A small cleanup. 'struct lwp_info'::thread_known is only useful for
thread-db.c.
gdbserver/
2013-04-16 Pedro Alves <palves@redhat.com>
* linux-low.h (struct lwp_info) <thread_known>: Move under
the USE_THREAD_DB #ifdef.
The previous patches are still not sufficient to build gdbserver with
our copy of thread_db.h.
../../../src/gdb/gdbserver/thread-db.c: In function ‘find_one_thread’:
../../../src/gdb/gdbserver/thread-db.c:316:6: error: ‘struct lwp_info’ has no member named ‘th’
../../../src/gdb/gdbserver/thread-db.c: In function ‘attach_thread’:
../../../src/gdb/gdbserver/thread-db.c:341:6: error: ‘struct lwp_info’ has no member named ‘th’
../../../src/gdb/gdbserver/thread-db.c: In function ‘thread_db_get_tls_address’:
../../../src/gdb/gdbserver/thread-db.c:514:47: error: ‘struct lwp_info’ has no member named ‘th’
make: *** [thread-db.o] Error 1
First, linux-low.h is including <thread_db.h> directly instead of our
gdb_thread_db.h, although thread-db.c includes the latter. Then the
'th' field of struct lwp_info is only defined if HAVE_THREAD_DB_H is
defined, which is not true if we're using our replacement copy of
thread_db.h. We have a USE_THREAD_DB symbol defined if we're building
thread-db.c that's ideal for this, however, it's currently only
defined when compiling linux-low.c (through a Makefile rule). The
patch makes it defined when compiling any file.
gdb/gdbserver/
2013-04-16 Pedro Alves <palves@redhat.com>
* Makefile.in (INTERNAL_CFLAGS): Add @USE_THREAD_DB@.
(linux-low.o): Delete rule.
* linux-low.h: Always include "gdb_thread_db.h" instead of
conditionally including thread_db.h.
(struct lwp_info) <th>: Guard with #ifdef USE_THREAD_DB instead of
HAVE_THREAD_DB_H.
Oleg Nesterov told me that the Linux kernel copies the parent's ptrace
options to fork/clone children, so there's no need for GDB to do that
manually.
I was actually a bit surprised, since I thought the ptracer had to
always set the ptrace options itself, and GDB is indeed calling
PTRACE_SETOPTIONS for each new fork child, if it'll stay attached.
Looking at the history of that code, I found that is was actually I
who added that set-ptrace-options-in-children bit, back in
http://sourceware.org/ml/gdb-patches/2009-05/msg00656.html. But,
honestly, I don't recall why I needed that. I think I may have just
blindly believed it was necessary.
I then looked back at the history of all the PTRACE_SETOPTIONS code we
have, and found that gdb never did copy the ptrace options before my
patch. But, when gdbserver learnt to use PTRACE_EVENT_CLONE, at
http://sourceware.org/ml/gdb-patches/2007-10/msg00547.html, it was
made to do 'ptrace (PTRACE_SETOPTIONS, new_pid, 0,
PTRACE_O_TRACECLONE)' for all new clones. Hmmm. But, GDB itself
never did that, so it can't really ever have been necessary, I
believe, otherwise GDB should have been doing it too.
(GDBserver doesn't support following forks, and so naturally doesn't
do any PTRACE_SETOPTIONS on fork children.)
So this patch removes the -I believe- unnecessary ptrace syscalls.
Tested on x86_64 Fedora 17, native/gdbserver, and on x86_64 RHEL5
native/gdbserver (Linux 2.6.18, I think a ptrace-on-utrace kernel).
No regressions.
gdb/
2013-03-22 Pedro Alves <palves@redhat.com>
* linux-nat.c (linux_child_follow_fork): Don't call
linux_enable_event_reporting.
(linux_handle_extended_wait): Don't call
linux_enable_event_reporting.
gdb/gdbserver/
2013-03-22 Pedro Alves <palves@redhat.com>
* linux-low.c (handle_extended_wait): Don't call
linux_enable_event_reporting.
This fixes the followin error when HAVE_LINUX_BTRACE is not defined:
linux-low.c:5943: error: excess elements in struct initializer
linux-low.c:5943: error: (near initialization for 'linux_target_ops')
gdb/gdbserver/ChangeLog:
* linux-low.c (linux_target_ops) [!HAVE_LINUX_BTRACE]:
Remove extraneous NULL element.
the last matched 'V' blcok in trace frame.
gdb/gdbserver:
* tracepoint.c (traceframe_read_tsv): Look for the last matched
'V' block in trace frame.
gdb/testsuite:
* gdb.trace/tsv.exp (check_tsv): New.
(top level): Save a tfile on current trace session. Call
check_tsv on live target. Load the tfile with target tfile
and call check_tsv again.
We define the following packets:
Qbtrace:bts enable branch tracing for the current thread
returns "OK" or "Enn"
Qbtrace:off disable branch tracing for the current thread
returns "OK" or "Enn"
qXfer:btrace:read read the full branch trace data for the current thread
gdb/
* target.h (enum target_object): Add TARGET_OBJECT_BTRACE.
