diff options
Diffstat (limited to 'gdb/infrun.c')
| -rw-r--r-- | gdb/infrun.c | 1534 |
1 files changed, 1428 insertions, 106 deletions
diff --git a/gdb/infrun.c b/gdb/infrun.c index ffd95cb..2f1df81 100644 --- a/gdb/infrun.c +++ b/gdb/infrun.c @@ -51,6 +51,31 @@ static int hook_stop_stub PARAMS ((char *)); static void delete_breakpoint_current_contents PARAMS ((PTR)); +int inferior_ignoring_startup_exec_events = 0; +int inferior_ignoring_leading_exec_events = 0; + +#ifdef HPUXHPPA +/* wait_for_inferior and normal_stop use this to notify the user + when the inferior stopped in a different thread than it had been + running in. */ +static int switched_from_inferior_pid; +#endif + +/* resume and wait_for_inferior use this to ensure that when + stepping over a hit breakpoint in a threaded application + only the thread that hit the breakpoint is stepped and the + other threads don't continue. This prevents having another + thread run past the breakpoint while it is temporarily + removed. + + This is not thread-specific, so it isn't saved as part of + the infrun state. + + Versions of gdb which don't use the "step == this thread steps + and others continue" model but instead use the "step == this + thread steps and others wait" shouldn't do this. */ +static int thread_step_needed = 0; + void _initialize_infrun PARAMS ((void)); /* GET_LONGJMP_TARGET returns the PC at which longjmp() will resume the @@ -212,6 +237,350 @@ static int breakpoints_failed; static int stop_print_frame; +static struct breakpoint *step_resume_breakpoint = NULL; +static struct breakpoint *through_sigtramp_breakpoint = NULL; + +/* On some platforms (e.g., HP-UX), hardware watchpoints have bad + interactions with an inferior that is running a kernel function + (aka, a system call or "syscall"). wait_for_inferior therefore + may have a need to know when the inferior is in a syscall. This + is a count of the number of inferior threads which are known to + currently be running in a syscall. */ +static int number_of_threads_in_syscalls; + +/* This is used to remember when a fork, vfork or exec event + was caught by a catchpoint, and thus the event is to be + followed at the next resume of the inferior, and not + immediately. */ +static struct { + enum target_waitkind kind; + struct { + int parent_pid; + int saw_parent_fork; + int child_pid; + int saw_child_fork; + int saw_child_exec; + } fork_event; + char * execd_pathname; +} pending_follow; + +/* Some platforms don't allow us to do anything meaningful with a + vforked child until it has exec'd. Vforked processes on such + platforms can only be followed after they've exec'd. + + When this is set to 0, a vfork can be immediately followed, + and an exec can be followed merely as an exec. When this is + set to 1, a vfork event has been seen, but cannot be followed + until the exec is seen. + + (In the latter case, inferior_pid is still the parent of the + vfork, and pending_follow.fork_event.child_pid is the child. The + appropriate process is followed, according to the setting of + follow-fork-mode.) */ +static int follow_vfork_when_exec; + +static char * follow_fork_mode_kind_names [] = { +/* ??rehrauer: The "both" option is broken, by what may be a 10.20 + kernel problem. It's also not terribly useful without a GUI to + help the user drive two debuggers. So for now, I'm disabling + the "both" option. + "parent", "child", "both", "ask" }; + */ + "parent", "child", "ask" }; + +static char * follow_fork_mode_string = NULL; + + +#if defined(HPUXHPPA) +static void +follow_inferior_fork (parent_pid, child_pid, has_forked, has_vforked) + int parent_pid; + int child_pid; + int has_forked; + int has_vforked; +{ + int followed_parent = 0; + int followed_child = 0; + int ima_clone = 0; + + /* Which process did the user want us to follow? */ + char * follow_mode = + savestring (follow_fork_mode_string, strlen (follow_fork_mode_string)); + + /* Or, did the user not know, and want us to ask? */ + if (STREQ (follow_fork_mode_string, "ask")) + { + char requested_mode[100]; + + free (follow_mode); + error ("\"ask\" mode NYI"); + follow_mode = savestring (requested_mode, strlen (requested_mode)); + } + + /* If we're to be following the parent, then detach from child_pid. + We're already following the parent, so need do nothing explicit + for it. */ + if (STREQ (follow_mode, "parent")) + { + followed_parent = 1; + + /* We're already attached to the parent, by default. */ + + /* Before detaching from the child, remove all breakpoints from + it. (This won't actually modify the breakpoint list, but will + physically remove the breakpoints from the child.) */ + if (! has_vforked || ! follow_vfork_when_exec) + { + detach_breakpoints (child_pid); + SOLIB_REMOVE_INFERIOR_HOOK (child_pid); + } + + /* Detach from the child. */ + dont_repeat (); + + target_require_detach (child_pid, "", 1); + } + + /* If we're to be following the child, then attach to it, detach + from inferior_pid, and set inferior_pid to child_pid. */ + else if (STREQ (follow_mode, "child")) + { + char child_pid_spelling [100]; /* Arbitrary length. */ + + followed_child = 1; + + /* Before detaching from the parent, detach all breakpoints from + the child. But only if we're forking, or if we follow vforks + as soon as they happen. (If we're following vforks only when + the child has exec'd, then it's very wrong to try to write + back the "shadow contents" of inserted breakpoints now -- they + belong to the child's pre-exec'd a.out.) */ + if (! has_vforked || ! follow_vfork_when_exec) + { + detach_breakpoints (child_pid); + } + + /* Before detaching from the parent, remove all breakpoints from it. */ + remove_breakpoints (); + + /* Also reset the solib inferior hook from the parent. */ + SOLIB_REMOVE_INFERIOR_HOOK (inferior_pid); + + /* Detach from the parent. */ + dont_repeat (); + target_detach (NULL, 1); + + /* Attach to the child. */ + inferior_pid = child_pid; + sprintf (child_pid_spelling, "%d", child_pid); + dont_repeat (); + + target_require_attach (child_pid_spelling, 1); + + /* Was there a step_resume breakpoint? (There was if the user + did a "next" at the fork() call.) If so, explicitly reset its + thread number. + + step_resumes are a form of bp that are made to be per-thread. + Since we created the step_resume bp when the parent process + was being debugged, and now are switching to the child process, + from the breakpoint package's viewpoint, that's a switch of + "threads". We must update the bp's notion of which thread + it is for, or it'll be ignored when it triggers... */ + if (step_resume_breakpoint && (! has_vforked || ! follow_vfork_when_exec)) + breakpoint_re_set_thread (step_resume_breakpoint); + + /* Reinsert all breakpoints in the child. (The user may've set + breakpoints after catching the fork, in which case those + actually didn't get set in the child, but only in the parent.) */ + if (! has_vforked || ! follow_vfork_when_exec) + { + breakpoint_re_set (); + insert_breakpoints (); + } + } + + /* If we're to be following both parent and child, then fork ourselves, + and attach the debugger clone to the child. */ + else if (STREQ (follow_mode, "both")) + { + char pid_suffix [100]; /* Arbitrary length. */ + + /* Clone ourselves to follow the child. This is the end of our + involvement with child_pid; our clone will take it from here... */ + dont_repeat (); + target_clone_and_follow_inferior (child_pid, &followed_child); + followed_parent = !followed_child; + + /* We continue to follow the parent. To help distinguish the two + debuggers, though, both we and our clone will reset our prompts. */ + sprintf (pid_suffix, "[%d] ", inferior_pid); + set_prompt (strcat (get_prompt (), pid_suffix)); + } + + /* The parent and child of a vfork share the same address space. + Also, on some targets the order in which vfork and exec events + are received for parent in child requires some delicate handling + of the events. + + For instance, on ptrace-based HPUX we receive the child's vfork + event first, at which time the parent has been suspended by the + OS and is essentially untouchable until the child's exit or second + exec event arrives. At that time, the parent's vfork event is + delivered to us, and