Nowadays, when --wrapper is used, GDBserver skips extra traps/stops
in the wrapper program, and stops at the first instruction of the
program to be debugged. However, GDBserver created target description
in the first stop of inferior, and the executable of the inferior
is the wrapper program rather than the program to be debugged. In
this way, the target description can be wrong if the architectures
of wrapper program and program to be debugged are different. This
is shown by some fails in gdb.server/wrapper.exp on buildbot.
We are testing i686-linux GDB (Fedora-i686) on an x86_64-linux box
(fedora-x86-64-4) in buildbot, such configuration causes fails in
gdb.server/wrapper.exp like this:
spawn /home/gdb-buildbot-2/fedora-x86-64-4/fedora-i686/build/gdb/testsuite/../../gdb/gdbserver/gdbserver --once --wrapper env TEST=1 -- :2346 /home/gdb-buildbot-2/fedora-x86-64-4/fedora-i686/build/gdb/testsuite/outputs/gdb.server/wrapper/wrapper
Process /home/gdb-buildbot-2/fedora-x86-64-4/fedora-i686/build/gdb/testsuite/outputs/gdb.server/wrapper/wrapper created; pid = 8795
Can't debug 64-bit process with 32-bit GDBserver
Exiting
target remote localhost:2346
localhost:2346: Connection timed out.
(gdb) FAIL: gdb.server/wrapper.exp: setting breakpoint at marker
See https://sourceware.org/ml/gdb-testers/2015-q3/msg01541.html
In this case, program to be debugged ("wrapper") is 32-bit but wrapper
program ("/usr/bin/env") is 64-bit, so GDBserver gets the 64-bit
target description instead of 32-bit.
The root cause of this problem is that GDBserver creates target
description too early, and the rationale of fix could be creating
target description once the GDBserver skips extra traps and inferior
stops at the first instruction of the program we want to debug. IOW,
when GDBserver skips extra traps, the inferior's tdesc is NULL, and
mywait and its callees shouldn't use inferior's tdesc, so in this
patch, we skip code that requires register access, see changes in
linux_resume_one_lwp_throw and need_step_over_p.
In linux_low_filter_event, if target description isn't initialised and
GDBserver attached the process, we create target description immediately,
because GDBserver don't have to skip extra traps for attach, IOW, it
makes no sense to use --attach and --wrapper together. Otherwise, the
process is launched by GDBserver, we keep the status pending, and return.
After GDBserver skipped extra traps in start_inferior, we call a
target_ops hook arch_setup to initialise target description there.
gdb/gdbserver:
2015-07-24 Yao Qi <yao.qi@linaro.org>
* linux-low.c (linux_arch_setup): New function.
(linux_low_filter_event): If proc->tdesc is NULL and
proc->attached is true, call the_low_target.arch_setup.
Otherwise, keep status pending, and return.
(linux_resume_one_lwp_throw): Don't call get_pc if
thread->while_stepping isn't NULL. Don't call
get_thread_regcache if proc->tdesc is NULL.
(need_step_over_p): Return 0 if proc->tdesc is NULL.
(linux_target_ops): Install arch_setup.
* server.c (start_inferior): Call the_target->arch_setup.
* target.h (struct target_ops) <arch_setup>: New field.
(target_arch_setup): New marco.
* lynx-low.c (lynx_target_ops): Update.
* nto-low.c (nto_target_ops): Update.
* spu-low.c (spu_target_ops): Update.
* win32-low.c (win32_target_ops): Update.
This patch is to refactor function start_inferior that signal_pid
is return in one place.
gdb/gdbserver:
2015-07-24 Yao Qi <yao.qi@linaro.org>
* server.c (start_inferior): Code refactor.
When I run gdb.server/ext-restart.exp, I get the following GDB internal
error,
run^M
The program being debugged has been started already.^M
Start it from the beginning? (y or n) y^M
Sending packet: $vKill;53c5#3d...Packet received: OK^M
Packet vKill (kill) is supported^M
Sending packet: $vFile:close:6#b6...Packet received: F0^M
Sending packet: $vFile:close:3#b3...Packet received: F0^M
Starting program: /scratch/yao/gdb/build-git/x86_64/gdb/testsuite/gdb.server/ext-restart ^M
Sending packet: $QDisableRandomization:1#cf...Packet received: OK^M
Sending packet: $R0#82...Sending packet: $qC#b4...Packet received: QCp53c5.53c5^M <-- [1]
Sending packet: $qAttached:53c5#c9...Packet received: E01^M
warning: Remote failure reply: E01^M
....
0x00002aaaaaaac2d0 in ?? () from target:/lib64/ld-linux-x86-64.so.2^M
/home/yao/SourceCode/gnu/gdb/git/gdb/thread.c:88: internal-error: inferior_thread: Assertion `tp' failed.^M
A problem internal to GDB has been detected,^M
further debugging may prove unreliable.^M
Quit this debugging session? (y or n) FAIL: gdb.server/ext-restart.exp: run to main (GDB internal error)
Resyncing due to internal error.
the test is to restart the program, to make sure GDBserver handles
packet 'R' correctly. From the GDBserver output, we can see,
Remote debugging from host 127.0.0.1^M
Process /scratch/yao/gdb/build-git/x86_64/gdb/testsuite/gdb.server/ext-restart created; pid = 21445^M
GDBserver restarting^M
Process /scratch/yao/gdb/build-git/x86_64/gdb/testsuite/gdb.server/ext-restart created; pid = 21446^M
Killing process(es): 21446
we first start process 21445(0x53c5), kill it and restart a new process
21446. However, in the gdb output above [1], we can see that the reply
of qC is still the old process id rather than the new one. Looks
general_thread isn't up to date after GDBserver receives R packet.
This patch is to update general_thread after call start_inferior.
gdb/gdbserver:
2015-07-24 Yao Qi <yao.qi@linaro.org>
* server.c (process_serial_event): Set general_thread.
gdb/testsuite:
2015-07-24 Yao Qi <yao.qi@linaro.org>
* gdb.server/ext-restart.exp: New file.
Adds a new command "record btrace pt" to configure the kernel to use
Intel(R) Processor Trace instead of Branch Trace Strore.
The "record btrace" command chooses the tracing format automatically.
Intel(R) Processor Trace support requires Linux 4.1 and libipt.
gdb/
* NEWS: Announce new commands "record btrace pt" and "record pt".
Announce new options "set|show record btrace pt buffer-size".
* btrace.c: Include "rsp-low.h".
Include "inttypes.h".
(btrace_add_pc): Add forward declaration.
(pt_reclassify_insn, ftrace_add_pt, btrace_pt_readmem_callback)
(pt_translate_cpu_vendor, btrace_finalize_ftrace_pt)
(btrace_compute_ftrace_pt): New.
(btrace_compute_ftrace): Support BTRACE_FORMAT_PT.
(check_xml_btrace_version): Update version check.
(parse_xml_raw, parse_xml_btrace_pt_config_cpu)
(parse_xml_btrace_pt_raw, parse_xml_btrace_pt)
(btrace_pt_config_cpu_attributes, btrace_pt_config_children)
(btrace_pt_children): New.
(btrace_children): Add support for "pt".
(parse_xml_btrace_conf_pt, btrace_conf_pt_attributes): New.
(btrace_conf_children): Add support for "pt".
* btrace.h: Include "intel-pt.h".
(btrace_pt_error): New.
* common/btrace-common.c (btrace_format_string, btrace_data_fini)
(btrace_data_empty): Support BTRACE_FORMAT_PT.
* common/btrace-common.h (btrace_format): Add BTRACE_FORMAT_PT.
(struct btrace_config_pt): New.
(struct btrace_config)<pt>: New.
(struct btrace_data_pt_config, struct btrace_data_pt): New.
(struct btrace_data)<pt>: New.
* features/btrace-conf.dtd (btrace-conf)<pt>: New.
(pt): New.
* features/btrace.dtd (btrace)<pt>: New.
(pt, pt-config, cpu): New.
* nat/linux-btrace.c (perf_event_read, perf_event_read_all)
(perf_event_pt_event_type, kernel_supports_pt)
(linux_supports_pt): New.
(linux_supports_btrace): Support BTRACE_FORMAT_PT.
(linux_enable_bts): Free tinfo on error.
(linux_enable_pt): New.
(linux_enable_btrace): Support BTRACE_FORMAT_PT.
(linux_disable_pt): New.
(linux_disable_btrace): Support BTRACE_FORMAT_PT.
(linux_fill_btrace_pt_config, linux_read_pt): New.
(linux_read_btrace): Support BTRACE_FORMAT_PT.
* nat/linux-btrace.h (struct btrace_tinfo_pt): New.
(struct btrace_target_info)<pt>: New.
* record-btrace.c (set_record_btrace_pt_cmdlist)
(show_record_btrace_pt_cmdlist): New.
(record_btrace_print_pt_conf): New.
(record_btrace_print_conf): Support BTRACE_FORMAT_PT.
(btrace_ui_out_decode_error): Support BTRACE_FORMAT_PT.
(cmd_record_btrace_pt_start): New.
(cmd_record_btrace_start): Support BTRACE_FORMAT_PT.
(cmd_set_record_btrace_pt, cmd_show_record_btrace_pt): New.
(_initialize_record_btrace): Add new commands.
* remote.c (PACKET_Qbtrace_pt, PACKET_Qbtrace_conf_pt_size): New.
(remote_protocol_features): Add "Qbtrace:pt".
Add "Qbtrace-conf:pt:size".
(remote_supports_btrace): Support BTRACE_FORMAT_PT.
(btrace_sync_conf): Support PACKET_Qbtrace_conf_pt_size.
(remote_enable_btrace): Support BTRACE_FORMAT_PT.
(_initialize_remote): Add new commands.
gdbserver/
* linux-low.c: Include "rsp-low.h"
(linux_low_encode_pt_config, linux_low_encode_raw): New.
(linux_low_read_btrace): Support BTRACE_FORMAT_PT.
(linux_low_btrace_conf): Support BTRACE_FORMAT_PT.
(handle_btrace_enable_pt): New.
(handle_btrace_general_set): Support "pt".
(handle_btrace_conf_general_set): Support "pt:size".
doc/
* gdb.texinfo (Process Record and Replay): Spell out that variables
and registers are not available during btrace replay.
Describe the new "record btrace pt" command.
Describe the new "set|show record btrace pt buffer-size" options.
(General Query Packets): Describe the new Qbtrace:pt and
Qbtrace-conf:pt:size packets.
Expand "bts" to "Branch Trace Store".
Update the branch trace DTD.
Adapt code in remote.c to take into account addressable unit size when
reading/writing memory.
A few variables are renamed and suffixed with _bytes or _units. This
way, it's more obvious if there is any place where we add or compare
values of different kinds (which would be a mistake).
gdb/ChangeLog:
* common/rsp-low.c (needs_escaping): New.
(remote_escape_output): Add unit_size parameter. Refactor to
support multi-byte addressable units. Rename parameters.
* common/rsp-low.h (remote_escape_output): Add unit_size
parameter and rename others. Update doc.
* remote.c (align_for_efficient_write): New.
(remote_write_bytes_aux): Add unit_size parameter and use it.
Rename some variables. Update doc.
(remote_xfer_partial): Get unit size and use it.
(remote_read_bytes_1): Add unit_size parameter and use it.
Rename some variables. Update doc.
(remote_write_bytes): Same.
(remote_xfer_live_readonly_partial): Same.
(remote_read_bytes): Same.
(remote_flash_write): Update call to remote_write_bytes_aux.
(remote_write_qxfer): Update call to remote_escape_output.
