kernel July 2011

kernel@lists.fedoraproject.org

19 participants
30 discussions

[PATCH 10/19] teach wake_up_quiescent() to do "selective" wake_up
by Oleg Nesterov 01 Jul '11

01 Jul '11

Both utrace and ptrace can want the same thread to be quiescent, in this case its state is TASK_TRACED | TASK_UTRACED. And this also means that this task must not run unless both utrace and ptrace resume it. Change wake_up_quiescent(p, state) to do "p->state &= ~state" and return false unless there is no more "quiescent" bits in task->state. Signed-off-by: Oleg Nesterov <oleg(a)redhat.com> --- kernel/signal.c | 15 +++++++++++++++ 1 files changed, 15 insertions(+), 0 deletions(-) diff --git a/kernel/signal.c b/kernel/signal.c index a7979ad..57552e6 100644 --- a/kernel/signal.c +++ b/kernel/signal.c @@ -651,11 +651,26 @@ void signal_wake_up(struct task_struct *t, int resume) kick_process(t); } +#define STATE_QUIESCENT (__TASK_STOPPED | __TASK_TRACED | __TASK_UTRACED) /* * wakes up the STOPPED/TRACED task, must be called with ->siglock held. */ int wake_up_quiescent(struct task_struct *p, unsigned int state) { + unsigned int quiescent = (p->state & STATE_QUIESCENT); + + WARN_ON(state & ~(STATE_QUIESCENT | TASK_INTERRUPTIBLE)); + + if (quiescent) { + state &= ~TASK_INTERRUPTIBLE; + if ((quiescent & ~state) != 0) { + p->state &= ~state; + WARN_ON(!(p->state & STATE_QUIESCENT)); + WARN_ON(!(p->state & TASK_WAKEKILL)); + return 0; + } + } + return wake_up_state(p, state); } -- 1.5.5.1

1 0

[PATCH 09/19] tracehooks: kill tracehook_finish_jctl(), add tracehook_finish_stop()
by Oleg Nesterov 01 Jul '11

01 Jul '11

tracehook_finish_jctl() is needed to avoid the races with SIGKILL which wakes up UTRACED task, and thus it should be called every time after the STOPPED/TRACED/UTRACED returns from schedule(), remember that TASK_UTRACED can be added while the task is STOPPED/UTRACED. - rename it to tracehook_finish_stop(),jctl no longer matches the reality. - change do_signal_state() to call this helper right after schedule(), otherwise this logic is broken by the upstream changes - now that utrace doesn't control TASK_TRACED bit, ptrace_stop() must call this helper too. Signed-off-by: Oleg Nesterov <oleg(a)redhat.com> --- include/linux/tracehook.h | 6 +++--- kernel/signal.c | 5 +++-- kernel/utrace.c | 2 +- 3 files changed, 7 insertions(+), 6 deletions(-) diff --git a/include/linux/tracehook.h b/include/linux/tracehook.h index 7d7bdde..3c7b6b3 100644 --- a/include/linux/tracehook.h +++ b/include/linux/tracehook.h @@ -528,11 +528,11 @@ static inline int tracehook_get_signal(struct task_struct *task, } /** - * tracehook_finish_jctl - report about return from job control stop + * tracehook_finish_stop - report about return from STOPPED/TRACED * - * This is called by do_signal_stop() after wakeup. + * This is called by do_signal_stop() and ptrace_stop after wakeup. */ -static inline void tracehook_finish_jctl(void) +static inline void tracehook_finish_stop(void) { if (task_utrace_flags(current)) utrace_finish_stop(); diff --git a/kernel/signal.c b/kernel/signal.c index 4fcf1c7..a7979ad 100644 --- a/kernel/signal.c +++ b/kernel/signal.c @@ -1816,6 +1816,7 @@ static void ptrace_stop(int exit_code, int why, int clear_code, siginfo_t *info) read_unlock(&tasklist_lock); } + tracehook_finish_stop(); /* * While in TASK_TRACED, we were considered "frozen enough". * Now that we woke up, it's crucial if we're supposed to be @@ -1952,6 +1953,8 @@ retry: /* Now we don't run again until woken by SIGCONT or SIGKILL */ schedule(); + tracehook_finish_stop(); + spin_lock_irq(&current->sighand->siglock); } else { ptrace_stop(current->group_stop & GROUP_STOP_SIGMASK, @@ -1974,8 +1977,6 @@ retry: spin_unlock_irq(&current->sighand->siglock); - tracehook_finish_jctl(); - return 1; } diff --git a/kernel/utrace.c b/kernel/utrace.c index be98607..daa96b9 100644 --- a/kernel/utrace.c +++ b/kernel/utrace.c @@ -741,7 +741,7 @@ static bool utrace_reset(struct task_struct *task, struct utrace *utrace) void utrace_finish_stop(void) { /* - * If we were task_is_traced() and then SIGKILL'ed, make + * If we were task_is_utraced() and then SIGKILL'ed, make * sure we do nothing until the tracer drops utrace->lock. */ if (unlikely(__fatal_signal_pending(current))) { -- 1.5.5.1

1 0

[PATCH 08/19] utrace: use TASK_UTRACED instead of TASK_TRACED
by Oleg Nesterov 01 Jul '11

01 Jul '11

Change utrace.c to use TASK_UTRACED instead of TASK_TRACED. - utrace_stop/utrace_wakeup: simply use the new state - utrace_do_stop: do not clear STOPPED/TRACED, but add the new __TASK_UTRACED bit to state the fact that both ptrace and utrace want this task to be stopped - naturally, do not use task_is_traced() to check if this task was stopped by utrace, use the new task_is_utraced() helper which checks __TASK_UTRACED. Signed-off-by: Oleg Nesterov <oleg(a)redhat.com> --- kernel/utrace.c | 26 ++++++++++++++------------ 1 files changed, 14 insertions(+), 12 deletions(-) diff --git a/kernel/utrace.c b/kernel/utrace.c index d7c547c..be98607 100644 --- a/kernel/utrace.c +++ b/kernel/utrace.c @@ -462,6 +462,8 @@ static void put_detached_list(struct list_head *list) */ #define ENGINE_STOP (1UL << _UTRACE_NEVENTS) +#define task_is_utraced(task) ((task->state & __TASK_UTRACED) != 0) + static void mark_engine_wants_stop(struct task_struct *task, struct utrace_engine *engine) { @@ -576,7 +578,7 @@ int utrace_set_events(struct task_struct *target, ret = 0; if ((old_flags & ~events) && target != current && - !task_is_stopped_or_traced(target) && !target->exit_state) { + !task_is_utraced(target) && !target->exit_state) { /* * This barrier ensures that our engine->flags changes * have hit before we examine utrace->reporting, @@ -623,21 +625,21 @@ static void mark_engine_detached(struct utrace_engine *engine) */ static bool utrace_do_stop(struct task_struct *target, struct utrace *utrace) { - if (task_is_stopped(target)) { + if (task_is_stopped_or_traced(target)) { /* * Stopped is considered quiescent; when it wakes up, it will * go through utrace_finish_stop() before doing anything else. */ spin_lock_irq(&target->sighand->siglock); - if (likely(task_is_stopped(target))) - __set_task_state(target, TASK_TRACED); + if (likely(task_is_stopped_or_traced(target))) + target->state |= TASK_UTRACED; spin_unlock_irq(&target->sighand->siglock); } else if (utrace->resume > UTRACE_REPORT) { utrace->resume = UTRACE_REPORT; set_notify_resume(target); } - return task_is_traced(target); + return task_is_utraced(target); } /* @@ -648,7 +650,7 @@ static void utrace_wakeup(struct task_struct *target, struct utrace *utrace) { lockdep_assert_held(&utrace->lock); spin_lock_irq(&target->sighand->siglock); - wake_up_quiescent(target, __TASK_TRACED); + wake_up_quiescent(target, __TASK_UTRACED); spin_unlock_irq(&target->sighand->siglock); } @@ -710,7 +712,7 @@ static bool utrace_reset(struct task_struct *task, struct utrace *utrace) /* * If no more engines want it stopped, wake it up. */ - if (task_is_traced(task) && !(flags & ENGINE_STOP)) { + if (task_is_utraced(task) && !(flags & ENGINE_STOP)) { /* * It just resumes, so make sure single-step * is not left set. @@ -749,7 +751,7 @@ void utrace_finish_stop(void) } /* - * Perform %UTRACE_STOP, i.e. block in TASK_TRACED until woken up. + * Perform %UTRACE_STOP, i.e. block in TASK_UTRACED until woken up. * @task == current, @utrace == current->utrace, which is not locked. * Return true if we were woken up by SIGKILL even though some utrace * engine may still want us to stay stopped. @@ -799,7 +801,7 @@ relock: return; } - __set_current_state(TASK_TRACED); + __set_current_state(TASK_UTRACED); spin_unlock_irq(&task->sighand->siglock); spin_unlock(&utrace->lock); @@ -809,14 +811,14 @@ relock: utrace_finish_stop(); /* - * While in TASK_TRACED, we were considered "frozen enough". + * While in TASK_UTRACED, we were considered "frozen enough". * Now that we woke up, it's crucial if we're supposed to be * frozen that we freeze now before running anything substantial. */ try_to_freeze(); /* - * While we were in TASK_TRACED, complete_signal() considered + * While we were in TASK_UTRACED, complete_signal() considered * us "uninterested" in signal wakeups. Now make sure our * TIF_SIGPENDING state is correct for normal running. */ @@ -1087,7 +1089,7 @@ int utrace_control(struct task_struct *target, if (unlikely(IS_ERR(utrace))) return PTR_ERR(utrace); - reset = task_is_traced(target); + reset = task_is_utraced(target); ret = 0; /* -- 1.5.5.1

1 0

[PATCH 07/19] introduce TASK_UTRACED state
by Oleg Nesterov 01 Jul '11

01 Jul '11

Introduce TASK_UTRACED state, will be used by utrace instead of TASK_TRACED. Note: this state is reported as "t (tracing stop)" to the user-space to avoid the confusion. IOW, it looks like TASK_TRACED in /proc/pid/status. Signed-off-by: Oleg Nesterov <oleg(a)redhat.com> --- fs/proc/array.c | 11 ++++++----- include/linux/sched.h | 20 +++++++++++--------- 2 files changed, 17 insertions(+), 14 deletions(-) diff --git a/fs/proc/array.c b/fs/proc/array.c index 496fef3..bfaa998 100644 --- a/fs/proc/array.c +++ b/fs/proc/array.c @@ -138,11 +138,12 @@ static const char * const task_state_array[] = { "D (disk sleep)", /* 2 */ "T (stopped)", /* 4 */ "t (tracing stop)", /* 8 */ - "Z (zombie)", /* 16 */ - "X (dead)", /* 32 */ - "x (dead)", /* 64 */ - "K (wakekill)", /* 128 */ - "W (waking)", /* 256 */ + "t (tracing stop)", /* 16 (stopped by utrace) */ + "Z (zombie)", /* 32 */ + "X (dead)", /* 64 */ + "x (dead)", /* 128 */ + "K (wakekill)", /* 256 */ + "W (waking)", /* 512 */ }; static inline const char *get_task_state(struct task_struct *tsk) diff --git a/include/linux/sched.h b/include/linux/sched.h index b87de83..7a0008c 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -184,16 +184,17 @@ print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq) #define TASK_UNINTERRUPTIBLE 2 #define __TASK_STOPPED 4 #define __TASK_TRACED 8 +#define __TASK_UTRACED 16 /* in tsk->exit_state */ -#define EXIT_ZOMBIE 16 -#define EXIT_DEAD 32 +#define EXIT_ZOMBIE 32 +#define EXIT_DEAD 64 /* in tsk->state again */ -#define TASK_DEAD 64 -#define TASK_WAKEKILL 128 -#define TASK_WAKING 256 -#define TASK_STATE_MAX 512 +#define TASK_DEAD 128 +#define TASK_WAKEKILL 256 +#define TASK_WAKING 512 +#define TASK_STATE_MAX 1024 -#define TASK_STATE_TO_CHAR_STR "RSDTtZXxKW" +#define TASK_STATE_TO_CHAR_STR "RSDTtUZXxKW" extern char ___assert_task_state[1 - 2*!!( sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)]; @@ -202,15 +203,16 @@ extern char ___assert_task_state[1 - 2*!!( #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE) #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED) #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED) +#define TASK_UTRACED (TASK_WAKEKILL | __TASK_UTRACED) /* Convenience macros for the sake of wake_up */ #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE) -#define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED) +#define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED | __TASK_UTRACED) /* get_task_state() */ #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \ TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \ - __TASK_TRACED) + __TASK_TRACED | __TASK_UTRACED) #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0) #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0) -- 1.5.5.1

1 0

[PATCH 06/19] wait_task_inactive: treat task->state and match_state as bitmasks
by Oleg Nesterov 01 Jul '11

01 Jul '11

Change wait_task_inactive() to check "state & match_state" instead of "state == match_state". This should not make any difference, but this allows us to add more "stopped" bits which can be set or cleared independently. IOW. wait_task_inactive() assumes that if task->state != 0, it can only be changed to TASK_RUNNING. Currently this is true, and in this case "state & match_state" continues to work. But, unlike the current check, it also works if task->state has other bits set while the caller is only interested in, say, __TASK_TRACED. Note: I think wait_task_inactive() should be cleanuped upstrean anyway, nowadays we have TASK_WAKING and task->state != 0 doesn't necessarily mean it is TASK_RUNNING. It also makes sense to exclude the !TASK_REPORT bits during the check. Finally, probably this patch makes sense anyway even without utrace. For example, a stopped _and_ traced thread could have task->state = TASK_STOPPED | TASK_TRACED, this can be useful. Signed-off-by: Oleg Nesterov <oleg(a)redhat.com> --- kernel/sched.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/kernel/sched.c b/kernel/sched.c index 3f2e502..ade7997 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -2277,7 +2277,7 @@ unsigned long wait_task_inactive(struct task_struct *p, long match_state) * is actually now running somewhere else! */ while (task_running(rq, p)) { - if (match_state && unlikely(p->state != match_state)) + if (match_state && !likely(p->state & match_state)) return 0; cpu_relax(); } -- 1.5.5.1