that's when we see and decide how to follow + the vfork. But to get to that point, we must continue the child + until it execs or exits. To do that smoothly, all breakpoints + must be removed from the child, in case there are any set between + the vfork() and exec() calls. But removing them from the child + also removes them from the parent, due to the shared-address-space + nature of a vfork'd parent and child. On HPUX, therefore, we must + take care to restore the bp's to the parent before we continue it. + Else, it's likely that we may not stop in the expected place. (The + worst scenario is when the user tries to step over a vfork() call; + the step-resume bp must be restored for the step to properly stop + in the parent after the call completes!) + + Sequence of events, as reported to gdb from HPUX: + + Parent Child Action for gdb to take + ------------------------------------------------------- + 1 VFORK Continue child + 2 EXEC + 3 EXEC or EXIT + 4 VFORK */ + if (has_vforked) + { + target_post_follow_vfork (parent_pid, + followed_parent, + child_pid, + followed_child); + } + + pending_follow.fork_event.saw_parent_fork = 0; + pending_follow.fork_event.saw_child_fork = 0; + + free (follow_mode); +} + +static void +follow_fork (parent_pid, child_pid) + int parent_pid; + int child_pid; +{ + follow_inferior_fork (parent_pid, child_pid, 1, 0); +} + + +/* Forward declaration. */ +static void follow_exec PARAMS((int, char *)); + +static void +follow_vfork (parent_pid, child_pid) + int parent_pid; + int child_pid; +{ + follow_inferior_fork (parent_pid, child_pid, 0, 1); + + /* Did we follow the child? Had it exec'd before we saw the parent vfork? */ + if (pending_follow.fork_event.saw_child_exec && (inferior_pid == child_pid)) + { + pending_follow.fork_event.saw_child_exec = 0; + pending_follow.kind = TARGET_WAITKIND_SPURIOUS; + follow_exec (inferior_pid, pending_follow.execd_pathname); + free (pending_follow.execd_pathname); + } +} +#endif /* HPUXHPPA */ + +static void +follow_exec (pid, execd_pathname) + int pid; + char * execd_pathname; +{ +#ifdef HPUXHPPA + int saved_pid = pid; + extern struct target_ops child_ops; + + /* Did this exec() follow a vfork()? If so, we must follow the + vfork now too. Do it before following the exec. */ + if (follow_vfork_when_exec && (pending_follow.kind == TARGET_WAITKIND_VFORKED)) + { + pending_follow.kind = TARGET_WAITKIND_SPURIOUS; + follow_vfork (inferior_pid, pending_follow.fork_event.child_pid); + follow_vfork_when_exec = 0; + saved_pid = inferior_pid; + + /* Did we follow the parent? If so, we're done. If we followed + the child then we must also follow its exec(). */ + if (inferior_pid == pending_follow.fork_event.parent_pid) + return; + } + + /* This is an exec event that we actually wish to pay attention to. + Refresh our symbol table to the newly exec'd program, remove any + momentary bp's, etc. + + If there are breakpoints, they aren't really inserted now, + since the exec() transformed our inferior into a fresh set + of instructions. + + We want to preserve symbolic breakpoints on the list, since + we have hopes that they can be reset after the new a.out's + symbol table is read. + + However, any "raw" breakpoints must be removed from the list + (e.g., the solib bp's), since their address is probably invalid + now. + + And, we DON'T want to call delete_breakpoints() here, since + that may write the bp's "shadow contents" (the instruction + value that was overwritten witha TRAP instruction). Since + we now have a new a.out, those shadow contents aren't valid. */ + update_breakpoints_after_exec (); + + /* If there was one, it's gone now. We cannot truly step-to-next + statement through an exec(). */ + step_resume_breakpoint = NULL; + step_range_start = 0; + step_range_end = 0; + + /* If there was one, it's gone now. */ + through_sigtramp_breakpoint = NULL; + + /* What is this a.out's name? */ + printf_unfiltered ("Executing new program: %s\n", execd_pathname); + + /* We've followed the inferior through an exec. Therefore, the + inferior has essentially been killed & reborn. */ + gdb_flush (gdb_stdout); + target_mourn_inferior (); + inferior_pid = saved_pid; /* Because mourn_inferior resets inferior_pid. */ + push_target (&child_ops); + + /* That a.out is now the one to use. */ + exec_file_attach (execd_pathname, 0); + + /* And also is where symbols can be found. */ + symbol_file_command (execd_pathname, 0); + + /* Reset the shared library package. This ensures that we get + a shlib event when the child reaches "_start", at which point + the dld will have had a chance to initialize the child. */ + SOLIB_RESTART (); + SOLIB_CREATE_INFERIOR_HOOK (inferior_pid); + + /* Reinsert all breakpoints. (Those which were symbolic have + been reset to the proper address in the new a.out, thanks + to symbol_file_command...) */ + insert_breakpoints (); + + /* The next resume of this inferior should bring it to the shlib + startup breakpoints. (If the user had also set bp's on + "main" from the old (parent) process, then they'll auto- + matically get reset there in the new process.) */ +#endif +} + /* Non-zero if we just simulating a single-step. This is needed because we cannot remove the breakpoints in the inferior process until after the `wait' in `wait_for_inferior'. */ @@ -240,6 +609,7 @@ resume (step, sig) int step; enum target_signal sig; { + int should_resume = 1; struct cleanup *old_cleanups = make_cleanup ((make_cleanup_func) resume_cleanups, 0); QUIT; @@ -268,10 +638,84 @@ resume (step, sig) DO_DEFERRED_STORES; #endif +#ifdef HPUXHPPA + /* If there were any forks/vforks/execs that were caught and are + now to be followed, then do so. */ + switch (pending_follow.kind) + { + case (TARGET_WAITKIND_FORKED): + pending_follow.kind = TARGET_WAITKIND_SPURIOUS; + follow_fork (inferior_pid, pending_follow.fork_event.child_pid); + break; + + case (TARGET_WAITKIND_VFORKED): + { + int saw_child_exec = pending_follow.fork_event.saw_child_exec; + + pending_follow.kind = TARGET_WAITKIND_SPURIOUS; + follow_vfork (inferior_pid, pending_follow.fork_event.child_pid); + + /* Did we follow the child, but not yet see the child's exec event? + If so, then it actually ought to be waiting for us; we respond to + parent vfork events. We don't actually want to resume the child + in this situation; we want to just get its exec event. */ + if (! saw_child_exec && (inferior_pid == pending_follow.fork_event.child_pid)) + should_resume = 0; + } + break; + + case (TARGET_WAITKIND_EXECD): + /* If we saw a vfork event but couldn't follow it until we saw + an exec, then now might be the time! */ + pending_follow.kind = TARGET_WAITKIND_SPURIOUS; + /* follow_exec is called as soon as the exec event is seen. */ + break; + + default: + break; + } +#endif /* HPUXHPPA */ + /* Install inferior's terminal modes. */ target_terminal_inferior (); - target_resume (-1, step, sig); + if (should_resume) + { +#ifdef HPUXHPPA + if (thread_step_needed) + { + /* We stopped on a BPT instruction; + don't continue other threads and + just step this thread. */ + thread_step_needed = 0; + + if (!breakpoint_here_p(read_pc())) + { + /* Breakpoint deleted: ok to do regular resume + where all the threads either step or continue. */ + target_resume (-1, step, sig); + } + else + { + if (!step) + { + warning ( "Internal error, changing continue to step." ); + remove_breakpoints (); + breakpoints_inserted = 0; + trap_expected = 1; + step = 1; + } + + target_resume (inferior_pid, step, sig); + } + } + else +#endif /* HPUXHPPA */ + + /* Vanilla resume. */ + target_resume (-1, step, sig); + } + discard_cleanups (old_cleanups); } @@ -325,7 +769,8 @@ proceed (addr, siggnal, step) { /* If there is a breakpoint at the address we will resume at, step one instruction before inserting breakpoints - so that we do not stop right away. */ + so that we do not stop right away (and report a second + hit at this breakpoint). */ if (read_pc () == stop_pc && breakpoint_here_p (read_pc ())) oneproc = 1; @@ -343,24 +788,38 @@ proceed (addr, siggnal, step) oneproc = 1; } else - write_pc (addr); + { + write_pc (addr); + + /* New address; we don't need to single-step a thread + over a breakpoint we just hit, 'cause we aren't + continuing from there. + + It's not worth worrying about the case where a user + asks for a "jump" at the current PC--if they get the + hiccup of re-hiting a hit breakpoint, what else do + they expect? */ + thread_step_needed = 0; + } #ifdef PREPARE_TO_PROCEED - /* In a multi-threaded task we may select another thread and then continue. - - In this case the thread that stopped at a breakpoint will immediately - cause another stop, if it is not stepped over first. On the other hand, - if (ADDR != -1) we only want to single step over the breakpoint if we did - switch to another thread. - - If we are single stepping, don't do any of the above. - (Note that in the current implementation single stepping another - thread after a breakpoint and then continuing will cause the original - breakpoint to be hit again, but you can always continue, so it's not - a big deal.) */ - - if (! step && PREPARE_TO_PROCEED (1) && breakpoint_here_p (read_pc ())) - oneproc = 1; + /* In a multi-threaded task we may select another thread + and then continue or step. + + But if the old thread was stopped at a breakpoint, it + will immediately cause another breakpoint stop without + any execution (i.e. it will report a breakpoint hit + incorrectly). So we must step over it first. + + PREPARE_TO_PROCEED checks the current thread against the thread + that reported the most recent event. If a step-over is required + it returns TRUE and sets the current thread to the old thread. */ + if (PREPARE_TO_PROCEED() && breakpoint_here_p (read_pc())) + { + oneproc = 1; + thread_step_needed = 1; + } + #endif /* PREPARE_TO_PROCEED */ #ifdef HP_OS_BUG @@ -388,6 +847,7 @@ proceed (addr, siggnal, step) error ("Cannot insert breakpoints.\n\ The same program may be running in another process."); } + breakpoints_inserted = 1; } @@ -430,7 +890,6 @@ start_remote () { init_thread_list (); init_wait_for_inferior (); - clear_proceed_status (); stop_soon_quietly = 1; trap_expected = 0; wait_for_inferior (); @@ -451,10 +910,21 @@ init_wait_for_inferior () trap_expected_after_continue = 0; #endif breakpoints_inserted = 0; - breakpoint_init_inferior (); + breakpoint_init_inferior (inf_starting); /* Don't confuse first call to proceed(). */ stop_signal = TARGET_SIGNAL_0; + + /* The first resume is not following a fork/vfork/exec. */ + pending_follow.kind = TARGET_WAITKIND_SPURIOUS; /* I.e., none. */ + pending_follow.fork_event.saw_parent_fork = 0; + pending_follow.fork_event.saw_child_fork = 0; + pending_follow.fork_event.saw_child_exec = 0; + + /* See wait_for_inferior's handling of SYSCALL_ENTRY/RETURN events. */ + number_of_threads_in_syscalls = 0; + + clear_proceed_status (); } static void @@ -462,8 +932,11 @@ delete_breakpoint_current_contents (arg) PTR arg; { struct breakpoint **breakpointp = (struct breakpoint **)arg; - if (*breakpointp != NULL) - delete_breakpoint (*breakpointp); + if (*breakpointp != NULL) + { + delete_breakpoint (*breakpointp); + *breakpointp = NULL; + } } /* Wait for control to return from inferior to debugger. @@ -491,10 +964,18 @@ wait_for_inferior () int current_line; struct symtab *current_symtab; int handling_longjmp = 0; /* FIXME */ - struct breakpoint *step_resume_breakpoint = NULL; - struct breakpoint *through_sigtramp_breakpoint = NULL; int pid; + int saved_inferior_pid; int update_step_sp = 0; + int stepping_through_solib_after_catch = 0; + bpstat stepping_through_solib_catchpoints = NULL; + int enable_hw_watchpoints_after_wait = 0; + int stepping_through_sigtramp = 0; + int new_thread_event; + +#ifdef HAVE_NONSTEPPABLE_WATCHPOINT + int stepped_after_stopped_by_watchpoint; +#endif old_cleanups = make_cleanup (delete_breakpoint_current_contents, &step_resume_breakpoint); @@ -510,8 +991,16 @@ wait_for_inferior () && !handling_longjmp \ && ((step_range_end && step_resume_breakpoint == NULL) \ || trap_expected)) \ + || stepping_through_solib_after_catch \ || bpstat_should_step ()) +thread_step_needed = 0; + +#ifdef HPUXHPPA + /* We'll update this if & when we switch to a new thread. */ + switched_from_inferior_pid = inferior_pid; +#endif + while (1) { extern int overlay_cache_invalid; /* declared in symfile.h */ @@ -530,6 +1019,24 @@ wait_for_inferior () else pid = target_wait (-1, &w); + /* Since we've done a wait, we have a new event. Don't carry + over any expectations about needing to step over a + breakpoint. */ + thread_step_needed = 0; + + /* See comments where a TARGET_WAITKIND_SYSCALL_RETURN event is + serviced in this loop, below. */ + if (enable_hw_watchpoints_after_wait) + { + TARGET_ENABLE_HW_WATCHPOINTS(inferior_pid); + enable_hw_watchpoints_after_wait = 0; + } + + +#ifdef HAVE_NONSTEPPABLE_WATCHPOINT +stepped_after_stopped_by_watchpoint = 0; +#endif + /* Gross. We goto this label from elsewhere in wait_for_inferior when we want @@ -541,14 +1048,23 @@ wait_for_inferior () /* If it's a new process, add it to the thread database */ + new_thread_event = ((pid != inferior_pid) && !in_thread_list (pid)); + if (w.kind != TARGET_WAITKIND_EXITED && w.kind != TARGET_WAITKIND_SIGNALLED - && pid != inferior_pid - && !in_thread_list (pid)) + && new_thread_event) { add_thread (pid); + + +#ifdef HPUXHPPA + fprintf_unfiltered (gdb_stderr, "[New %s]\n", target_pid_or_tid_to_str (pid)); + +#else printf_filtered ("[New %s]\n", target_pid_to_str (pid)); +#endif +#if 0 /* We may want to consider not doing a resume here in order to give the user a chance to play with the new thread. It might be good to make that a user-settable option. */ @@ -559,7 +1075,8 @@ wait_for_inferior () target_resume (-1, 0, TARGET_SIGNAL_0); continue; - } +#endif + } switch (w.kind) { @@ -648,13 +1165,235 @@ wait_for_inferior () singlestep_breakpoints_inserted_p = 0; /*SOFTWARE_SINGLE_STEP_P*/ goto stop_stepping; + /* The following are the only cases in which we keep going; + the above cases end in a continue or goto. */ + case TARGET_WAITKIND_FORKED: + stop_signal = TARGET_SIGNAL_TRAP; + pending_follow.kind = w.kind; + + /* Ignore fork events reported for the parent; we're only + interested in reacting to forks of the child. Note that + we expect the child's fork event to be available if we + waited for it now. */ + if (inferior_pid == pid) + { + pending_follow.fork_event.saw_parent_fork = 1; + pending_follow.fork_event.parent_pid = pid; + pending_follow.fork_event.child_pid = w.value.related_pid; + continue; + } + else + { + pending_follow.fork_event.saw_child_fork = 1; + pending_follow.fork_event.child_pid = pid; + pending_follow.fork_event.parent_pid = w.value.related_pid; + } + + stop_pc = read_pc_pid (pid); + saved_inferior_pid = inferior_pid; + inferior_pid = pid; + stop_bpstat = bpstat_stop_status (&stop_pc, +#if DECR_PC_AFTER_BREAK + (prev_pc != stop_pc - DECR_PC_AFTER_BREAK + && CURRENTLY_STEPPING ()) +#else /* DECR_PC_AFTER_BREAK zero */ + 0 +#endif /* DECR_PC_AFTER_BREAK zero */ + ); + random_signal = ! bpstat_explains_signal (stop_bpstat); + inferior_pid = saved_inferior_pid; + goto process_event_stop_test; + + /* If this a platform which doesn't allow a debugger to touch a + vfork'd inferior until after it exec's, then we'd best keep + our fingers entirely off the inferior, other than continuing + it. This has the unfortunate side-effect that catchpoints + of vforks will be ignored. But since the platform doesn't + allow the inferior be touched at vfork time, there's really + little choice. */ + case TARGET_WAITKIND_VFORKED: + stop_signal = TARGET_SIGNAL_TRAP; + pending_follow.kind = w.kind; + + /* Is this a vfork of the parent? If so, then give any + vfork catchpoints a chance to trigger now. (It's + dangerous to do so if the child canot be touched until + it execs, and the child has not yet exec'd. We probably + should warn the user to that effect when the catchpoint + triggers...) */ + if (pid == inferior_pid) + { + pending_follow.fork_event.saw_parent_fork = 1; + pending_follow.fork_event.parent_pid = pid; + pending_follow.fork_event.child_pid = w.value.related_pid; + } + + /* If we've seen the child's vfork event but cannot really touch + the child until it execs, then we must continue the child now. + Else, give any vfork catchpoints a chance to trigger now. */ + else + { + pending_follow.fork_event.saw_child_fork = 1; + pending_follow.fork_event.child_pid = pid; + pending_follow.fork_event.parent_pid = w.value.related_pid; + target_post_startup_inferior (pending_follow.fork_event.child_pid); + follow_vfork_when_exec = ! target_can_follow_vfork_prior_to_exec (); + if (follow_vfork_when_exec) + { + target_resume (pid, 0, TARGET_SIGNAL_0); + continue; + } + } + + stop_pc = read_pc (); + stop_bpstat = bpstat_stop_status (&stop_pc, +#if DECR_PC_AFTER_BREAK + (prev_pc != stop_pc - DECR_PC_AFTER_BREAK + && CURRENTLY_STEPPING ()) +#else /* DECR_PC_AFTER_BREAK zero */ + 0 +#endif /* DECR_PC_AFTER_BREAK zero */ + ); + random_signal = ! bpstat_explains_signal (stop_bpstat); + goto process_event_stop_test; + + case TARGET_WAITKIND_EXECD: + stop_signal = TARGET_SIGNAL_TRAP; + + /* Is this a target which reports multiple exec events per actual + call to exec()? (HP-UX using ptrace does, for example.) If so, + ignore all but the last one. Just resume the exec'r, and wait + for the next exec event. */ + if (inferior_ignoring_leading_exec_events) + { + inferior_ignoring_leading_exec_events--; + if (pending_follow.kind == TARGET_WAITKIND_VFORKED) + ENSURE_VFORKING_PARENT_REMAINS_STOPPED(pending_follow.fork_event.parent_pid); + target_resume (pid, 0, TARGET_SIGNAL_0); + continue; + } + inferior_ignoring_leading_exec_events = + target_reported_exec_events_per_exec_call () - 1; + + pending_follow.execd_pathname = savestring (w.value.execd_pathname, + strlen (w.value.execd_pathname)); + + /* Did inferior_pid exec, or did a (possibly not-yet-followed) + child of a vfork exec? + + ??rehrauer: This is unabashedly an HP-UX specific thing. On + HP-UX, events associated with a vforking inferior come in + threes: a vfork event for the child (always first), followed + a vfork event for the parent and an exec event for the child. + The latter two can come in either order. + + If we get the parent vfork event first, life's good: We follow + either the parent or child, and then the child's exec event is + a "don't care". + + But if we get the child's exec event first, then we delay + responding to it until we handle the parent's vfork. Because, + otherwise we can't satisfy a "catch vfork". */ + if (pending_follow.kind == TARGET_WAITKIND_VFORKED) + { + pending_follow.fork_event.saw_child_exec = 1; + + /* On some targets, the child must be resumed before + the parent vfork event is delivered. A single-step + suffices. */ + if (RESUME_EXECD_VFORKING_CHILD_TO_GET_PARENT_VFORK()) + target_resume (pid, 1, TARGET_SIGNAL_0); + continue; /* We expect the parent vfork event to be available now. */ + } + + /* This causes the eventpoints and symbol table to be reset. Must + do this now, before trying to determine whether to stop. */ + follow_exec (inferior_pid, pending_follow.execd_pathname); + free (pending_follow.execd_pathname); + + stop_pc = read_pc_pid (pid); + saved_inferior_pid = inferior_pid; + inferior_pid = pid; + stop_bpstat = bpstat_stop_status (&stop_pc, +#if DECR_PC_AFTER_BREAK + (prev_pc != stop_pc - DECR_PC_AFTER_BREAK + && CURRENTLY_STEPPING ()) +#else /* DECR_PC_AFTER_BREAK zero */ + 0 +#endif /* DECR_PC_AFTER_BREAK zero */ + ); + random_signal = ! bpstat_explains_signal (stop_bpstat); + inferior_pid = saved_inferior_pid; + goto process_event_stop_test; + + /* These syscall events are returned on HP-UX, as part of its + implementation of page-protection-based "hardware" watchpoints. + HP-UX has unfortunate interactions between page-protections and + some system calls. Our solution is to disable hardware watches + when a system call is entered, and reenable them when the syscall + completes. The downside of this is that we may miss the precise + point at which a watched piece of memory is modified. "Oh well." + + Note that we may have multiple threads running, which may each + enter syscalls at roughly the same time. Since we don't have a + good notion currently of whether a watched piece of memory is + thread-private, we'd best not have any page-protections active + when any thread is in a syscall. Thus, we only want to reenable + hardware watches when no threads are in a syscall. + + Also, be careful not to try to gather much state about a thread + that's in a syscall. It's frequently a losing proposition. */ + case TARGET_WAITKIND_SYSCALL_ENTRY: + number_of_threads_in_syscalls++; + if (number_of_threads_in_syscalls == 1) + { + TARGET_DISABLE_HW_WATCHPOINTS(inferior_pid); + } + resume (0, TARGET_SIGNAL_0); + continue; + + /* Before examining the threads further, step this thread to + get it entirely out of the syscall. (We get notice of the + event when the thread is just on the verge of exiting a + syscall. Stepping one instruction seems to get it back + into user code.) + + Note that although the logical place to reenable h/w watches + is here, we cannot. We cannot reenable them before stepping + the thread (this causes the next wait on the thread to hang). + + Nor can we enable them after stepping until we've done a + wait. Thus, we simply set the flag enable_hw_watchpoints_after_wait + here, which will be serviced immediately after the target + is waited on. */ + case TARGET_WAITKIND_SYSCALL_RETURN: + target_resume (pid, 1, TARGET_SIGNAL_0); + + if (number_of_threads_in_syscalls > 0) + { + number_of_threads_in_syscalls--; + enable_hw_watchpoints_after_wait = + (number_of_threads_in_syscalls == 0); + } + continue; + case TARGET_WAITKIND_STOPPED: - /* This is the only case in which we keep going; the above cases - end in a continue or goto. */ + stop_signal = w.value.sig; break; } - stop_signal = w.value.sig; + /* We may want to consider not doing a resume here in order to give + the user a chance to play with the new thread. It might be good + to make that a user-settable option. */ + + /* At this point, all threads are stopped (happens automatically in + either the OS or the native code). Therefore we need to continue + all threads in order to make progress. */ + if (new_thread_event) + { + target_resume (-1, 0, TARGET_SIGNAL_0); + continue; + } stop_pc = read_pc_pid (pid); @@ -673,10 +1412,24 @@ wait_for_inferior () random_signal = 0; if (!breakpoint_thread_match (stop_pc - DECR_PC_AFTER_BREAK, pid)) { + int remove_status; + /* Saw a breakpoint, but it was hit by the wrong thread. Just continue. */ write_pc_pid (stop_pc - DECR_PC_AFTER_BREAK, pid); - remove_breakpoints (); + remove_status = remove_breakpoints (); + /* Did we fail to remove breakpoints? If so, try to set the + PC past the bp. (There's at least one situation in which + we can fail to remove the bp's: On HP-UX's that use ttrace, + we can't change the address space of a vforking child process + until the child exits (well, okay, not then either :-) or + execs. */ + if (remove_status != 0) + { + write_pc_pid (stop_pc - DECR_PC_AFTER_BREAK + 4, pid); + } + else + { target_resume (pid, 1, TARGET_SIGNAL_0); /* Single step */ /* FIXME: What if a signal arrives instead of the single-step happening? */ @@ -686,36 +1439,60 @@ wait_for_inferior () else target_wait (pid, &w); insert_breakpoints (); + } /* We need to restart all the threads now. */ target_resume (-1, 0, TARGET_SIGNAL_0); continue; } - } + else + { + /* This breakpoint matches--either it is the right + thread or it's a generic breakpoint for all threads. + Remember that we'll need to step just _this_ thread + on any following user continuation! */ + thread_step_needed = 1; + } + } } else random_signal = 1; /* See if something interesting happened to the non-current thread. If so, then switch to that thread, and eventually give control back to - the user. */ + the user. - if (pid != inferior_pid) + Note that if there's any kind of pending follow (i.e., of a fork, + vfork or exec), we don't want to do this now. Rather, we'll let + the next resume handle it. */ + if ((pid != inferior_pid) && + (pending_follow.kind == TARGET_WAITKIND_SPURIOUS)) { int printed = 0; /* If it's a random signal for a non-current thread, notify user - if he's expressed an interest. */ - + if he's expressed an interest. */ if (random_signal && signal_print[stop_signal]) { +/* ??rehrauer: I don't understand the rationale for this code. If the + inferior will stop as a result of this signal, then the act of handling + the stop ought to print a message that's couches the stoppage in user + terms, e.g., "Stopped for breakpoint/watchpoint". If the inferior + won't stop as a result of the signal -- i.e., if the signal is merely + a side-effect of something GDB's doing "under the covers" for the + user, such as stepping threads over a breakpoint they shouldn't stop + for -- then the message seems to be a serious annoyance at best. + + For now, remove the message altogether. */ +#if 0 printed = 1; target_terminal_ours_for_output (); printf_filtered ("\nProgram received signal %s, %s.