(remote_search_memory): Same.
(remote_hostio_pwrite): Same.
gdb/gdbserver/ChangeLog:
* server.c (write_qxfer_response): Update call to
remote_escape_output.
While reimplementing <library-list/> I found from expat-2.0.1-11.fc15.x86_64:
warning: while parsing target library list (at line 1): Required attribute "version" of <library-list-svr4> not specified
I believe the same bug has to apply for existing FSF gdbserver but I do not
have any <library-list/> platform to test it (I did not try to build MinGW).
features/library-list.dtd:
<!ATTLIST library-list version CDATA #FIXED "1.0">
http://www.xml.com/pub/a/98/10/guide0.html?page=3 says:
In this case, the attribute is not required, but if it occurs, it must
have the specified value.
Which would suggest gdbserver is right but solib-target.c is wrong. One could
also make gdbserver explicit for the version (if those 14 bytes are not of
a concern).
gdb/ChangeLog
2015-06-10 Jan Kratochvil <jan.kratochvil@redhat.com>
* solib-target.c (library_list_start_list): Do not dereference
variable version in its initialization. Make the VERSION check handle
NULL.
(library_list_attributes): Make "version" GDB_XML_AF_OPTIONAL.
gdb/gdbserver/ChangeLog
2015-06-10 Jan Kratochvil <jan.kratochvil@redhat.com>
* server.c (handle_qxfer_libraries): Set `version' attribute for
<library-list>.
This commit implements the "vFile:setfs" packet in gdbserver.
gdb/gdbserver/ChangeLog:
* target.h (struct target_ops) <multifs_open>: New field.
<multifs_unlink>: Likewise.
<multifs_readlink>: Likewise.
* linux-low.c (nat/linux-namespaces.h): New include.
(linux_target_ops): Initialize the_target->multifs_open,
the_target->multifs_unlink and the_target->multifs_readlink.
* hostio.h (hostio_handle_new_gdb_connection): New declaration.
* hostio.c (hostio_fs_pid): New static variable.
(hostio_handle_new_gdb_connection): New function.
(handle_setfs): Likewise.
(handle_open): Use the_target->multifs_open as appropriate.
(handle_unlink): Use the_target->multifs_unlink as appropriate.
(handle_readlink): Use the_target->multifs_readlink as
appropriate.
(handle_vFile): Handle vFile:setfs packets.
* server.c (handle_query): Call hostio_handle_new_gdb_connection
after target_handle_new_gdb_connection.
This patch implements basic support for follow-fork and detach-on-fork on
extended-remote Linux targets. Only 'fork' is supported in this patch;
'vfork' support is added n a subsequent patch. This patch depends on
the previous patches in the patch series.
Sufficient extended-remote functionality has been implemented here to pass
gdb.base/multi-forks.exp, as well as gdb.base/foll-fork.exp with the
catchpoint tests commented out. Some other fork tests fail with this
patch because it doesn't provide the architecture support needed for
watchpoint inheritance or fork catchpoints.
The implementation follows the same general structure as for the native
implementation as much as possible.
This implementation includes:
* enabling fork events in linux-low.c in initialize_low and
linux_enable_extended_features
* handling fork events in gdbserver/linux-low.c:handle_extended_wait
- when a fork event occurs in gdbserver, we must do the full creation
of the new process, thread, lwp, and breakpoint lists. This is
required whether or not the new child is destined to be
detached-on-fork, because GDB will make target calls that require all
the structures. In particular we need the breakpoint lists in order
to remove the breakpoints from a detaching child. If we are not
detaching the child we will need all these structures anyway.
- as part of this event handling we store the target_waitstatus in a new
member of the parent lwp_info structure, 'waitstatus'. This
is used to store extended event information for reporting to GDB.
- handle_extended_wait is given a return value, denoting whether the
handled event should be reported to GDB. Previously it had only
handled clone events, which were never reported.
* using a new predicate in gdbserver to control handling of the fork event
(and eventually all extended events) in linux_wait_1. The predicate,
extended_event_reported, checks a target_waitstatus.kind for an
extended ptrace event.
* implementing a new RSP 'T' Stop Reply Packet stop reason: "fork", in
gdbserver/remote-utils.c and remote.c.
* implementing new target and RSP support for target_follow_fork with
target extended-remote. (The RSP components were actually defined in
patch 1, but they see their first use here).
- remote target routine remote_follow_fork, which just sends the 'D;pid'
detach packet to detach the new fork child cleanly. We can't just
call target_detach because the data structures for the forked child
have not been allocated on the host side.
Tested on x64 Ubuntu Lucid, native, remote, extended-remote.
gdb/gdbserver/ChangeLog:
* linux-low.c (handle_extended_wait): Implement return value,
rename argument 'event_child' to 'event_lwp', handle
PTRACE_EVENT_FORK, call internal_error for unrecognized event.
(linux_low_ptrace_options): New function.
(linux_low_filter_event): Call linux_low_ptrace_options,
use different argument fo linux_enable_event_reporting,
use return value from handle_extended_wait.
(extended_event_reported): New function.
(linux_wait_1): Call extended_event_reported and set
status to report fork events.
(linux_write_memory): Add pid to debug message.
(reset_lwp_ptrace_options_callback): New function.
(linux_handle_new_gdb_connection): New function.
(linux_target_ops): Initialize new structure member.
* linux-low.h (struct lwp_info) <waitstatus>: New member.
* lynx-low.c: Initialize new structure member.
* remote-utils.c (prepare_resume_reply): Implement stop reason
"fork" for "T" stop message.
* server.c (handle_query): Call handle_new_gdb_connection.
* server.h (report_fork_events): Declare global flag.
* target.h (struct target_ops) <handle_new_gdb_connection>:
New member.
(target_handle_new_gdb_connection): New macro.
* win32-low.c: Initialize new structure member.
gdb/ChangeLog:
* linux-nat.c (linux_nat_ptrace_options): New function.
(linux_init_ptrace, wait_lwp, linux_nat_filter_event):
Call linux_nat_ptrace_options and use different argument to
linux_enable_event_reporting.
(_initialize_linux_nat): Delete call to
linux_ptrace_set_additional_flags.
* nat/linux-ptrace.c (current_ptrace_options): Rename to
supported_ptrace_options.
(additional_flags): Delete variable.
(linux_check_ptrace_features): Use supported_ptrace_options.
(linux_test_for_tracesysgood, linux_test_for_tracefork):
Likewise, and remove additional_flags check.
(linux_enable_event_reporting): Change 'attached' argument to
'options'. Use supported_ptrace_options.
(ptrace_supports_feature): Change comment. Use
supported_ptrace_options.
(linux_ptrace_set_additional_flags): Delete function.
* nat/linux-ptrace.h (linux_ptrace_set_additional_flags):
Delete function prototype.
* remote.c (remote_fork_event_p): New function.
(remote_detach_pid): New function.
(remote_detach_1): Call remote_detach_pid, don't mourn inferior
if doing detach-on-fork.
(remote_follow_fork): New function.
(remote_parse_stop_reply): Handle new "T" stop reason "fork".
(remote_pid_to_str): Print "process" strings for pid/0/0 ptids.
(init_extended_remote_ops): Initialize to_follow_fork.
This patch implements a mechanism for GDB to determine whether fork
events are supported in gdbserver. This is a preparatory patch for
remote fork and exec event support.
Two new RSP packets are defined to represent fork and vfork event
support. These packets are used just like PACKET_multiprocess_feature
to denote whether the corresponding event is supported. GDB sends
fork-events+ and vfork-events+ to gdbserver to inquire about fork
event support. If the response enables these packets, then GDB
knows that gdbserver supports the corresponding events and will
enable them.
Target functions used to query for support are included along with
each new packet.
In order for gdbserver to know whether the events are supported at the
point where the qSupported packet arrives, the code in nat/linux-ptrace.c
had to be reorganized. Previously it would test for fork/exec event
support, then enable the events using the pid of the inferior. When the
qSupported packet arrives there may not be an inferior. So the mechanism
was split into two parts: a function that checks whether the events are
supported, called when gdbserver starts up, and another that enables the
events when the inferior stops for the first time.
Another gdbserver change was to add some global variables similar to
multi_process, one per new packet. These are used to control whether
the corresponding fork events are enabled. If GDB does not inquire
about the event support in the qSupported packet, then gdbserver will
not set these "report the event" flags. If the flags are not set, the
events are ignored like they were in the past. Thus, gdbserver will
never send fork event notification to an older GDB that doesn't
recognize fork events.
Tested on Ubuntu x64, native/remote/extended-remote, and as part of
subsequent patches in the series.
gdb/gdbserver/ChangeLog:
* linux-low.c (linux_supports_fork_events): New function.
(linux_supports_vfork_events): New function.
(linux_target_ops): Initialize new structure members.
(initialize_low): Call linux_check_ptrace_features.
* lynx-low.c (lynx_target_ops): Initialize new structure
members.
* server.c (report_fork_events, report_vfork_events):
New global flags.
(handle_query): Add new features to qSupported packet and
response.
(captured_main): Initialize new global variables.
* target.h (struct target_ops) <supports_fork_events>:
New member.
<supports_vfork_events>: New member.
(target_supports_fork_events): New macro.
(target_supports_vfork_events): New macro.
* win32-low.c (win32_target_ops): Initialize new structure
members.
gdb/ChangeLog:
* nat/linux-ptrace.c (linux_check_ptrace_features): Change
from static to extern.
* nat/linux-ptrace.h (linux_check_ptrace_features): Declare.
* remote.c (anonymous enum): <PACKET_fork_event_feature,
* PACKET_vfork_event_feature>: New enumeration constants.
(remote_protocol_features): Add table entries for new packets.
(remote_query_supported): Add new feature queries to qSupported
packet.
(_initialize_remote): Exempt new packets from the requirement
to have 'set remote' commands.
This commit allows GDB to determine filenames of main executables
when debugging using remote stubs without multiprocess extensions.
The qXfer:exec-file:read packet is extended to allow an empty
annex, with the meaning that the remote stub should supply the
filename of whatever it thinks is the current process.
gdb/ChangeLog:
* remote.c (remote_add_inferior): Call exec_file_locate_attach
for fake PIDs as well as real ones.
(remote_pid_to_exec_file): Send empty annex if PID is fake.
gdb/doc/ChangeLog:
* gdb.texinfo (General Query Packets): Document
qXfer:exec-file:read with empty annex.
gdb/gdbserver/ChangeLog:
* server.c (handle_qxfer_exec_file): Use current process
if annex is empty.
GDBserver steps over breakpoint if the condition is false, but if target
doesn't support hardware single step, the step over is very simple, if
not incorrect, in linux-arm-low.c:
/* We only place breakpoints in empty marker functions, and thread locking
is outside of the function. So rather than importing software single-step,
we can just run until exit. */
static CORE_ADDR
arm_reinsert_addr (void)
{
struct regcache *regcache = get_thread_regcache (current_thread, 1);
unsigned long pc;
collect_register_by_name (regcache, "lr", &pc);
return pc;
}
and linux-mips-low.c does the same. GDBserver sets a breakpoint at the
return address of the current function, resume and wait the program hits
the breakpoint in order to achieve "breakpoint step over". What if
program hits other user breakponits during this "step over"?
It is worse if the arm/thumb interworking is considered. Nowadays,
GDBserver arm backend unconditionally inserts arm breakpoint,
/* Define an ARM-mode breakpoint; we only set breakpoints in the C
library, which is most likely to be ARM. If the kernel supports
clone events, we will never insert a breakpoint, so even a Thumb
C library will work; so will mixing EABI/non-EABI gdbserver and
application. */
(const unsigned char *) &arm_breakpoint,
(const unsigned char *) &arm_eabi_breakpoint,
note that the comments are no longer valid as C library can be compiled
in thumb mode.