1 0

[PATCH 05/19] introduce ptrace_signal_wake_up()
by Oleg Nesterov 01 Jul '11

01 Jul '11

Add the new helper, ptrace_signal_wake_up(), change ptrace.c to use it instead of signal_wake_up(). The new helper does almost the same, except: - it doesn't use the TASK_WAKEKILL bit to wake up the TRACED or STOPPED task, it uses __TASK_STOPPED | __TASK_TRACED explicitly. This is what ptrace actually wants, it should never wake up a TASK_KILLABLE task. This should be cleanuped upatream, signal_wake_up() should take the state as an argument, not a boolean. Until then we add a new static helper. - it uses wake_up_quiescent() instead of wake_up_state(). Thereafter every change from STOPPED/TRACED to RUNNING is done via wake_up_quiescent(). Signed-off-by: Oleg Nesterov <oleg(a)redhat.com> --- kernel/ptrace.c | 16 ++++++++++++++-- 1 files changed, 14 insertions(+), 2 deletions(-) diff --git a/kernel/ptrace.c b/kernel/ptrace.c index 26ae214..0b2aba5 100644 --- a/kernel/ptrace.c +++ b/kernel/ptrace.c @@ -24,6 +24,18 @@ #include <linux/regset.h> #include <linux/hw_breakpoint.h> +static void ptrace_signal_wake_up(struct task_struct *p, int quiescent) +{ + unsigned int state; + + set_tsk_thread_flag(p, TIF_SIGPENDING); + + state = TASK_INTERRUPTIBLE; + if (quiescent) + state |= (__TASK_STOPPED | __TASK_TRACED); + if (!wake_up_quiescent(p, state)) + kick_process(p); +} /* * ptrace a task: make the debugger its new parent and @@ -92,7 +104,7 @@ void __ptrace_unlink(struct task_struct *child) * TASK_KILLABLE sleeps. */ if (child->group_stop & GROUP_STOP_PENDING || task_is_traced(child)) - signal_wake_up(child, task_is_traced(child)); + ptrace_signal_wake_up(child, task_is_traced(child)); spin_unlock(&child->sighand->siglock); } @@ -245,7 +257,7 @@ static int ptrace_attach(struct task_struct *task) */ if (task_is_stopped(task)) { task->group_stop |= GROUP_STOP_PENDING | GROUP_STOP_TRAPPING; - signal_wake_up(task, 1); + ptrace_signal_wake_up(task, 1); wait_trap = true; } -- 1.5.5.1

1 0

[PATCH 04/19] introduce wake_up_quiescent()
by Oleg Nesterov 01 Jul '11

01 Jul '11

No functional changes. Add the new helper, wake_up_quiescent(task, state), which simply returns wake_up_state(task, state). Change all callers which do wake_up_state(STOPPED/TRACED) to use the new helper. ptrace_stop() is a bit special, it does __set_current_state(RUNNING) in the very unlikely case, change it as well. Signed-off-by: Oleg Nesterov <oleg(a)redhat.com> --- include/linux/signal.h | 2 ++ kernel/ptrace.c | 2 +- kernel/signal.c | 12 ++++++++++-- kernel/utrace.c | 2 +- 4 files changed, 14 insertions(+), 4 deletions(-) diff --git a/include/linux/signal.h b/include/linux/signal.h index a822300..2be3712 100644 --- a/include/linux/signal.h +++ b/include/linux/signal.h @@ -239,6 +239,8 @@ static inline int valid_signal(unsigned long sig) struct timespec; struct pt_regs; +extern int wake_up_quiescent(struct task_struct *p, unsigned int state); + extern int next_signal(struct sigpending *pending, sigset_t *mask); extern int do_send_sig_info(int sig, struct siginfo *info, struct task_struct *p, bool group); diff --git a/kernel/ptrace.c b/kernel/ptrace.c index 9988b13..26ae214 100644 --- a/kernel/ptrace.c +++ b/kernel/ptrace.c @@ -566,7 +566,7 @@ static int ptrace_resume(struct task_struct *child, long request, child->exit_code = data; if (lock_task_sighand(child, &flags)) { - wake_up_state(child, __TASK_TRACED); + wake_up_quiescent(child, __TASK_TRACED); unlock_task_sighand(child, &flags); } diff --git a/kernel/signal.c b/kernel/signal.c index 2138cee..4fcf1c7 100644 --- a/kernel/signal.c +++ b/kernel/signal.c @@ -652,6 +652,14 @@ void signal_wake_up(struct task_struct *t, int resume) } /* + * wakes up the STOPPED/TRACED task, must be called with ->siglock held. + */ +int wake_up_quiescent(struct task_struct *p, unsigned int state) +{ + return wake_up_state(p, state); +} + +/* * Remove signals in mask from the pending set and queue. * Returns 1 if any signals were found. * @@ -811,7 +819,7 @@ static int prepare_signal(int sig, struct task_struct *p, int from_ancestor_ns) do { task_clear_group_stop_pending(t); rm_from_queue(SIG_KERNEL_STOP_MASK, &t->pending); - wake_up_state(t, __TASK_STOPPED); + wake_up_quiescent(t, __TASK_STOPPED); } while_each_thread(p, t); /* @@ -1800,7 +1808,7 @@ static void ptrace_stop(int exit_code, int why, int clear_code, siginfo_t *info) do_notify_parent_cldstop(current, false, why); spin_lock_irq(&current->sighand->siglock); - __set_current_state(TASK_RUNNING); + wake_up_quiescent(current, __TASK_TRACED); spin_unlock_irq(&current->sighand->siglock); if (clear_code) diff --git a/kernel/utrace.c b/kernel/utrace.c index 6e7fafb..d7c547c 100644 --- a/kernel/utrace.c +++ b/kernel/utrace.c @@ -648,7 +648,7 @@ static void utrace_wakeup(struct task_struct *target, struct utrace *utrace) { lockdep_assert_held(&utrace->lock); spin_lock_irq(&target->sighand->siglock); - wake_up_state(target, __TASK_TRACED); + wake_up_quiescent(target, __TASK_TRACED); spin_unlock_irq(&target->sighand->siglock); } -- 1.5.5.1

1 0

[PATCH 03/19] ptrace: take ->siglock around s/TRACED/RUNNING/
by Oleg Nesterov 01 Jul '11

01 Jul '11

change ptrace_resume() and ptrace_stop() to take ->siglock around changing task->state from TRACED to RUNNING. With this patch __TASK_TRACED/STOPPED bits are fully protected by ->siglock, nobody can set or clear these bits without ->siglock held. Signed-off-by: Oleg Nesterov <oleg(a)redhat.com> --- kernel/ptrace.c | 8 +++++++- kernel/signal.c | 3 +++ 2 files changed, 10 insertions(+), 1 deletions(-) diff --git a/kernel/ptrace.c b/kernel/ptrace.c index 2df1157..9988b13 100644 --- a/kernel/ptrace.c +++ b/kernel/ptrace.c @@ -534,6 +534,8 @@ static int ptrace_setsiginfo(struct task_struct *child, const siginfo_t *info) static int ptrace_resume(struct task_struct *child, long request, unsigned long data) { + unsigned long flags; + if (!valid_signal(data)) return -EIO; @@ -562,7 +564,11 @@ static int ptrace_resume(struct task_struct *child, long request, } child->exit_code = data; - wake_up_state(child, __TASK_TRACED); + + if (lock_task_sighand(child, &flags)) { + wake_up_state(child, __TASK_TRACED); + unlock_task_sighand(child, &flags); + } return 0; } diff --git a/kernel/signal.c b/kernel/signal.c index ff76786..2138cee 100644 --- a/kernel/signal.c +++ b/kernel/signal.c @@ -1799,7 +1799,10 @@ static void ptrace_stop(int exit_code, int why, int clear_code, siginfo_t *info) if (gstop_done) do_notify_parent_cldstop(current, false, why); + spin_lock_irq(&current->sighand->siglock); __set_current_state(TASK_RUNNING); + spin_unlock_irq(&current->sighand->siglock); + if (clear_code) current->exit_code = 0; read_unlock(&tasklist_lock); -- 1.5.5.1

1 0

[PATCH 02/19] utrace: remove jobctl bits
by Oleg Nesterov 01 Jul '11

01 Jul '11

- change utrace_get_signal() to check GROUP_STOP_PENDING instead of signal->group_stop_count. With the recent changes group_stop_count doesn't necessarily mean this task should participate in group stop. - remove the "participate in group stop" code from utrace_wakeup() and utrace_stop(), this is no longer needed and wrong. Signed-off-by: Oleg Nesterov <oleg(a)redhat.com> --- kernel/utrace.c | 16 ++-------------- 1 files changed, 2 insertions(+), 14 deletions(-) diff --git a/kernel/utrace.c b/kernel/utrace.c index f332d65..6e7fafb 100644 --- a/kernel/utrace.c +++ b/kernel/utrace.c @@ -648,11 +648,7 @@ static void utrace_wakeup(struct task_struct *target, struct utrace *utrace) { lockdep_assert_held(&utrace->lock); spin_lock_irq(&target->sighand->siglock); - if (target->signal->flags & SIGNAL_STOP_STOPPED || - target->signal->group_stop_count) - target->state = TASK_STOPPED; - else - wake_up_state(target, __TASK_TRACED); + wake_up_state(target, __TASK_TRACED); spin_unlock_irq(&target->sighand->siglock); } @@ -805,14 +801,6 @@ relock: __set_current_state(TASK_TRACED); - /* - * If there is a group stop in progress, - * we must participate in the bookkeeping. - */ - if (unlikely(task->signal->group_stop_count) && - !--task->signal->group_stop_count) - task->signal->flags = SIGNAL_STOP_STOPPED; - spin_unlock_irq(&task->sighand->siglock); spin_unlock(&utrace->lock); @@ -2036,7 +2024,7 @@ int utrace_get_signal(struct task_struct *task, struct pt_regs *regs, ka = NULL; memset(return_ka, 0, sizeof *return_ka); } else if (!(task->utrace_flags & UTRACE_EVENT_SIGNAL_ALL) || - unlikely(task->signal->group_stop_count)) { + unlikely(task->group_stop & GROUP_STOP_PENDING)) { /* * If no engine is interested in intercepting signals or * we must stop, let the caller just dequeue them normally -- 1.5.5.1