\n", target_signal_to_name (stop_signal), target_signal_to_string (stop_signal)); gdb_flush (gdb_stdout); +#endif } /* If it's not SIGTRAP and not a signal we want to stop for, then @@ -745,7 +1522,14 @@ wait_for_inferior () through_sigtramp_breakpoint, step_range_start, step_range_end, step_frame_address, handling_longjmp, - another_trap); + another_trap, + stepping_through_solib_after_catch, + stepping_through_solib_catchpoints, + stepping_through_sigtramp); + +#ifdef HPUXHPPA + switched_from_inferior_pid = inferior_pid; +#endif inferior_pid = pid; @@ -756,7 +1540,10 @@ wait_for_inferior () &through_sigtramp_breakpoint, &step_range_start, &step_range_end, &step_frame_address, &handling_longjmp, - &another_trap); + &another_trap, + &stepping_through_solib_after_catch, + &stepping_through_solib_catchpoints, + &stepping_through_sigtramp); if (context_hook) context_hook (pid_to_thread_id (pid)); @@ -776,10 +1563,16 @@ wait_for_inferior () it so that the user won't be confused when GDB appears to be ready to execute it. */ +#if 0 /* XXX DEBUG */ + printf ("infrun.c:1607: pc = 0x%x\n", read_pc ()); +#endif + /* if (INSTRUCTION_NULLIFIED && CURRENTLY_STEPPING ()) */ if (INSTRUCTION_NULLIFIED) { struct target_waitstatus tmpstatus; - +#if 0 + all_registers_info ((char *)0, 0); +#endif registers_changed (); target_resume (pid, 1, TARGET_SIGNAL_0); @@ -792,7 +1585,6 @@ wait_for_inferior () else target_wait (pid, &tmpstatus); - goto have_waited; } @@ -839,9 +1631,11 @@ wait_for_inferior () else target_wait (pid, &w); insert_breakpoints (); + /* FIXME-maybe: is this cleaner than setting a flag? Does it handle things like signals arriving and other things happening in combination correctly? */ + stepped_after_stopped_by_watchpoint = 1; goto have_waited; } #endif @@ -927,7 +1721,7 @@ wait_for_inferior () (CURRENTLY_STEPPING () && prev_pc != stop_pc - DECR_PC_AFTER_BREAK && !(step_range_end - && read_sp () INNER_THAN (step_sp - 16))) : + && INNER_THAN (read_sp (), (step_sp - 16)))) : 0) ); /* Following in case break condition called a @@ -944,6 +1738,7 @@ wait_for_inferior () FRAME_FP (get_current_frame ())) #endif /* No CALL_DUMMY_BREAKPOINT_OFFSET. */ || (step_range_end && step_resume_breakpoint == NULL)); + else { random_signal @@ -960,8 +1755,57 @@ wait_for_inferior () stop_signal = TARGET_SIGNAL_TRAP; } } + + /* When we reach this point, we've pretty much decided + that the reason for stopping must've been a random + (unexpected) signal. */ + else random_signal = 1; + /* If a fork, vfork or exec event was seen, then there are two + possible responses we can make: + + 1. If a catchpoint triggers for the event (random_signal == 0), + then we must stop now and issue a prompt. We will resume + the inferior when the user tells us to. + 2. If no catchpoint triggers for the event (random_signal == 1), + then we must resume the inferior now and keep checking. + + In either case, we must take appropriate steps to "follow" the + the fork/vfork/exec when the inferior is resumed. For example, + if follow-fork-mode is "child", then we must detach from the + parent inferior and follow the new child inferior. + + In either case, setting pending_follow causes the next resume() + to take the appropriate following action. */ +process_event_stop_test: + if (w.kind == TARGET_WAITKIND_FORKED) + { + if (random_signal) /* I.e., no catchpoint triggered for this. */ + { + trap_expected = 1; + stop_signal = TARGET_SIGNAL_0; + goto keep_going; + } + } + else if (w.kind == TARGET_WAITKIND_VFORKED) + { + if (random_signal) /* I.e., no catchpoint triggered for this. */ + { + stop_signal = TARGET_SIGNAL_0; + goto keep_going; + } + } + else if (w.kind == TARGET_WAITKIND_EXECD) + { + pending_follow.kind = w.kind; + if (random_signal) /* I.e., no catchpoint triggered for this. */ + { + trap_expected = 1; + stop_signal = TARGET_SIGNAL_0; + goto keep_going; + } + } /* For the program's own signals, act according to the signal handling tables. */ @@ -1000,9 +1844,16 @@ wait_for_inferior () if (signal_program[stop_signal] == 0) stop_signal = TARGET_SIGNAL_0; - /* I'm not sure whether this needs to be check_sigtramp2 or - whether it could/should be keep_going. */ - goto check_sigtramp2; + /* If we're in the middle of a "next" command, let the code for + stepping over a function handle this. pai/1997-09-10 + + A previous comment here suggested it was possible to change + this to jump to keep_going in all cases. */ + + if (step_over_calls > 0) + goto step_over_function; + else + goto check_sigtramp2; } /* Handle cases caused by hitting a breakpoint. */ @@ -1064,8 +1915,8 @@ wait_for_inferior () #if 0 /* FIXME - Need to implement nested temporary breakpoints */ if (step_over_calls - && (FRAME_FP (get_current_frame ()) - INNER_THAN step_frame_address)) + && (INNER_THAN (FRAME_FP (get_current_frame ()), + step_frame_address))) { another_trap = 1; goto keep_going; @@ -1079,7 +1930,10 @@ wait_for_inferior () case BPSTAT_WHAT_SINGLE: if (breakpoints_inserted) - remove_breakpoints (); + { + thread_step_needed = 1; + remove_breakpoints (); + } breakpoints_inserted = 0; another_trap = 1; /* Still need to check other stuff, at least the case @@ -1105,6 +1959,21 @@ wait_for_inferior () goto stop_stepping; case BPSTAT_WHAT_STEP_RESUME: + /* This proably demands a more elegant solution, but, yeah + right... + + This function's use of the simple variable step_resume_breakpoint + doesn't seem to accomodate simultaneously active step-resume bp's, + although the breakpoint list certainly can. + + If we reach here and step_resume_breakpoint is already NULL, then + apparently we have multiple active step-resume bp's. We'll just + delete the breakpoint we stopped at, and carry on. */ + if (step_resume_breakpoint == NULL) + { + step_resume_breakpoint = + bpstat_find_step_resume_breakpoint (stop_bpstat); + } delete_breakpoint (step_resume_breakpoint); step_resume_breakpoint = NULL; break; @@ -1121,6 +1990,7 @@ wait_for_inferior () break; case BPSTAT_WHAT_CHECK_SHLIBS: + case BPSTAT_WHAT_CHECK_SHLIBS_RESUME_FROM_HOOK: #ifdef SOLIB_ADD { extern int auto_solib_add; @@ -1156,6 +2026,40 @@ wait_for_inferior () stop_print_frame = 0; goto stop_stepping; } + + /* If we stopped due to an explicit catchpoint, then the + (see above) call to SOLIB_ADD pulled in any symbols + from a newly-loaded library, if appropriate. + + We do want the inferior to stop, but not where it is + now, which is in the dynamic linker callback. Rather, + we would like it stop in the user's program, just after + the call that caused this catchpoint to trigger. That + gives the user a more useful vantage from which to + examine their program's state. */ + else if (what.main_action == BPSTAT_WHAT_CHECK_SHLIBS_RESUME_FROM_HOOK) + { + /* ??rehrauer: If I could figure out how to get the + right return PC from here, we could just set a temp + breakpoint and resume. I'm not sure we can without + cracking open the dld's shared libraries and sniffing + their unwind tables and text/data ranges, and that's + not a terribly portable notion. + + Until that time, we must step the inferior out of the + dld callback, and also out of the dld itself (and any + code or stubs in libdld.sl, such as "shl_load" and + friends) until we reach non-dld code. At that point, + we can stop stepping. */ + bpstat_get_triggered_catchpoints (stop_bpstat, + &stepping_through_solib_catchpoints); + stepping_through_solib_after_catch = 1; + + /* Be sure to lift all breakpoints, so the inferior does + actually step past this point... */ + another_trap = 1; + break; + } else { /* We want to step over this breakpoint, then keep