* remote.c: Include btrace.h.
(struct btrace_target_info): New struct.
(remote_supports_btrace): New function.
(send_Qbtrace): New function.
(remote_enable_btrace): New function.
(remote_disable_btrace): New function.
(remote_teardown_btrace): New function.
(remote_read_btrace): New function.
(init_remote_ops): Add btrace ops.
(enum <unnamed>): Add btrace packets.
(struct protocol_feature remote_protocol_features[]): Add btrace packets.
(_initialize_remote): Add packet configuration for branch tracing.
gdbserver/
* target.h (struct target_ops): Add btrace ops.
(target_supports_btrace): New macro.
(target_enable_btrace): New macro.
(target_disable_btrace): New macro.
(target_read_btrace): New macro.
* gdbthread.h (struct thread_info): Add btrace field.
* server.c: Include btrace-common.h.
(handle_btrace_general_set): New function.
(handle_btrace_enable): New function.
(handle_btrace_disable): New function.
(handle_general_set): Call handle_btrace_general_set.
(handle_qxfer_btrace): New function.
(struct qxfer qxfer_packets[]): Add btrace entry.
* inferiors.c (remove_thread): Disable btrace.
* linux-low: Include linux-btrace.h.
(linux_low_enable_btrace): New function.
(linux_low_read_btrace): New function.
(linux_target_ops): Add btrace ops.
* configure.srv (i[34567]86-*-linux*): Add linux-btrace.o.
Add srv_linux_btrace=yes.
(x86_64-*-linux*): Add linux-btrace.o.
Add srv_linux_btrace=yes.
* configure.ac: Define HAVE_LINUX_BTRACE.
* config.in: Regenerated.
* configure: Regenerated.
Hafiz Abid Qadeer <abidh@codesourcery.com>
gdb/
* NEWS: Mention set and show trace-buffer-size commands.
Mention new packet.
* target.h (struct target_ops): New method
to_set_trace_buffer_size.
(target_set_trace_buffer_size): New macro.
* target.c (update_current_target): Set up new method.
* tracepoint.c (trace_buffer_size): New global.
(start_tracing): Send it to the target.
(set_trace_buffer_size): New function.
(_initialize_tracepoint): Add new setshow for trace-buffer-size.
* remote.c (remote_set_trace_buffer_size): New function.
(_initialize_remote): Use it.
(QTBuffer:size) New remote command.
(PACKET_QTBuffer_size): New enum.
(remote_protocol_features): Add an entry for
PACKET_QTBuffer_size.
gdb/gdbserver/
* tracepoint.c (trace_buffer_size): New global.
(DEFAULT_TRACE_BUFFER_SIZE): New define.
(init_trace_buffer): Change to one-argument function. Allocate
trace buffer memory.
(handle_tracepoint_general_set): Call cmd_bigqtbuffer_size to
handle QTBuffer:size packet.
(cmd_bigqtbuffer_size): New function.
(initialize_tracepoint): Call init_trace_buffer with
DEFAULT_TRACE_BUFFER_SIZE.
* server.c (handle_query): Add QTBuffer:size in the
supported packets.
gdb/doc/
* gdb.texinfo (Starting and Stopping Trace Experiments): Document
trace-buffer-size set and show commands.
(Tracepoint Packets): Document QTBuffer:size.
(General Query Packets): Document QTBuffer:size.
gdb/testsuite/
* gdb.trace/trace-buffer-size.exp: New file.
* gdb.trace/trace-buffer-size.c: New file.
* tracepoint.c (cur_action, cur_step_action): Make them unsigned.
(cmd_qtfp): Initialize cur_action and cur_step_action 0 instead
of -1.
(cmd_qtsp): Adjust condition. Do post increment.
Set cur_action and cur_step_action back to 0.
PROBLEM:
The function linux_write_memory () in linux-low.c allocates a buffer
on the stack to hold a copy of the data to be written.
register PTRACE_XFER_TYPE *buffer = (PTRACE_XFER_TYPE *)
alloca (count * sizeof (PTRACE_XFER_TYPE));
"count" is the number of bytes to be written, rounded up to the
nearest multiple of sizeof (PTRACE_XFER_TYPE) and allowing for not
being an aligned address. The function later uses
buffer[0] = ptrace (PTRACE_PEEKTEXT, pid,
(PTRACE_ARG3_TYPE) (uintptr_t) addr, 0);
The problem is that this function can be called to write zero bytes on
an aligned address, for example when receiving an X packet of length 0
(used to test if 8-bit write is supported). Under these circumstances,
count can be zero.
Since in this case, buffer[0] may never have been allocated, the stack
is corrupted and gdbserver may crash.
SOLUTION:
Writing zero bytes should always succeed. The patch below returns
successfully early if the length is zero, so avoiding the stack
corruption.
Verified on the ARC GDB 7.5.1 port.