1 0

[PATCH 01/19] utrace core
by Oleg Nesterov 01 Jul '11

01 Jul '11

From: Roland McGrath <roland(a)redhat.com> This adds the utrace facility, a new modular interface in the kernel for implementing user thread tracing and debugging. This fits on top of the tracehook_* layer, so the new code is well-isolated. The new interface is in <linux/utrace.h> and the DocBook utrace book describes it. It allows for multiple separate tracing engines to work in parallel without interfering with each other. Higher-level tracing facilities can be implemented as loadable kernel modules using this layer. The new facility is made optional under CONFIG_UTRACE. When this is not enabled, no new code is added. It can only be enabled on machines that have all the prerequisites and select CONFIG_HAVE_ARCH_TRACEHOOK. In this initial version, utrace and ptrace do not play together at all, the next patches try to fix this. This is is same/old utrace-core patch except: - use group_stop/GROUP_STOP_DEQUEUED instead of removed signal->flags/SIGNAL_STOP_DEQUEUED in utrace_get_signal() - rediff the changes in tracehook.h against the current code without PT_PTRACED tweaks which were needed for ptrace-utrace Signed-off-by: Roland McGrath <roland(a)redhat.com> Signed-off-by: Oleg Nesterov <oleg(a)redhat.com> --- Documentation/DocBook/Makefile | 2 +- Documentation/DocBook/utrace.tmpl | 589 +++++++++ fs/proc/array.c | 3 + include/linux/sched.h | 5 + include/linux/tracehook.h | 85 ++- include/linux/utrace.h | 692 +++++++++++ init/Kconfig | 9 + kernel/Makefile | 1 + kernel/fork.c | 3 + kernel/utrace.c | 2440 +++++++++++++++++++++++++++++++++++++ 10 files changed, 3827 insertions(+), 2 deletions(-) create mode 100644 Documentation/DocBook/utrace.tmpl create mode 100644 include/linux/utrace.h create mode 100644 kernel/utrace.c diff --git a/Documentation/DocBook/Makefile b/Documentation/DocBook/Makefile index 3cebfa0..86c288b 100644 --- a/Documentation/DocBook/Makefile +++ b/Documentation/DocBook/Makefile @@ -14,7 +14,7 @@ DOCBOOKS := z8530book.xml mcabook.xml device-drivers.xml \ genericirq.xml s390-drivers.xml uio-howto.xml scsi.xml \ 80211.xml debugobjects.xml sh.xml regulator.xml \ alsa-driver-api.xml writing-an-alsa-driver.xml \ - tracepoint.xml media.xml drm.xml + tracepoint.xml utrace.xml media.xml drm.xml ### # The build process is as follows (targets): diff --git a/Documentation/DocBook/utrace.tmpl b/Documentation/DocBook/utrace.tmpl new file mode 100644 index 0000000..0c40add --- /dev/null +++ b/Documentation/DocBook/utrace.tmpl @@ -0,0 +1,589 @@ +<?xml version="1.0" encoding="UTF-8"?> +<!DOCTYPE book PUBLIC "-//OASIS//DTD DocBook XML V4.1.2//EN" +"http://www.oasis-open.org/docbook/xml/4.1.2/docbookx.dtd" []> + +<book id="utrace"> + <bookinfo> + <title>The utrace User Debugging Infrastructure</title> + </bookinfo> + + <toc></toc> + + <chapter id="concepts"><title>utrace concepts</title> + + <sect1 id="intro"><title>Introduction</title> + + <para> + <application>utrace</application> is infrastructure code for tracing + and controlling user threads. This is the foundation for writing + tracing engines, which can be loadable kernel modules. + </para> + + <para> + The basic actors in <application>utrace</application> are the thread + and the tracing engine. A tracing engine is some body of code that + calls into the <filename><linux/utrace.h></filename> + interfaces, represented by a <structname>struct + utrace_engine_ops</structname>. (Usually it's a kernel module, + though the legacy <function>ptrace</function> support is a tracing + engine that is not in a kernel module.) The interface operates on + individual threads (<structname>struct task_struct</structname>). + If an engine wants to treat several threads as a group, that is up + to its higher-level code. + </para> + + <para> + Tracing begins by attaching an engine to a thread, using + <function>utrace_attach_task</function> or + <function>utrace_attach_pid</function>. If successful, it returns a + pointer that is the handle used in all other calls. + </para> + + </sect1> + + <sect1 id="callbacks"><title>Events and Callbacks</title> + + <para> + An attached engine does nothing by default. An engine makes something + happen by requesting callbacks via <function>utrace_set_events</function> + and poking the thread with <function>utrace_control</function>. + The synchronization issues related to these two calls + are discussed further below in <xref linkend="teardown"/>. + </para> + + <para> + Events are specified using the macro + <constant>UTRACE_EVENT(<replaceable>type</replaceable>)</constant>. + Each event type is associated with a callback in <structname>struct + utrace_engine_ops</structname>. A tracing engine can leave unused + callbacks <constant>NULL</constant>. The only callbacks required + are those used by the event flags it sets. + </para> + + <para> + Many engines can be attached to each thread. When a thread has an + event, each engine gets a callback if it has set the event flag for + that event type. For most events, engines are called in the order they + attached. Engines that attach after the event has occurred do not get + callbacks for that event. This includes any new engines just attached + by an existing engine's callback function. Once the sequence of + callbacks for that one event has completed, such new engines are then + eligible in the next sequence that starts when there is another event. + </para> + + <para> + Event reporting callbacks have details particular to the event type, + but are all called in similar environments and have the same + constraints. Callbacks are made from safe points, where no locks + are held, no special resources are pinned (usually), and the + user-mode state of the thread is accessible. So, callback code has + a pretty free hand. But to be a good citizen, callback code should + never block for long periods. It is fine to block in + <function>kmalloc</function> and the like, but never wait for i/o or + for user mode to do something. If you need the thread to wait, use + <constant>UTRACE_STOP</constant> and return from the callback + quickly. When your i/o finishes or whatever, you can use + <function>utrace_control</function> to resume the thread. + </para> + + <para> + The <constant>UTRACE_EVENT(SYSCALL_ENTRY)</constant> event is a special + case. While other events happen in the kernel when it will return to + user mode soon, this event happens when entering the kernel before it + will proceed with the work requested from user mode. Because of this + difference, the <function>report_syscall_entry</function> callback is + special in two ways. For this event, engines are called in reverse of + the normal order (this includes the <function>report_quiesce</function> + call that precedes a <function>report_syscall_entry</function> call). + This preserves the semantics that the last engine to attach is called + "closest to user mode"--the engine that is first to see a thread's user + state when it enters the kernel is also the last to see that state when + the thread returns to user mode. For the same reason, if these + callbacks use <constant>UTRACE_STOP</constant> (see the next section), + the thread stops immediately after callbacks rather than only when it's + ready to return to user mode; when allowed to resume, it will actually + attempt the system call indicated by the register values at that time. + </para> + + </sect1> + + <sect1 id="safely"><title>Stopping Safely</title> + + <sect2 id="well-behaved"><title>Writing well-behaved callbacks</title> + + <para> + Well-behaved callbacks are important to maintain two essential + properties of the interface. The first of these is that unrelated + tracing engines should not interfere with each other. If your engine's + event callback does not return quickly, then another engine won't get + the event notification in a timely manner. The second important + property is that tracing should be as noninvasive as possible to the + normal operation of the system overall and of the traced thread in + particular. That is, attached tracing engines should not perturb a + thread's behavior, except to the extent that changing its user-visible + state is explicitly what you want to do. (Obviously some perturbation + is unavoidable, primarily timing changes, ranging from small delays due + to the overhead of tracing, to arbitrary pauses in user code execution + when a user stops a thread with a debugger for examination.) Even when + you explicitly want the perturbation of making the traced thread block, + just blocking directly in your callback has more unwanted effects. For + example, the <constant>CLONE</constant> event callbacks are called when + the new child thread has been created but not yet started running; the + child can never be scheduled until the <constant>CLONE</constant> + tracing callbacks return. (This allows engines tracing the parent to + attach to the child.) If a <constant>CLONE</constant> event callback + blocks the parent thread, it also prevents the child thread from + running (even to process a <constant>SIGKILL</constant>). If what you + want is to make both the parent and child block, then use + <function>utrace_attach_task</function> on the child and then use + <constant>UTRACE_STOP</constant> on both threads. A more crucial + problem with blocking in callbacks is that it can prevent + <constant>SIGKILL</constant> from working. A thread that is blocking + due to <constant>UTRACE_STOP</constant> will still wake up and die + immediately when sent a <constant>SIGKILL</constant>, as all threads + should. Relying on the <application>utrace</application> + infrastructure rather than on private synchronization calls in event + callbacks is an important way to help keep tracing robustly + noninvasive. + </para> + + </sect2> + + <sect2 id="UTRACE_STOP"><title>Using <constant>UTRACE_STOP</constant></title> + + <para> + To control another thread and access its state, it must be stopped + with <constant>UTRACE_STOP</constant>. This means that it is + stopped and won't start running again while we access it. When a + thread is not already stopped, <function>utrace_control</function> + returns <constant>-EINPROGRESS</constant> and an engine must wait + for an event callback when the thread is ready to stop. The thread + may be running on another CPU or may be blocked. When it is ready + to be examined, it will make callbacks to engines that set the + <constant>UTRACE_EVENT(QUIESCE)</constant> event bit. To wake up an + interruptible wait, use <constant>UTRACE_INTERRUPT</constant>. + </para> + + <para> + As long as some engine has used <constant>UTRACE_STOP</constant> and + not called <function>utrace_control</function> to resume the thread, + then the thread will remain stopped. <constant>SIGKILL</constant> + will wake it up, but it will not run user code. When the stop is + cleared with <function>utrace_control</function> or a callback + return value, the thread starts running again. + (See also <xref linkend="teardown"/>.) + </para> + + </sect2> + + </sect1> + + <sect1 id="teardown"><title>Tear-down Races</title> + + <sect2 id="SIGKILL"><title>Primacy of <constant>SIGKILL</constant></title> + <para> + Ordinarily synchronization issues for tracing engines are kept fairly + straightforward by using <constant>UTRACE_STOP</constant>. You ask a + thread to stop, and then once it makes the + <function>report_quiesce</function> callback it cannot do anything else + that would result in another callback, until you let it with a + <function>utrace_control</function> call. This simple arrangement + avoids complex and error-prone code in each one of a tracing engine's + event callbacks to keep them serialized with the engine's other + operations done on that thread from another thread of control. + However, giving tracing engines complete power to keep a traced thread + stuck in place runs afoul of a more important kind of simplicity that + the kernel overall guarantees: nothing can prevent or delay + <constant>SIGKILL</constant> from making a thread die and release its + resources. To preserve this important property of + <constant>SIGKILL</constant>, it as a special case can break + <constant>UTRACE_STOP</constant> like nothing else normally can. This + includes both explicit <constant>SIGKILL</constant> signals and the + implicit <constant>SIGKILL</constant> sent to each other thread in the + same thread group by a thread doing an exec, or processing a fatal + signal, or making an <function>exit_group</function> system call. A + tracing engine can prevent a thread from beginning the exit or exec or + dying by signal (other than <constant>SIGKILL</constant>) if it is + attached to that thread, but once the operation begins, no tracing + engine can prevent or delay all other threads in the same thread group + dying. + </para> + </sect2> + + <sect2 id="reap"><title>Final callbacks</title> + <para> + The <function>report_reap</function> callback is always the final event + in the life cycle of a traced thread. Tracing engines can use this as + the trigger to clean up their own data structures. The + <function>report_death</function> callback is always the penultimate + event a tracing engine might see; it's seen unless the thread was + already in the midst of dying when the engine attached. Many tracing + engines will have no interest in when a parent reaps a dead process, + and nothing they want to do with a zombie thread once it dies; for + them, the <function>report_death</function> callback is the natural + place to clean up data structures and detach. To facilitate writing + such engines robustly, given the asynchrony of + <constant>SIGKILL</constant>, and without error-prone manual + implementation of synchronization schemes, the + <application>utrace</application> infrastructure provides some special + guarantees about the <function>report_death</function> and + <function>report_reap</function> callbacks. It still takes some care + to be sure your tracing engine is robust to tear-down races, but these + rules make it reasonably straightforward and concise to handle a lot of + corner cases correctly. + </para> + </sect2> + + <sect2 id="refcount"><title>Engine and task pointers</title> + <para> + The first sort of guarantee concerns the core data structures + themselves. <structname>struct utrace_engine</structname> is + a reference-counted data structure. While you hold a reference, an + engine pointer will always stay valid so that you can safely pass it to + any <application>utrace</application> call. Each call to + <function>utrace_attach_task</function> or + <function>utrace_attach_pid</function> returns an engine pointer with a + reference belonging to the caller. You own that reference until you + drop it using <function>utrace_engine_put</function>. There is an + implicit reference on the engine while it is attached. So if you drop + your only reference, and then use + <function>utrace_attach_task</function> without + <constant>UTRACE_ATTACH_CREATE</constant> to look up that same engine, + you will get the same pointer with a new reference to replace the one + you dropped, just like calling <function>utrace_engine_get</function>. + When an engine has been detached, either explicitly with + <constant>UTRACE_DETACH</constant> or implicitly after + <function>report_reap</function>, then any references you hold are all + that keep the old engine pointer alive. + </para> + + <para> + There is nothing a kernel module can do to keep a <structname>struct + task_struct</structname> alive outside of + <function>rcu_read_lock</function>. When the task dies and is reaped + by its parent (or itself), that structure can be freed so that any + dangling pointers you have stored become invalid. + <application>utrace</application> will not prevent this, but it can + help you detect it safely. By definition, a task that has been reaped + has had all its engines detached. All + <application>utrace</application> calls can be safely called on a + detached engine if the caller holds a reference on that engine pointer, + even if the task pointer passed in the call is invalid. All calls + return <constant>-ESRCH</constant> for a detached engine, which tells + you that the task pointer you passed could be invalid now. Since + <function>utrace_control</function> and + <function>utrace_set_events</function> do not block, you can call those + inside a <function>rcu_read_lock</function> section and be sure after + they don't return <constant>-ESRCH</constant> that the task pointer is + still valid until <function>rcu_read_unlock</function>. The + infrastructure never holds task references of its own. Though neither + <function>rcu_read_lock</function> nor any other lock is held while + making a callback, it's always guaranteed that the <structname>struct + task_struct</structname> and the <structname>struct + utrace_engine</structname> passed as arguments remain valid + until the callback function returns. + </para> + + <para> + The common means for safely holding task pointers that is available to + kernel modules is to use <structname>struct pid</structname>, which + permits <function>put_pid</function> from kernel modules. When using + that, the calls <function>utrace_attach_pid</function>, + <function>utrace_control_pid</function>, + <function>utrace_set_events_pid</function>, and + <function>utrace_barrier_pid</function> are available. + </para> + </sect2> + + <sect2 id="reap-after-death"> + <title> + Serialization of <constant>DEATH</constant> and <constant>REAP</constant> + </title> + <para> + The second guarantee is the serialization of + <constant>DEATH</constant> and <constant>REAP</constant> event + callbacks for a given thread. The actual reaping by the parent + (<function>release_task</function> call) can occur simultaneously + while the thread is still doing the final steps of dying, including + the <function>report_death</function> callback. If a tracing engine + has requested both <constant>DEATH</constant> and + <constant>REAP</constant> event reports, it's guaranteed that the + <function>report_reap</function> callback will not be made until + after the <function>report_death</function> callback has returned. + If the <function>report_death</function> callback itself detaches + from the thread, then the <function>report_reap</function> callback + will never be made. Thus it is safe for a + <function>report_death</function> callback to clean up data + structures and detach. + </para> + </sect2> + + <sect2 id="interlock"><title>Interlock with final callbacks</title> + <para> + The final sort of guarantee is that a tracing engine will know for sure + whether or not the <function>report_death</function> and/or + <function>report_reap</function> callbacks will be made for a certain + thread. These tear-down races are disambiguated by the error return + values of <function>utrace_set_events</function> and + <function>utrace_control</function>. Normally + <function>utrace_control</function> called with + <constant>UTRACE_DETACH</constant> returns zero, and this means that no + more callbacks will be made. If the thread is in the midst of dying, + it returns <constant>-EALREADY</constant> to indicate that the + <constant>report_death</constant> callback may already be in progress; + when you get this error, you know that any cleanup your + <function>report_death</function> callback does is about to happen or + has just happened--note that if the <function>report_death</function> + callback does not detach, the engine remains attached until the thread + gets reaped. If the thread is in the midst of being reaped, + <function>utrace_control</function> returns <constant>-ESRCH</constant> + to indicate that the <function>report_reap</function> callback may + already be in progress; this means the engine is implicitly detached + when the callback completes. This makes it possible for a tracing + engine that has decided asynchronously to detach from a thread to + safely clean up its data structures, knowing that no + <function>report_death</function> or <function>report_reap</function> + callback will try to do the same. <constant>utrace_detach</constant> + returns <constant>-ESRCH</constant> when the <structname>struct + utrace_engine</structname> has already been detached, but is + still a valid pointer because of its reference count. A tracing engine + can use this to safely synchronize its own independent multiple threads + of control with each other and with its event callbacks that detach. + </para> + + <para> + In the same vein, <function>utrace_set_events</function> normally + returns zero; if the target thread was stopped before the call, then + after a successful call, no event callbacks not requested in the new + flags will be made. It fails with <constant>-EALREADY</constant> if + you try to clear <constant>UTRACE_EVENT(DEATH)</constant> when the + <function>report_death</function> callback may already have begun, or if + you try to newly set <constant>UTRACE_EVENT(DEATH)</constant> or + <constant>UTRACE_EVENT(QUIESCE)</constant> when the target is already + dead or dying. Like <function>utrace_control</function>, it returns + <constant>-ESRCH</constant> when the <function>report_reap</function> + callback may already have begun, or the thread has already been detached + (including forcible detach on reaping). This lets the tracing engine + know for sure which event callbacks it will or won't see after + <function>utrace_set_events</function> has returned. By checking for + errors, it can know whether to clean up its data structures immediately + or to let its callbacks do the work. + </para> + </sect2> + + <sect2 id="barrier"><title>Using <function>utrace_barrier</function></title> + <para> + When a thread is safely stopped, calling + <function>utrace_control</function> with <constant>UTRACE_DETACH</constant> + or calling <function>utrace_set_events</function> to disable some events + ensures synchronously that your engine won't get any more of the callbacks + that have been disabled (none at all when detaching). But these can also + be used while the thread is not stopped, when it might be simultaneously + making a callback to your engine. For this situation, these calls return + <constant>-EINPROGRESS</constant> when it's possible a callback is in + progress. If you are not prepared to have your old callbacks still run, + then you can synchronize to be sure all the old callbacks are finished, + using <function>utrace_barrier</function>. This is necessary if the + kernel module containing your callback code is going to be unloaded. + </para> + <para> + After using <constant>UTRACE_DETACH</constant> once, further calls to + <function>utrace_control</function> with the same engine pointer will + return <constant>-ESRCH</constant>. In contrast, after getting + <constant>-EINPROGRESS</constant> from + <function>utrace_set_events</function>, you can call + <function>utrace_set_events</function> again later and if it returns zero + then know the old callbacks have finished. + </para> + <para> + Unlike all other calls, <function>utrace_barrier</function> (and + <function>utrace_barrier_pid</function>) will accept any engine pointer you + hold a reference on, even if <constant>UTRACE_DETACH</constant> has already + been used. After any <function>utrace_control</function> or + <function>utrace_set_events</function> call (these do not block), you can + call <function>utrace_barrier</function> to block until callbacks have + finished. This returns <constant>-ESRCH</constant> only if the engine is + completely detached (finished all callbacks). Otherwise it waits + until the thread is definitely not in the midst of a callback to this + engine and then returns zero, but can return + <constant>-ERESTARTSYS</constant> if its wait is interrupted. + </para> + </sect2> + +</sect1> + +</chapter> + +<chapter id="core"><title>utrace core API</title> + +<para> + The utrace API is declared in <filename><linux/utrace.h></filename>. +</para> + +!Iinclude/linux/utrace.h +!Ekernel/utrace.c + +</chapter> + +<chapter id="machine"><title>Machine State</title> + +<para> + The <function>task_current_syscall</function> function can be used on any + valid <structname>struct task_struct</structname> at any time, and does + not even require that <function>utrace_attach_task</function> was used at all. +</para> + +<para> + The other ways to access the registers and other machine-dependent state of + a task can only be used on a task that is at a known safe point. The safe + points are all the places where <function>utrace_set_events</function> can + request callbacks (except for the <constant>DEATH</constant> and + <constant>REAP</constant> events). So at any event callback, it is safe to + examine <varname>current</varname>. +</para> + +<para> + One task can examine another only after a callback in the target task that + returns <constant>UTRACE_STOP</constant> so that task will not return to user + mode after the safe point. This guarantees that the task will not resume + until the same engine uses <function>utrace_control</function>, unless the + task dies suddenly. To examine safely, one must use a pair of calls to + <function>utrace_prepare_examine</function> and + <function>utrace_finish_examine</function> surrounding the calls to + <structname>struct user_regset</structname> functions or direct examination + of task data structures. <function>utrace_prepare_examine</function> returns + an error if the task is not properly stopped, or is dead. After a + successful examination, the paired <function>utrace_finish_examine</function> + call returns an error if the task ever woke up during the examination. If + so, any data gathered may be scrambled and should be discarded. This means + there was a spurious wake-up (which should not happen), or a sudden death. +</para> + +<sect1 id="regset"><title><structname>struct user_regset</structname></title> + +<para> + The <structname>struct user_regset</structname> API + is declared in <filename><linux/regset.h></filename>. +</para> + +!Finclude/linux/regset.h + +</sect1> + +<sect1 id="task_current_syscall"> + <title><filename>System Call Information</filename></title> + +<para> + This function is declared in <filename><linux/ptrace.h></filename>. +</para> + +!Elib/syscall.c + +</sect1> + +<sect1 id="syscall"><title><filename>System Call Tracing</filename></title> + +<para> + The arch API for system call information is declared in + <filename><asm/syscall.h></filename>. + Each of these calls can be used only at system call entry tracing, + or can be used only at system call exit and the subsequent safe points + before returning to user mode. + At system call entry tracing means either during a + <structfield>report_syscall_entry</structfield> callback, + or any time after that callback has returned <constant>UTRACE_STOP</constant>. +</para> + +!Finclude/asm-generic/syscall.h + +</sect1> + +</chapter> + +<chapter id="internals"><title>Kernel Internals</title> + +<para> + This chapter covers the interface to the tracing infrastructure + from the core of the kernel and the architecture-specific code. + This is for maintainers of the kernel and arch code, and not relevant + to using the tracing facilities described in preceding chapters. +</para> + +<sect1 id="tracehook"><title>Core Calls In</title> + +<para> + These calls are declared in <filename><linux/tracehook.h></filename>. + The core kernel calls these functions at various important places. +</para> + +!Finclude/linux/tracehook.h + +</sect1> + +<sect1 id="arch"><title>Architecture Calls Out</title> + +<para> + An arch that has done all these things sets + <constant>CONFIG_HAVE_ARCH_TRACEHOOK</constant>. + This is required to enable the <application>utrace</application> code. +</para> + +<sect2 id="arch-ptrace"><title><filename><asm/ptrace.h></filename></title> + +<para> + An arch defines these in <filename><asm/ptrace.h></filename> + if it supports hardware single-step or block-step features. +</para> + +!Finclude/linux/ptrace.h arch_has_single_step arch_has_block_step +!Finclude/linux/ptrace.h user_enable_single_step user_enable_block_step +!Finclude/linux/ptrace.h user_disable_single_step + +</sect2> + +<sect2 id="arch-syscall"> + <title><filename><asm/syscall.h></filename></title> + + <para> + An arch provides <filename><asm/syscall.h></filename> that + defines these as inlines, or declares them as exported functions. + These interfaces are described in <xref linkend="syscall"/>. + </para> + +</sect2> + +<sect2 id="arch-tracehook"> + <title><filename><linux/tracehook.h></filename></title> + + <para> + An arch must define <constant>TIF_NOTIFY_RESUME</constant> + and <constant>TIF_SYSCALL_TRACE</constant> + in its <filename><asm/thread_info.h></filename>. + The arch code must call the following functions, all declared + in <filename><linux/tracehook.h></filename> and + described in <xref linkend="tracehook"/>: + + <itemizedlist> + <listitem> + <para><function>tracehook_notify_resume</function></para> + </listitem> + <listitem> + <para><function>tracehook_report_syscall_entry</function></para> + </listitem> + <listitem> + <para><function>tracehook_report_syscall_exit</function></para> + </listitem> + <listitem> + <para><function>tracehook_signal_handler</function></para> + </listitem> + </itemizedlist> + + </para> + +</sect2> + +</sect1> + +</chapter> + +</book> diff --git a/fs/proc/array.c b/fs/proc/array.c index 9b45ee8..496fef3 100644 --- a/fs/proc/array.c +++ b/fs/proc/array.c @@ -81,6 +81,7 @@ #include <linux/pid_namespace.h> #include <linux/ptrace.h> #include <linux/tracehook.h> +#include <linux/utrace.h> #include <asm/pgtable.h> #include <asm/processor.h> @@ -192,6 +193,8 @@ static inline void task_state(struct seq_file *m, struct pid_namespace *ns, cred->uid, cred->euid, cred->suid, cred->fsuid, cred->gid, cred->egid, cred->sgid, cred->fsgid); + task_utrace_proc_status(m, p); + task_lock(p); if (p->files) fdt = files_fdtable(p->files); diff --git a/include/linux/sched.h b/include/linux/sched.h index a837b20..b87de83 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -1397,6 +1397,11 @@ struct task_struct { #endif seccomp_t seccomp; +#ifdef CONFIG_UTRACE + struct utrace *utrace; + unsigned long utrace_flags; +#endif + /* Thread group tracking */ u32 parent_exec_id; u32 self_exec_id; diff --git a/include/linux/tracehook.h b/include/linux/tracehook.h index e95f523..7d7bdde 100644 --- a/include/linux/tracehook.h +++ b/include/linux/tracehook.h @@ -49,6 +49,7 @@ #include <linux/sched.h> #include <linux/ptrace.h> #include <linux/security.h> +#include <linux/utrace.h> struct linux_binprm; /** @@ -63,6 +64,8 @@ struct linux_binprm; */ static inline int tracehook_expect_breakpoints(struct task_struct *task) { + if (unlikely(task_utrace_flags(task) & UTRACE_EVENT(SIGNAL_CORE))) + return 1; return (task_ptrace(task) & PT_PTRACED) != 0; } @@ -111,6 +114,9 @@ static inline void ptrace_report_syscall(struct pt_regs *regs) static inline __must_check int tracehook_report_syscall_entry( struct pt_regs *regs) { + if ((task_utrace_flags(current) & UTRACE_EVENT(SYSCALL_ENTRY)) && + utrace_report_syscall_entry(regs)) + return 1; ptrace_report_syscall(regs); return 0; } @@ -134,6 +140,9 @@ static inline __must_check int tracehook_report_syscall_entry( */ static inline void tracehook_report_syscall_exit(struct pt_regs *regs, int step) { + if (task_utrace_flags(current) & UTRACE_EVENT(SYSCALL_EXIT)) + utrace_report_syscall_exit(regs); + if (step) { siginfo_t info; user_single_step_siginfo(current, regs, &info); @@ -201,6 +210,8 @@ static inline void tracehook_report_exec(struct linux_binfmt *fmt, struct linux_binprm *bprm, struct pt_regs *regs) { + if (unlikely(task_utrace_flags(current) & UTRACE_EVENT(EXEC))) + utrace_report_exec(fmt, bprm, regs); if (!ptrace_event(PT_TRACE_EXEC, PTRACE_EVENT_EXEC, 0) && unlikely(task_ptrace(current) & PT_PTRACED)) send_sig(SIGTRAP, current, 0); @@ -218,10 +229,37 @@ static inline void tracehook_report_exec(struct linux_binfmt *fmt, */ static inline void tracehook_report_exit(long *exit_code) { + if (unlikely(task_utrace_flags(current) & UTRACE_EVENT(EXIT))) + utrace_report_exit(exit_code); ptrace_event(PT_TRACE_EXIT, PTRACE_EVENT_EXIT, *exit_code); } /** + * tracehook_init_task - task_struct has just been copied + * @task: new &struct task_struct just copied from parent + * + * Called from do_fork() when @task has just been duplicated. + * After this, @task will be passed to tracehook_free_task() + * even if the rest of its setup fails before it is fully created. + */ +static inline void tracehook_init_task(struct task_struct *task) +{ + utrace_init_task(task); +} + +/** + * tracehook_free_task - task_struct is being freed + * @task: dead &struct task_struct being freed + * + * Called from free_task() when @task is no longer in use. + */ +static inline void tracehook_free_task(struct task_struct *task) +{ + if (task_utrace_struct(task)) + utrace_free_task(task); +} + +/** * tracehook_prepare_clone - prepare for new child to be cloned * @clone_flags: %CLONE_* flags from clone/fork/vfork system call * @@ -285,6 +323,8 @@ static inline void tracehook_report_clone(struct pt_regs *regs, unsigned long clone_flags, pid_t pid, struct task_struct *child) { + if (unlikely(task_utrace_flags(current) & UTRACE_EVENT(CLONE))) + utrace_report_clone(clone_flags, child); if (unlikely(task_ptrace(child))) { /* * It doesn't matter who attached/attaching to this @@ -317,6 +357,9 @@ static inline void tracehook_report_clone_complete(int trace, pid_t pid, struct task_struct *child) { + if (unlikely(task_utrace_flags(current) & UTRACE_EVENT(CLONE)) && + (clone_flags & CLONE_VFORK)) + utrace_finish_vfork(current); if (unlikely(trace)) ptrace_event(0, trace, pid); } @@ -351,6 +394,10 @@ static inline void tracehook_report_vfork_done(struct task_struct *child, */ static inline void tracehook_prepare_release_task(struct task_struct *task) { + /* see utrace_add_engine() about this barrier */ + smp_mb(); + if (task_utrace_flags(task)) + utrace_maybe_reap(task, task_utrace_struct(task), true); } /** @@ -365,6 +412,7 @@ static inline void tracehook_prepare_release_task(struct task_struct *task) static inline void tracehook_finish_release_task(struct task_struct *task) { ptrace_release_task(task); + BUG_ON(task->exit_state != EXIT_DEAD); } /** @@ -386,6 +434,8 @@ static inline void tracehook_signal_handler(int sig, siginfo_t *info, const struct k_sigaction *ka, struct pt_regs *regs, int stepping) { + if (task_utrace_flags(current)) + utrace_signal_handler(current, stepping); if (stepping) ptrace_notify(SIGTRAP); } @@ -403,6 +453,8 @@ static inline void tracehook_signal_handler(int sig, siginfo_t *info, static inline int tracehook_consider_ignored_signal(struct task_struct *task, int sig) { + if (unlikely(task_utrace_flags(task) & UTRACE_EVENT(SIGNAL_IGN))) + return 1; return (task_ptrace(task) & PT_PTRACED) != 0; } @@ -422,6 +474,9 @@ static inline int tracehook_consider_ignored_signal(struct task_struct *task, static inline int tracehook_consider_fatal_signal(struct task_struct *task, int sig) { + if (unlikely(task_utrace_flags(task) & (UTRACE_EVENT(SIGNAL_TERM) | + UTRACE_EVENT(SIGNAL_CORE)))) + return 1; return (task_ptrace(task) & PT_PTRACED) != 0; } @@ -436,6 +491,8 @@ static inline int tracehook_consider_fatal_signal(struct task_struct *task, */ static inline int tracehook_force_sigpending(void) { + if (unlikely(task_utrace_flags(current))) + return utrace_interrupt_pending(); return 0; } @@ -465,6 +522,8 @@ static inline int tracehook_get_signal(struct task_struct *task, siginfo_t *info, struct k_sigaction *return_ka) { + if (unlikely(task_utrace_flags(task))) + return utrace_get_signal(task, regs, info, return_ka); return 0; } @@ -475,6 +534,8 @@ static inline int tracehook_get_signal(struct task_struct *task, */ static inline void tracehook_finish_jctl(void) { + if (task_utrace_flags(current)) + utrace_finish_stop(); } #define DEATH_REAP -1 @@ -497,6 +558,8 @@ static inline void tracehook_finish_jctl(void) static inline int tracehook_notify_death(struct task_struct *task, void **death_cookie, int group_dead) { + *death_cookie = task_utrace_struct(task); + if (task_detached(task)) return task->ptrace ? SIGCHLD : DEATH_REAP; @@ -533,6 +596,15 @@ static inline void tracehook_report_death(struct task_struct *task, int signal, void *death_cookie, int group_dead) { + /* + * If utrace_set_events() was just called to enable + * UTRACE_EVENT(DEATH), then we are obliged to call + * utrace_report_death() and not miss it. utrace_set_events() + * checks @task->exit_state under tasklist_lock to synchronize + * with exit_notify(), the caller. + */ + if (task_utrace_flags(task) & _UTRACE_DEATH_EVENTS) + utrace_report_death(task, death_cookie, group_dead, signal); } #ifdef TIF_NOTIFY_RESUME @@ -562,10 +634,21 @@ static inline void set_notify_resume(struct task_struct *task) * asynchronously, this will be called again before we return to * user mode. * - * Called without locks. + * Called without locks. However, on some machines this may be + * called with interrupts disabled. */ static inline void tracehook_notify_resume(struct pt_regs *regs) { + struct task_struct *task = current; + /* + * Prevent the following store/load from getting ahead of the + * caller which clears TIF_NOTIFY_RESUME. This pairs with the + * implicit mb() before setting TIF_NOTIFY_RESUME in + * set_notify_resume(). + */ + smp_mb(); + if (task_utrace_flags(task)) + utrace_resume(task, regs); } #endif /* TIF_NOTIFY_RESUME */ diff --git a/include/linux/utrace.h b/include/linux/utrace.h new file mode 100644 index 0000000..f251efe --- /dev/null +++ b/include/linux/utrace.h @@ -0,0 +1,692 @@ +/* + * utrace infrastructure interface for debugging user processes + * + * Copyright (C) 2006-2009 Red Hat, Inc. All rights reserved. + * + * This copyrighted material is made available to anyone wishing to use, + * modify, copy, or redistribute it subject to the terms and conditions + * of the GNU General Public License v.2. + * + * Red Hat Author: Roland McGrath. + * + * This interface allows for notification of interesting events in a + * thread. It also mediates access to thread state such as registers. + * Multiple unrelated users can be associated with a single thread. + * We call each of these a tracing engine. + * + * A tracing engine starts by calling utrace_attach_task() or + * utrace_attach_pid() on the chosen thread, passing in a set of hooks + * (&struct utrace_engine_ops), and some associated data. This produces a + * &struct utrace_engine, which is the handle used for all other + * operations. An attached engine has its ops vector, its data, and an + * event mask controlled by utrace_set_events(). + * + * For each event bit that is set, that engine will get the + * appropriate ops->report_*() callback when the event occurs. The + * &struct utrace_engine_ops need not provide callbacks for an event + * unless the engine sets one of the associated event bits. + */ + +#ifndef _LINUX_UTRACE_H +#define _LINUX_UTRACE_H 1 + +#include <linux/list.h> +#include <linux/kref.h> +#include <linux/signal.h> +#include <linux/sched.h> + +struct linux_binprm; +struct pt_regs; +struct utrace; +struct user_regset; +struct user_regset_view; + +/* + * Event bits passed to utrace_set_events(). + * These appear in &struct task_struct.@utrace_flags + * and &struct utrace_engine.@flags. + */ +enum utrace_events { + _UTRACE_EVENT_QUIESCE, /* Thread is available for examination. */ + _UTRACE_EVENT_REAP, /* Zombie reaped, no more tracing possible. */ + _UTRACE_EVENT_CLONE, /* Successful clone/fork/vfork just done. */ + _UTRACE_EVENT_EXEC, /* Successful execve just completed. */ + _UTRACE_EVENT_EXIT, /* Thread exit in progress. */ + _UTRACE_EVENT_DEATH, /* Thread has died. */ + _UTRACE_EVENT_SYSCALL_ENTRY, /* User entered kernel for system call. */ + _UTRACE_EVENT_SYSCALL_EXIT, /* Returning to user after system call. */ + _UTRACE_EVENT_SIGNAL, /* Signal delivery will run a user handler. */ + _UTRACE_EVENT_SIGNAL_IGN, /* No-op signal to be delivered. */ + _UTRACE_EVENT_SIGNAL_STOP, /* Signal delivery will suspend. */ + _UTRACE_EVENT_SIGNAL_TERM, /* Signal delivery will terminate. */ + _UTRACE_EVENT_SIGNAL_CORE, /* Signal delivery will dump core. */ + _UTRACE_EVENT_JCTL, /* Job control stop or continue completed. */ + _UTRACE_NEVENTS +}; +#define UTRACE_EVENT(type) (1UL << _UTRACE_EVENT_##type) + +/* + * All the kinds of signal events. + * These all use the @report_signal() callback. + */ +#define UTRACE_EVENT_SIGNAL_ALL (UTRACE_EVENT(SIGNAL) \ + | UTRACE_EVENT(SIGNAL_IGN) \ + | UTRACE_EVENT(SIGNAL_STOP) \ + | UTRACE_EVENT(SIGNAL_TERM) \ + | UTRACE_EVENT(SIGNAL_CORE)) +/* + * Both kinds of syscall events; these call the @report_syscall_entry() + * and @report_syscall_exit() callbacks, respectively. + */ +#define UTRACE_EVENT_SYSCALL \ + (UTRACE_EVENT(SYSCALL_ENTRY) | UTRACE_EVENT(SYSCALL_EXIT)) + +/* + * The event reports triggered synchronously by task death. + */ +#define _UTRACE_DEATH_EVENTS (UTRACE_EVENT(DEATH) | UTRACE_EVENT(QUIESCE)) + +/* + * Hooks in <linux/tracehook.h> call these entry points to the utrace dispatch. + */ +void utrace_free_task(struct task_struct *); +bool utrace_interrupt_pending(void); +void utrace_resume(struct task_struct *, struct pt_regs *); +void utrace_finish_stop(void); +void utrace_maybe_reap(struct task_struct *, struct utrace *, bool); +int utrace_get_signal(struct task_struct *, struct pt_regs *, + siginfo_t *, struct k_sigaction *); +void utrace_report_clone(unsigned long, struct task_struct *); +void utrace_finish_vfork(struct task_struct *); +void utrace_report_exit(long *exit_code); +void utrace_report_death(struct task_struct *, struct utrace *, bool, int); +void utrace_report_jctl(int notify, int type); +void utrace_report_exec(struct linux_binfmt *, struct linux_binprm *, + struct pt_regs *regs); +bool utrace_report_syscall_entry(struct pt_regs *); +void utrace_report_syscall_exit(struct pt_regs *); +void utrace_signal_handler(struct task_struct *, int); + +#ifndef CONFIG_UTRACE + +/* + * <linux/tracehook.h> uses these accessors to avoid #ifdef CONFIG_UTRACE. + */ +static inline unsigned long task_utrace_flags(struct task_struct *task) +{ + return 0; +} +static inline struct utrace *task_utrace_struct(struct task_struct *task) +{ + return NULL; +} +static inline void utrace_init_task(struct task_struct *child) +{ +} + +static inline void task_utrace_proc_status(struct seq_file *m, + struct task_struct *p) +{ +} + +#else /* CONFIG_UTRACE */ + +static inline unsigned long task_utrace_flags(struct task_struct *task) +{ + return task->utrace_flags; +} + +static inline struct utrace *task_utrace_struct(struct task_struct *task) +{ + struct utrace *utrace; + + /* + * This barrier ensures that any prior load of task->utrace_flags + * is ordered before this load of task->utrace. We use those + * utrace_flags checks in the hot path to decide to call into + * the utrace code. The first attach installs task->utrace before + * setting task->utrace_flags nonzero with implicit barrier in + * between, see utrace_add_engine(). + */ + smp_rmb(); + utrace = task->utrace; + + smp_read_barrier_depends(); /* See utrace_task_alloc(). */ + return utrace; +} + +static inline void utrace_init_task(struct task_struct *task) +{ + task->utrace_flags = 0; + task->utrace = NULL; +} + +void task_utrace_proc_status(struct seq_file *m, struct task_struct *p); + + +/* + * Version number of the API defined in this file. This will change + * whenever a tracing engine's code would need some updates to keep + * working. We maintain this here for the benefit of tracing engine code + * that is developed concurrently with utrace API improvements before they + * are merged into the kernel, making LINUX_VERSION_CODE checks unwieldy. + */ +#define UTRACE_API_VERSION 20091216 + +/** + * enum utrace_resume_action - engine's choice of action for a traced task + * @UTRACE_STOP: Stay quiescent after callbacks. + * @UTRACE_INTERRUPT: Make @report_signal() callback soon. + * @UTRACE_REPORT: Make some callback soon. + * @UTRACE_SINGLESTEP: Resume in user mode for one instruction. + * @UTRACE_BLOCKSTEP: Resume in user mode until next branch. + * @UTRACE_RESUME: Resume normally in user mode. + * @UTRACE_DETACH: Detach my engine (implies %UTRACE_RESUME). + * + * See utrace_control() for detailed descriptions of each action. This is + * encoded in the @action argument and the return value for every callback + * with a &u32 return value. + * + * The order of these is important. When there is more than one engine, + * each supplies its choice and the smallest value prevails. + */ +enum utrace_resume_action { + UTRACE_STOP, + UTRACE_INTERRUPT, + UTRACE_REPORT, + UTRACE_SINGLESTEP, + UTRACE_BLOCKSTEP, + UTRACE_RESUME, + UTRACE_DETACH, + UTRACE_RESUME_MAX +}; +#define UTRACE_RESUME_BITS (ilog2(UTRACE_RESUME_MAX) + 1) +#define UTRACE_RESUME_MASK ((1 << UTRACE_RESUME_BITS) - 1) + +/** + * utrace_resume_action - &enum utrace_resume_action from callback action + * @action: &u32 callback @action argument or return value + * + * This extracts the &enum utrace_resume_action from @action, + * which is the @action argument to a &struct utrace_engine_ops + * callback or the return value from one. + */ +static inline enum utrace_resume_action utrace_resume_action(u32 action) +{ + return action & UTRACE_RESUME_MASK; +} + +/** + * enum utrace_signal_action - disposition of signal + * @UTRACE_SIGNAL_DELIVER: Deliver according to sigaction. + * @UTRACE_SIGNAL_IGN: Ignore the signal. + * @UTRACE_SIGNAL_TERM: Terminate the process. + * @UTRACE_SIGNAL_CORE: Terminate with core dump. + * @UTRACE_SIGNAL_STOP: Deliver as absolute stop. + * @UTRACE_SIGNAL_TSTP: Deliver as job control stop. + * @UTRACE_SIGNAL_REPORT: Reporting before pending signals. + * @UTRACE_SIGNAL_HANDLER: Reporting after signal handler setup. + * + * This is encoded in the @action argument and the return value for + * a @report_signal() callback. It says what will happen to the + * signal described by the &siginfo_t parameter to the callback. + * + * The %UTRACE_SIGNAL_REPORT value is used in an @action argument when + * a tracing report is being made before dequeuing any pending signal. + * If this is immediately after a signal handler has been set up, then + * %UTRACE_SIGNAL_HANDLER is used instead. A @report_signal callback + * that uses %UTRACE_SIGNAL_DELIVER|%UTRACE_SINGLESTEP will ensure + * it sees a %UTRACE_SIGNAL_HANDLER report. + */ +enum utrace_signal_action { + UTRACE_SIGNAL_DELIVER = 0x00, + UTRACE_SIGNAL_IGN = 0x10, + UTRACE_SIGNAL_TERM = 0x20, + UTRACE_SIGNAL_CORE = 0x30, + UTRACE_SIGNAL_STOP = 0x40, + UTRACE_SIGNAL_TSTP = 0x50, + UTRACE_SIGNAL_REPORT = 0x60, + UTRACE_SIGNAL_HANDLER = 0x70 +}; +#define UTRACE_SIGNAL_MASK 0xf0 +#define UTRACE_SIGNAL_HOLD 0x100 /* Flag, push signal back on queue. */ + +/** + * utrace_signal_action - &enum utrace_signal_action from callback action + * @action: @report_signal callback @action argument or return value + * + * This extracts the &enum utrace_signal_action from @action, which + * is the @action argument to a @report_signal callback or the + * return value from one. + */ +static inline enum utrace_signal_action utrace_signal_action(u32 action) +{ + return action & UTRACE_SIGNAL_MASK; +} + +/** + * enum utrace_syscall_action - disposition of system call attempt + * @UTRACE_SYSCALL_RUN: Run the system call. + * @UTRACE_SYSCALL_ABORT: Don't run the system call. + * + * This is encoded in the @action argument and the return value for + * a @report_syscall_entry callback. + */ +enum utrace_syscall_action { + UTRACE_SYSCALL_RUN = 0x00, + UTRACE_SYSCALL_ABORT = 0x10 +}; +#define UTRACE_SYSCALL_MASK 0xf0 +#define UTRACE_SYSCALL_RESUMED 0x100 /* Flag, report_syscall_entry() repeats */ + +/** + * utrace_syscall_action - &enum utrace_syscall_action from callback action + * @action: @report_syscall_entry callback @action or return value + * + * This extracts the &enum utrace_syscall_action from @action, which + * is the @action argument to a @report_syscall_entry callback or the + * return value from one. + */ +static inline enum utrace_syscall_action utrace_syscall_action(u32 action) +{ + return action & UTRACE_SYSCALL_MASK; +} + +/* + * Flags for utrace_attach_task() and utrace_attach_pid(). + */ +#define UTRACE_ATTACH_MATCH_OPS 0x0001 /* Match engines on ops. */ +#define UTRACE_ATTACH_MATCH_DATA 0x0002 /* Match engines on data. */ +#define UTRACE_ATTACH_MATCH_MASK 0x000f +#define UTRACE_ATTACH_CREATE 0x0010 /* Attach a new engine. */ +#define UTRACE_ATTACH_EXCLUSIVE 0x0020 /* Refuse if existing match. */ + +/** + * struct utrace_engine - per-engine structure + * @ops: &struct utrace_engine_ops pointer passed to utrace_attach_task() + * @data: engine-private &void * passed to utrace_attach_task() + * @flags: event mask set by utrace_set_events() plus internal flag bits + * + * The task itself never has to worry about engines detaching while + * it's doing event callbacks. These structures are removed from the + * task's active list only when it's stopped, or by the task itself. + * + * utrace_engine_get() and utrace_engine_put() maintain a reference count. + * When it drops to zero, the structure is freed. One reference is held + * implicitly while the engine is attached to its task. + */ +struct utrace_engine { +/* private: */ + struct kref