going. */ @@ -1180,6 +2084,29 @@ wait_for_inferior () test for stepping. But, if not stepping, do not stop. */ + /* Are we stepping to get the inferior out of the dynamic + linker's hook (and possibly the dld itself) after catching + a shlib event? */ + if (stepping_through_solib_after_catch) + { +#if defined(SOLIB_ADD) + /* Have we reached our destination? If not, keep going. */ + if (SOLIB_IN_DYNAMIC_LINKER (pid, stop_pc)) + { + another_trap = 1; + goto keep_going; + } +#endif + /* Else, stop and report the catchpoint(s) whose triggering + caused us to begin stepping. */ + stepping_through_solib_after_catch = 0; + bpstat_clear (&stop_bpstat); + stop_bpstat = bpstat_copy (stepping_through_solib_catchpoints); + bpstat_clear (&stepping_through_solib_catchpoints); + stop_print_frame = 1; + goto stop_stepping; + } + #ifndef CALL_DUMMY_BREAKPOINT_OFFSET /* This is the old way of detecting the end of the stack dummy. An architecture which defines CALL_DUMMY_BREAKPOINT_OFFSET gets @@ -1190,7 +2117,7 @@ wait_for_inferior () just stop silently, unless the user was doing an si/ni, in which case she'd better know what she's doing. */ - if (PC_IN_CALL_DUMMY (stop_pc, read_sp (), FRAME_FP (get_current_frame ())) + if (CALL_DUMMY_HAS_COMPLETED (stop_pc, read_sp (), FRAME_FP (get_current_frame ())) && !step_range_end) { stop_print_frame = 0; @@ -1216,7 +2143,11 @@ wait_for_inferior () whether it could/should be keep_going. */ goto check_sigtramp2; - /* If stepping through a line, keep going if still within it. */ + /* If stepping through a line, keep going if still within it. + + Note that step_range_end is the address of the first instruction + beyond the step range, and NOT the address of the last instruction + within it! */ if (stop_pc >= step_range_start && stop_pc < step_range_end #if 0 @@ -1229,7 +2160,7 @@ wait_for_inferior () just changed, we've stepped outside */ && !(stop_pc == step_range_start && FRAME_FP (get_current_frame ()) - && (read_sp () INNER_THAN step_sp + && (INNER_THAN (read_sp (), step_sp) || FRAME_FP (get_current_frame ()) != step_frame_address)) #endif ) @@ -1256,34 +2187,70 @@ wait_for_inferior () /* Did we just take a signal? */ if (IN_SIGTRAMP (stop_pc, stop_func_name) && !IN_SIGTRAMP (prev_pc, prev_func_name) - && read_sp () INNER_THAN step_sp) + && INNER_THAN (read_sp (), step_sp)) { /* We've just taken a signal; go until we are back to the point where we took it and one more. */ - /* This code is needed at least in the following case: - The user types "next" and then a signal arrives (before - the "next" is done). */ + /* Note: The test above succeeds not only when we stepped + into a signal handler, but also when we step past the last + statement of a signal handler and end up in the return stub + of the signal handler trampoline. To distinguish between + these two cases, check that the frame is INNER_THAN the + previous one below. pai/1997-09-11 */ - /* Note that if we are stopped at a breakpoint, then we need - the step_resume breakpoint to override any breakpoints at - the same location, so that we will still step over the - breakpoint even though the signal happened. */ { - struct symtab_and_line sr_sal; - - INIT_SAL (&sr_sal); /* initialize to zeroes */ - sr_sal.pc = prev_pc; - sr_sal.section = find_pc_overlay (sr_sal.pc); - /* We could probably be setting the frame to - step_frame_address; I don't think anyone thought to try it. */ - step_resume_breakpoint = - set_momentary_breakpoint (sr_sal, NULL, bp_step_resume); - if (breakpoints_inserted) - insert_breakpoints (); + CORE_ADDR current_frame = FRAME_FP (get_current_frame()); + + if (INNER_THAN (current_frame, step_frame_address)) + { + /* We have just taken a signal; go until we are back to + the point where we took it and one more. */ + + /* This code is needed at least in the following case: + The user types "next" and then a signal arrives (before + the "next" is done). */ + + /* Note that if we are stopped at a breakpoint, then we need + the step_resume breakpoint to override any breakpoints at + the same location, so that we will still step over the + breakpoint even though the signal happened. */ + struct symtab_and_line sr_sal; + + INIT_SAL(&sr_sal); + sr_sal.symtab = NULL; + sr_sal.line = 0; + sr_sal.pc = prev_pc; + /* We could probably be setting the frame to + step_frame_address; I don't think anyone thought to try it. */ + step_resume_breakpoint = + set_momentary_breakpoint (sr_sal, NULL, bp_step_resume); + if (breakpoints_inserted) + insert_breakpoints (); + } + else + { + /* We just stepped out of a signal handler and into its calling + trampoline. + + Normally, we'd jump to step_over_function from here, + but for some reason GDB can't unwind the stack correctly to find + the real PC for the point user code where the signal trampoline will + return -- FRAME_SAVED_PC fails, at least on HP-UX 10.20. But signal + trampolines are pretty small stubs of code, anyway, so it's OK instead + to just single-step out. Note: assuming such trampolines don't + exhibit recursion on any platform... */ + find_pc_partial_function (stop_pc, &stop_func_name, &stop_func_start, + &stop_func_end); + /* Readjust stepping range */ + step_range_start = stop_func_start; + step_range_end = stop_func_end; + stepping_through_sigtramp = 1; + } } + /* If this is stepi or nexti, make sure that the stepping range gets us past that instruction. */ if (step_range_end == 1) @@ -1320,7 +2287,7 @@ wait_for_inferior () SKIP_PROLOGUE (prologue_pc); } - if (!(step_sp INNER_THAN read_sp ()) /* don't mistake (sig)return as a call */ + if (!(INNER_THAN (step_sp, read_sp ())) /* don't mistake (sig)return as a call */ && (/* Might be a non-recursive call. If the symbols are missing enough that stop_func_start == prev_func_start even though they are really two functions, we will treat some calls as @@ -1373,7 +2340,8 @@ wait_for_inferior () Mips, PA, and Power architectures with good results. */ if (stop_pc == stop_func_start /* Quick test */ - || in_prologue (stop_pc, stop_func_start) + || (in_prologue (stop_pc, stop_func_start) && + ! IN_SOLIB_RETURN_TRAMPOLINE (stop_pc, stop_func_name)) || IN_SOLIB_CALL_TRAMPOLINE (stop_pc, stop_func_name) || stop_func_name == 0) #endif @@ -1440,15 +2408,47 @@ step_over_function: /* Set a special breakpoint after the return */ struct symtab_and_line sr_sal; - INIT_SAL (&sr_sal); /* initialize to zeroes */ - sr_sal.pc = - ADDR_BITS_REMOVE (SAVED_PC_AFTER_CALL (get_current_frame ())); - sr_sal.section = find_pc_overlay (sr_sal.pc); - step_resume_breakpoint = - set_momentary_breakpoint (sr_sal, get_current_frame (), - bp_step_resume); - if (!IN_SOLIB_DYNSYM_RESOLVE_CODE (sr_sal.pc)) - step_resume_breakpoint->frame = step_frame_address; + INIT_SAL(&sr_sal); + sr_sal.symtab = NULL; + sr_sal.line = 0; + + /* If we came here after encountering a signal in the middle of + a "next", use the stashed-away previous frame pc */ + sr_sal.pc + = stopped_by_random_signal + ? prev_pc + : ADDR_BITS_REMOVE (SAVED_PC_AFTER_CALL (get_current_frame ())); + + step_resume_breakpoint = + set_momentary_breakpoint (sr_sal, + stopped_by_random_signal ? NULL : get_current_frame (), + bp_step_resume); + + /* We've just entered a callee, and we wish to resume until + it returns to the caller. Setting a step_resume bp on + the return PC will catch a return from the callee. + + However, if the callee is recursing, we want to be careful + not to catch returns of those recursive calls, but of THIS + instance of the call. + + To do this, we set the step_resume bp's frame to our current + caller's frame (step_frame_address, which is set by the "next" + or "until" command, before execution begins). + + But ... don't do it if we're single-stepping out of a sigtramp, + because the reason we're single-stepping is precisely because + unwinding is a problem (HP-UX 10.20, e.g.) and the frame address + is likely to be incorrect. No danger of sigtramp recursion */ + + if (stepping_through_sigtramp) + { + step_resume_breakpoint->frame = NULL; + stepping_through_sigtramp = 0; + } + else if (!IN_SOLIB_DYNSYM_RESOLVE_CODE (sr_sal.pc)) + step_resume_breakpoint->frame = step_frame_address; + if (breakpoints_inserted) insert_breakpoints (); } @@ -1510,7 +2510,7 @@ step_into_function: /* We've wandered out of the step range. */ sal = find_pc_line(stop_pc, 0); - + if (step_range_end == 1) { /* It is stepi or nexti. We always want to stop stepping after @@ -1561,7 +2561,7 @@ step_into_function: break; } - if (stop_pc == sal.pc + if ((stop_pc == sal.pc) && (current_line != sal.line || current_symtab != sal.symtab)) { /* We are at the start of a different line. So stop. Note that @@ -1594,13 +2594,24 @@ step_into_function: step_frame_address = FRAME_FP (get_current_frame ()); current_line = sal.line; current_symtab = sal.symtab; + + /* In the case where we just stepped out of a function into the middle + of a line of the caller, continue stepping, but step_frame_address + must be modified to current frame */ + { + CORE_ADDR current_frame = FRAME_FP (get_current_frame()); + if (!