTIONS_BLOCK, PARSING_SECTION_CMDS)
(PARSING_MEMORY_DEF): New tokens.
(top): Add new productions for INCLUDE files.
(file_cmd): Replace file_or_sections_cmd with copy of its productions.
Pass PARSING_LINKER_SCRIPT to script_include_directive.
(section_block_cmd): Likewise; pass PARSING_SECTIONS_BLOCK.
(section_cmd): Pass PARSING_SECTION_CMDS.
(file_or_sections_cmd): Remove.
(memory_def): Pass PARSING_MEMORY_DEF.
* testsuite/Makefile.am (memory_test_2): New test.
* testsuite/Makefile.in: Regenerate.
* testsuite/memory_test_inc.t: New script file.
* testsuite/memory_test_inc_1.t.src: New script file.
* testsuite/memory_test_inc_2.t.src: New script file.
* testsuite/memory_test_inc_3.t.src: New script file.
Cary Coutant [Sun, 22 Mar 2015 01:50:11 +0000 (18:50 -0700)]
Support compressed debug sections in dynamic object files.
This patch adds support for reading compressed debug info in
shared objects. It actually simplifies things, by moving the
support for compressed sections all the way up to the top-level
Object class, eliminating the need for several virtual methods.
gold/
* dwp.cc (Sized_relobj_dwo::do_section_contents): Delete.
(Sized_relobj_dwo::setup): Build compressed section map.
(Sized_relobj_dwo::do_decompressed_section_contents): Delete.
* dynobj.cc (Sized_dynobj::base_read_symbols): Build compressed
section map.
* object.cc (Sized_relobj_file::Sized_relobj_file): Remove
compressed_sections_ field.
(build_compressed_section_map): Take Object instead of
Sized_relobj_file parameter; add decompress_if_needed parameter.
(Sized_relobj_file::do_find_special_sections): Store compressed
section map in parent Object.
(Sized_relobj_file::do_decompressed_section_contents): Move
implementation to Object::decompressed_section_contents.
(Sized_relobj_file::do_discard_decompressed_sections): Move
implementation to Object::discard_decompressed_sections.
* object.h (build_compressed_section_map): Declare.
(Object::Object): Add compressed_sections_ field.
(Object::section_is_compressed): Move implementation here.
(Object::decompressed_section_contents): De-virtualize.
(Object::discard_decompressed_sections): De-virtualize.
(Object::do_section_is_compressed): Delete.
(Object::do_decompressed_section_contents): Delete.
(Object::set_compressed_sections): New method.
(Object::compressed_sections): New method.
(Object::compressed_sections_): New data member.
(Compressed_section_info, Compressed_section_map): Move to top of file.
(Sized_relobj_file::do_section_is_compressed): Delete.
(Sized_relobj_file::do_decompressed_section_contents): Delete.
(Sized_relobj_file::do_discard_decompressed_sections): Delete.
(Sized_relobj_file::compressed_sections_): Move to Object class.
Cary Coutant [Sun, 22 Mar 2015 00:47:49 +0000 (17:47 -0700)]
Fix internal error in do_relocate_sections when using plugins.
In a previous patch for PR 14675, to fix a problem with
the .eh_frame section when static linking, I added a step to
finalize the .eh_frame section at the end of the first link
pass. This patch caused PR 18152, where a plugin-claimed
object caused a non-claimed object's layout to be deferred
until replacement files were read. The call to
finalize_eh_frame_section() is happening before the layout of
the deferred objects, leading to the internal error in
do_relocate_sections.
This patch moves the finalization of the .eh_frame section to
after deferred objects have been processed.
gold/
PR gold/14675
PR gold/18152
* gold.cc (queue_middle_tasks): Finalize .eh_frame after laying out
deferred objects.
Eli Zaretskii [Sat, 21 Mar 2015 08:48:34 +0000 (10:48 +0200)]
Fix undefined behavior in TUI's TAB expansion
gdb/ChangeLog:
* tui/tui-io.c (tui_expand_tabs): Reinitialize the column counter
before the second loop, to avoid undefined behavior. Reported by
Anton Blanchard <anton@samba.org>.
Keven Boell [Fri, 20 Mar 2015 21:45:20 +0000 (17:45 -0400)]
[gdb/DWARF] Introduce linked list for dynamic attributes
This patch introduces a linked list for dynamic attributes of a type.
This is a pre-work for the Fortran dynamic array support. The Fortran
dynamic array support will add more dynamic attributes to a type.
As only a few types will have such dynamic attributes set, a linked
list is more efficient in terms of memory consumption than adding
multiple attributes to main_type.
gdb/ChangeLog:
* gdbtypes.c (resolve_dynamic_type_internal): Adapt
data_location usage to linked list.
(resolve_dynamic_type_internal): Adapt data_location to
linked list.
(get_dyn_prop, add_dyn_prop, copy_dynamic_prop_list): New function.