kref; + void (*release)(void *); + struct list_head entry; + +/* public: */ + const struct utrace_engine_ops *ops; + void *data; + + unsigned long flags; +}; + +/** + * utrace_engine_get - acquire a reference on a &struct utrace_engine + * @engine: &struct utrace_engine pointer + * + * You must hold a reference on @engine, and you get another. + */ +static inline void utrace_engine_get(struct utrace_engine *engine) +{ + kref_get(&engine->kref); +} + +void __utrace_engine_release(struct kref *); + +/** + * utrace_engine_put - release a reference on a &struct utrace_engine + * @engine: &struct utrace_engine pointer + * + * You must hold a reference on @engine, and you lose that reference. + * If it was the last one, @engine becomes an invalid pointer. + */ +static inline void utrace_engine_put(struct utrace_engine *engine) +{ + kref_put(&engine->kref, __utrace_engine_release); +} + +/** + * struct utrace_engine_ops - tracing engine callbacks + * + * Each @report_*() callback corresponds to an %UTRACE_EVENT(*) bit. + * utrace_set_events() calls on @engine choose which callbacks will + * be made to @engine from @task. + * + * Most callbacks take an @action argument, giving the resume action + * chosen by other tracing engines. All callbacks take an @engine + * argument. The @report_reap callback takes a @task argument that + * might or might not be @current. All other @report_* callbacks + * report an event in the @current task. + * + * For some calls, @action also includes bits specific to that event + * and utrace_resume_action() is used to extract the resume action. + * This shows what would happen if @engine wasn't there, or will if + * the callback's return value uses %UTRACE_RESUME. This always + * starts as %UTRACE_RESUME when no other tracing is being done on + * this task. + * + * All return values contain &enum utrace_resume_action bits. For + * some calls, other bits specific to that kind of event are added to + * the resume action bits with OR. These are the same bits used in + * the @action argument. The resume action returned by a callback + * does not override previous engines' choices, it only says what + * @engine wants done. What @current actually does is the action that's + * most constrained among the choices made by all attached engines. + * See utrace_control() for more information on the actions. + * + * When %UTRACE_STOP is used in @report_syscall_entry, then @current + * stops before attempting the system call. In this case, another + * @report_syscall_entry callback will follow after @current resumes if + * %UTRACE_REPORT or %UTRACE_INTERRUPT was returned by some callback + * or passed to utrace_control(). In a second or later callback, + * %UTRACE_SYSCALL_RESUMED is set in the @action argument to indicate + * a repeat callback still waiting to attempt the same system call + * invocation. This repeat callback gives each engine an opportunity + * to reexamine registers another engine might have changed while + * @current was held in %UTRACE_STOP. + * + * In other cases, the resume action does not take effect until @current + * is ready to check for signals and return to user mode. If there + * are more callbacks to be made, the last round of calls determines + * the final action. A @report_quiesce callback with @event zero, or + * a @report_signal callback, will always be the last one made before + * @current resumes. Only %UTRACE_STOP is "sticky"--if @engine returned + * %UTRACE_STOP then @current stays stopped unless @engine returns + * different from a following callback. + * + * The report_death() and report_reap() callbacks do not take @action + * arguments, and only %UTRACE_DETACH is meaningful in the return value + * from a report_death() callback. None of the resume actions applies + * to a dead thread. + * + * All @report_*() hooks are called with no locks held, in a generally + * safe environment when we will be returning to user mode soon (or just + * entered the kernel). It is fine to block for memory allocation and + * the like, but all hooks are asynchronous and must not block on + * external events! If you want the thread to block, use %UTRACE_STOP + * in your hook's return value; then later wake it up with utrace_control(). + * + * @report_quiesce: + * Requested by %UTRACE_EVENT(%QUIESCE). + * This does not indicate any event, but just that @current is in a + * safe place for examination. This call is made before each specific + * event callback, except for @report_reap. The @event argument gives + * the %UTRACE_EVENT(@which) value for the event occurring. This + * callback might be made for events @engine has not requested, if + * some other engine is tracing the event; calling utrace_set_events() + * call here can request the immediate callback for this occurrence of + * @event. @event is zero when there is no other event, @current is + * now ready to check for signals and return to user mode, and some + * engine has used %UTRACE_REPORT or %UTRACE_INTERRUPT to request this + * callback. For this case, if @report_signal is not %NULL, the + * @report_quiesce callback may be replaced with a @report_signal + * callback passing %UTRACE_SIGNAL_REPORT in its @action argument, + * whenever @current is entering the signal-check path anyway. + * + * @report_signal: + * Requested by %UTRACE_EVENT(%SIGNAL_*) or %UTRACE_EVENT(%QUIESCE). + * Use utrace_signal_action() and utrace_resume_action() on @action. + * The signal action is %UTRACE_SIGNAL_REPORT when some engine has + * used %UTRACE_REPORT or %UTRACE_INTERRUPT; the callback can choose + * to stop or to deliver an artificial signal, before pending signals. + * It's %UTRACE_SIGNAL_HANDLER instead when signal handler setup just + * finished (after a previous %UTRACE_SIGNAL_DELIVER return); this + * serves in lieu of any %UTRACE_SIGNAL_REPORT callback requested by + * %UTRACE_REPORT or %UTRACE_INTERRUPT, and is also implicitly + * requested by %UTRACE_SINGLESTEP or %UTRACE_BLOCKSTEP into the + * signal delivery. The other signal actions indicate a signal about + * to be delivered; the previous engine's return value sets the signal + * action seen by the the following engine's callback. The @info data + * can be changed at will, including @info->si_signo. The settings in + * @return_ka determines what %UTRACE_SIGNAL_DELIVER does. @orig_ka + * is what was in force before other tracing engines intervened, and + * it's %NULL when this report began as %UTRACE_SIGNAL_REPORT or + * %UTRACE_SIGNAL_HANDLER. For a report without a new signal, @info + * is left uninitialized and must be set completely by an engine that + * chooses to deliver a signal; if there was a previous @report_signal + * callback ending in %UTRACE_STOP and it was just resumed using + * %UTRACE_REPORT or %UTRACE_INTERRUPT, then @info is left unchanged + * from the previous callback. In this way, the original signal can + * be left in @info while returning %UTRACE_STOP|%UTRACE_SIGNAL_IGN + * and then found again when resuming with %UTRACE_INTERRUPT. + * The %UTRACE_SIGNAL_HOLD flag bit can be OR'd into the return value, + * and might be in @action if the previous engine returned it. This + * flag asks that the signal in @info be pushed back on @current's queue + * so that it will be seen again after whatever action is taken now. + * + * @report_clone: + * Requested by %UTRACE_EVENT(%CLONE). + * Event reported for parent, before the new task @child might run. + * @clone_flags gives the flags used in the clone system call, or + * equivalent flags for a fork() or vfork() system call. This + * function can use utrace_attach_task() on @child. Then passing + * %UTRACE_STOP to utrace_control() on @child here keeps the child + * stopped before it ever runs in user mode, %UTRACE_REPORT or + * %UTRACE_INTERRUPT ensures a callback from @child before it + * starts in user mode. + * + * @report_jctl: + * Requested by %UTRACE_EVENT(%JCTL). + * Job control event; @type is %CLD_STOPPED or %CLD_CONTINUED, + * indicating whether we are stopping or resuming now. If @notify + * is nonzero, @current is the last thread to stop and so will send + * %SIGCHLD to its parent after this callback; @notify reflects + * what the parent's %SIGCHLD has in @si_code, which can sometimes + * be %CLD_STOPPED even when @type is %CLD_CONTINUED. + * + * @report_exec: + * Requested by %UTRACE_EVENT(%EXEC). + * An execve system call has succeeded and the new program is about to + * start running. The initial user register state is handy to be tweaked + * directly in @regs. @fmt and @bprm gives the details of this exec. + * + * @report_syscall_entry: + * Requested by %UTRACE_EVENT(%SYSCALL_ENTRY). + * Thread has entered the kernel to request a system call. + * The user register state is handy to be tweaked directly in @regs. + * The @action argument contains an &enum utrace_syscall_action, + * use utrace_syscall_action() to extract it. The return value + * overrides the last engine's action for the system call. + * If the final action is %UTRACE_SYSCALL_ABORT, no system call + * is made. The details of the system call being attempted can + * be fetched here with syscall_get_nr() and syscall_get_arguments(). + * The parameter registers can be changed with syscall_set_arguments(). + * See above about the %UTRACE_SYSCALL_RESUMED flag in @action. + * Use %UTRACE_REPORT in the return value to guarantee you get + * another callback (with %UTRACE_SYSCALL_RESUMED flag) in case + * @current stops with %UTRACE_STOP before attempting the system call. + * + * @report_syscall_exit: + * Requested by %UTRACE_EVENT(%SYSCALL_EXIT). + * Thread is about to leave the kernel after a system call request. + * The user register state is handy to be tweaked directly in @regs. + * The results of the system call attempt can be examined here using + * syscall_get_error() and syscall_get_return_value(). It is safe + * here to call syscall_set_return_value() or syscall_rollback(). + * + * @report_exit: + * Requested by %UTRACE_EVENT(%EXIT). + * Thread is exiting and cannot be prevented from doing so, + * but all its state is still live. The @code value will be + * the wait result seen by the parent, and can be changed by + * this engine or others. The @orig_code value is the real + * status, not changed by any tracing engine. Returning %UTRACE_STOP + * here keeps @current stopped before it cleans up its state and dies, + * so it can be examined by other processes. When @current is allowed + * to run, it will die and get to the @report_death callback. + * + * @report_death: + * Requested by %UTRACE_EVENT(%DEATH). + * Thread is really dead now. It might be reaped by its parent at + * any time, or self-reap immediately. Though the actual reaping + * may happen in parallel, a report_reap() callback will always be + * ordered after a report_death() callback. + * + * @report_reap: + * Requested by %UTRACE_EVENT(%REAP). + * Called when someone reaps the dead task (parent, init, or self). + * This means the parent called wait, or else this was a detached + * thread or a process whose parent ignores SIGCHLD. + * No more callbacks are made after this one. + * The engine is always detached. + * There is nothing more a tracing engine can do about this thread. + * After this callback, the @engine pointer will become invalid. + * The @task pointer may become invalid if get_task_struct() hasn't + * been used to keep it alive. + * An engine should always request this callback if it stores the + * @engine pointer or stores any pointer in @engine->data, so it + * can clean up its data structures. + * Unlike other callbacks, this can be called from the parent's context + * rather than from the traced thread itself--it must not delay the + * parent by blocking. + * + * @release: + * If not %NULL, this is called after the last utrace_engine_put() + * call for a &struct utrace_engine, which could be implicit after + * a %UTRACE_DETACH return from another callback. Its argument is + * the engine's @data member. + */ +struct utrace_engine_ops { + u32 (*report_quiesce)(u32 action, struct utrace_engine *engine, + unsigned long event); + u32 (*report_signal)(u32 action, struct utrace_engine *engine, + struct pt_regs *regs, + siginfo_t *info, + const struct k_sigaction *orig_ka, + struct k_sigaction *return_ka); + u32 (*report_clone)(u32 action, struct utrace_engine *engine, + unsigned long clone_flags, + struct task_struct *child); + u32 (*report_jctl)(u32 action, struct utrace_engine *engine, + int type, int notify); + u32 (*report_exec)(u32 action, struct utrace_engine *engine, + const struct linux_binfmt *fmt, + const struct linux_binprm *bprm, + struct pt_regs *regs); + u32 (*report_syscall_entry)(u32 action, struct utrace_engine *engine, + struct pt_regs *regs); + u32 (*report_syscall_exit)(u32 action, struct utrace_engine *engine, + struct pt_regs *regs); + u32 (*report_exit)(u32 action, struct utrace_engine *engine, + long orig_code, long *code); + u32 (*report_death)(struct utrace_engine *engine, + bool group_dead, int signal); + void (*report_reap)(struct utrace_engine *engine, + struct task_struct *task); + void (*release)(void *data); +}; + +/** + * struct utrace_examiner - private state for using utrace_prepare_examine() + * + * The members of &struct utrace_examiner are private to the implementation. + * This data type holds the state from a call to utrace_prepare_examine() + * to be used by a call to utrace_finish_examine(). + */ +struct utrace_examiner { +/* private: */ + long state; + unsigned long ncsw; +}; + +/* + * These are the exported entry points for tracing engines to use. + * See kernel/utrace.c for their kerneldoc comments with interface details. + */ +struct utrace_engine *utrace_attach_task(struct task_struct *, int, + const struct utrace_engine_ops *, + void *); +struct utrace_engine *utrace_attach_pid(struct pid *, int, + const struct utrace_engine_ops *, + void *); +int __must_check utrace_control(struct task_struct *, + struct utrace_engine *, + enum utrace_resume_action); +int __must_check utrace_set_events(struct task_struct *, + struct utrace_engine *, + unsigned long eventmask); +int __must_check utrace_barrier(struct task_struct *, + struct utrace_engine *); +int __must_check utrace_prepare_examine(struct task_struct *, + struct utrace_engine *, + struct utrace_examiner *); +int __must_check utrace_finish_examine(struct task_struct *, + struct utrace_engine *, + struct utrace_examiner *); + +/** + * utrace_control_pid - control a thread being traced by a tracing engine + * @pid: thread to affect + * @engine: attached engine to affect + * @action: &enum utrace_resume_action for thread to do + * + * This is the same as utrace_control(), but takes a &struct pid + * pointer rather than a &struct task_struct pointer. The caller must + * hold a ref on @pid, but does not need to worry about the task + * staying valid. If it's been reaped so that @pid points nowhere, + * then this call returns -%ESRCH. + */ +static inline __must_check int utrace_control_pid( + struct pid *pid, struct utrace_engine *engine, + enum utrace_resume_action action) +{ + /* + * We don't bother with rcu_read_lock() here to protect the + * task_struct pointer, because utrace_control will return + * -ESRCH without looking at that pointer if the engine is + * already detached. A task_struct pointer can't die before + * all the engines are detached in release_task() first. + */ + struct task_struct *task = pid_task(pid, PIDTYPE_PID); + return unlikely(!task) ? -ESRCH : utrace_control(task, engine, action); +} + +/** + * utrace_set_events_pid - choose which event reports a tracing engine gets + * @pid: thread to affect + * @engine: attached engine to affect + * @eventmask: new event mask + * + * This is the same as utrace_set_events(), but takes a &struct pid + * pointer rather than a &struct task_struct pointer. The caller must + * hold a ref on @pid, but does not need to worry about the task + * staying valid. If it's been reaped so that @pid points nowhere, + * then this call returns -%ESRCH. + */ +static inline __must_check int utrace_set_events_pid( + struct pid *pid, struct utrace_engine *engine, unsigned long eventmask) +{ + struct task_struct *task = pid_task(pid, PIDTYPE_PID); + return unlikely(!task) ? -ESRCH : + utrace_set_events(task, engine, eventmask); +} + +/** + * utrace_barrier_pid - synchronize with simultaneous tracing callbacks + * @pid: thread to affect + * @engine: engine to affect (can be detached) + * + * This is the same as utrace_barrier(), but