(INNER_THAN (current_frame, step_frame_address))) + step_frame_address = current_frame; + } + + goto keep_going; check_sigtramp2: if (trap_expected && IN_SIGTRAMP (stop_pc, stop_func_name) && !IN_SIGTRAMP (prev_pc, prev_func_name) - && read_sp () INNER_THAN step_sp) + && INNER_THAN (read_sp (), step_sp)) { /* What has happened here is that we have just stepped the inferior with a signal (because it is a signal which shouldn't make @@ -1634,6 +2645,17 @@ step_into_function: It's really much cleaner to do a goto than a maze of if-else conditions. */ + /* ??rehrauer: ttrace on HP-UX theoretically allows one to debug + a vforked child beetween its creation and subsequent exit or + call to exec(). However, I had big problems in this rather + creaky exec engine, getting that to work. The fundamental + problem is that I'm trying to debug two processes via an + engine that only understands a single process with possibly + multiple threads. + + Hence, this spot is known to have problems when + target_can_follow_vfork_prior_to_exec returns 1. */ + /* Save the pc before execution, to compare with pc after stop. */ prev_pc = read_pc (); /* Might have been DECR_AFTER_BREAK */ prev_func_start = stop_func_start; /* Ok, since if DECR_PC_AFTER @@ -1714,6 +2736,28 @@ step_into_function: stop_stepping: if (target_has_execution) { + /* Are we stopping for a vfork event? We only stop when we see + the child's event. However, we may not yet have seen the + parent's event. And, inferior_pid is still set to the parent's + pid, until we resume again and follow either the parent or child. + + To ensure that we can really touch inferior_pid (aka, the + parent process) -- which calls to functions like read_pc + implicitly do -- wait on the parent if necessary. */ + if ((pending_follow.kind == TARGET_WAITKIND_VFORKED) + && ! pending_follow.fork_event.saw_parent_fork) + { + int parent_pid; + + do { + if (target_wait_hook) + parent_pid = target_wait_hook (-1, &w); + else + parent_pid = target_wait (-1, &w); + } while (parent_pid != inferior_pid); + } + + /* Assuming the inferior still exists, set these up for next time, just like we did above if we didn't break out of the loop. */ @@ -1723,6 +2767,72 @@ step_into_function: } do_cleanups (old_cleanups); } + +/* This function returns TRUE if ep is an internal breakpoint + set to catch generic shared library (aka dynamically-linked + library) events. (This is *NOT* the same as a catchpoint for a + shlib event. The latter is something a user can set; this is + something gdb sets for its own use, and isn't ever shown to a + user.) */ +static int +is_internal_shlib_eventpoint (ep) + struct breakpoint * ep; +{ + return + (ep->type == bp_shlib_event) + ; +} + +/* This function returns TRUE if bs indicates that the inferior + stopped due to a shared library (aka dynamically-linked library) + event. */ +static int +stopped_for_internal_shlib_event (bs) + bpstat bs; +{ + /* Note that multiple eventpoints may've caused the stop. Any + that are associated with shlib events will be accepted. */ + for (;bs != NULL; bs = bs->next) + { + if ((bs->breakpoint_at != NULL) + && is_internal_shlib_eventpoint (bs->breakpoint_at)) + return 1; + } + + /* If we get here, then no candidate was found. */ + return 0; +} + +/* This function returns TRUE if bs indicates that the inferior + stopped due to a shared library (aka dynamically-linked library) + event caught by a catchpoint. + + If TRUE, cp_p is set to point to the catchpoint. + + Else, the value of cp_p is undefined. */ +static int +stopped_for_shlib_catchpoint (bs, cp_p) + bpstat bs; + struct breakpoint ** cp_p; +{ + /* Note that multiple eventpoints may've caused the stop. Any + that are associated with shlib events will be accepted. */ + *cp_p = NULL; + + for (;bs != NULL; bs = bs->next) + { + if ((bs->breakpoint_at != NULL) + && ep_is_shlib_catchpoint (bs->breakpoint_at)) + { + *cp_p = bs->breakpoint_at; + return 1; + } + } + + /* If we get here, then no candidate was found. */ + return 0; +} + /* Here to return control to GDB when the inferior stops for real. Print appropriate messages, remove breakpoints, give terminal our modes. @@ -1735,6 +2845,23 @@ step_into_function: void normal_stop () { + +#ifdef HPUXHPPA + /* As with the notification of thread events, we want to delay + notifying the user that we've switched thread context until + the inferior actually stops. + + (Note that there's no point in saying anything if the inferior + has exited!) */ + if ((switched_from_inferior_pid != inferior_pid) && + target_has_execution) + { + target_terminal_ours_for_output (); + printf_filtered ("[Switched to %s]\n", target_pid_or_tid_to_str (inferior_pid)); + switched_from_inferior_pid = inferior_pid; + } +#endif + /* Make sure that the current_frame's pc is correct. This is a correction for setting up the frame info before doing DECR_PC_AFTER_BREAK */ @@ -1750,6 +2877,7 @@ The same program may be running in another process.\n"); } if (target_has_execution && breakpoints_inserted) + { if (remove_breakpoints ()) { target_terminal_ours_for_output (); @@ -1757,7 +2885,7 @@ The same program may be running in another process.\n"); It might be running in another process.\n\ Further execution is probably impossible.\n"); } - + } breakpoints_inserted = 0; /* Delete the breakpoint we stopped at, if it wants to be deleted. @@ -1771,46 +2899,80 @@ Further execution is probably impossible.\n"); if (stopped_by_random_signal) disable_current_display (); + /* Don't print a message if in the middle of doing a "step n" + operation for n > 1 */ if (step_multi && stop_step) goto done; target_terminal_ours (); - if (stop_bpstat - && stop_bpstat->breakpoint_at - && stop_bpstat->breakpoint_at->type == bp_shlib_event) - printf_filtered ("Stopped due to shared library event\n"); + /* Did we stop because the user set the stop_on_solib_events + variable? (If so, we report this as a generic, "Stopped due + to shlib event" message.) */ + if (stopped_for_internal_shlib_event (stop_bpstat)) + { + printf_filtered ("Stopped due to shared library event\n"); + } /* Look up the hook_stop and run it if it exists. */ - if (stop_command->hook) + if (stop_command && stop_command->hook) { catch_errors (hook_stop_stub, (char *)stop_command->hook, "Error while running hook_stop:\n", RETURN_MASK_ALL); } if (!target_has_stack) + { + goto done; + } + + /* Select innermost stack frame - i.e., current frame is frame 0, + and current location is based on that. + Don't do this on return from a stack dummy routine, + or if the program has exited. */ - /* Select innermost stack frame except on return from a stack dummy routine, - or if the program has exited. Print it without a level number if - we have changed functions or hit a breakpoint. Print source line - if we have one. */ if (!stop_stack_dummy) { select_frame (get_current_frame (), 0); + /* Print current location without a level number, if + we have changed functions or hit a breakpoint. + Print source line if we have one. + bpstat_print() contains the logic deciding in detail + what to print, based on the event(s) that just occurred. */ + if (stop_print_frame) { - int source_only; - - source_only = bpstat_print (stop_bpstat); - source_only = source_only || - ( stop_step - && step_frame_address == FRAME_FP (get_current_frame ()) - && step_start_function == find_pc_function (stop_pc)); - - print_stack_frame (selected_frame, -1, source_only? -1: 1); + int bpstat_ret; + int source_flag; + + bpstat_ret = bpstat_print (stop_bpstat); + /* bpstat_print() returned one of: + -1: Didn't print anything + 0: Printed preliminary "Breakpoint n, " message, desires + location tacked on + 1: Printed something, don't tack on location */ + + if (bpstat_ret == -1) + if ( stop_step + && step_frame_address == FRAME_FP (get_current_frame ()) + && step_start_function == find_pc_function (stop_pc)) + source_flag = -1; /* finished step, just print source line */ + else + source_flag = 1; /* print location and source line */ + else if (bpstat_ret == 0) /* hit bpt, desire location */ + source_flag = 1; /* print location and source line */ + else /* bpstat_ret == 1, hit bpt, do not desire location */ + source_flag = -1; /* just print source line */ + + /* The behavior of this routine with respect to the source + flag is: + -1: Print only source line + 0: Print only location + 1: Print location and source line */ + show_and_print_stack_frame (selected_frame, -1, source_flag); /* Display the auto-display expressions. */ do_displays (); @@ -1834,6 +2996,10 @@ Further execution is probably impossible.