(copy_type_recursive, copy_type): Add copy of linked list.
* gdbtypes.h (enum dynamic_prop_node_kind): New enum.
(struct dynamic_prop_list): New struct.
* dwarf2read.c (set_die_type): Set data_location data.
Pedro Alves [Mon, 9 Mar 2015 11:58:23 +0000 (11:58 +0000)]
constify i386-sol2-tdep.c and machoread.c
/home/pedro/gdb/mygit/src/gdb/i386-sol2-tdep.c: In function ‘const char* i386_sol2_static_transform_name(const char*)’:
/home/pedro/gdb/mygit/src/gdb/i386-sol2-tdep.c:93:29: error: invalid conversion from ‘const char*’ to ‘char*’ [-fpermissive]
p = strrchr (name, '.');
^
gdb:
2015-03-20 Pedro Alves <palves@redhat.com>
* i386-sol2-tdep.c (i386_sol2_static_transform_name): Move "p" to
inner block and make it const.
* machoread.c (get_archive_prefix_len): Make "lparen" const.
Pedro Alves [Mon, 9 Mar 2015 11:58:20 +0000 (11:58 +0000)]
constify xcoffread.c
/home/pedro/gdb/mygit/src/gdb/xcoffread.c: In function ‘void scan_xcoff_symtab(objfile*)’:
/home/pedro/gdb/mygit/src/gdb/xcoffread.c:2644:33: error: invalid conversion from ‘const char*’ to ‘char*’ [-fpermissive]
p = strchr (namestring, ':');
^
gdb:
2015-03-20 Pedro Alves <palves@redhat.com>
* xcoffread.c (scan_xcoff_symtab): Make "p" and "q" const.
Pedro Alves [Mon, 9 Mar 2015 11:58:09 +0000 (11:58 +0000)]
constify gdbserver/tracepoint.c
gdb/gdbserver/tracepoint.c:3647:7: error: invalid conversion from 'const char*' to 'char*' [-fpermissive]
gdb/gdbserver/tracepoint.c:3652:7: error: invalid conversion from 'const char*' to 'char*' [-fpermissive]
gdb/gdbserver/tracepoint.c:3657:7: error: invalid conversion from 'const char*' to 'char*' [-fpermissive]
H.J. Lu [Fri, 20 Mar 2015 11:39:04 +0000 (04:39 -0700)]
Limit multi-byte nop instructions to 10 bytes
There is no performance advantage to use multi-byte nop instructions
greater than 10 bytes. This patch limits multi-byte nop instructions
to 10 bytes. Since there is only one way to encode multi-byte nop
instructions now, it also removed redundant nop tests.
gas/
* config/tc-i386.c (i386_align_code): Limit multi-byte nop
instructions to 10 bytes.
Nick Clifton [Fri, 20 Mar 2015 10:29:46 +0000 (10:29 +0000)]
Fixes implementation of hardware multiply/divide peripheral in RL78 G13 cores.
* load.c (rl78_load): If the G10, G13 or G14 flag bits are set in
the ELF header use them to select the proper emulation mode.
* mem.c (mem_put_byte): Use mem_put_hi to store a value into the
MDB register.
(mem_get_byte): Use mem_get_hi to extract a value from the MDB
register.
Yao Qi [Fri, 20 Mar 2015 10:15:21 +0000 (10:15 +0000)]
Don't call set_gdbarch_cannot_step_breakpoint in aarch64_gdbarch_init
Hi,
I am looking at the following fails in aarch64-linux,
stepi^M
47 NOP; /* after permanent bp */^M
(gdb) FAIL: gdb.base/bp-permanent.exp: always_inserted=off, sw_watchpoint=0: stepi signal with handler: single-step to handler
the test expects GDB single step into signal handler, but GDB doesn't.
The code in infrun.c:resume
/* Most targets can step a breakpoint instruction, thus
executing it normally. But if this one cannot, just
continue and we will hit it anyway. */
if (gdbarch_cannot_step_breakpoint (gdbarch))
step = 0;
change the intended action from "step" to "continue". The gdbarch method
cannot_step_breakpoint isn't documented well, and I don't get much clue
after explore the history. However, from the comments above,
aarch64-linux can step a breakpoint instruction, so don't need to call
set_gdbarch_cannot_step_breakpoint.
Pedro Alves [Thu, 19 Mar 2015 15:12:33 +0000 (15:12 +0000)]
Fix race exposed by gdb.threads/killed.exp
On GNU/Linux, this test sometimes FAILs like this:
(gdb) run
Starting program: /home/pedro/gdb/mygit/build/gdb/testsuite/gdb.threads/killed
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib64/libthread_db.so.1".
ptrace: No such process.
(gdb)
Program terminated with signal SIGKILL, Killed.
The program no longer exists.
FAIL: gdb.threads/killed.exp: run program to completion (timeout)
Note the suspicious "No such process" line (that's errno==ESRCH).
