diff options
author | Antoine Tremblay <antoine.tremblay@ericsson.com> | 2015-11-19 11:29:10 -0500 |
---|---|---|
committer | Antoine Tremblay <antoine.tremblay@ericsson.com> | 2015-11-30 15:18:57 -0500 |
commit | 9b4c5f878ff39e04127a1ad95f6b3832afe6d278 (patch) | |
tree | 419f45d3ab06c0c444ed36d58835f17bffa165b3 /gdb/gdbserver | |
parent | 7d00775ece9e2364da5cfd65ebbfce515859667f (diff) | |
download | gdb-9b4c5f878ff39e04127a1ad95f6b3832afe6d278.zip gdb-9b4c5f878ff39e04127a1ad95f6b3832afe6d278.tar.gz gdb-9b4c5f878ff39e04127a1ad95f6b3832afe6d278.tar.bz2 |
Remove support for thread events without PTRACE_EVENT_CLONE in GDBServer.
This patch removes support for thread events if PTRACE_EVENT_CLONE is not
supported in GDBServer.
Before, on systems that did not support PTRACE_EVENT_CLONE, both GDB and
GDBServer coordinated with libthread_db.so to insert breakpoints at magic
locations in libpthread.so, in order to break at thread creation and thread
death.
Simple software single stepping support was implemented to step over these
breakpoints in case there was no hardware single stepping support. However,
these simple software single stepping implementations were not fit for any other
use as discussed in :
https://sourceware.org/ml/gdb-patches/2015-04/msg01110.html
These too simple implementations conflict with ongoing work to make proper
implementations of software single stepping in GDBServer.
The problem is that if some implementations are correct and others are not and
only there for the thread magic breakpoint, we can't enable features based
solely software single step support since some would be broken.
To keep the incorrect implementations and allow the new proper ones at the same
time we would need to implement fallback code and it quickly becomes ugly and
confusing with multiple checks for legacy software single step or proper
software single step.
However, PTRACE_EVENT_CLONE was first introduced in Linux 2.5.46,
released in November 2002.
So I think it's reasonable to just remove support for kernels that don't support
PTRACE_EVENT_CLONE, and sidestep the libthread_db breakpoints issues entirely.
This thread on the mailling list discusses the issue :
https://sourceware.org/ml/gdb/2015-10/msg00078.html
No regressions, tested on ubuntu 14.04 ARMv7 and x86.
With gdbserver-{native,extended} / { -marm -mthumb }
gdb/gdbserver/ChangeLog:
* linux-low.c (linux_look_up_symbols): Don't call
linux_supports_traceclone.
* linux-low.h (thread_db_init): Remove use_events argument.
* thread-db.c (thread_db_use_event): Remove global variable.
(struct thread_db) <td_thr_event_enable_p>: Remove field.
(struct thread_db) <td_create_bp>: Remove field.
(thread_db_create_event): Remove function.
(thread_db_enable_reporting): Likewise.
(find_one_thread): Don't check for thread_db_use_events.
(attach_thread): Likewise.
(thread_db_load_search): Remove td_thr_event_enable_p initialization.
(try_thread_db_load_1): Don't check for thread_db_use_events.
(thread_db_init): Remove use_events argument and thread events
handling.
(remove_thread_event_breakpoints): Remove function.
(thread_db_detach): Remove call to remove_thred_event_breakpoints.
Diffstat (limited to 'gdb/gdbserver')
-rw-r--r-- | gdb/gdbserver/ChangeLog | 19 | ||||
-rw-r--r-- | gdb/gdbserver/linux-low.c | 5 | ||||
-rw-r--r-- | gdb/gdbserver/linux-low.h | 2 | ||||
-rw-r--r-- | gdb/gdbserver/thread-db.c | 153 |
4 files changed, 25 insertions, 154 deletions
diff --git a/gdb/gdbserver/ChangeLog b/gdb/gdbserver/ChangeLog index 4f25205..719ac2b 100644 --- a/gdb/gdbserver/ChangeLog +++ b/gdb/gdbserver/ChangeLog @@ -1,5 +1,24 @@ 2015-11-30 Antoine Tremblay <antoine.tremblay@ericsson.com> + * linux-low.c (linux_look_up_symbols): Don't call + linux_supports_traceclone. + * linux-low.h (thread_db_init): Remove use_events argument. + * thread-db.c (thread_db_use_event): Remove global variable. + (struct thread_db) <td_thr_event_enable_p>: Remove field. + (struct thread_db) <td_create_bp>: Remove field. + (thread_db_create_event): Remove function. + (thread_db_enable_reporting): Likewise. + (find_one_thread): Don't check for thread_db_use_events. + (attach_thread): Likewise. + (thread_db_load_search): Remove td_thr_event_enable_p initialization. + (try_thread_db_load_1): Don't check for thread_db_use_events. + (thread_db_init): Remove use_events argument and thread events + handling. + (remove_thread_event_breakpoints): Remove