\n"); stop_pc = read_pc(); select_frame (get_current_frame (), 0); } + + + TUIDO (((TuiOpaqueFuncPtr)tui_vCheckDataValues, selected_frame)); + done: annotate_stopped (); } @@ -2071,6 +3237,75 @@ Are you sure you want to change it? ", do_cleanups (old_chain); } +static void +xdb_handle_command (args, from_tty) + char *args; + int from_tty; +{ + char **argv; + struct cleanup *old_chain; + + /* Break the command line up into args. */ + + argv = buildargv (args); + if (argv == NULL) + { + nomem (0); + } + old_chain = make_cleanup (freeargv, (char *) argv); + if (argv[1] != (char *)NULL) + { + char *argBuf; + int bufLen; + + bufLen = strlen(argv[0]) + 20; + argBuf = (char *)xmalloc(bufLen); + if (argBuf) + { + int validFlag = 1; + enum target_signal oursig; + + oursig = target_signal_from_name (argv[0]); + memset(argBuf, 0, bufLen); + if (strcmp(argv[1], "Q") == 0) + sprintf(argBuf, "%s %s", argv[0], "noprint"); + else + { + if (strcmp(argv[1], "s") == 0) + { + if (!signal_stop[oursig]) + sprintf(argBuf, "%s %s", argv[0], "stop"); + else + sprintf(argBuf, "%s %s", argv[0], "nostop"); + } + else if (strcmp(argv[1], "i") == 0) + { + if (!signal_program[oursig]) + sprintf(argBuf, "%s %s", argv[0], "pass"); + else + sprintf(argBuf, "%s %s", argv[0], "nopass"); + } + else if (strcmp(argv[1], "r") == 0) + { + if (!signal_print[oursig]) + sprintf(argBuf, "%s %s", argv[0], "print"); + else + sprintf(argBuf, "%s %s", argv[0], "noprint"); + } + else + validFlag = 0; + } + if (validFlag) + handle_command(argBuf, from_tty); + else + printf_filtered("Invalid signal handling flag.\n"); + if (argBuf) + free(argBuf); + } + } + do_cleanups (old_chain); +} + /* Print current contents of the tables set by the handle command. It is possible we should just be printing signals actually used by the current target (but for things to work right when switching @@ -2180,13 +3415,19 @@ restore_selected_frame (args) /* If inf_status->selected_frame_address is NULL, there was no previously selected frame. */ if (frame == NULL || - FRAME_FP (frame) != fr->frame_address || + /* FRAME_FP (frame) != fr->frame_address ||*/ /* elz: deleted this check as a quick fix + to the problem that for function called by hand + gdb creates no internal frame structure + and the real stack and gdb's idea of stack + are different if nested calls by hands are made*/ level != 0) - { + { warning ("Unable to restore previously selected frame.\n"); return 0; } + select_frame (frame, fr->level); + return(1); } @@ -2241,16 +3482,40 @@ restore_inferior_status (inf_status) RETURN_MASK_ERROR) == 0) /* Error in restoring the selected frame. Select the innermost frame. */ - select_frame (get_current_frame (), 0); + + + select_frame (get_current_frame (), 0); + } } + + +void +set_follow_fork_mode_command (arg, from_tty, c) + char * arg; + int from_tty; + struct cmd_list_element * c; +{ + if (! STREQ (arg, "parent") && + ! STREQ (arg, "child") && + ! STREQ (arg, "both") && + ! STREQ (arg, "ask")) + error ("follow-fork-mode must be one of \"parent\", \"child\", \"both\" or \"ask\"."); + + if (follow_fork_mode_string != NULL) + free (follow_fork_mode_string); + follow_fork_mode_string = savestring (arg, strlen (arg)); +} + + void _initialize_infrun () { register int i; register int numsigs; + struct cmd_list_element * c; add_info ("signals", signals_info, "What debugger does when program gets various signals.\n\ @@ -2272,8 +3537,31 @@ Print means print a message if this signal happens.\n\ Pass means let program see this signal; otherwise program doesn't know.\n\ Ignore is a synonym for nopass and noignore is a synonym for pass.\n\ Pass and Stop may be combined.", NULL)); + if (xdb_commands) + { + add_com("lz", class_info, signals_info, + "What debugger does when program gets various signals.\n\ +Specify a signal as argument to print info on that signal only."); + add_com("z", class_run, xdb_handle_command, + concat ("Specify how to handle a signal.\n\ +Args are signals and actions to apply to those signals.\n\ +Symbolic signals (e.g. SIGSEGV) are recommended but numeric signals\n\ +from 1-15 are allowed for compatibility with old versions of GDB.\n\ +Numeric ranges may be specified with the form LOW-HIGH (e.g. 1-5).\n\ +The special arg \"all\" is recognized to mean all signals except those\n\ +used by the debugger, typically SIGTRAP and SIGINT.\n", +"Recognized actions include \"s\" (toggles between stop and nostop), \n\ +\"r\" (toggles between print and noprint), \"i\" (toggles between pass and \ +nopass), \"Q\" (noprint)\n\ +Stop means reenter debugger if this signal happens (implies print).\n\ +Print means print a message if this signal happens.\n\ +Pass means let program see this signal; otherwise program doesn't know.\n\ +Ignore is a synonym for nopass and noignore is a synonym for pass.\n\ +Pass and Stop may be combined.", NULL)); + } - stop_command = add_cmd ("stop", class_obscure, not_just_help_class_command, + if (!dbx_commands) + stop_command = add_cmd ("stop", class_obscure, not_just_help_class_command, "There is no `stop' command, but you can set a hook on `stop'.\n\ This allows you to set a list of commands to be run each time execution\n\ of the program stops.", &cmdlist); @@ -2324,4 +3612,38 @@ to the user would be loading/unloading of a new library.\n", &setlist), &showlist); #endif + + c = add_set_enum_cmd ("follow-fork-mode", + class_run, + follow_fork_mode_kind_names, + (char *) &follow_fork_mode_string, +/* ??rehrauer: The "both" option is broken, by what may be a 10.20 + kernel problem. It's also not terribly useful without a GUI to + help the user drive two debuggers. So for now, I'm disabling + the "both" option. +/* "Set debugger response to a program call of fork or vfork.\n\ +A fork or vfork creates a new process. follow-fork-mode can be:\n\ + parent - the original process is debugged after a fork\n\ + child - the new process is debugged after a fork\n\ + both - both the parent and child are debugged after a fork\n\ + ask - the debugger will ask for one of the above choices\n\ +For \"both\", another copy of the debugger will be started to follow\n\ +the new child process. The original debugger will continue to follow\n\ +the original parent process. To distinguish their prompts, the\n\ +debugger copy's prompt will be changed.\n\ +For \"parent\" or \"child\", the unfollowed process will run free.\n\ +By default, the debugger will follow the parent process.", +*/ + "Set debugger response to a program call of fork or vfork.\n\ +A fork or vfork creates a new process. follow-fork-mode can be:\n\ + parent - the original process is debugged after a fork\n\ + child - the new process is debugged after a fork\n\ + ask - the debugger will ask for one of the above choices\n\ +For \"parent\" or \"child\", the unfollowed process will run free.\n\ +By default, the debugger will follow the parent process.", + &setlist); +/* c->function.sfunc = ;*/ + add_show_from_set (c, &showlist); + + set_follow_fork_mode_command ("parent", 0, NULL); } |