Adding debug output we see:
linux_nat_wait: [process -1], [TARGET_WNOHANG]
LLW: enter
LNW: waitpid(-1, ...) returned 18465, ERRNO-OK
LLW: waitpid 18465 received Stopped (signal) (stopped)
LNW: waitpid(-1, ...) returned 18461, ERRNO-OK
LLW: waitpid 18461 received Trace/breakpoint trap (stopped)
LLW: Handling extended status 0x03057f
LHEW: Got clone event from LWP 18461, new child is LWP 18465
LNW: waitpid(-1, ...) returned 0, ERRNO-OK
RSRL: resuming stopped-resumed LWP LWP 18465 at 0x3b36af4b51: step=0
RSRL: resuming stopped-resumed LWP LWP 18461 at 0x3b36af4b51: step=0
sigchld
ptrace: No such process.
(gdb) linux_nat_wait: [process -1], [TARGET_WNOHANG]
LLW: enter
LNW: waitpid(-1, ...) returned 18465, ERRNO-OK
LLW: waitpid 18465 received Killed (terminated)
LLW: LWP 18465 exited.
LNW: waitpid(-1, ...) returned 18461, No child processes
LLW: waitpid 18461 received Killed (terminated)
Process 18461 exited
LNW: waitpid(-1, ...) returned -1, No child processes
LLW: exit
sigchld
infrun: target_wait (-1, status) =
infrun: 18461 [process 18461],
infrun: status->kind = signalled, signal = GDB_SIGNAL_KILL
infrun: TARGET_WAITKIND_SIGNALLED
Program terminated with signal SIGKILL, Killed.
The program no longer exists.
infrun: stop_waiting
FAIL: gdb.threads/killed.exp: run program to completion (timeout)
The issue is that here:
RSRL: resuming stopped-resumed LWP LWP 18465 at 0x3b36af4b51: step=0
RSRL: resuming stopped-resumed LWP LWP 18461 at 0x3b36af4b51: step=0
The first line shows we had just resumed LWP 18465, which does:
So if the kernel manages to schedule that thread fast enough, the
process may be killed before GDB has a chance to resume LWP 18461.
GDBserver has code at the tail end of linux_resume_one_lwp to cope
with this:
~~~
ptrace (step ? PTRACE_SINGLESTEP : PTRACE_CONT, lwpid_of (thread),
(PTRACE_TYPE_ARG3) 0,
/* Coerce to a uintptr_t first to avoid potential gcc warning
of coercing an 8 byte integer to a 4 byte pointer. */
(PTRACE_TYPE_ARG4) (uintptr_t) signal);
current_thread = saved_thread;
if (errno)
{
/* ESRCH from ptrace either means that the thread was already
running (an error) or that it is gone (a race condition). If
it's gone, we will get a notification the next time we wait,
so we can ignore the error. We could differentiate these
two, but it's tricky without waiting; the thread still exists
as a zombie, so sending it signal 0 would succeed. So just
ignore ESRCH. */
if (errno == ESRCH)
return;
perror_with_name ("ptrace");
}
~~~
However, that's not a complete fix, because between starting to handle
the resume request and getting that PTRACE_CONTINUE, we run other
ptrace calls that can also fail with ESRCH, and that end up throwing
an error (with perror_with_name).
In the case above, I indeed sometimes see resume_stopped_resumed_lwps
fail in the registers read:
Or e.g., in 32-bit mode, i386_linux_resume has several calls that can
throw too.
Whether to ignore ptrace errors or not depends on context that is only
available somewhere up the call chain. So the fix is to let ptrace
errors throw as they do today, and wrap the resume request in a
TRY/CATCH that swallows it iff the lwp that we were trying to resume
is no longer ptrace-stopped.
gdb/gdbserver/ChangeLog:
2015-03-19 Pedro Alves <palves@redhat.com>
* linux-low.c (linux_resume_one_lwp): Rename to ...
(linux_resume_one_lwp_throw): ... this. Don't handle ESRCH here,
instead call perror_with_name.
(check_ptrace_stopped_lwp_gone): New function.
(linux_resume_one_lwp): Reimplement as wrapper around
linux_resume_one_lwp_throw that swallows errors if the LWP is
gone.
gdb/ChangeLog:
2015-03-19 Pedro Alves <palves@redhat.com>
* linux-nat.c (linux_resume_one_lwp): Rename to ...
(linux_resume_one_lwp_throw): ... this. Don't handle ESRCH here,
instead call perror_with_name.
(check_ptrace_stopped_lwp_gone): New function.
(linux_resume_one_lwp): Reimplement as wrapper around
linux_resume_one_lwp_throw that swallows errors if the LWP is
gone.
(resume_stopped_resumed_lwps): Try register reads in TRY/CATCH and
swallows errors if the LWP is gone. Use
linux_resume_one_lwp_throw instead of linux_resume_one_lwp.