takes a &struct pid + * pointer rather than a &struct task_struct pointer. The caller must + * hold a ref on @pid, but does not need to worry about the task + * staying valid. If it's been reaped so that @pid points nowhere, + * then this call returns -%ESRCH. + */ +static inline __must_check int utrace_barrier_pid(struct pid *pid, + struct utrace_engine *engine) +{ + struct task_struct *task = pid_task(pid, PIDTYPE_PID); + return unlikely(!task) ? -ESRCH : utrace_barrier(task, engine); +} + +#endif /* CONFIG_UTRACE */ + +#endif /* linux/utrace.h */ diff --git a/init/Kconfig b/init/Kconfig index 412c21b..a03ae51 100644 --- a/init/Kconfig +++ b/init/Kconfig @@ -372,6 +372,15 @@ config AUDIT_TREE depends on AUDITSYSCALL select FSNOTIFY +config UTRACE + bool "Infrastructure for tracing and debugging user processes" + depends on EXPERIMENTAL + depends on HAVE_ARCH_TRACEHOOK + help + Enable the utrace process tracing interface. This is an internal + kernel interface exported to kernel modules, to track events in + user threads, extract and change user thread state. + source "kernel/irq/Kconfig" menu "RCU Subsystem" diff --git a/kernel/Makefile b/kernel/Makefile index 2d64cfc..4a22e81 100644 --- a/kernel/Makefile +++ b/kernel/Makefile @@ -68,6 +68,7 @@ obj-$(CONFIG_IKCONFIG) += configs.o obj-$(CONFIG_RESOURCE_COUNTERS) += res_counter.o obj-$(CONFIG_SMP) += stop_machine.o obj-$(CONFIG_KPROBES_SANITY_TEST) += test_kprobes.o +obj-$(CONFIG_UTRACE) += utrace.o obj-$(CONFIG_AUDIT) += audit.o auditfilter.o obj-$(CONFIG_AUDITSYSCALL) += auditsc.o obj-$(CONFIG_AUDIT_WATCH) += audit_watch.o diff --git a/kernel/fork.c b/kernel/fork.c index 0276c30..62caf3d 100644 --- a/kernel/fork.c +++ b/kernel/fork.c @@ -168,6 +168,7 @@ void free_task(struct task_struct *tsk) free_thread_info(tsk->stack); rt_mutex_debug_task_free(tsk); ftrace_graph_exit_task(tsk); + tracehook_free_task(tsk); free_task_struct(tsk); } EXPORT_SYMBOL(free_task); @@ -1095,6 +1096,8 @@ static struct task_struct *copy_process(unsigned long clone_flags, if (!p) goto fork_out; + tracehook_init_task(p); + ftrace_graph_init_task(p); rt_mutex_init_task(p); diff --git a/kernel/utrace.c b/kernel/utrace.c new file mode 100644 index 0000000..f332d65 --- /dev/null +++ b/kernel/utrace.c @@ -0,0 +1,2440 @@ +/* + * utrace infrastructure interface for debugging user processes + * + * Copyright (C) 2006-2010 Red Hat, Inc. All rights reserved. + * + * This copyrighted material is made available to anyone wishing to use, + * modify, copy, or redistribute it subject to the terms and conditions + * of the GNU General Public License v.2. + * + * Red Hat Author: Roland McGrath. + */ + +#include <linux/utrace.h> +#include <linux/tracehook.h> +#include <linux/regset.h> +#include <asm/syscall.h> +#include <linux/ptrace.h> +#include <linux/err.h> +#include <linux/sched.h> +#include <linux/freezer.h> +#include <linux/module.h> +#include <linux/init.h> +#include <linux/slab.h> +#include <linux/seq_file.h> + + +/* + * Per-thread structure private to utrace implementation. + * If task_struct.utrace_flags is nonzero, task_struct.utrace + * has always been allocated first. Once allocated, it is + * never freed until free_task(). + * + * The common event reporting loops are done by the task making the + * report without ever taking any locks. To facilitate this, the two + * lists @attached and @attaching work together for smooth asynchronous + * attaching with low overhead. Modifying either list requires @lock. + * The @attaching list can be modified any time while holding @lock. + * New engines being attached always go on this list. + * + * The @attached list is what the task itself uses for its reporting + * loops. When the task itself is not quiescent, it can use the + * @attached list without taking any lock. Nobody may modify the list + * when the task is not quiescent. When it is quiescent, that means + * that it won't run again without taking @lock itself before using + * the list. + * + * At each place where we know the task is quiescent (or it's current), + * while holding @lock, we call splice_attaching(), below. This moves + * the @attaching list members on to the end of the @attached list. + * Since this happens at the start of any reporting pass, any new + * engines attached asynchronously go on the stable @attached list + * in time to have their callbacks seen. + */ +struct utrace { + spinlock_t lock; + struct list_head attached, attaching; + + struct task_struct *cloning; + + struct utrace_engine *reporting; + + enum utrace_resume_action resume:UTRACE_RESUME_BITS; + unsigned int signal_handler:1; + unsigned int vfork_stop:1; /* need utrace_stop() before vfork wait */ + unsigned int death:1; /* in utrace_report_death() now */ + unsigned int reap:1; /* release_task() has run */ + unsigned int pending_attach:1; /* need splice_attaching() */ +}; + +static struct kmem_cache *utrace_cachep; +static struct kmem_cache *utrace_engine_cachep; +static const struct utrace_engine_ops utrace_detached_ops; /* forward decl */ + +static int __init utrace_init(void) +{ + utrace_cachep = KMEM_CACHE(utrace, SLAB_PANIC); + utrace_engine_cachep = KMEM_CACHE(utrace_engine, SLAB_PANIC); + return 0; +} +module_init(utrace_init); + +/* + * Set up @task.utrace for the first time. We can have races + * between two utrace_attach_task() calls here. The task_lock() + * governs installing the new pointer. If another one got in first, + * we just punt the new one we allocated. + * + * This returns false only in case of a memory allocation failure. + */ +static bool utrace_task_alloc(struct task_struct *task) +{ + struct utrace *utrace = kmem_cache_zalloc(utrace_cachep, GFP_KERNEL); + if (unlikely(!utrace)) + return false; + spin_lock_init(&utrace->lock); + INIT_LIST_HEAD(&utrace->attached); + INIT_LIST_HEAD(&utrace->attaching); + utrace->resume = UTRACE_RESUME; + task_lock(task); + if (likely(!task->utrace)) { + /* + * This barrier makes sure the initialization of the struct + * precedes the installation of the pointer. This pairs + * with smp_read_barrier_depends() in task_utrace_struct(). + */ + smp_wmb(); + task->utrace = utrace; + } + task_unlock(task); + + if (unlikely(task->utrace != utrace)) + kmem_cache_free(utrace_cachep, utrace); + return true; +} + +/* + * This is called via tracehook_free_task() from free_task() + * when @task is being deallocated. + */ +void utrace_free_task(struct task_struct *task) +{ + kmem_cache_free(utrace_cachep, task->utrace); +} + +/* + * This is calledwhen the task is safely quiescent, i.e. it won't consult + * utrace->attached without the lock. Move any engines attached + * asynchronously from @utrace->attaching onto the @utrace->attached list. + */ +static void splice_attaching(struct utrace *utrace) +{ + lockdep_assert_held(&utrace->lock); + list_splice_tail_init(&utrace->attaching, &utrace->attached); + utrace->pending_attach = 0; +} + +/* + * This is the exported function used by the utrace_engine_put() inline. + */ +void __utrace_engine_release(struct kref *kref) +{ + struct utrace_engine *engine = container_of(kref, struct utrace_engine, + kref); + BUG_ON(!list_empty(&engine->entry)); + if (engine->release) + (*engine->release)(engine->data); + kmem_cache_free(utrace_engine_cachep, engine); +} +EXPORT_SYMBOL_GPL(__utrace_engine_release); + +static bool engine_matches(struct utrace_engine *engine, int flags, + const struct utrace_engine_ops *ops, void *data) +{ + if ((flags & UTRACE_ATTACH_MATCH_OPS) && engine->ops != ops) + return false; + if ((flags & UTRACE_ATTACH_MATCH_DATA) && engine->data != data) + return false; + return engine->ops && engine->ops != &utrace_detached_ops; +} + +static struct utrace_engine *find_matching_engine( + struct utrace *utrace, int flags, + const struct utrace_engine_ops *ops, void *data) +{ + struct utrace_engine *engine; + list_for_each_entry(engine, &utrace->attached, entry) + if (engine_matches(engine, flags, ops, data)) + return engine; + list_for_each_entry(engine, &utrace->attaching, entry) + if (engine_matches(engine, flags, ops, data)) + return engine; + return NULL; +} + +/* + * Enqueue @engine, or maybe don't if UTRACE_ATTACH_EXCLUSIVE. + */ +static int utrace_add_engine(struct task_struct *target, + struct utrace *utrace, + struct utrace_engine *engine, + int flags, + const struct utrace_engine_ops *ops, + void *data) +{ + int ret; + + spin_lock(&utrace->lock); + + ret = -EEXIST; + if ((flags & UTRACE_ATTACH_EXCLUSIVE) && + unlikely(find_matching_engine(utrace, flags, ops, data))) + goto unlock; + + /* + * In case we had no engines before, make sure that + * utrace_flags is not zero. Since we did unlock+lock + * at least once after utrace_task_alloc() installed + * ->utrace, we have the necessary barrier which pairs + * with rmb() in task_utrace_struct(). + */ + ret = -ESRCH; + if (!target->utrace_flags) { + target->utrace_flags = UTRACE_EVENT(REAP); + /* + * If we race with tracehook_prepare_release_task() + * make sure that either it sees utrace_flags != 0 + * or we see exit_state == EXIT_DEAD. + */ + smp_mb(); + if (unlikely(target->exit_state == EXIT_DEAD)) { + target->utrace_flags = 0; + goto unlock; + } + } + + /* + * Put the new engine on the pending ->attaching list. + * Make sure it gets onto the ->attached list by the next + * time it's examined. Setting ->pending_attach ensures + * that start_report() takes the lock and splices the lists + * before the next new reporting pass. + * + * When target == current, it would be safe just to call + * splice_attaching() right here. But if we're inside a + * callback, that would mean the new engine also gets + * notified about the event that precipitated its own + * creation. This is not what the user wants. + */ + list_add_tail(&engine->entry, &utrace->attaching); + utrace->pending_attach = 1; + utrace_engine_get(engine); + ret = 0; +unlock: + spin_unlock(&utrace->lock); + + return ret; +} + +/** + * utrace_attach_task - attach new engine, or look up an attached engine + * @target: thread to attach to + * @flags: flag bits combined with OR, see below + * @ops: callback table for new engine + * @data: engine private data pointer + * + * The caller must ensure that the @target thread does not get freed, + * i.e. hold a ref or be its parent. It is always safe to call this + * on @current, or on the @child pointer in a @report_clone callback. + * For most other cases, it's easier to use utrace_attach_pid() instead. + * + * UTRACE_ATTACH_CREATE: + * Create a new engine. If %UTRACE_ATTACH_CREATE is not specified, you + * only look up an existing engine already attached to the thread. + * + * UTRACE_ATTACH_EXCLUSIVE: + * Attempting to attach a second (matching) engine fails with -%EEXIST. + * + * UTRACE_ATTACH_MATCH_OPS: Only consider engines matching @ops. + * UTRACE_ATTACH_MATCH_DATA: Only consider engines matching @data. + * + * Calls with neither %UTRACE_ATTACH_MATCH_OPS nor %UTRACE_ATTACH_MATCH_DATA + * match the first among any engines attached to @target. That means that + * %UTRACE_ATTACH_EXCLUSIVE in such a call fails with -%EEXIST if there + * are any engines on @target at all. + */ +struct utrace_engine *utrace_attach_task( + struct task_struct *target, int flags, + const struct utrace_engine_ops *ops, void *data) +{ + struct utrace *utrace = task_utrace_struct(target); + struct utrace_engine *engine; + int ret; + + if (!(flags & UTRACE_ATTACH_CREATE)) { + if (unlikely(!utrace)) + return ERR_PTR(-ENOENT); + spin_lock(&utrace->lock); + engine = find_matching_engine(utrace, flags, ops, data); + if (engine) + utrace_engine_get(engine); + spin_unlock(&utrace->lock); + return engine ?: ERR_PTR(-ENOENT); + } + + if (unlikely(!ops) || unlikely(ops == &utrace_detached_ops)) + return ERR_PTR(-EINVAL); + + if (unlikely(target->flags & PF_KTHREAD)) + /* + * Silly kernel, utrace is for users! + */ + return ERR_PTR(-EPERM); + + if (!utrace) { + if (unlikely(!utrace_task_alloc(target))) + return ERR_PTR(-ENOMEM); + utrace = task_utrace_struct(target); + } + + engine = kmem_cache_alloc(utrace_engine_cachep, GFP_KERNEL); + if (unlikely(!engine)) + return ERR_PTR(-ENOMEM); + + /* + * Initialize the new engine structure. It starts out with one ref + * to return. utrace_add_engine() adds another for being attached. + */ + kref_init(&engine->kref); + engine->flags = 0; + engine->ops = ops; + engine->data = data; + engine->release = ops->release; + + ret = utrace_add_engine(target, utrace, engine, flags, ops, data); + + if (unlikely(ret)) { + kmem_cache_free(utrace_engine_cachep, engine); + engine = ERR_PTR(ret); + } + + + return engine; +} +EXPORT_SYMBOL_GPL(utrace_attach_task); + +/** + * utrace_attach_pid - attach new engine, or look up an attached engine + * @pid: &struct pid pointer representing thread to attach to + * @flags: flag bits combined with OR, see utrace_attach_task() + * @ops: callback table for new engine + * @data: engine private data pointer + * + * This is the same as utrace_attach_task(), but takes a &struct pid + * pointer rather than a &struct task_struct pointer. The caller must + * hold a ref on @pid, but does not need to worry about the task + * staying valid. If it's been reaped so that @pid points nowhere, + * then this call returns -%ESRCH. + */ +struct utrace_engine *utrace_attach_pid( + struct pid *pid, int flags, + const struct utrace_engine_ops *ops, void *data) +{ + struct utrace_engine *engine = ERR_PTR(-ESRCH); + struct task_struct *task = get_pid_task(pid, PIDTYPE_PID); + if (task) { + engine = utrace_attach_task(task, flags, ops, data); + put_task_struct(task); + } + return engine; +} +EXPORT_SYMBOL_GPL(utrace_attach_pid); + +/* + * When an engine is detached, the target thread may still see it and + * make callbacks until it quiesces. We install a special ops vector + * with these two callbacks. When the target thread quiesces, it can + * safely free the engine itself. For any event we will always get + * the report_quiesce() callback first, so we only need this one + * pointer to be set. The only exception is report_reap(), so we + * supply that callback too. + */ +static u32 utrace_detached_quiesce(u32 action, struct utrace_engine *engine, + unsigned long event) +{ + return UTRACE_DETACH; +} + +static void utrace_detached_reap(struct utrace_engine *engine, + struct task_struct *task) +{ +} + +static const struct utrace_engine_ops utrace_detached_ops = { + .report_quiesce = &utrace_detached_quiesce, + .report_reap = &utrace_detached_reap +}; + +/* + * The caller has to hold a ref on the engine. If the attached flag is + * true (all but utrace_barrier() calls), the engine is supposed to be + * attached. If the attached flag is false (utrace_barrier() only), + * then return -ERESTARTSYS for an engine marked for detach but not yet + * fully detached. The task pointer can be invalid if the engine is + * detached. + * + * Get the utrace lock for the target task. + * Returns the struct if locked, or ERR_PTR(-errno). + * + * This has to be robust against races with: + * utrace_control(target, UTRACE_DETACH) calls + * UTRACE_DETACH after reports + * utrace_report_death + * utrace_release_task + */ +static struct utrace *get_utrace_lock(struct task_struct *target, + struct utrace_engine *engine, + bool attached) + __acquires(utrace->lock) +{ + struct utrace *utrace; + + rcu_read_lock(); + + /* + * If this engine was already detached, bail out before we look at + * the task_struct pointer at all. If it's detached after this + * check, then RCU is still keeping this task_struct pointer valid. + * + * The ops pointer is NULL when the engine is fully detached. + * It's &utrace_detached_ops when it's marked detached but still + * on the list. In the latter case, utrace_barrier() still works, + * since the target might be in the middle of an old callback. + */ + if (unlikely(!engine->ops)) { + rcu_read_unlock(); + return ERR_PTR(-ESRCH); + } + + if (unlikely(engine->ops == &utrace_detached_ops)) { + rcu_read_unlock(); + return attached ? ERR_PTR(-ESRCH) : ERR_PTR(-ERESTARTSYS); + } + + utrace = task_utrace_struct(target); + spin_lock(&utrace->lock); + if (unlikely(utrace->reap) || unlikely(!engine->ops) || + unlikely(engine->ops == &utrace_detached_ops)) { + /* + * By the time we got the utrace lock, + * it had been reaped or detached already. + */ + spin_unlock(&utrace->lock); + utrace = ERR_PTR(-ESRCH); + if (!attached && engine->ops == &utrace_detached_ops) + utrace = ERR_PTR(-ERESTARTSYS); + } + rcu_read_unlock(); + + return utrace; +} + +/* + * Now that we don't