When GDBserver steps over a breakpoint in arm mode function, which
returns to thumb mode, GDBserver will insert arm mode breakpoint by
mistake and the program will crash. GDBserver alone is unable to
determine the arm/thumb mode given a PC address. See how GDB does
it in arm-tdep.c:arm_pc_is_thumb.
After thinking about how to teach GDBserver inserting right breakpoint
(arm or thumb) for a while, I reconsider it from a different direction
that it may be unreasonable to run target-side conditional breakpoint for
targets without hardware single step. Pedro also pointed this out here
https://sourceware.org/ml/gdb-patches/2015-04/msg00337.html
This patch is to add a new target_ops hook
supports_conditional_breakpoints, and only reply
";ConditionalBreakpoints+" if it is true. On linux targets,
supports_conditional_breakpoints returns true if target has hardware
single step, on other targets, (win32, lynx, nto, spu), set it to NULL,
because conditional breakpoint is a linux-specific feature.
gdb/gdbserver:
2015-05-08 Yao Qi <yao.qi@linaro.org>
* linux-low.c (linux_supports_conditional_breakpoints): New
function.
(linux_target_ops): Install new target method.
* lynx-low.c (lynx_target_ops): Install NULL hook for
supports_conditional_breakpoints.
* nto-low.c (nto_target_ops): Likewise.
* spu-low.c (spu_target_ops): Likewise.
* win32-low.c (win32_target_ops): Likewise.
* server.c (handle_query): Check
target_supports_conditional_breakpoints.
* target.h (struct target_ops) <supports_conditional_breakpoints>:
New field.
(target_supports_conditional_breakpoints): New macro.
$ ./gdbserver :1234 blah
Process blah created; pid = 16471
Cannot exec blah: No such file or directory.
Child exited with status 127
Killing process(es): 16471
../../../../src/binutils-gdb/gdb/gdbserver/linux-low.c:920: A problem internal to GDBserver has been detected.
kill_wait_lwp: Assertion `res > 0' failed.
GDBserver shouldn't even be trying to kill that process. GDBserver
kills or detaches from all processes on exit, and due to a missing
mourn_inferior call, GDBserver tries to kill the process that it had
already seen exit.
Tested on x86_64 Fedora 20. New test included. I emulated what
Windows outputs by hacking an error call in linux_create_inferior.
gdb/gdbserver/ChangeLog:
2015-05-06 Pedro Alves <palves@redhat.com>
PR server/18081
* server.c (start_inferior): If the process exits, mourn it.
gdb/testsuite/ChangeLog:
2015-05-06 Pedro Alves <palves@redhat.com>
PR server/18081
* gdb.server/non-existing-program.exp: New file.
--attach/--multi are currently only mentioned on the usage info first
lines, the meaning of PROG is completely absent and the COMM text does
not mention '-/stdio'.
A few options are missing:
. --disable-randomization / --no-disable-randomization is not mentioned.
Although the manual has a comment saying these are superceded by
QDisableRandomization, that only makes sense for "run" in
extended-remote mode. When we start gdbserver passing it a PROG,
--disable-randomization / --no-disable-randomization do take effect.
So I think we should document these.
. We show --debug / --remote-debug, so might as well show --disable-packet too.
GDB's --help has this "For more information, consult the GDB manual"
blurb that is missing in GDBserver's --help.
Then shuffle things around a bit into "Operating modes", "Other
options" and "Debug options" sections, similarly to GDB's --help
structure.
Before:
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
$ ./gdbserver/gdbserver --help
Usage: gdbserver [OPTIONS] COMM PROG [ARGS ...]
gdbserver [OPTIONS] --attach COMM PID
gdbserver [OPTIONS] --multi COMM
COMM may either be a tty device (for serial debugging), or
HOST:PORT to listen for a TCP connection.
Options:
--debug Enable general debugging output.
--debug-format=opt1[,opt2,...]
Specify extra content in debugging output.
Options:
all
none
timestamp
--remote-debug Enable remote protocol debugging output.
--version Display version information and exit.
--wrapper WRAPPER -- Run WRAPPER to start new programs.
--once Exit after the first connection has closed.
Report bugs to "<http://www.gnu.org/software/gdb/bugs/>".
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
After:
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
$ ./gdbserver/gdbserver --help
Usage: gdbserver [OPTIONS] COMM PROG [ARGS ...]
gdbserver [OPTIONS] --attach COMM PID
gdbserver [OPTIONS] --multi COMM
COMM may either be a tty device (for serial debugging),
HOST:PORT to listen for a TCP connection, or '-' or 'stdio' to use
stdin/stdout of gdbserver.
PROG is the executable program. ARGS are arguments passed to inferior.
PID is the process ID to attach to, when --attach is specified.
Operating modes:
--attach Attach to running process PID.
--multi Start server without a specific program, and
only quit when explicitly commanded.
--once Exit after the first connection has closed.
--help Print this message and then exit.
--version Display version information and exit.
Other options:
--wrapper WRAPPER -- Run WRAPPER to start new programs.
--disable-randomization
Run PROG with address space randomization disabled.
--no-disable-randomization
Don't disable address space randomization when
starting PROG.
Debug options:
--debug Enable general debugging output.
--debug-format=opt1[,opt2,...]
Specify extra content in debugging output.
Options:
all
none
timestamp
--remote-debug Enable remote protocol debugging output.
--disable-packet=opt1[,opt2,...]
Disable support for RSP packets or features.
Options:
vCont, Tthread, qC, qfThreadInfo and
threads (disable all threading packets).
For more information, consult the GDB manual (available as on-line
info or a printed manual).
Report bugs to "<http://www.gnu.org/software/gdb/bugs/>".
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
gdb/gdbserver/ChangeLog:
2015-04-01 Pedro Alves <palves@redhat.com>
Cleber Rosa <crosa@redhat.com>
* server.c (gdbserver_usage): Reorganize and extend the usage
message.
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.
gdb/doc/ChangeLog:
* gdb.texinfo (General Query Packets): Remove documentation
for now-removed vFile:fstat qSupported features.
gdb/gdbserver/ChangeLog:
* server.c (handle_query): Do not report vFile:fstat as supported.
This patch splits the TRY_CATCH macro into three, so that we go from
this:
~~~
volatile gdb_exception ex;
TRY_CATCH (ex, RETURN_MASK_ERROR)
{
}
if (ex.reason < 0)
{
}
~~~
to this:
~~~
TRY
{
}
CATCH (ex, RETURN_MASK_ERROR)
{
}
END_CATCH
~~~
Thus, we'll be getting rid of the local volatile exception object, and
declaring the caught exception in the catch block.
This allows reimplementing TRY/CATCH in terms of C++ exceptions when
building in C++ mode, while still allowing to build GDB in C mode
(using setjmp/longjmp), as a transition step.
TBC, after this patch, is it _not_ valid to have code between the TRY
and the CATCH blocks, like:
TRY
{
}
// some code here.
CATCH (ex, RETURN_MASK_ERROR)
{
}
END_CATCH
Just like it isn't valid to do that with C++'s native try/catch.
By switching to creating the exception object inside the CATCH block
scope, we can get rid of all the explicitly allocated volatile
exception objects all over the tree, and map the CATCH block more
directly to C++'s catch blocks.
The majority of the TRY_CATCH -> TRY+CATCH+END_CATCH conversion was
done with a script, rerun from scratch at every rebase, no manual
editing involved. After the mechanical conversion, a few places
needed manual intervention, to fix preexisting cases where we were
using the exception object outside of the TRY_CATCH block, and cases
where we were using "else" after a 'if (ex.reason) < 0)' [a CATCH
after this patch]. The result was folded into this patch so that GDB
still builds at each incremental step.
END_CATCH is necessary for two reasons:
First, because we name the exception object in the CATCH block, which
requires creating a scope, which in turn must be closed somewhere.
Declaring the exception variable in the initializer field of a for
block, like:
#define CATCH(EXCEPTION, mask) \
for (struct gdb_exception EXCEPTION; \
exceptions_state_mc_catch (&EXCEPTION, MASK); \
EXCEPTION = exception_none)
would avoid needing END_CATCH, but alas, in C mode, we build with C90,
which doesn't allow mixed declarations and code.
Second, because when TRY/CATCH are wired to real C++ try/catch, as
long as we need to handle cleanup chains, even if there's no CATCH
block that wants to catch the exception, we need for stop at every
frame in the unwind chain and run cleanups, then rethrow. That will
be done in END_CATCH.
After we require C++, we'll still need TRY/CATCH/END_CATCH until
cleanups are completely phased out -- TRY/CATCH in C++ mode will
save/restore the current cleanup chain, like in C mode, and END_CATCH
catches otherwise uncaugh exceptions, runs cleanups and rethrows, so
that C++ cleanups and exceptions can coexist.
IMO, this still makes the TRY/CATCH code look a bit more like a
newcomer would expect, so IMO worth it even if we weren't considering
C++.
gdb/ChangeLog.
2015-03-07 Pedro Alves <palves@redhat.com>
* common/common-exceptions.c (struct catcher) <exception>: No
longer a pointer to volatile exception. Now an exception value.
<mask>: Delete field.
(exceptions_state_mc_init): Remove all parameters. Adjust.
(exceptions_state_mc): No longer pop the catcher here.
(exceptions_state_mc_catch): New function.
(throw_exception): Adjust.
* common/common-exceptions.h (exceptions_state_mc_init): Remove
all parameters.
(exceptions_state_mc_catch): Declare.
(TRY_CATCH): Rename to ...
(TRY): ... this. Remove EXCEPTION and MASK parameters.
(CATCH, END_CATCH): New.
All callers adjusted.
gdb/gdbserver/ChangeLog:
2015-03-07 Pedro Alves <palves@redhat.com>
Adjust all callers of TRY_CATCH to use TRY/CATCH/END_CATCH
instead.
This commit introduces a new inline common function "startswith"
which takes two string arguments and returns nonzero if the first
string starts with the second. It also updates the 295 places
where this logic was written out longhand to use the new function.
gdb/ChangeLog:
* common/common-utils.h (startswith): New inline function.
All places where this logic was used updated to use the above.
This patch teaches the core of gdbserver about the new "swbreak" and
"hwbreak" stop reasons, and adds the necessary hooks a backend needs
to implement to support the feature.
gdb/gdbserver/ChangeLog:
2015-03-04 Pedro Alves <palves@redhat.com>
* remote-utils.c (prepare_resume_reply): Report swbreak/hbreak.
* server.c (swbreak_feature, hwbreak_feature): New globals.
(handle_query) <qSupported>: Handle "swbreak+" and "hwbreak+".
(captured_main): Clear swbreak_feature and hwbreak_feature.
* server.h (swbreak_feature, hwbreak_feature): Declare.
* target.h (struct target_ops) <stopped_by_sw_breakpoint,
supports_stopped_by_sw_breakpoint, stopped_by_hw_breakpoint,
supports_stopped_by_hw_breakpoint>: New fields.
(target_supports_stopped_by_sw_breakpoint)
(target_stopped_by_sw_breakpoint)
(target_supports_stopped_by_hw_breakpoint)
(target_stopped_by_hw_breakpoint): Declare.
Allow the size of the branch trace ring buffer to be defined by the
user. The specified buffer size will be used when BTS tracing is
enabled for new threads.
The obtained buffer size may differ from the requested size. The
actual buffer size for the current thread is shown in the "info record"
command.