2013-03-07 Jeremy Bennett <jeremy.bennett@embecosm.com>
PR server/15236
* linux-low.c (linux_write_memory): Return early success if LEN is
zero.
printf-like functions to avoid type related warnings on all
platforms.
(get_child_debug_event): Print dwDebugEventCode as hex since
that's how it's usually documented.
* regformats/reg-tilegx.dat (name): Change abi name to "tilegx".
* regformats/reg-tilegx32.dat: New.
gdbserver/
* Makefile.in (clean): Remove reg-tilegx.c, reg-tilegx32.c.
(reg-tilegx32.c): New rule.
* configure.srv (tilegx-*-linux*): Add reg-tilegx32.o to srv_regobj.
* linux-tile-low.c (tile_arch_setup): New function. Invoke
different register info initializer according to elf class.
(init_registers_tilgx32): New function. The tilegx32 register info
initializer.
(tile_fill_gregset): Use "uint_reg_t" to represent register size.
(tile_store_gregset): Likewise.
Addresses, as most numbers in the RSP are hex encoded, with variable
length (that just means the width isn't specified, and there's no top
cap. So they should be extracted with unpack_varlen_hex.
A couple spots in server.c are using strto(u)l, which doesn't work on
LLP64 targets.
This patch fixes it.
Tested on x86_64 Fedora 17.
2013-02-19 Pedro Alves <palves@redhat.com>
Kai Tietz <ktietz@redhat.com>
PR gdb/15161
* server.c (handle_query) <CRC check>: Use unpack_varlen_hex
instead of strtoul to extract address from packet.
(process_serial_event) <'z'>: Likewise.
The previous notes aren't being released before setting new ones.
Tested on x86_64 Fedora 17.
gdb/gdbserver/
2013-02-14 Pedro Alves <palves@redhat.com>
Plug memory leak.
* tracepoint.c (cmd_qtnotes): Free TRACING_USER_NAME,
TRACING_NOTES and TRACING_STOP_NOTE before clobbering.
An obvious use case for savestring.
Tested on x86_64 Fedora 17.
2013-02-14 Pedro Alves <palves@redhat.com>
* tracepoint.c (cmd_qtdpsrc): Use savestring.
This makes gdbserver share gdb's savestring, instead of baking its own.
Tested on x86_64 Fedora 17.
gdb/
2013-02-14 Pedro Alves <palves@redhat.com>
* utils.c (savestring): Don't #undef it. Move function to
common/common-utils.c.
* common/common-utils.c: Include gdb_string.h.
(savestring): Move here from utils.c.
* common/common-utils.h (savestring): Declare.
gdb/gdbserver/
2013-02-14 Pedro Alves <palves@redhat.com>
* tracepoint.c (save_string): Delete.
(add_tracepoint_action): Use savestring instead of save_string.
null ptr check to prevent gdbserver from crashing
Evaluating a thread local storage variable in a remote scenario crashes
gdbserver if libthread-db could not be loaded.
2013-02-12 Sanimir Agovic <sanimir.agovic@intel.com>
gdbserver/
* thread-db.c (thread_db_get_tls_address):
NULL pointer check thread_db.
testsuite/
* gdb.server/no-thread-db.exp: New file.
* gdb.server/no-thread-db.c: New file.
* gdb.server/Makefile.in (EXECUTABLES): Add no-thread-db.
* linux-aarch64-low.c (aarch64_arch_setup): Clamp
aarch64_num_wp_regs and aarch64_num_bp_regs to
AARCH64_HWP_MAX_NUM and AARCH64_HBP_MAX_NUM respectively.
... following Pedro's advice of using a temporary macro.
gdb/gdbserver/ChangeLog:
* lynx-low.c (ptrace_request_to_str): Define a temporary
macro and use it to simplify this function's implementation.
This is not strictly needed, since both GDB and GDBserver seem
to agree on the register numbering without this. But this allows
us to make sure that this is always going to be the case.
gdb/gdbserver/ChangeLog:
* configure.srv (i[34567]86-*-lynxos*): Set srv_xmlfiles.
... back to GDB. The transfer occurs when GDB sends the
'qXfer:features:read:target.xml' packet. This allows us to make
sure that GDB and GDBserver use the same register numbering.
This is important on Lynx 178, where GDB selects the rs6000:6000
architecture by default instead of the powerpc:common architecture.
gdb/gdbserver/ChangeLog:
* configure.srv (powerpc-*-lynxos*): Set srv_xmlfiles.
Before this patch, the ptid passed to lynx_resume was completely
ignored, and we used the current_inferior. This resulted in trying
to resume the inferior execution using the wrong ptid after having
received a thread create/exit event, because the inferior_ptid
was still set to the ptid prior to receiving the signal.
gdb/gdbserver/ChangeLog:
* lynx-low.c (lynx_resume): Use the resume_info parameter
to determine the ptid for the lynx_ptrace call, unless
it is equal to minus_one_ptid, in which case we use the
ptid of the current_inferior.
(lynx_wait_1): After having received a thread create/exit
event, resume the inferior's execution using the signaling
thread's ptid, rather than the old ptid.