/home/pedro/gdb/mygit/src/gdb/gdbserver/linux-low.c:2329: A problem internal to GDBserver has been detected.
select_event_lwp: Assertion `num_events > 0' failed.
=thread-group-exited,id="i1"
Certainly select_event_lwp_callback should always at least find one
event, as it's only called because an event triggered (though we may
have more than one: the point of the function is randomly picking
one).
An LWP that GDB previously asked to continue/step (thus is resumed)
and gets a vCont;t request ends up with last_resume_kind ==
resume_stop. These functions in gdbserver used to filter out events
that weren't going to be reported to GDB; I think the last_resume_kind
kind check used to make sense at that point, but it no longer does.
gdb/gdbserver/ChangeLog:
2015-03-19 Pedro Alves <palves@redhat.com>
* linux-low.c (count_events_callback, select_event_lwp_callback):
No longer check whether the thread has resume_stop as last resume
kind.
bin * readelf.c (get_machine_flags): Decode RL78's G13 and G14 flags.
gas * config/tc-rl78.c (enum options): Add G13 and G14.
(md_longopts): Add -mg13 and -mg14.
(md_parse_option): Handle -mg13 and -mg14.
(md_show_usage): List -mg13 and -mg14.
* doc/c-rl78.texi: Add description of -mg13 and -mg14 options.
bfd * elf32-rl78.c (rl78_cpu_name): New function. Prints the name of
the RL78 core based upon the flags.
(rl78_elf_merge_private_bfd_data): Handle merging of G13 and G14
flags.
(rl78_elf_print_private_bfd_data): Use rl78_cpu_name.
(elf32_rl78_machine): Always return bfd_mach_rl78.
Wanting to make sure the new continue-pending-status.exp test tests
both cases of threads 2 and 3 reporting an event, I added counters to
the test, to make it FAIL if events for both threads aren't seen.
Assuming a well behaved backend, and given a reasonable number of
iterations, it should PASS.
However, running that against GNU/Linux gdbserver, I found that
surprisingly, that FAILed. GDBserver always reported the breakpoint
hit for the same thread.
Turns out that I broke gdbserver's thread event randomization
recently, with git commit 582511be ([gdbserver] linux-low.c: better
starvation avoidance, handle non-stop mode too). In that commit I
missed that the thread structure also has a status_pending_p field...
The end result was that count_events_callback always returns 0, and
then if no thread is stepping, select_event_lwp always returns the
event thread. IOW, no randomization is happening at all. Quite
curious how all the other changes in that patch were sufficient to fix
non-stop-fair-events.exp anyway even with that broken.
Tested on x86_64 Fedora 20, native and gdbserver.
gdb/gdbserver/ChangeLog:
2015-03-19 Pedro Alves <palves@redhat.com>
* linux-low.c (count_events_callback, select_event_lwp_callback):
Use the lwp's status_pending_p field, not the thread's.
gdb/testsuite/ChangeLog:
2015-03-19 Pedro Alves <palves@redhat.com>
* gdb.threads/continue-pending-status.exp (saw_thread_2)
(saw_thread_3): New globals.
(top level): Increment them when an event for the corresponding
thread is seen.
(no thread starvation): New test.
Pedro Alves [Thu, 19 Mar 2015 12:20:25 +0000 (12:20 +0000)]
native/Linux: internal error if resume is short-circuited
If the linux_nat_resume's short-circuits the resume because the
current thread has a pending status, and, a thread with a higher
number was previously stopped for a breakpoint, GDB internal errors,
like:
Pedro Alves [Wed, 18 Mar 2015 19:28:08 +0000 (19:28 +0000)]
Tighten gdb.base/disp-step-syscall.exp
This fixes several problems with this test.
E.g,. with --target_board=native-extended-gdbserver on x86_64 Fedora
20, I get:
Running /home/pedro/gdb/mygit/src/gdb/testsuite/gdb.base/disp-step-syscall.exp ...
FAIL: gdb.base/disp-step-syscall.exp: vfork: get hexadecimal valueof "$pc" (timeout)
FAIL: gdb.base/disp-step-syscall.exp: vfork: single step over vfork final pc
FAIL: gdb.base/disp-step-syscall.exp: vfork: delete break vfork insn
FAIL: gdb.base/disp-step-syscall.exp: vfork: continue to marker (vfork) (the program is no longer running)
And with --target=native-gdbserver, I get:
Running /home/pedro/gdb/mygit/src/gdb/testsuite/gdb.base/disp-step-syscall.exp ...
KPASS: gdb.base/disp-step-syscall.exp: vfork: single step over vfork (PRMS server/13796)
FAIL: gdb.base/disp-step-syscall.exp: vfork: get hexadecimal valueof "$pc" (timeout)
FAIL: gdb.base/disp-step-syscall.exp: vfork: single step over vfork final pc
FAIL: gdb.base/disp-step-syscall.exp: vfork: delete break vfork insn
FAIL: gdb.base/disp-step-syscall.exp: vfork: continue to marker (vfork) (the program is no longer running)
First, the lack of fork support on remote targets is supposed to be
kfailed, so the KPASS is obviously bogus. The extended-remote board
should have KFAILed too.