Bigger buffers mean longer traces, but also longer processing time.
2015-02-09 Markus Metzger <markus.t.metzger@intel.com>
* btrace.c (parse_xml_btrace_conf_bts): Add size.
(btrace_conf_bts_attributes): New.
(btrace_conf_children): Add attributes.
* common/btrace-common.h (btrace_config_bts): New.
(btrace_config)<bts>: New.
(btrace_config): Update comment.
* nat/linux-btrace.c (linux_enable_btrace, linux_enable_bts):
Use config.
* features/btrace-conf.dtd: Increment version. Add size
attribute to bts element.
* record-btrace.c (set_record_btrace_bts_cmdlist,
show_record_btrace_bts_cmdlist): New.
(record_btrace_adjust_size, record_btrace_print_bts_conf,
record_btrace_print_conf, cmd_set_record_btrace_bts,
cmd_show_record_btrace_bts): New.
(record_btrace_info): Call record_btrace_print_conf.
(_initialize_record_btrace): Add commands.
* remote.c: Add PACKET_Qbtrace_conf_bts_size enum.
(remote_protocol_features): Add Qbtrace-conf:bts:size packet.
(btrace_sync_conf): Synchronize bts size.
(_initialize_remote): Add Qbtrace-conf:bts:size packet.
* NEWS: Announce new commands and new packets.
doc/
* gdb.texinfo (Branch Trace Configuration Format): Add size.
(Process Record and Replay): Describe new set|show commands.
(General Query Packets): Describe Qbtrace-conf:bts:size packet.
testsuite/
* gdb.btrace/buffer-size: New.
gdbserver/
* linux-low.c (linux_low_btrace_conf): Print size.
* server.c (handle_btrace_conf_general_set): New.
(hanle_general_set): Call handle_btrace_conf_general_set.
(handle_query): Report Qbtrace-conf:bts:size as supported.
Add a struct to describe the branch trace configuration and use it for
enabling branch tracing.
The user will be able to set configuration fields for each tracing format
to be used for new threads.
The actual configuration that is active for a given thread will be shown
in the "info record" command.
At the moment, the configuration struct only contains a format field
that is set to the only available format.
The format is the only configuration option that can not be set via set
commands. It is given as argument to the "record btrace" command when
starting recording.
2015-02-09 Markus Metzger <markus.t.metzger@intel.com>
* Makefile.in (XMLFILES): Add btrace-conf.dtd.
* x86-linux-nat.c (x86_linux_enable_btrace): Update parameters.
(x86_linux_btrace_conf): New.
(x86_linux_create_target): Initialize to_btrace_conf.
* nat/linux-btrace.c (linux_enable_btrace): Update parameters.
Check format. Split into this and ...
(linux_enable_bts): ... this.
(linux_btrace_conf): New.
(perf_event_skip_record): Renamed into ...
(perf_event_skip_bts_record): ... this. Updated users.
(linux_disable_btrace): Split into this and ...
(linux_disable_bts): ... this.
(linux_read_btrace): Check format.
* nat/linux-btrace.h (linux_enable_btrace): Update parameters.
(linux_btrace_conf): New.
(btrace_target_info)<ptid>: Moved.
(btrace_target_info)<conf>: New.
(btrace_target_info): Split into this and ...
(btrace_tinfo_bts): ... this. Updated users.
* btrace.c (btrace_enable): Update parameters.
(btrace_conf, parse_xml_btrace_conf_bts, parse_xml_btrace_conf)
(btrace_conf_children, btrace_conf_attributes)
(btrace_conf_elements): New.
* btrace.h (btrace_enable): Update parameters.
(btrace_conf, parse_xml_btrace_conf): New.
* common/btrace-common.h (btrace_config): New.
* feature/btrace-conf.dtd: New.
* record-btrace.c (record_btrace_conf): New.
(record_btrace_cmdlist): New.
(record_btrace_enable_warn, record_btrace_open): Pass
&record_btrace_conf.
(record_btrace_info): Print recording format.
(cmd_record_btrace_bts_start): New.
(cmd_record_btrace_start): Call cmd_record_btrace_bts_start.
(_initialize_record_btrace): Add "record btrace bts" subcommand.
Add "record bts" alias command.
* remote.c (remote_state)<btrace_config>: New.
(remote_btrace_reset, PACKET_qXfer_btrace_conf): New.
(remote_protocol_features): Add qXfer:btrace-conf:read.
(remote_open_1): Call remote_btrace_reset.
(remote_xfer_partial): Handle TARGET_OBJECT_BTRACE_CONF.
(btrace_target_info)<conf>: New.
(btrace_sync_conf, btrace_read_config): New.
(remote_enable_btrace): Update parameters. Call btrace_sync_conf and
btrace_read_conf.
(remote_btrace_conf): New.
(init_remote_ops): Initialize to_btrace_conf.
(_initialize_remote): Add qXfer:btrace-conf packet.
* target.c (target_enable_btrace): Update parameters.
(target_btrace_conf): New.
* target.h (target_enable_btrace): Update parameters.
(target_btrace_conf): New.
(target_object)<TARGET_OBJECT_BTRACE_CONF>: New.
(target_ops)<to_enable_btrace>: Update parameters and comment.
(target_ops)<to_btrace_conf>: New.
* target-delegates: Regenerate.
* target-debug.h (target_debug_print_const_struct_btrace_config_p)
(target_debug_print_const_struct_btrace_target_info_p): New.
NEWS: Announce new command and new packet.
doc/
* gdb.texinfo (Process Record and Replay): Describe the "record
btrace bts" command.
(General Query Packets): Describe qXfer:btrace-conf:read packet.
(Branch Trace Configuration Format): New.
gdbserver/
* linux-low.c (linux_low_enable_btrace): Update parameters.
(linux_low_btrace_conf): New.
(linux_target_ops)<to_btrace_conf>: Initialize.
* server.c (current_btrace_conf): New.
(handle_btrace_enable): Rename to ...
(handle_btrace_enable_bts): ... this. Pass ¤t_btrace_conf
to target_enable_btrace. Update comment. Update users.
(handle_qxfer_btrace_conf): New.
(qxfer_packets): Add btrace-conf entry.
(handle_query): Report qXfer:btrace-conf:read as supported packet.
* target.h (target_ops)<enable_btrace>: Update parameters and comment.
(target_ops)<read_btrace_conf>: New.
(target_enable_btrace): Update parameters.
(target_read_btrace_conf): New.
testsuite/
* gdb.btrace/delta.exp: Update "info record" output.
* gdb.btrace/enable.exp: Update "info record" output.
* gdb.btrace/finish.exp: Update "info record" output.
* gdb.btrace/instruction_history.exp: Update "info record" output.
* gdb.btrace/next.exp: Update "info record" output.
* gdb.btrace/nexti.exp: Update "info record" output.
* gdb.btrace/step.exp: Update "info record" output.
* gdb.btrace/stepi.exp: Update "info record" output.
* gdb.btrace/nohist.exp: Update "info record" output.
Add a format argument to the various supports_btrace functions to check
for support of a specific btrace format. This is to prepare for a new
format.
Removed two redundant calls. The check will be made in the subsequent
btrace_enable call.
2015-02-09 Markus Metzger <markus.t.metzger@intel.com>
* btrace.c (btrace_enable): Pass BTRACE_FORMAT_BTS.
* record-btrace.c (record_btrace_open): Remove call to
target_supports_btrace.
* remote.c (remote_supports_btrace): Update parameters.
* target.c (target_supports_btrace): Update parameters.
* target.h (to_supports_btrace, target_supports_btrace): Update
parameters.
* target-delegates.c: Regenerate.
* target-debug.h (target_debug_print_enum_btrace_format): New.
* nat/linux-btrace.c
(kernel_supports_btrace): Rename into ...
(kernel_supports_bts): ... this. Update users. Update warning text.
(intel_supports_btrace): Rename into ...
(intel_supports_bts): ... this. Update users.
(cpu_supports_btrace): Rename into ...
(cpu_supports_bts): ... this. Update users.
(linux_supports_btrace): Update parameters. Split into this and ...
(linux_supports_bts): ... this.
* nat/linux-btrace.h (linux_supports_btrace): Update parameters.
gdbserver/
* server.c (handle_btrace_general_set): Remove call to
target_supports_btrace.
(supported_btrace_packets): New.
(handle_query): Call supported_btrace_packets.
* target.h: include btrace-common.h.
(btrace_target_info): Removed.
(supports_btrace, target_supports_btrace): Update parameters.
As no place in the backends check cont_thread anymore, we can stop
setting and clearing it in places that resume the target and wait for
events. Instead simply clear it whenever a new GDB connects.
gdb/gdbserver/
2014-11-12 Pedro Alves <palves@redhat.com>
* server.c (cont_thread): Update comment.
(start_inferior, attach_inferior): No longer clear cont_thread.
(handle_v_cont): No longer set cont_thread.
(captured_main): Clear cont_thread each time a GDB connects.
GDB has a function named "current_inferior" and gdbserver has a global
variable named "current_inferior", but the two are not equivalent;
indeed, gdbserver does not have any real equivalent of what GDB calls
an inferior. What gdbserver's "current_inferior" is actually pointing
to is a structure describing the current thread. This commit renames
current_inferior as current_thread in gdbserver to clarify this. It
also renames the function "set_desired_inferior" to "set_desired_thread"
and renames various local variables from foo_inferior to foo_thread.
gdb/gdbserver/ChangeLog:
* inferiors.h (current_inferior): Renamed as...
(current_thread): New variable. All uses updated.
* linux-low.c (get_pc): Renamed saved_inferior as saved_thread.
(maybe_move_out_of_jump_pad): Likewise.
(cancel_breakpoint): Likewise.
(linux_low_filter_event): Likewise.
(wait_for_sigstop): Likewise.
(linux_resume_one_lwp): Likewise.
(need_step_over_p): Likewise.
(start_step_over): Likewise.
(linux_stabilize_threads): Renamed save_inferior as saved_thread.
* linux-x86-low.c (x86_linux_update_xmltarget): Likewise.
* proc-service.c (ps_lgetregs): Renamed reg_inferior as reg_thread
and save_inferior as saved_thread.
* regcache.c (get_thread_regcache): Renamed saved_inferior as
saved_thread.
(regcache_invalidate_thread): Likewise.
* remote-utils.c (prepare_resume_reply): Likewise.
* thread-db.c (thread_db_get_tls_address): Likewise.
(disable_thread_event_reporting): Likewise.
(remove_thread_event_breakpoints): Likewise.
* tracepoint.c (gdb_agent_about_to_close): Renamed save_inferior
as saved_thread.
* target.h (set_desired_inferior): Renamed as...
(set_desired_thread): New declaration. All uses updated.
* server.c (myresume): Updated comment to reference thread instead
of inferior.
(handle_serial_event): Likewise.
(handle_target_event): Likewise.
This commit adds a new global flag show_debug_regs to common-debug.h
to replace the flag debug_hw_points used by gdbserver and by the
Linux x86 and AArch64 ports, and to replace the flag maint_show_dr
used by the Linux MIPS port.
Note that some debug printing in the AArch64 port was enabled only if
debug_hw_points > 1 but no way to set debug_hw_points to values other
than 0 and 1 was provided; that code was effectively dead. This
commit enables all debug printing if show_debug_regs is nonzero, so
the AArch64 output will be more verbose than previously.
gdb/ChangeLog:
* common/common-debug.h (show_debug_regs): Declare.
* common/common-debug.c (show_debug_regs): Define.
* aarch64-linux-nat.c (debug_hw_points): Don't define. Replace
all uses with show_debug_regs. Replace all uses that considered
debug_hw_points as a multi-value integer with straight boolean
uses.