function. + (thread_db_detach): Remove call to remove_thred_event_breakpoints. + +2015-11-30 Antoine Tremblay <antoine.tremblay@ericsson.com> + * linux-aarch64-low.c (aarch64_supports_hardware_single_step): New function. (struct linux_target_ops) <supports_hardware_single_step>: Initialize. diff --git a/gdb/gdbserver/linux-low.c b/gdb/gdbserver/linux-low.c index cfc5987..ec5337a 100644 --- a/gdb/gdbserver/linux-low.c +++ b/gdb/gdbserver/linux-low.c @@ -5600,10 +5600,7 @@ linux_look_up_symbols (void) if (proc->priv->thread_db != NULL) return; - /* If the kernel supports tracing clones, then we don't need to - use the magic thread event breakpoint to learn about - threads. */ - thread_db_init (!linux_supports_traceclone ()); + thread_db_init (); #endif } diff --git a/gdb/gdbserver/linux-low.h b/gdb/gdbserver/linux-low.h index 24ea2d3..c211a37 100644 --- a/gdb/gdbserver/linux-low.h +++ b/gdb/gdbserver/linux-low.h @@ -379,7 +379,7 @@ void initialize_regsets_info (struct regsets_info *regsets_info); void initialize_low_arch (void); /* From thread-db.c */ -int thread_db_init (int use_events); +int thread_db_init (void); void thread_db_detach (struct process_info *); void thread_db_mourn (struct process_info *); int thread_db_handle_monitor_command (char *); diff --git a/gdb/gdbserver/thread-db.c b/gdb/gdbserver/thread-db.c index 3df10ff..8b3057a 100644 --- a/gdb/gdbserver/thread-db.c +++ b/gdb/gdbserver/thread-db.c @@ -24,8 +24,6 @@ extern int debug_threads; -static int thread_db_use_events; - #include "gdb_proc_service.h" #include "nat/gdb_thread_db.h" #include "gdb_vecs.h" @@ -56,16 +54,6 @@ struct thread_db void *handle; #endif - /* Thread creation event breakpoint. The code at this location in - the child process will be called by the pthread library whenever - a new thread is created. By setting a special breakpoint at this - location, GDB can detect when a new thread is created. We obtain - this location via the td_ta_event_addr call. Note that if the - running kernel supports tracing clones, then we don't need to use - (and in fact don't use) this magic thread event breakpoint to - learn about threads. */ - struct breakpoint *td_create_bp; - /* Addresses of libthread_db functions. */ td_ta_new_ftype *td_ta_new_p; td_ta_event_getmsg_ftype * td_ta_event_getmsg_p; @@ -73,7 +61,6 @@ struct thread_db td_ta_event_addr_ftype *td_ta_event_addr_p; td_ta_map_lwp2thr_ftype *td_ta_map_lwp2thr_p; td_thr_get_info_ftype *td_thr_get_info_p; - td_thr_event_enable_ftype *td_thr_event_enable_p; td_ta_thr_iter_ftype *td_ta_thr_iter_p; td_thr_tls_get_addr_ftype *td_thr_tls_get_addr_p; td_thr_tlsbase_ftype *td_thr_tlsbase_p; @@ -172,84 +159,6 @@ thread_db_state_str (td_thr_state_e state) #endif static int -thread_db_create_event (CORE_ADDR where) -{ - td_event_msg_t msg; - td_err_e err; - struct lwp_info *lwp; - struct thread_db *thread_db = current_process ()->priv->thread_db; - - gdb_assert (thread_db->td_ta_event_getmsg_p != NULL); - - if (debug_threads) - debug_printf ("Thread creation event.\n"); - - /* FIXME: This assumes we don't get another event. - In the LinuxThreads implementation, this is safe, - because all events come from the manager thread - (except for its own creation, of course). */ - err = thread_db->td_ta_event_getmsg_p (thread_db->thread_agent, &msg); - if (err != TD_OK) - fprintf (stderr, "thread getmsg err: %s\n", - thread_db_err_str (err)); - - /* If we do not know about the main thread yet, this would be a good time to - find it. We need to do this to pick up the main thread before any newly - created threads. */ - lwp = get_thread_lwp (current_thread); - if (lwp->thread_known == 0) - find_one_thread (current_thread->entry.id); - - /* msg.event == TD_EVENT_CREATE */ - - find_new_threads_callback (msg.th_p, NULL); - - return 0; -} - -static int -thread_db_enable_reporting (void) -{ - td_thr_events_t events; - td_notify_t notify; - td_err_e err; - struct thread_db *thread_db = current_process ()->priv->thread_db; - - if (thread_db->td_ta_set_event_p == NULL - || thread_db->td_ta_event_addr_p == NULL - || thread_db->td_ta_event_getmsg_p == NULL) - /* This libthread_db is missing required support. */ - return 0; - - /* Set the process wide mask saying which events we're interested in. */ - td_event_emptyset (&events); - td_event_addset (&events, TD_CREATE); - - err = thread_db->td_ta_set_event_p (thread_db->thread_agent, &events); - if (err != TD_OK) - { - warning ("Unable to set