The problem is that the test is using "is_remote" instead of
gdb_is_target_remote.
And then, I get:
(gdb) PASS: gdb.base/disp-step-syscall.exp: vfork: set displaced-stepping on
stepi
Program terminated with signal SIGSEGV, Segmentation fault.
The program no longer exists.
(gdb) PASS: gdb.base/disp-step-syscall.exp: vfork: single step over vfork
Obviously, that should be a FAIL. The problem is that the test only
expects SIGILL, not SIGSEGV. It also doesn't bail correctly if an
internal error or some other pattern caught by gdb_test_multiple
matches. The test doesn't really need to match specific exits/crashes
patterns, if the PASS regex is improved, like in ...
... this and the other "stepi" tests are a bit too lax, passing on
".*". This tightens those up to expect "x/i" and the "=>" current PC
indicator, like in:
1: x/i $pc
=> 0x3b36abc9e2 <vfork+34>: syscall
On x86_64 Fedora 20, I now get a quick KFAIL instead of timeouts with
both the native-extended-gdbserver and native-gdbserver boards:
PASS: gdb.base/disp-step-syscall.exp: vfork: delete break vfork
PASS: gdb.base/disp-step-syscall.exp: vfork: continue to syscall insn vfork
PASS: gdb.base/disp-step-syscall.exp: vfork: set displaced-stepping on
KFAIL: gdb.base/disp-step-syscall.exp: vfork: single step over vfork (PRMS: server/13796)
and a full pass with native testing.
gdb/testsuite/
2015-03-18 Pedro Alves <palves@redhat.com>
* gdb.base/disp-step-syscall.exp (disp_step_cross_syscall):
Use gdb_is_target_remote instead of is_remote. Use
gdb_test_multiple instead of gdb_expect. Exit early if
gdb_test_multiple hits its internal matches. Tighten stepi tests
expected output. Fail on exit with any signal, instead of just
SIGILL.
Jon Turney [Wed, 18 Mar 2015 15:47:13 +0000 (15:47 +0000)]
Fix debug section compression so that it is only performed if it would make the section smaller.
PR binutils/18087
gas * doc/as.texinfo: Note that when gas compresses debug sections the
compression is only performed if it makes the section smaller.
* write.c (compress_debug): Do not compress a debug section if
doing so would make it larger.
tests * gas/i386/dw2-compress-1.d: Do not expect the .debug_abbrev or
.debug_info sections to be compressed.
binu * doc/binutils.texi: Note that when objcopy compresses debug
sections the compression is only performed if it makes the section
smaller.
bfd * coffgen.c (make_a_section_from_file): Only prepend a z to a
debug section's name if the section was actually compressed.
* elf.c (_bfd_elf_make_section_from_shdr): Likewise.
* compress.c (bfd_init_section_compress_status): Do not compress
the section if doing so would make it bigger. In such cases leave
the section alone and return COMPRESS_SECTION_NONE.
Tristan Gingold [Thu, 5 Mar 2015 13:36:32 +0000 (14:36 +0100)]
Fix amd64 windows unwinding issues within MS dlls.
Unwind info in system dlls uses almost all possible codes, contrary to unwind
info generated by gcc. A few issues have been discovered: incorrect handling
of SAVE_NONVOL opcodes and incorrect in prologue range checks. Furthermore I
added comments not to forget what has been investigated.
gdb/ChangeLog:
* amd64-windows-tdep.c (amd64_windows_find_unwind_info): Move
redirection code to ...
(amd64_windows_frame_decode_insns): ... Here. Fix in prologue
checks. Fix SAVE_NONVOL operations. Add debug code and comments.
Gary Benson [Wed, 18 Mar 2015 11:19:02 +0000 (11:19 +0000)]
Reimplement "vFile:fstat" without qSupported
This commit makes support for the "vFile:fstat" packet be detected
by probing rather than using qSupported, for consistency with the
other vFile: packets.
gdb/ChangeLog:
(remote_protocol_features): Remove the "vFile:fstat" feature.
(remote_hostio_fstat): Probe for "vFile:fstat" support.
Yao Qi [Wed, 18 Mar 2015 10:47:45 +0000 (10:47 +0000)]
Support catch syscall on aarch64 linux
Hi,
This patch is to support catch syscall on aarch64 linux. We
implement gdbarch method get_syscall_number for aarch64-linux,
and add aarch64-linux.xml file, which looks straightforward, however
the changes to test case doesn't.
First of all, we enable catch-syscall.exp on aarch64-linux target,
but skip the multi_arch testing on current stage. I plan to touch
multi arch debugging on aarch64-linux later.