* x86-nat.c (debug_hw_points): Don't define. Replace all uses
with show_debug_regs.
* nat/x86-dregs.c (debug_hw_points): Don't declare. Replace
all uses with show_debug_regs.
* mips-linux-nat.c (maint_show_dr): Don't define. Replace all
uses with show_debug_regs.
gdb/gdbserver/ChangeLog:
* server.h (debug_hw_points): Don't declare.
* server.c (debug_hw_points): Don't define. Replace all uses
with show_debug_regs.
* linux-aarch64-low.c (debug_hw_points): Don't define. Replace
all uses with show_debug_regs.
This commit replaces the hacky "exception" system in gdbserver with
the exceptions and cleanups subsystem from GDB.
Only the catch/cleanup code in what was "main" has been updated to
use the new system. Other parts of gdbserver can now be converted
to use TRY_CATCH and cleanups on an as-needed basis.
A side-effect of this commit is that some error messages will change
slightly, and in cases with multiple errors the error messages will
be printed in a different order.
gdb/gdbserver/ChangeLog:
* server.h (setjmp.h): Do not include.
(toplevel): Do not declare.
(common-exceptions.h): Include.
(cleanups.h): Likewise.
* server.c (toplevel): Do not define.
(exit_code): New static global.
(detach_or_kill_for_exit_cleanup): New function.
(main): New function. Original main renamed to...
(captured_main): New function.
* utils.c (verror) [!IN_PROCESS_AGENT]: Use throw_verror.
This builds a libiberty just for gdbserver and arranges for gdbserver
to use it. I've tripped across the lack of libiberty in gdbserver at
least once, and I have seen other threads where it would have been
useful.
2014-06-12 Tom Tromey <tromey@redhat.com>
* debug.c (debug_printf): Remove HAVE_GETTIMEOFDAY checks.
* server.c (monitor_show_help): Remove HAVE_GETTIMEOFDAY check.
(parse_debug_format_options): Likewise.
(gdbserver_usage): Likewise.
* Makefile.in (LIBIBERTY_BUILDDIR, LIBIBERTY): New variables.
(SUBDIRS, REQUIRED_SUBDIRS): Add libiberty.
(gdbserver$(EXEEXT), gdbreplay$(EXEEXT)): Depend on and link
against libiberty.
($(LIBGNU)): Depend on libiberty.
(all-lib): Recurse into all subdirs.
(install-only): Invoke "install" target in subdirs.
(vasprintf.o, vsnprintf.o, safe-ctype.o, lbasename.o): Remove
targets.
* configure: Rebuild.
* configure.ac: Add ACX_CONFIGURE_DIR for libiberty. Don't check
for vasprintf, vsnprintf, or gettimeofday.
* configure.srv: Don't add safe-ctype.o or lbasename.o to
srv_tgtobj.
If GDB decides to change the breakpoint's conditions or commands,
it'll reinsert the same breakpoint again, with the new options
attached, without deleting the previous breakpoint. E.g.,
(gdb) set breakpoint always-inserted on
(gdb) b main if 0
Breakpoint 1 at 0x400594: file foo.c, line 21.
Sending packet: $Z0,400594,1;X3,220027#68...Packet received: OK
(gdb) b main
Breakpoint 15 at 0x400594: file foo.c, line 21.
Sending packet: $Z0,400594,1#49...Packet received: OK
GDBserver understands this and deletes the breakpoint's previous
conditions. But, it forgets to delete the previous commands.
gdb/gdbserver/
2014-06-02 Pedro Alves <palves@redhat.com>
* ax.c (gdb_free_agent_expr): New function.
* ax.h (gdb_free_agent_expr): New declaration.
* mem-break.c (delete_gdb_breakpoint_1): Also clear the commands
list.
(clear_breakpoint_conditions, clear_breakpoint_commands): Make
static.
(clear_breakpoint_conditions_and_commands): New function.
* mem-break.h (clear_breakpoint_conditions): Delete declaration.
(clear_breakpoint_conditions_and_commands): New declaration.
This patch fixes hardware breakpoint regressions exposed by my fix for
"PR breakpoints/7143 - Watchpoint does not trigger when first set", at
https://sourceware.org/ml/gdb-patches/2014-03/msg00167.html
The testsuite caught them on Linux/x86_64, at least. gdb.sum:
gdb.sum:
FAIL: gdb.base/hbreak2.exp: next over recursive call
FAIL: gdb.base/hbreak2.exp: backtrace from factorial(5.1)
FAIL: gdb.base/hbreak2.exp: continue until exit at recursive next test
gdb.log:
(gdb) next
Program received signal SIGTRAP, Trace/breakpoint trap.
factorial (value=4) at ../../../src/gdb/testsuite/gdb.base/break.c:113
113 if (value > 1) { /* set breakpoint 7 here */
(gdb) FAIL: gdb.base/hbreak2.exp: next over recursive call
Actually, that patch just exposed a latent issue to "breakpoints
always-inserted off" mode, not really caused it. After that patch,
GDB no longer removes breakpoints at each internal event, thus making
some scenarios behave like breakpoint always-inserted on. The bug is
easy to trigger with always-inserted on.
The issue is that since the target-side breakpoint conditions support,
if the stub/server supports evaluating breakpoint conditions on the
target side, then GDB is sending duplicate Zx packets to the target
without removing them before, and GDBserver is not really expecting
that for Z packets other than Z0/z0. E.g., with "set breakpoint
always-inserted on" and "set debug remote 1":
(gdb) b main
Sending packet: $m410943,1#ff...Packet received: 48
Breakpoint 4 at 0x410943: file ../../../src/gdb/gdbserver/server.c, line 3028.
Sending packet: $Z0,410943,1#48...Packet received: OK
^^^^^^^^^^^^
(gdb) b main
Note: breakpoint 4 also set at pc 0x410943.
Sending packet: $m410943,1#ff...Packet received: 48
Breakpoint 5 at 0x410943: file ../../../src/gdb/gdbserver/server.c, line 3028.
Sending packet: $Z0,410943,1#48...Packet received: OK
^^^^^^^^^^^^
(gdb) b main
Note: breakpoints 4 and 5 also set at pc 0x410943.
Sending packet: $m410943,1#ff...Packet received: 48
Breakpoint 6 at 0x410943: file ../../../src/gdb/gdbserver/server.c, line 3028.
Sending packet: $Z0,410943,1#48...Packet received: OK
^^^^^^^^^^^^
(gdb) del
Delete all breakpoints? (y or n) y
Sending packet: $Z0,410943,1#48...Packet received: OK
Sending packet: $Z0,410943,1#48...Packet received: OK
Sending packet: $z0,410943,1#68...Packet received: OK
And for Z1, similarly:
(gdb) hbreak main
Sending packet: $m410943,1#ff...Packet received: 48
Hardware assisted breakpoint 4 at 0x410943: file ../../../src/gdb/gdbserver/server.c, line 3028.
Sending packet: $Z1,410943,1#49...Packet received: OK
^^^^^^^^^^^^
Packet Z1 (hardware-breakpoint) is supported
(gdb) hbreak main
Note: breakpoint 4 also set at pc 0x410943.
Sending packet: $m410943,1#ff...Packet received: 48
Hardware assisted breakpoint 5 at 0x410943: file ../../../src/gdb/gdbserver/server.c, line 3028.
Sending packet: $Z1,410943,1#49...Packet received: OK
^^^^^^^^^^^^
(gdb) hbreak main
Note: breakpoints 4 and 5 also set at pc 0x410943.
Sending packet: $m410943,1#ff...Packet received: 48
Hardware assisted breakpoint 6 at 0x410943: file ../../../src/gdb/gdbserver/server.c, line 3028.
Sending packet: $Z1,410943,1#49...Packet received: OK
^^^^^^^^^^^^
(gdb) del
Delete all breakpoints? (y or n) y
Sending packet: $Z1,410943,1#49...Packet received: OK
^^^^^^^^^^^^
Sending packet: $Z1,410943,1#49...Packet received: OK
^^^^^^^^^^^^
Sending packet: $z1,410943,1#69...Packet received: OK
^^^^^^^^^^^^
So GDB sent a bunch of Z1 packets, and then when finally removing the
breakpoint, only one z1 packet was sent. On the GDBserver side (with
monitor set debug-hw-points 1), in the Z1 case, we see:
$ ./gdbserver :9999 ./gdbserver
Process ./gdbserver created; pid = 8629
Listening on port 9999
Remote debugging from host 127.0.0.1
insert_watchpoint (addr=410943, len=1, type=instruction-execute):
CONTROL (DR7): 00000101 STATUS (DR6): 00000000
DR0: addr=0x410943, ref.count=1 DR1: addr=0x0, ref.count=0
DR2: addr=0x0, ref.count=0 DR3: addr=0x0, ref.count=0
insert_watchpoint (addr=410943, len=1, type=instruction-execute):
CONTROL (DR7): 00000101 STATUS (DR6): 00000000
DR0: addr=0x410943, ref.count=2 DR1: addr=0x0, ref.count=0
DR2: addr=0x0, ref.count=0 DR3: addr=0x0, ref.count=0
insert_watchpoint (addr=410943, len=1, type=instruction-execute):
CONTROL (DR7): 00000101 STATUS (DR6): 00000000
DR0: addr=0x410943, ref.count=3 DR1: addr=0x0, ref.count=0
DR2: addr=0x0, ref.count=0 DR3: addr=0x0, ref.count=0
insert_watchpoint (addr=410943, len=1, type=instruction-execute):
CONTROL (DR7): 00000101 STATUS (DR6): 00000000
DR0: addr=0x410943, ref.count=4 DR1: addr=0x0, ref.count=0
DR2: addr=0x0, ref.count=0 DR3: addr=0x0, ref.count=0
insert_watchpoint (addr=410943, len=1, type=instruction-execute):
CONTROL (DR7): 00000101 STATUS (DR6): 00000000
DR0: addr=0x410943, ref.count=5 DR1: addr=0x0, ref.count=0
DR2: addr=0x0, ref.count=0 DR3: addr=0x0, ref.count=0
remove_watchpoint (addr=410943, len=1, type=instruction-execute):
CONTROL (DR7): 00000101 STATUS (DR6): 00000000
DR0: addr=0x410943, ref.count=4 DR1: addr=0x0, ref.count=0
DR2: addr=0x0, ref.count=0 DR3: addr=0x0, ref.count=0
That's one insert_watchpoint call for each Z1 packet, and then one
remove_watchpoint call for the z1 packet. Notice how ref.count
increased for each insert_watchpoint call, and then in the end, after
GDB told GDBserver to forget about the hardware breakpoint, GDBserver
ends with the the first debug register still with ref.count=4! IOW,
the hardware breakpoint is left armed on the target, while on the GDB
end it's gone. If the program happens to execute 0x410943 afterwards,
then the CPU traps, GDBserver reports the trap to GDB, and GDB not
having a breakpoint set at that address anymore, reports to the user a
spurious SIGTRAP.
This is exactly what is happening in the hbreak2.exp test, though in
that case, it's a shared library event that triggers a
breakpoint_re_set, when breakpoints are still inserted (because
nowadays GDB doesn't remove breakpoints while handling internal
events), and that recreates breakpoint locations, which likewise
forces breakpoint reinsertion and Zx packet resends...