global thread event mask: %s", - thread_db_err_str (err)); - return 0; - } - - /* Get address for thread creation breakpoint. */ - err = thread_db->td_ta_event_addr_p (thread_db->thread_agent, TD_CREATE, - ¬ify); - if (err != TD_OK) - { - warning ("Unable to get location for thread creation breakpoint: %s", - thread_db_err_str (err)); - return 0; - } - thread_db->td_create_bp - = set_breakpoint_at ((CORE_ADDR) (unsigned long) notify.u.bptaddr, - thread_db_create_event); - - return 1; -} - -static int find_one_thread (ptid_t ptid) { td_thrhandle_t th; @@ -287,14 +196,6 @@ find_one_thread (ptid_t ptid) return 0; } - if (thread_db_use_events) - { - err = thread_db->td_thr_event_enable_p (&th, 1); - if (err != TD_OK) - error ("Cannot enable thread event reporting for %d: %s", - ti.ti_lid, thread_db_err_str (err)); - } - /* If the new thread ID is zero, a final thread ID will be available later. Do not enable thread debugging yet. */ if (ti.ti_tid == 0) @@ -334,17 +235,6 @@ attach_thread (const td_thrhandle_t *th_p, td_thrinfo_t *ti_p) lwp->thread_known = 1; lwp->th = *th_p; - if (thread_db_use_events) - { - td_err_e err; - struct thread_db *thread_db = proc->priv->thread_db; - - err = thread_db->td_thr_event_enable_p (th_p, 1); - if (err != TD_OK) - error ("Cannot enable thread event reporting for %d: %s", - ti_p->ti_lid, thread_db_err_str (err)); - } - return 1; } @@ -584,9 +474,6 @@ thread_db_load_search (void) tdb->td_ta_thr_iter_p = &td_ta_thr_iter; tdb->td_symbol_list_p = &td_symbol_list; - /* This is required only when thread_db_use_events is on. */ - tdb->td_thr_event_enable_p = &td_thr_event_enable; - /* These are not essential. */ tdb->td_ta_event_addr_p = &td_ta_event_addr; tdb->td_ta_set_event_p = &td_ta_set_event; @@ -654,9 +541,6 @@ try_thread_db_load_1 (void *handle) CHK (1, TDB_DLSYM (tdb, td_ta_thr_iter)); CHK (1, TDB_DLSYM (tdb, td_symbol_list)); - /* This is required only when thread_db_use_events is on. */ - CHK (thread_db_use_events, TDB_DLSYM (tdb, td_thr_event_enable)); - /* These are not essential. */ CHK (0, TDB_DLSYM (tdb, td_ta_event_addr)); CHK (0, TDB_DLSYM (tdb, td_ta_set_event)); @@ -824,7 +708,7 @@ thread_db_load_search (void) #endif /* USE_LIBTHREAD_DB_DIRECTLY */ int -thread_db_init (int use_events) +thread_db_init (void) { struct process_info *proc = current_process (); @@ -839,31 +723,21 @@ thread_db_init (int use_events) This isn't the only place in gdbserver that assumes that the first process in the list is the thread group leader. */ - thread_db_use_events = use_events; - if (thread_db_load_search ()) { - if (use_events && thread_db_enable_reporting () == 0) - { - /* Keep trying; maybe event reporting will work later. */ - thread_db_mourn (proc); - return 0; - } - /* It's best to avoid td_ta_thr_iter if possible. That walks data structures in the inferior's address space that may be corrupted, or, if the target is running, the list may change while we walk it. In the latter case, it's possible that a thread exits just at the exact time that causes GDBserver to - get stuck in an infinite loop. If the kernel supports clone - events, and /proc/PID/task/ exits, then we already know about + get stuck in an infinite loop. As the kernel supports clone + events and /proc/PID/task/ exists, then we already know about all threads in the process. When we need info out of thread_db on a given thread (e.g., for TLS), we'll use find_one_thread then. That uses thread_db entry points that do not walk libpthread's thread list, so should be safe, as well as more efficient. */ - if (use_events - || !linux_proc_task_list_dir_exists (pid_of (proc))) + if (!linux_proc_task_list_dir_exists (pid_of (proc))) thread_db_find_new_threads (); thread_db_look_up_symbols (); return 1; @@ -929,24 +803,6 @@ disable_thread_event_reporting (struct process_info *proc) } } -static void -remove_thread_event_breakpoints (struct process_info *proc) -{ - struct thread_db *thread_db = proc->priv->thread_db; - - if (thread_db->td_create_bp != NULL) - { - struct thread_info *saved_thread = current_thread; - - switch_to_process (proc); - - delete_breakpoint (thread_db->td_create_bp); - thread_db->td_create_bp = NULL; - - current_thread = saved_thread; - } -} - void thread_db_detach (struct process_info *proc) { @@ -955,7 +811,6 @@ thread_db_detach (struct process_info *proc) if (thread_db) { disable_thread_event_reporting (proc); - remove_thread_event_breakpoints (proc); } } |