Then, when I run catch-syscall.exp on aarch64-linux, gcc errors that
SYS_pipe isn't defined. We find that aarch64 kernel only has pipe2
syscall and libc already convert pipe to pipe2. As a result, I change
catch-syscall.c to use SYS_pipe if it is defined, otherwise use
SYS_pipe2 instead. The vector all_syscalls in catch-syscall.exp can't
be pre-determined, so I add a new proc setup_all_syscalls to fill it,
according to the availability of SYS_pipe.
Regression tested on {x86_64, aarch64}-linux x {native, gdbserver}.
gdb:
2015-03-18 Yao Qi <yao.qi@linaro.org>
PR tdep/18107
* aarch64-linux-tdep.c: Include xml-syscall.h
(aarch64_linux_get_syscall_number): New function.
(aarch64_linux_init_abi): Call
set_gdbarch_get_syscall_number.
* syscalls/aarch64-linux.xml: New file.
gdb/testsuite:
2015-03-18 Yao Qi <yao.qi@linaro.org>
PR tdep/18107
* gdb.base/catch-syscall.c [!SYS_pipe] (pipe2_syscall): New
variable.
* gdb.base/catch-syscall.exp: Don't skip it on
aarch64*-*-linux* target. Remove elements in all_syscalls.
(test_catch_syscall_multi_arch): Skip it on aarch64*-linux*
target.
(setup_all_syscalls): New proc.
Gary Benson [Tue, 17 Mar 2015 11:26:09 +0000 (11:26 +0000)]
Remove forward declaration of struct stat
Forward declarations of struct stat break the Windows build.
This commit removes a forward declaration of struct stat and
includes sys/stat.h directly instead.
Yao Qi [Fri, 13 Mar 2015 10:56:01 +0000 (10:56 +0000)]
watchpoint-reuse-slot.exp: skip setting HW breakpoints on some address
We see some fails in watchpoint-reuse-slot.exp on aarch64-linux, because
it sets some HW breakpoint on some address doesn't meet the alignment
requirements by kernel, kernel will reject the
ptrace (PTRACE_SETHBPREGS) call, and some fails are caused, for example:
(gdb) PASS: gdb.base/watchpoint-reuse-slot.exp: always-inserted off: watch x hbreak: : width 1, iter 0: base + 0: delete $bpnum
hbreak *(buf.byte + 0 + 1)^M
Hardware assisted breakpoint 80 at 0x410a61^M
(gdb) PASS: gdb.base/watchpoint-reuse-slot.exp: always-inserted off: watch x hbreak: : width 1, iter 0: base + 1: hbreak *(buf.byte + 0 + 1)
stepi^M
Warning:^M
Cannot insert hardware breakpoint 80.^M
Could not insert hardware breakpoints:^M
You may have requested too many hardware breakpoints/watchpoints.^M
^M
(gdb) FAIL: gdb.base/watchpoint-reuse-slot.exp: always-inserted off: watch x hbreak: : width 1, iter 0: base + 1: stepi advanced
hbreak *(buf.byte + 0 + 1)^M
Hardware assisted breakpoint 440 at 0x410a61^M
Warning:^M
Cannot insert hardware breakpoint 440.^M
Could not insert hardware breakpoints:^M
You may have requested too many hardware breakpoints/watchpoints.^M
^M
(gdb) FAIL: gdb.base/watchpoint-reuse-slot.exp: always-inserted on: watch x hbreak: : width 1, iter 0: base + 1: hbreak *(buf.byte + 0 + 1)
This patch is to skip some tests by checking proc valid_addr_p.
We can handle other targets in valid_addr_p too.
gdb/testsuite:
2015-03-16 Yao Qi <yao.qi@linaro.org>
* gdb.base/watchpoint-reuse-slot.exp (valid_addr_p): New proc.
(top level): Skip tests if valid_addr_p returns false for
$cmd1 or $cmd2.
John Baldwin [Sat, 14 Mar 2015 17:47:20 +0000 (13:47 -0400)]
Fetch all registers before writing the core register notes.
Without this, not all registers were present in the core generated by
gcore. For example, running 'gcore' on a program without examining
the vector registers (SSE or AVX) would store all the vector registers
as zeros because they were not pulled into the regcache. Running
'info vector' before 'gcore' would store the correct values in the
core since it populated the regcache. For Linux processes, a similar
operation is achieved by having the thread iterator callback invoke
target_fetch_registers on each thread before its corresponding
register notes are dumped.
gdb/ChangeLog:
* fbsd-tdep.c (fbsd_make_corefile_notes): Fetch all target registers
before writing core register notes.