That is a lot of bogus Zx duplication that should possibly be
addressed on the GDB side. GDB resends Zx packets because the way to
change the target-side condition, is to resend the breakpoint to the
server with the new condition. (That's an option in the packet: e.g.,
"Z1,410943,1;X3,220027" for "hbreak main if 0". The packets in the
examples above are shorter because the breakpoints don't have
conditions attached). GDB doesn't remove the breakpoint first before
reinserting it because that'd be bad for non-stop, as it'd open a
window where the inferior could miss the breakpoint. The conditions
actually haven't changed between the resends, but GDB isn't smart
enough to realize that.
(TBC, if the target doesn't support target-side conditions, then GDB
doesn't trigger these resends (init_bp_location calls
mark_breakpoint_location_modified, and that does nothing if condition
evaluation is on the host side. The resends are caused by the
'loc->condition_changed = condition_modified.' line.)
But, even if GDB was made smarter, GDBserver should really still
handle the resends anyway. So target-side conditions also aren't
really to blame. The documentation of the Z/z packets says:
"To avoid potential problems with duplicate packets, the operations
should be implemented in an idempotent way."
As such, we may want to fix GDB, but we should definitely fix
GDBserver. The fix is a prerequisite for target-side conditions on
hardware breakpoints anyway (and while at it, on watchpoints too).
GDBserver indeed already treats duplicate Z0 packets in an idempotent
way. mem-break.c has the concept of high-level and low-level
breakpoints, somewhat similar to GDB's split of breakpoints vs
breakpoint locations, and keeps track of multiple breakpoints
referencing the same address/location, for the case of an internal
GDBserver breakpoint or a tracepoint being set at the same address as
a GDB breakpoint. But, it only allows GDB to ever contribute one
reference to a software breakpoint location. IOW, if gdbserver sees a
Z0 packet for the same address where it already had a GDB breakpoint
set, then GDBserver won't create another high-level GDB breakpoint.
However, mem-break.c only tracks GDB Z0 breakpoints. The same logic
should apply to all kinds of Zx packets. Currently, gdbserver passes
down each duplicate Zx (other than Z0) request directly to the
target->insert_point routine. The x86 watchpoint support itself
refcounts watchpoint / hw breakpoint requests, to handle overlapping
watchpoints, and save debug registers. But that code doesn't (and
really shouldn't) handle the duplicate requests, assuming that for
each insert there will be a corresponding remove.
So the fix is to generalize mem-break.c to track all kinds of Zx
breakpoints, and filter out duplicates. As mentioned, this ends up
adding support for target-side conditions on hardware breakpoints and
watchpoints too (though GDB itself doesn't support the latter yet).
Probably the least obvious change in the patch is that it kind of
turns the breakpoint insert/remove APIs inside out. Before, the
target methods were only called for GDB breakpoints. The internal
breakpoint set/delete methods inserted memory breakpoints directly
bypassing the insert/remove target methods. That's not good when the
target should use a debug API to set software breakpoints, instead of
relying on GDBserver patching memory with breakpoint instructions, as
is the case of NTO.
Now removal/insertion of all kinds of breakpoints/watchpoints, either
internal, or from GDB, always go through the target methods. The
insert_point/remove_point methods no longer get passed a Z packet
type, but an internal/raw breakpoint type. They're also passed a
pointer to the raw breakpoint itself (note that's still opaque outside
mem-break.c), so that insert_memory_breakpoint /
remove_memory_breakpoint have access to the breakpoint's shadow
buffer. I first tried passing down a new structure based on GDB's
"struct bp_target_info" (actually with that name exactly), but then
decided against it as unnecessary complication.
As software/memory breakpoints work by poking at memory, when setting
a GDB Z0 breakpoint (but not internal breakpoints, as those can assume
the conditions are already right), we need to tell the target to
prepare to access memory (which on Linux means stop threads). If that
operation fails, we need to return error to GDB. Seeing an error, if
this is the first breakpoint of that type that GDB tries to insert,
GDB would then assume the breakpoint type is supported, but it may
actually not be. So we need to check whether the type is supported at
all before preparing to access memory. And to solve that, the patch
adds a new target->supports_z_point_type method that is called before
actually trying to insert the breakpoint.
Other than that, hopefully the change is more or less obvious.
New test added that exercises the hbreak2.exp regression in a more
direct way, without relying on a breakpoint re-set happening before
main is reached.
Tested by building GDBserver for:
aarch64-linux-gnu
arm-linux-gnueabihf
i686-pc-linux-gnu
i686-w64-mingw32
m68k-linux-gnu
mips-linux-gnu
mips-uclinux
nios2-linux-gnu
powerpc-linux-gnu
sh-linux-gnu
tilegx-unknown-linux-gnu
x86_64-redhat-linux
x86_64-w64-mingw32
And also regression tested on x86_64 Fedora 20.
gdb/gdbserver/
2014-05-20 Pedro Alves <palves@redhat.com>
* linux-aarch64-low.c (aarch64_insert_point)
(aarch64_remove_point): No longer check whether the type is
supported here. Adjust to new interface.
(the_low_target): Install aarch64_supports_z_point_type as
supports_z_point_type method.
* linux-arm-low.c (raw_bkpt_type_to_arm_hwbp_type): New function.
(arm_linux_hw_point_initialize): Take an enum raw_bkpt_type
instead of a Z packet char. Adjust.
(arm_supports_z_point_type): New function.
(arm_insert_point, arm_remove_point): Adjust to new interface.
(the_low_target): Install arm_supports_z_point_type.
* linux-crisv32-low.c (cris_supports_z_point_type): New function.
(cris_insert_point, cris_remove_point): Adjust to new interface.
Don't check whether the type is supported here.
(the_low_target): Install cris_supports_z_point_type.
* linux-low.c (linux_supports_z_point_type): New function.
(linux_insert_point, linux_remove_point): Adjust to new interface.
* linux-low.h (struct linux_target_ops) <insert_point,
remove_point>: Take an enum raw_bkpt_type instead of a char. Add
raw_breakpoint pointer parameter.
<supports_z_point_type>: New method.
* linux-mips-low.c (mips_supports_z_point_type): New function.
(mips_insert_point, mips_remove_point): Adjust to new interface.
Use mips_supports_z_point_type.
(the_low_target): Install mips_supports_z_point_type.
* linux-ppc-low.c (the_low_target): Install NULL as
supports_z_point_type method.
* linux-s390-low.c (the_low_target): Install NULL as
supports_z_point_type method.
* linux-sparc-low.c (the_low_target): Install NULL as
supports_z_point_type method.
* linux-x86-low.c (x86_supports_z_point_type): New function.
(x86_insert_point): Adjust to new insert_point interface. Use
insert_memory_breakpoint. Adjust to new
i386_low_insert_watchpoint interface.
(x86_remove_point): Adjust to remove_point interface. Use
remove_memory_breakpoint. Adjust to new
i386_low_remove_watchpoint interface.
(the_low_target): Install x86_supports_z_point_type.
* lynx-low.c (lynx_target_ops): Install NULL as
supports_z_point_type callback.
* nto-low.c (nto_supports_z_point_type): New.
(nto_insert_point, nto_remove_point): Adjust to new interface.
(nto_target_ops): Install nto_supports_z_point_type.
* mem-break.c: Adjust intro comment.
(struct raw_breakpoint) <raw_type, size>: New fields.
<inserted>: Update comment.
<shlib_disabled>: Delete field.
(enum bkpt_type) <gdb_breakpoint>: Delete value.
<gdb_breakpoint_Z0, gdb_breakpoint_Z1, gdb_breakpoint_Z2,
gdb_breakpoint_Z3, gdb_breakpoint_Z4>: New values.
(raw_bkpt_type_to_target_hw_bp_type): New function.
(find_enabled_raw_code_breakpoint_at): New function.
(find_raw_breakpoint_at): New type and size parameters. Use them.
(insert_memory_breakpoint): New function, based off
set_raw_breakpoint_at.
(remove_memory_breakpoint): New function.
(set_raw_breakpoint_at): Reimplement.
(set_breakpoint): New, based on set_breakpoint_at.
(set_breakpoint_at): Reimplement.
(delete_raw_breakpoint): Go through the_target->remove_point
instead of assuming memory breakpoints.
(find_gdb_breakpoint_at): Delete.
(Z_packet_to_bkpt_type, Z_packet_to_raw_bkpt_type): New functions.
(find_gdb_breakpoint): New function.
(set_gdb_breakpoint_at): Delete.
(z_type_supported): New function.
(set_gdb_breakpoint_1): New function, loosely based off
set_gdb_breakpoint_at.
(check_gdb_bp_preconditions, set_gdb_breakpoint): New functions.
(delete_gdb_breakpoint_at): Delete.
(delete_gdb_breakpoint_1): New function, loosely based off
delete_gdb_breakpoint_at.
(delete_gdb_breakpoint): New function.
(clear_gdb_breakpoint_conditions): Rename to ...
(clear_breakpoint_conditions): ... this. Don't handle a NULL
breakpoint.
(add_condition_to_breakpoint): Make static.
(add_breakpoint_condition): Take a struct breakpoint pointer
instead of an address. Adjust.
(gdb_condition_true_at_breakpoint): Rename to ...
(gdb_condition_true_at_breakpoint_z_type): ... this, and add
z_type parameter.
(gdb_condition_true_at_breakpoint): Reimplement.
(add_breakpoint_commands): Take a struct breakpoint pointer
instead of an address. Adjust.
(gdb_no_commands_at_breakpoint): Rename to ...
(gdb_no_commands_at_breakpoint_z_type): ... this. Add z_type
parameter. Return true if no breakpoint was found. Change debug
output.
(gdb_no_commands_at_breakpoint): Reimplement.
(run_breakpoint_commands): Rename to ...
(run_breakpoint_commands_z_type): ... this. Add z_type parameter,
and change return type to boolean.
(run_breakpoint_commands): New function.
(gdb_breakpoint_here): Also check for Z1 breakpoints.
(uninsert_raw_breakpoint): Don't try to reinsert a disabled
breakpoint. Go through the_target->remove_point instead of
assuming memory breakpoint.
(uninsert_breakpoints_at, uninsert_all_breakpoints): Uninsert
software and hardware breakpoints.
(reinsert_raw_breakpoint): Go through the_target->insert_point
instead of assuming memory breakpoint.
(reinsert_breakpoints_at, reinsert_all_breakpoints): Reinsert
software and hardware breakpoints.
(check_breakpoints, breakpoint_here, breakpoint_inserted_here):
Check both software and hardware breakpoints.
(validate_inserted_breakpoint): Assert the breakpoint is a
software breakpoint. Set the inserted flag to -1 instead of
setting shlib_disabled.
(delete_disabled_breakpoints): Adjust.
(validate_breakpoints): Only validate software breakpoints.
Adjust to inserted flag change.
(check_mem_read, check_mem_write): Skip breakpoint types other
than software breakpoints. Adjust to inserted flag change.
* mem-break.h (enum raw_bkpt_type): New enum.
(raw_breakpoint, struct process_info): Forward declare.
(Z_packet_to_target_hw_bp_type): Delete declaration.
(raw_bkpt_type_to_target_hw_bp_type, Z_packet_to_raw_bkpt_type)
(set_gdb_breakpoint, delete_gdb_breakpoint)
(clear_breakpoint_conditions): New declarations.
(set_gdb_breakpoint_at, clear_gdb_breakpoint_conditions): Delete.
(breakpoint_inserted_here): Update comment.
(add_breakpoint_condition, add_breakpoint_commands): Replace
address parameter with a breakpoint pointer parameter.
(gdb_breakpoint_here): Update comment.
(delete_gdb_breakpoint_at): Delete.
(insert_memory_breakpoint, remove_memory_breakpoint): Declare.
* server.c (process_point_options): Take a struct breakpoint
pointer instead of an address. Adjust.
(process_serial_event) <Z/z packets>: Use set_gdb_breakpoint and
delete_gdb_breakpoint.
* spu-low.c (spu_target_ops): Install NULL as
supports_z_point_type method.
* target.h: Include mem-break.h.
(struct target_ops) <prepare_to_access_memory>: Update comment.
<supports_z_point_type>: New field.
<insert_point, remove_point>: Take an enum raw_bkpt_type argument
instead of a char. Also take a raw breakpoint pointer.
* win32-arm-low.c (the_low_target): Install NULL as
supports_z_point_type.
* win32-i386-low.c (i386_supports_z_point_type): New function.
(i386_insert_point, i386_remove_point): Adjust to new interface.
(the_low_target): Install i386_supports_z_point_type.
* win32-low.c (win32_supports_z_point_type): New function.
(win32_insert_point, win32_remove_point): Adjust to new interface.
(win32_target_ops): Install win32_supports_z_point_type.
* win32-low.h (struct win32_target_ops):
<supports_z_point_type>: New method.
<insert_point, remove_point>: Take an enum raw_bkpt_type argument
instead of a char. Also take a raw breakpoint pointer.
gdb/testsuite/
2014-05-20 Pedro Alves <palves@redhat.com>
* gdb.base/break-idempotent.c: New file.
* gdb.base/break-idempotent.exp: New file.
While trying to fix hbreak2.exp against GDBserver I noticed this...
(gdb) hbreak main if 1
Sending packet: $m400580,40#2e...Packet received: e8d2ffffff5dc3554889e54883ec10c745fc00000000eb0eb800000000e8c1ffffff8345fc01817dfce70300007ee9b800000000c9c3662e0f1f840000000000
Sending packet: $m40058f,1#31...Packet received: c7
Hardware assisted breakpoint 1 at 0x40058f: file ../../../src/gdb/testsuite/gdb.base/break-idempotent.c, line 46.
Sending packet: $Z1,40058f,1;X3,220127#9b...
*hangs forever*
The issue is that nothing advances the packet pointer if
add_breakpoint_condition either fails to parse the agent expression,
or fails to find the breakpoint, resulting in an infinite loop in
process_point_options. The latter case should really be fixed by
GDBserver tracking GDB Z1 breakpoints in its breakpoint structures
like Z0 breakpoints are, but the latter case still needs handling.
add_breakpoint_commands has the same issue, though at present I don't
know any way to trigger it other than sending a manually cooked
packet.
Unbelievably, it doesn't look like we have any test that tries setting
a conditional hardware breakpoint. Looking at cond-eval-mode.exp, it
looks like the file was meant to actually test something, but it's
mostly empty today. This patch adds tests that tries all sorts of
conditional breakpoints and watchpoints. The test hangs/fails without
the GDBserver fix.
Tested on x86_64 Fedora 17.
gdb/gdbserver/
2014-04-10 Pedro Alves <palves@redhat.com>
* mem-break.c (add_breakpoint_condition, add_breakpoint_commands):
Check if the condition or command is NULL before checking if the
breakpoint is known. On success, return true.
* mem-break.h (add_breakpoint_condition): Document return.
(add_breakpoint_commands): Add describing comment.
* server.c (skip_to_semicolon): New function.
(process_point_options): Use it.
gdb/testsuite/
2014-04-10 Pedro Alves <palves@redhat.com>
* gdb.base/cond-eval-mode.c: New file.
* gdb.base/cond-eval-mode.exp: Use standard_testfile. Adjust
prepare_for_testing to build the new file. Check result of
runto_main.
(test_break, test_watch): New procedures.
(top level): Use them.
GDBserver currently hangs forever in waitpid if the leader thread
exits before other threads, or if all resumed threads exit - e.g.,
next over a thread exit with sched-locking on. This is exposed by
leader-exit.exp. leader-exit.exp is part of a series of tests for a
set of related problems. See
<http://www.sourceware.org/ml/gdb-patches/2011-10/msg00704.html>:
"
To recap, on the Linux kernel, ptrace/waitpid don't allow reaping the
leader thread until all other threads in the group are reaped. When
the leader exits, it goes zombie, but waitpid will not return an exit
status until the other threads are gone. This is presently exercised
by the gdb.threads/leader-exit.exp test. The fix for that test, in
linux-nat.c:wait_lwp, handles the case where we see the leader gone
when we're stopping all threads to report an event to some other
thread to the core.
(...)
The latter bit about not blocking if there no resumed threads in the
process also applies to some other thread exiting, not just the main
thread. E.g., this test starts a thread, and runs to a breakpoint in
that thread:
...
(gdb) c
Continuing.
[New Thread 0x7ffff75a4700 (LWP 23397)]
[Switching to Thread 0x7ffff75a4700 (LWP 23397)]
Breakpoint 2, thread_a (arg=0x0) at ../../../src/gdb/testsuite/gdb.threads/no-unwaited-for-left.c:28
28 return 0; /* break-here */
(gdb) info threads
* 2 Thread 0x7ffff75a4700 (LWP 23397) thread_a (arg=0x0) at ../../../src/gdb/testsuite/gdb.threads/no-unwaited-for-left.c:28
1 Thread 0x7ffff7fcb720 (LWP 23391) 0x00007ffff7bc606d in pthread_join (threadid=140737343276800, thread_return=0x0) at pthread_join.c:89
The thread will exit as soon as we resume it. But if we only resume
that thread, leaving the rest of the threads stopped:
(gdb) set scheduler-locking on
(gdb) c
Continuing.
^C^C^C^C^C^C^C^C
"
This patch fixes the issues by implementing TARGET_WAITKIND_NO_RESUMED
on GDBserver, similarly to what the patch above did for native
Linux GDB.
gdb.threads/leader-exit.exp now passes.
gdb.threads/no-unwaited-for-left.exp now at least errors out instead
of hanging:
continue
Continuing.
warning: Remote failure reply: E.No unwaited-for children left.
[Thread 15454] #1 stopped.
0x00000034cf408e60 in pthread_join (threadid=140737353922368, thread_return=0x0) at pthread_join.c:93
93 lll_wait_tid (pd->tid);
(gdb) FAIL: gdb.threads/no-unwaited-for-left.exp: continue stops when the main thread exits
The gdb.threads/non-ldr-exc-*.exp tests are skipped because GDBserver
unfortunately doesn't support fork/exec yet, but I'm confident this
fixes the related issues.
I'm leaving modeling TARGET_WAITKIND_NO_RESUMED in the RSP for a
separate pass.
(BTW, in case of error in response to a vCont, it would be better for
GDB to query the target for the current thread, or re-select one,
instead of assuming current inferior_ptid is still the selected
thread.)
This implementation is a little different from GDB's, because I'm
avoiding bringing in more of this broken use of waitpid(PID) into
GDBserver. Specifically, this avoids waitpid(PID) when stopping all
threads. There's really no need for wait_for_sigstop to wait for each
LWP in turn. Instead, with some refactoring, we make it reuse
linux_wait_for_event.
gdb/gdbserver/
2014-02-27 Pedro Alves <palves@redhat.com>
PR 12702
* inferiors.h (A_I_NEXT, ALL_INFERIORS_TYPE, ALL_PROCESSES): New
macros.
* linux-low.c (delete_lwp, handle_extended_wait): Add debug
output.
(last_thread_of_process_p): Take a PID argument instead of a
thread pointer.
(linux_wait_for_lwp): Delete.
(num_lwps, check_zombie_leaders, not_stopped_callback): New
functions.
(linux_low_filter_event): New function, party factored out from
linux_wait_for_event.
(linux_wait_for_event): Rename to ...
(linux_wait_for_event_filtered): ... this. Add new filter ptid
argument. Partly rewrite. Always use waitpid(-1, WNOHANG) and
sigsuspend. Check for zombie leaders.
(linux_wait_for_event): Reimplement as wrapper around
linux_wait_for_event_filtered.
(linux_wait_1): Handle TARGET_WAITKIND_NO_RESUMED. Assume that if
a normal or signal exit is seen, it's the whole process exiting.
(wait_for_sigstop): No longer a for_each_inferior callback.
Rewrite on top of linux_wait_for_event_filtered.
(stop_all_lwps): Call wait_for_sigstop directly.
* server.c (resume, handle_target_event): Handle
TARGET_WAITKIND_NO_RESUMED.
* dll.c (clear_dlls): Replace accessing list implemention details
with API function.
* gdbthread.h (get_first_thread): Declare.
* inferiors.c (for_each_inferior_with_data): New function.
(get_first_thread): New function.
(find_thread_ptid): Simplify.
(get_first_inferior): New function.
(clear_list): Delete.
(one_inferior_p): New function.
(clear_inferior_list): New function.
(clear_inferiors): Update.
* inferiors.h (for_each_inferior_with_data): Declare.
(clear_inferior_list): Declare.
(one_inferior_p): Declare.
(get_first_inferior): Declare.
* linux-low.c (linux_wait_for_event): Replace accessing list
implemention details with API function.
* server.c (target_running): Ditto.
(accumulate_file_name_length): New function.
(emit_dll_description): New function.
(handle_qxfer_libraries): Replace accessing list implemention
details with API function.
(handle_qxfer_threads_worker): New function.
(handle_qxfer_threads_proper): Replace accessing list implemention
details with API function.
(handle_query): Ditto.
(visit_actioned_threads_callback_ftype): New typedef.
(visit_actioned_threads_data): New struct.
(visit_actioned_threads): Rewrite to be find_inferior callback.
(resume): Call find_inferior.
(handle_status): Replace accessing list implemention
details with API function.
(process_serial_event): Replace accessing list implemention details
with API function.
* target.c (set_desired_inferior): Replace accessing list implemention
details with API function.
* tracepoint.c (same_process_p): New function.
(gdb_agent_about_to_close): Replace accessing list implemention
details with API function.
* win32-low.c (child_delete_thread): Replace accessing list
implemention details with API function.
(match_dll_by_basename): New function.
(dll_is_loaded_by_basename): New function.
(win32_ensure_ntdll_loaded): Replace accessing list implemention
details call to dll_is_loaded_by_basename.
unhexify and hex2bin are identical, so this removes unhexify. The
particular choice of which to keep was made on the basis of
parallelism with the earlier patch that removed hexify.
2014-02-12 Tom Tromey <tromey@redhat.com>
* common/rsp-low.h (unhexify): Don't declare.
* common/rsp-low.c (unhexify): Remove.
2014-02-12 Tom Tromey <tromey@redhat.com>
* server.c (handle_query, handle_v_run): Use hex2bin, not
unhexify.
* tracepoint.c (cmd_qtdpsrc, cmd_qtdv, cmd_qtnotes): Likewise.
convert_int_to_ascii is identical to bin2hex. This removes the
former. In this case I made the choice of which to keep on the basis
that I consider the name bin2hex to be superior to
convert_int_to_ascii.
2014-02-12 Tom Tromey <tromey@redhat.com>
* common/rsp-low.h (convert_int_to_ascii): Don't declare.
* common/rsp-low.c (convert_int_to_ascii): Remove.
2014-02-12 Tom Tromey <tromey@redhat.com>
* ax.c (gdb_unparse_agent_expr): Use bin2hex, not
convert_int_to_ascii.
* regcache.c (registers_to_string, collect_register_as_string):
Likewise.
* remote-utils.c (look_up_one_symbol, relocate_instruction):
Likewise.
* server.c (process_serial_event): Likewise.
* tracepoint.c (cmd_qtstatus, response_source, response_tsv)
(cmd_qtbuffer, cstr_to_hexstr): Likewise.