Yuanhui Zhang [Mon, 16 Mar 2015 11:28:24 +0000 (11:28 +0000)]
stub termcap, add extern "C"
Fixes linking an --enable-build-with-cxx build on mingw:
../readline/terminal.c:278: undefined reference to `tgetnum'
../readline/terminal.c:297: undefined reference to `tgetnum'
../readline/libreadline.a(terminal.o): In function `get_term_capabilities':
../readline/terminal.c:427: undefined reference to `tgetstr'
../readline/libreadline.a(terminal.o): In function `_rl_init_terminal_io':
[etc.]
gdb/ChangeLog:
2015-03-16 Yuanhui Zhang <asmwarrior@gmail.com>
Pedro Alves <palves@redhat.com>
* gdb_curses.h (tgetnum): Mark with EXTERN_C.
* stub-termcap.c (tgetent, tgetnum, tgetflag, tgetstr, tputs)
(tgoto): Wrap with extern "C".
Pedro Alves [Mon, 16 Mar 2015 11:28:23 +0000 (11:28 +0000)]
stub-termcap.c: prototype tputs's parameter's parameter, for C++ mode
src/gdb/stub-termcap.c: In function 'int tputs(char*, int, int (*)())':
src/gdb/stub-termcap.c:67:22: error: too many arguments to function
outfun (*string++);
^
gdb/ChangeLog:
2015-03-16 Pedro Alves <palves@redhat.com>
Yuanhui Zhang <asmwarrior@gmail.com>
Yuanhui Zhang [Mon, 16 Mar 2015 11:31:31 +0000 (11:31 +0000)]
windows-nat.c: conflicting declaration of struct thread_info in C++ mode
Building mingw GDB with --enable-build-with-cxx shows:
../../binutils-gdb/gdb/windows-nat.c: At global scope:
../../binutils-gdb/gdb/windows-nat.c:192:1: error: conflicting declaration 'typedef struct thread_info_struct thread_info'
thread_info;
^
In file included from ../../binutils-gdb/gdb/windows-nat.c:52:0:
../../binutils-gdb/gdb/gdbthread.h:160:8: error: 'struct thread_info' has a previous declaration as 'struct thread_info'
struct thread_info
^
Simply rename the structure to avoid the conflict.
gdb/ChangeLog:
2015-03-16 Yuanhui Zhang <asmwarrior@gmail.com>
Pedro Alves <palves@redhat.com>
* windows-nat.c (struct thread_info_struct): Rename to ...
(struct windows_thread_info_struct): ... this.
(thread_info): Rename to ...
(windows_thread_info): ... this.
All users updated.
Mike Frysinger [Sun, 15 Mar 2015 03:41:25 +0000 (23:41 -0400)]
sim: rename tconfig.in to tconfig.h
Rather than manually include tconfig.h when we think we'll need it (which
is error prone as it can define symbols we expect from config.h), have it
be included directly by config.h. Since we know we have to include that
header everywhere already, this will make sure tconfig.h isn't missed.
It should also be fine as tconfig.h is supposed to be simple and only set
up a few core defines for the target.
This allows us to stop symlinking it in place all the time and just use
it straight out of the respective source directory.
Mike Frysinger [Sun, 15 Mar 2015 06:55:08 +0000 (02:55 -0400)]
sim: dv-sockser: push module init prototype down
Pull out the duplicated dv_sockser_install prototype from the tconfig.in
files and put it in the one place it gets used -- sim-module.c. This is
still arguably incorrect, but it's better than the status quo where the
tconfig.in has to include header files and duplicate the dv-sockser func.
The tconfig header is meant to be simple and contain a target defines.
Mike Frysinger [Sun, 15 Mar 2015 00:52:24 +0000 (20:52 -0400)]
sim: delete unused sim-inline.c rule
This dates back to the start of the repo, but has never really been used.
The sim-inline.c file has been checked in to the source, and attempts to
build it in the build tree leads to a circular dep warning from make. It
also doesn't produce a file that is usable -- it can't be compiled. Punt!
Mike Frysinger [Thu, 19 Feb 2015 08:46:01 +0000 (03:46 -0500)]
sim: make nrun the default run program
We want people to stop using the run.c frontend, but it's hard to notice
when it's still set as the default. Lets flip things so nrun.c is the
default, and users of run.c will get an error by default. We turn that
error into a warning for existing sims so we don't break them -- this is
mostly meant for people starting new ports.
Andreas Krebbel [Sat, 14 Mar 2015 10:45:05 +0000 (11:45 +0100)]
S/390: Disable relocation sort against code sections.
When downgrading from GD to IE model we rewrite the call to
__tls_get_offset to a 64 bit load instruction. This relies on the fact
that the additional relocation for the call target has already been
executed when doing the rewrite.
Due to the reloc sorting the order changed an the PLT32DBL reloc is
executed after the rewrite and overwrites part of the load instruction
with garbage.
bfd/
2015-03-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
Mike Frysinger [Sat, 14 Mar 2015 07:42:08 +0000 (03:42 -0400)]
sim: move sim-io.h to sim-assert.h
Since sim-endian.c doesn't actually use sim_io funcs, it's weird to
include the sim-io.h header here. It's doing so only for the assert
header. So lets relocate the include to the right place.