This moves various low-level remote serial protocol bits into
common/rsp-low.[ch].
This is as close to a pure move as possible. There are some
redundancies remaining but those will be dealt with in a subsequent
patch.
Note that the two variants of remote_escape_output disagreed on the
treatment of "*". On the theory that quoting cannot hurt but the
absence possibly can, I chose the gdbserver variant to be the
canonical one.
2014-02-12 Tom Tromey <tromey@redhat.com>
* tracepoint.c: Include rsp-low.h.
* remote.h (hex2bin, bin2hex, unpack_varlen_hex): Don't declare.
* remote.c: Include rsp-low.h.
(hexchars, ishex, unpack_varlen_hex, pack_nibble, pack_hex_byte)
(fromhex, hex2bin, tohex, bin2hex, remote_escape_output)
(remote_unescape_input): Move to common/rsp-low.c.
* common/rsp-low.h: New file.
* common/rsp-low.c: New file.
* Makefile.in (SFILES): Add common/rsp-low.c.
(HFILES_NO_SRCDIR): Add common/rsp-low.h.
(COMMON_OBS): Add rsp-low.o.
(rsp-low.o): New target.
2014-02-12 Tom Tromey <tromey@redhat.com>
* tracepoint.c: Include rsp-low.h.
* server.c: Include rsp-low.h.
* remote-utils.h (convert_ascii_to_int, convert_int_to_ascii)
(unhexify, hexify, remote_escape_output, unpack_varlen_hex): Don't
declare.
* remote-utils.c: Include rsp-low.h.
(fromhex, hexchars, ishex, unhexify, tohex, hexify)
(remote_escape_output, remote_unescape_input, unpack_varlen_hex)
(convert_int_to_ascii, convert_ascii_to_int): Move to
common/rsp-low.c.
* regcache.c: Include rsp-low.h.
* ax.c: Include rsp-low.h.
* Makefile.in (SFILES): Add common/rsp-low.c.
(OBS): Add rsp-low.o.
(rsp-low.o): New target.
Read branch trace data incrementally and extend the current trace rather than
discarding it and reading the entire trace buffer each time.
If the branch trace buffer overflowed, we can't extend the current trace so we
discard it and start anew by reading the entire branch trace buffer.
2014-01-16 Markus Metzger <markus.t.metzger@intel.com>
* common/linux-btrace.c (perf_event_read_bts, linux_read_btrace):
Support delta reads.
(linux_disable_btrace): Change return type.
* common/linux-btrace.h (linux_read_btrace): Change parameters
and return type to allow error reporting. Update users.
(linux_disable_btrace): Change return type. Update users.
* common/btrace-common.h (btrace_read_type) <BTRACE_READ_DELTA>:
New.
(btrace_error): New.
(btrace_block) <begin>: Comment on BEGIN == 0.
* btrace.c (btrace_compute_ftrace): Start from the end of
the current trace.
(btrace_stitch_trace, btrace_clear_history): New.
(btrace_fetch): Read delta trace, return if replaying.
(btrace_clear): Move clear history code to btrace_clear_history.
(parse_xml_btrace): Throw an error if parsing failed.
* target.h (struct target_ops) <to_read_btrace>: Change parameters
and return type to allow error reporting.
(target_read_btrace): Change parameters and return type to allow
error reporting.
* target.c (target_read_btrace): Update.
* remote.c (remote_read_btrace): Support delta reads. Pass
errors on.
* NEWS: Announce it.
gdbserver/
* target.h (target_ops) <read_btrace>: Change parameters and
return type to allow error reporting.
* server.c (handle_qxfer_btrace): Support delta reads. Pass
trace reading errors on.
* linux-low.c (linux_low_read_btrace): Pass trace reading
errors on.
(linux_low_disable_btrace): New.
... not when a new GDB connection sends the status packet ('?').
Mainly just a cleanup/simplification, as GDB always sends '?' first.
Tested on x86_64 Fedora 17.
2014-01-08 Pedro Alves <palves@redhat.com>
* server.c (handle_status): Don't discard previous queued stop
replies or thread's pending status here.
(main) <disconnection>: Do it here instead.
Currently, when GDB connects in all-stop mode, GDBserver always
responds to the status packet with a GDB_SIGNAL_TRAP, even if the
program is actually stopped for some other signal.
(gdb) tar rem ...
...
(gdb) c
Program received signal SIGUSR1, User defined signal 1.
(gdb) disconnect
(gdb) tar rem ...
(gdb) c
(Or a GDB crash instead of an explicit disconnect.)
This results in the program losing that signal on that last continue,
because gdb will tell the target to resume with no signal (to suppress
the GDB_SIGNAL_TRAP, due to 'handle SISGTRAP nopass'), and that will
actually suppress the real signal the program had stopped for
(SIGUSR1). To fix that, I think we should make GDBserver report the
real signal the thread had stopped for in response to the status
packet:
@item ?
@cindex @samp{?} packet
Indicate the reason the target halted. The reply is the same as for
step and continue.
But, that raises the question -- which thread are we reporting the
status for? Due to how the RSP in all-stop works, we can only report
one status. The status packet's response is a stop reply packet, so
it includes the thread identifier, so it's not a problem packet-wise.
However, GDBserver is currently always reporting the status for first
thread in the thread list, even though that may well not be the thread
that got the signal that caused the program to stop. So the next
logical step would be to report the status for the
last_ptid/last_status thread (the last event reported to gdb), if it's
still around; and if not, fallback to some other thread.
There's an issue on the GDB side with that, though...
GDB currently always adds the thread reported in response to the
status query as the first thread in its list. That means that if we
start with e.g.,
(gdb) info threads
3 Thread 1003 ...
* 2 Thread 1002 ...
1 Thread 1001 ...
And reconnect:
(gdb) disconnect
(gdb) tar rem ...
We end up with:
(gdb) info threads
3 Thread 1003 ...
2 Thread 1001 ...
* 1 Thread 1002 ...
Not a real big issue, but it's reasonably fixable, by having GDB
fetch/sync the thread list before fetching the status/'?', and then
using the status to select the right thread as current on the GDB
side. Holes in the thread numbers are squashed before/after
reconnection (e.g., 2,3,5 becomes 1,2,3), but the order is preserved,
which I think is both good, and good enough.
However (yes, there's more...), the previous GDB that was connected
might have had gdbserver running in non-stop mode, or could have left
gdbserver doing disconnected tracing (which also forces non-stop), and
if the new gdb/connection is in all-stop mode, we can end up with more
than one thread with a signal to report back to gdb. As we can only
report one thread/status (in the all-stop RSP variant; the non-stop
variant doesn't have this issue), we get to do what we do at every
other place we have this situation -- leave events we can't report
right now as pending, so that the next resume picks them up.
Note all this ammounts to a QoI change, within the existing framework.
There's really no RSP change here.
The only user visible change (other than that the signal is program is
stopped at isn't lost / is passed to the program), is in "info
program", that now can show the signal the program stopped for. Of
course, the next resume will respect the pass/nopass setting for the
signal in question. It'd be reasonable to have the initial connection
tell the user the program was stopped with a signal, similar to when
we load a core to debug, but I'm leaving that out for a future change.
I think we'll need to either change how handle_inferior_event & co
handle stop_soon, or maybe bypass them completely (like
fork-child.c:startup_inferior) for that.
Tested on x86_64 Fedora 17.
gdb/gdbserver/
2014-01-08 Pedro Alves <palves@redhat.com>
* gdbthread.h (struct thread_info) <status_pending_p>: New field.
* server.c (visit_actioned_threads, handle_pending_status): New
function.
(handle_v_cont): Factor out parts to ...
(resume): ... this new function. If in all-stop, and a thread
being resumed has a pending status, report it without actually
resuming.
(myresume): Adjust to use the new 'resume' function.
(clear_pending_status_callback, set_pending_status_callback)
(find_status_pending_thread_callback): New functions.
(handle_status): Handle the case of multiple threads having
interesting statuses to report. Report threads' real last signal
instead of always reporting GDB_SIGNAL_TRAP. Look for a thread
with an interesting thread to report the status for, instead of
always reporting the status of the first thread.
gdb/
2014-01-08 Pedro Alves <palves@redhat.com>
* remote.c (remote_add_thread): Add threads silently if starting
up.
(remote_notice_new_inferior): If in all-stop, and starting up,
don't call notice_new_inferior.
(get_current_thread): New function, factored out from ...
(add_current_inferior_and_thread): ... this. Adjust.
(remote_start_remote) <all-stop>: Fetch the thread list. If we
found any thread, then select the remote's current thread as GDB's
current thread too.
gdb/testsuite/
2014-01-08 Pedro Alves <palves@redhat.com>
* gdb.threads/reconnect-signal.c: New file.
* gdb.threads/reconnect-signal.exp: New file.
gdb/ChangeLog:
* top.c (print_gdb_version): Set copyright year to 2014.
gdb/gdbserver/ChangeLog:
* gdbserver.c (gdbserver_version): Set copyright year to 2014.
* gdbreplay.c (gdbreplay_version): Likewise.
If we make gdbserver gdb_continue_to_end actually expect a process
exit with GDBserver, we get many testsuite failures with the remote
stdio board:
-PASS: gdb.arch/amd64-disp-step.exp: continue until exit at amd64-disp-step
+FAIL: gdb.arch/amd64-disp-step.exp: continue until exit at amd64-disp-step (the program exited)
-PASS: gdb.base/break.exp: continue until exit at recursive next test
+FAIL: gdb.base/break.exp: continue until exit at recursive next test (the program exited)
-PASS: gdb.base/chng-syms.exp: continue until exit at breakpoint first time through
+FAIL: gdb.base/chng-syms.exp: continue until exit at breakpoint first time through (the program exited)
... etc. ...
This is what the log shows for all of them:
(gdb) continue
Continuing.
Child exited with status 0
GDBserver exiting
[Inferior 1 (process 22721) exited normally]
(gdb) FAIL: gdb.arch/amd64-disp-step.exp: continue until exit (the program exited)
The problem is the whole "Child exited ... GDBserver exiting" output,
that comes out of GDBserver, and that the testsuite is not expecting.
I pondered somehow making the testsuite adjust to this. But,
testsuite aside, I think GDBserver should not be outputting this at
all when GDB is connected through stdio. GDBserver will be printing
this in GDB's console, but the user can already tell from the regular
output that the inferior is gone.
Again, manually:
(gdb) tar remote | ./gdbserver/gdbserver - program
Remote debugging using | ./gdbserver/gdbserver - program
Process program created; pid = 22486
stdin/stdout redirected
Remote debugging using stdio
done.
Loaded symbols for /lib64/ld-linux-x86-64.so.2
0x000000323d001530 in _start () from /lib64/ld-linux-x86-64.so.2
(gdb) c
Continuing.
Child exited with status 1
^^^^^^^^^^^^^^^^^^^^^^^^^^
GDBserver exiting
^^^^^^^^^^^^^^^^^
[Inferior 1 (process 22486) exited with code 01]
(gdb)
Suppressing those two lines makes the output be exactly like when
debugging against a remote tcp gdbserver:
(gdb) c
Continuing.
[Inferior 1 (process 22914) exited with code 01]
(gdb)
2013-10-02 Pedro Alves <palves@redhat.com>
* server.c (process_serial_event): Don't output "GDBserver
exiting" if GDB is connected through stdio.
* target.c (mywait): Likewise, be silent if GDB is connected
through stdio.
Yao Qi
Markus Metzger
Simon Marchi
Jan Kratochvil
Gary Benson
Don Breazeal
Pedro Alves
Joel Brobecker
Tom Tromey
Doug Evans
Yao Qi