Jan Kratochvil [Fri, 13 Mar 2015 19:24:22 +0000 (20:24 +0100)]
Remove HPUX
IIUC it is a pre-requisite for IPv6 support, some UNICes do not support
getaddrinfo required for IPv6. But coincidentally such UNICes are no longer
really supported by GDB. Therefore it was concluded we can remove all such
UNICes and then we can implement IPv6 easily with getaddrinfo.
In mail
Re: getaddrinfo available on all GDB hosts? [Re: [PATCH v2] Add IPv6 support for remote TCP connections]
Message-ID: <20140211034157.GG5485@adacore.com>
https://sourceware.org/ml/gdb-patches/2014-02/msg00333.html
Joel said:
So I chose HP-UX first for this patch.
gdb/ChangeLog
2014-10-16 Jan Kratochvil <jan.kratochvil@redhat.com>
Remove HPUX.
* Makefile.in (ALL_64_TARGET_OBS): Remove ia64-hpux-tdep.o.
(ALL_TARGET_OBS): Remove hppa-hpux-tdep.o, solib-som.o and solib-pa64.o.
(HFILES_NO_SRCDIR): Remove solib-som.h, inf-ttrace.h, solib-pa64.h and
ia64-hpux-tdep.h, solib-ia64-hpux.h.
(ALLDEPFILES): Remove hppa-hpux-tdep.c, hppa-hpux-nat.c,
ia64-hpux-nat.c, ia64-hpux-tdep.c, somread.c and solib-som.c.
* config/djgpp/fnchange.lst: Remove hppa-hpux-nat.c and
hppa-hpux-tdep.c.
* config/ia64/hpux.mh: Remove file.
* config/pa/hpux.mh: Remove file.
* configure: Rebuilt.
* configure.ac (dlgetmodinfo, somread.o): Remove.
* configure.host (hppa*-*-hpux*, ia64-*-hpux*): Make them obsolete.
(ia64-*-hpux*): Remove its float format exception.
* configure.tgt (hppa*-*-hpux*, ia64-*-hpux*): Make them obsolete.
* hppa-hpux-nat.c: Remove file.
* hppa-hpux-tdep.c: Remove file.
* hppa-tdep.c (struct hppa_unwind_info, struct hppa_objfile_private):
Move them here from hppa-tdep.h
(hppa_objfile_priv_data, hppa_init_objfile_priv_data): Make it static.
(hppa_frame_prev_register_helper): Remove HPPA_FLAGS_REGNUM exception.
* hppa-tdep.h (struct hppa_unwind_info, struct hppa_objfile_private):
Move them to hppa-tdep.c.
(hppa_objfile_priv_data, hppa_init_objfile_priv_data): Remove
declarations.
* ia64-hpux-nat.c: Remove file.
* ia64-hpux-tdep.c: Remove file.
* ia64-hpux-tdep.h: Remove file.
* inf-ttrace.c: Remove file.
* inf-ttrace.h: Remove file.
* solib-ia64-hpux.c: Remove file.
* solib-ia64-hpux.h: Remove file.
* solib-pa64.c: Remove file.
* solib-pa64.h: Remove file.
* solib-som.c: Remove file.
* solib-som.h: Remove file.
* somread.c: Remove file.
John Baldwin [Wed, 18 Feb 2015 00:15:45 +0000 (19:15 -0500)]
Use kinfo_getvmmap on FreeBSD to enumerate memory regions.
Use kinfo_getvmmap from libutil on FreeBSD to enumerate memory
regions in a running process instead of /proc/<pid>/map. FreeBSD systems
do not mount procfs by default, but kinfo_getvmmap uses a sysctl that
is always available.
Skip memory regions for devices as well as regions an application has
requested to not be dumped via the MAP_NOCORE flag to mmap or
MADV_NOCORE advice to madvise.
gdb/ChangeLog:
* configure.ac: AC_CHECK_LIB(util, kinfo_getvmmap).
* configure: Regenerate.
* config.in: Regenerate.
* fbsd-nat.c [!HAVE_KINFO_GETVMMAP] (fbsd_read_mapping): Don't
define.
(fbsd_find_memory_regions): Use kinfo_getvmmap to
enumerate memory regions if present.
Nick Clifton [Thu, 12 Mar 2015 15:58:37 +0000 (15:58 +0000)]
Fixes a problem generating relocs for thumb function calls to local symbols defined in other sections.
PR gas/17444
* config/tc-arm.h (MD_APPLY_SYM_VALUE): Pass the current segment
to arm_apply_sym_value. Update prototype.
* config/tc-arm.c (arm_apply_sym_value): Add segment argument.
Do not apply the value if the symbol is in a different segment to
the current segment.
projects / thirdparty / binutils-gdb.git / log