hold any locks, run through any + * detached engines and free their references. Each + * engine had one implicit ref while it was attached. + */ +static void put_detached_list(struct list_head *list) +{ + struct utrace_engine *engine, *next; + list_for_each_entry_safe(engine, next, list, entry) { + list_del_init(&engine->entry); + utrace_engine_put(engine); + } +} + +/* + * We use an extra bit in utrace_engine.flags past the event bits, + * to record whether the engine is keeping the target thread stopped. + * + * This bit is set in task_struct.utrace_flags whenever it is set in any + * engine's flags. Only utrace_reset() resets it in utrace_flags. + */ +#define ENGINE_STOP (1UL << _UTRACE_NEVENTS) + +static void mark_engine_wants_stop(struct task_struct *task, + struct utrace_engine *engine) +{ + engine->flags |= ENGINE_STOP; + task->utrace_flags |= ENGINE_STOP; +} + +static void clear_engine_wants_stop(struct utrace_engine *engine) +{ + engine->flags &= ~ENGINE_STOP; +} + +static bool engine_wants_stop(struct utrace_engine *engine) +{ + return (engine->flags & ENGINE_STOP) != 0; +} + +/** + * utrace_set_events - choose which event reports a tracing engine gets + * @target: thread to affect + * @engine: attached engine to affect + * @events: new event mask + * + * This changes the set of events for which @engine wants callbacks made. + * + * This fails with -%EALREADY and does nothing if you try to clear + * %UTRACE_EVENT(%DEATH) when the @report_death callback may already have + * begun, or if you try to newly set %UTRACE_EVENT(%DEATH) or + * %UTRACE_EVENT(%QUIESCE) when @target is already dead or dying. + * + * This fails with -%ESRCH if you try to clear %UTRACE_EVENT(%REAP) when + * the @report_reap callback may already have begun, or when @target has + * already been detached, including forcible detach on reaping. + * + * If @target was stopped before the call, then after a successful call, + * no event callbacks not requested in @events will be made; if + * %UTRACE_EVENT(%QUIESCE) is included in @events, then a + * @report_quiesce callback will be made when @target resumes. + * + * If @target was not stopped and @events excludes some bits that were + * set before, this can return -%EINPROGRESS to indicate that @target + * may have been making some callback to @engine. When this returns + * zero, you can be sure that no event callbacks you've disabled in + * @events can be made. If @events only sets new bits that were not set + * before on @engine, then -%EINPROGRESS will never be returned. + * + * To synchronize after an -%EINPROGRESS return, see utrace_barrier(). + * + * When @target is @current, -%EINPROGRESS is not returned. But note + * that a newly-created engine will not receive any callbacks related to + * an event notification already in progress. This call enables @events + * callbacks to be made as soon as @engine becomes eligible for any + * callbacks, see utrace_attach_task(). + * + * These rules provide for coherent synchronization based on %UTRACE_STOP, + * even when %SIGKILL is breaking its normal simple rules. + */ +int utrace_set_events(struct task_struct *target, + struct utrace_engine *engine, + unsigned long events) +{ + struct utrace *utrace; + unsigned long old_flags, old_utrace_flags; + int ret = -EALREADY; + + /* + * We just ignore the internal bit, so callers can use + * engine->flags to seed bitwise ops for our argument. + */ + events &= ~ENGINE_STOP; + + utrace = get_utrace_lock(target, engine, true); + if (unlikely(IS_ERR(utrace))) + return PTR_ERR(utrace); + + old_utrace_flags = target->utrace_flags; + old_flags = engine->flags & ~ENGINE_STOP; + + /* + * If utrace_report_death() is already progress now, + * it's too late to clear the death event bits. + */ + if (((old_flags & ~events) & _UTRACE_DEATH_EVENTS) && utrace->death) + goto unlock; + + /* + * When setting these flags, it's essential that we really + * synchronize with exit_notify(). They cannot be set after + * exit_notify() takes the tasklist_lock. By holding the read + * lock here while setting the flags, we ensure that the calls + * to tracehook_notify_death() and tracehook_report_death() will + * see the new flags. This ensures that utrace_release_task() + * knows positively that utrace_report_death() will be called or + * that it won't. + */ + if ((events & ~old_flags) & _UTRACE_DEATH_EVENTS) { + read_lock(&tasklist_lock); + if (unlikely(target->exit_state)) { + read_unlock(&tasklist_lock); + goto unlock; + } + target->utrace_flags |= events; + read_unlock(&tasklist_lock); + } + + engine->flags = events | (engine->flags & ENGINE_STOP); + target->utrace_flags |= events; + + if ((events & UTRACE_EVENT_SYSCALL) && + !(old_utrace_flags & UTRACE_EVENT_SYSCALL)) + set_tsk_thread_flag(target, TIF_SYSCALL_TRACE); + + ret = 0; + if ((old_flags & ~events) && target != current && + !task_is_stopped_or_traced(target) && !target->exit_state) { + /* + * This barrier ensures that our engine->flags changes + * have hit before we examine utrace->reporting, + * pairing with the barrier in start_callback(). If + * @target has not yet hit finish_callback() to clear + * utrace->reporting, we might be in the middle of a + * callback to @engine. + */ + smp_mb(); + if (utrace->reporting == engine) + ret = -EINPROGRESS; + } +unlock: + spin_unlock(&utrace->lock); + + return ret; +} +EXPORT_SYMBOL_GPL(utrace_set_events); + +/* + * Asynchronously mark an engine as being detached. + * + * This must work while the target thread races with us doing + * start_callback(), defined below. It uses smp_rmb() between checking + * @engine->flags and using @engine->ops. Here we change @engine->ops + * first, then use smp_wmb() before changing @engine->flags. This ensures + * it can check the old flags before using the old ops, or check the old + * flags before using the new ops, or check the new flags before using the + * new ops, but can never check the new flags before using the old ops. + * Hence, utrace_detached_ops might be used with any old flags in place. + * It has report_quiesce() and report_reap() callbacks to handle all cases. + */ +static void mark_engine_detached(struct utrace_engine *engine) +{ + engine->ops = &utrace_detached_ops; + smp_wmb(); + engine->flags = UTRACE_EVENT(QUIESCE); +} + +/* + * Get @target to stop and return true if it is already stopped now. + * If we return false, it will make some event callback soonish. + * Called with @utrace locked. + */ +static bool utrace_do_stop(struct task_struct *target, struct utrace *utrace) +{ + if (task_is_stopped(target)) { + /* + * Stopped is considered quiescent; when it wakes up, it will + * go through utrace_finish_stop() before doing anything else. + */ + spin_lock_irq(&target->sighand->siglock); + if (likely(task_is_stopped(target))) + __set_task_state(target, TASK_TRACED); + spin_unlock_irq(&target->sighand->siglock); + } else if (utrace->resume > UTRACE_REPORT) { + utrace->resume = UTRACE_REPORT; + set_notify_resume(target); + } + + return task_is_traced(target); +} + +/* + * If the target is not dead it should not be in tracing + * stop any more. Wake it unless it's in job control stop. + */ +static void utrace_wakeup(struct task_struct *target, struct utrace *utrace) +{ + lockdep_assert_held(&utrace->lock); + spin_lock_irq(&target->sighand->siglock); + if (target->signal->flags & SIGNAL_STOP_STOPPED || + target->signal->group_stop_count) + target->state = TASK_STOPPED; + else + wake_up_state(target, __TASK_TRACED); + spin_unlock_irq(&target->sighand->siglock); +} + +/* + * This is called when there might be some detached engines on the list or + * some stale bits in @task->utrace_flags. Clean them up and recompute the + * flags. Returns true if we're now fully detached. + * + * Called with @utrace->lock held, returns with it released. + * After this returns, @utrace might be freed if everything detached. + */ +static bool utrace_reset(struct task_struct *task, struct utrace *utrace) + __releases(utrace->lock) +{ + struct utrace_engine *engine, *next; + unsigned long flags = 0; + LIST_HEAD(detached); + + splice_attaching(utrace); + + /* + * Update the set of events of interest from the union + * of the interests of the remaining tracing engines. + * For any engine marked detached, remove it from the list. + * We'll collect them on the detached list. + */ + list_for_each_entry_safe(engine, next, &utrace->attached, entry) { + if (engine->ops == &utrace_detached_ops) { + engine->ops = NULL; + list_move(&engine->entry, &detached); + } else { + flags |= engine->flags | UTRACE_EVENT(REAP); + } + } + + if (task->exit_state) { + /* + * Once it's already dead, we never install any flags + * except REAP. When ->exit_state is set and events + * like DEATH are not set, then they never can be set. + * This ensures that utrace_release_task() knows + * positively that utrace_report_death() can never run. + */ + BUG_ON(utrace->death); + flags &= UTRACE_EVENT(REAP); + } else if (!(flags & UTRACE_EVENT_SYSCALL) && + test_tsk_thread_flag(task, TIF_SYSCALL_TRACE)) { + clear_tsk_thread_flag(task, TIF_SYSCALL_TRACE); + } + + if (!flags) { + /* + * No more engines, cleared out the utrace. + */ + utrace->resume = UTRACE_RESUME; + utrace->signal_handler = 0; + } + + /* + * If no more engines want it stopped, wake it up. + */ + if (task_is_traced(task) && !(flags & ENGINE_STOP)) { + /* + * It just resumes, so make sure single-step + * is not left set. + */ + if (utrace->resume == UTRACE_RESUME) + user_disable_single_step(task); + utrace_wakeup(task, utrace); + } + + /* + * In theory spin_lock() doesn't imply rcu_read_lock(). + * Once we clear ->utrace_flags this task_struct can go away + * because tracehook_prepare_release_task() path does not take + * utrace->lock when ->utrace_flags == 0. + */ + rcu_read_lock(); + task->utrace_flags = flags; + spin_unlock(&utrace->lock); + rcu_read_unlock(); + + put_detached_list(&detached); + + return !flags; +} + +void utrace_finish_stop(void) +{ + /* + * If we were task_is_traced() and then SIGKILL'ed, make + * sure we do nothing until the tracer drops utrace->lock. + */ + if (unlikely(__fatal_signal_pending(current))) { + struct utrace *utrace = task_utrace_struct(current); + spin_unlock_wait(&utrace->lock); + } +} + +/* + * Perform %UTRACE_STOP, i.e. block in TASK_TRACED until woken up. + * @task == current, @utrace == current->utrace, which is not locked. + * Return true if we were woken up by SIGKILL even though some utrace + * engine may still want us to stay stopped. + */ +static void utrace_stop(struct task_struct *task, struct utrace *utrace, + enum utrace_resume_action action) +{ +relock: + spin_lock(&utrace->lock); + + if (action < utrace->resume) { + /* + * Ensure a reporting pass when we're resumed. + */ + utrace->resume = action; + if (action == UTRACE_INTERRUPT) + set_thread_flag(TIF_SIGPENDING); + else + set_thread_flag(TIF_NOTIFY_RESUME); + } + + /* + * If the ENGINE_STOP bit is clear in utrace_flags, that means + * utrace_reset() ran after we processed some UTRACE_STOP return + * values from callbacks to get here. If all engines have detached + * or resumed us, we don't stop. This check doesn't require + * siglock, but it should follow the interrupt/report bookkeeping + * steps (this can matter for UTRACE_RESUME but not UTRACE_DETACH). + */ + if (unlikely(!(task->utrace_flags & ENGINE_STOP))) { + utrace_reset(task, utrace); + if (task->utrace_flags & ENGINE_STOP) + goto relock; + return; + } + + /* + * The siglock protects us against signals. As well as SIGKILL + * waking us up, we must synchronize with the signal bookkeeping + * for stop signals and SIGCONT. + */ + spin_lock_irq(&task->sighand->siglock); + + if (unlikely(__fatal_signal_pending(task))) { + spin_unlock_irq(&task->sighand->siglock); + spin_unlock(&utrace->lock); + return; + } + + __set_current_state(TASK_TRACED); + + /* + * If there is a group stop in progress, + * we must participate in the bookkeeping. + */ + if (unlikely(task->signal->group_stop_count) && + !--task->signal->group_stop_count) + task->signal->flags = SIGNAL_STOP_STOPPED; + + spin_unlock_irq(&task->sighand->siglock); + spin_unlock(&utrace->lock); + + schedule(); + + utrace_finish_stop(); + + /* + * While in TASK_TRACED, we were considered "frozen enough". + * Now that we woke up, it's crucial if we're supposed to be + * frozen that we freeze now before running anything substantial. + */ + try_to_freeze(); + + /* + * While we were in TASK_TRACED, complete_signal() considered + * us "uninterested" in signal wakeups. Now make sure our + * TIF_SIGPENDING state is correct for normal running. + */ + spin_lock_irq(&task->sighand->siglock); + recalc_sigpending(); + spin_unlock_irq(&task->sighand->siglock); +} + +/* + * Called by release_task() with @reap set to true. + * Called by utrace_report_death() with @reap set to false. + * On reap, make report_reap callbacks and clean out @utrace + * unless still making callbacks. On death, update bookkeeping + * and handle the reap work if release_task() came in first. + */ +void utrace_maybe_reap(struct task_struct *target, struct utrace *utrace, + bool reap) +{ + struct utrace_engine *engine, *next; + struct list_head attached; + + spin_lock(&utrace->lock); + + if (reap) { + /* + * If the target will do some final callbacks but hasn't + * finished them yet, we know because it clears these event + * bits after it's done. Instead of cleaning up here and + * requiring utrace_report_death() to cope with it, we + * delay the REAP report and the teardown until after the + * target finishes its death reports. + */ + utrace->reap = 1; + + if (target->utrace_flags & _UTRACE_DEATH_EVENTS) { + spin_unlock(&utrace->lock); + return; + } + } else { + /* + * After we unlock with this flag clear, any competing + * utrace_control/utrace_set_events calls know that we've + * finished our callbacks and any detach bookkeeping. + */ + utrace->death = 0; + + if (!utrace->reap) { + /* + * We're just dead, not reaped yet. This will + * reset @target->utrace_flags so the later call + * with @reap set won't hit the check above. + */ + utrace_reset(target, utrace); + return; + } + } + + /* + * utrace_add_engine() checks ->utrace_flags != 0. Since + * @utrace->reap is set, nobody can set or clear UTRACE_EVENT(REAP) + * in @engine->flags or change @engine->ops and nobody can change + * @utrace->attached after we drop the lock. + */ + target->utrace_flags = 0; + + /* + * We clear out @utrace->attached before we drop the lock so + * that find_matching_engine() can't come across any old engine + * while we are busy tearing it down. + */ + list_replace_init(&utrace->attached, &attached); + list_splice_tail_init(&utrace->attaching, &attached); + + spin_unlock(&utrace->lock); + + list_for_each_entry_safe(engine, next, &attached, entry) { + if (engine->flags & UTRACE_EVENT(REAP)) + engine->ops->report_reap(engine, target); + + engine->ops = NULL; + engine->flags = 0; + list_del_init(&engine->entry); + + utrace_engine_put(engine); + } +} + +/* + * You can't do anything to a dead task but detach it. + * If release_task() has been called, you can't do that. + * + * On the exit path, DEATH and QUIESCE event bits are set only + * before utrace_report_death() has taken the lock. At that point, + * the death report will come soon, so disallow detach until it's + * done. This prevents us from racing with it detaching itself. + * + * Called only when @target->exit_state is nonzero. + */ +static inline int utrace_control_dead(struct task_struct *target, + struct utrace *utrace, + enum utrace_resume_action action) +{ + lockdep_assert_held(&utrace->lock); + + if (action != UTRACE_DETACH || unlikely(utrace->reap)) + return -ESRCH; + + if (unlikely(utrace->death)) + /* + * We have already started the death report. We can't + * prevent the report_death and report_reap callbacks, + * so tell the caller they will happen. + */ + return -EALREADY; + + return 0; +} + +/** + * utrace_control - control a thread being traced by a tracing engine + * @target: thread to affect + * @engine: attached engine to affect + * @action: &enum utrace_resume_action for thread to do + * + * This is how a tracing engine asks a traced thread to do something. + * This call is controlled by the @action argument, which has the + * same meaning as the &enum utrace_resume_action value returned by + * event reporting callbacks. + * + * If @target is already dead (@target->exit_state nonzero), + * all actions except %UTRACE_DETACH fail with -%ESRCH. + * + * The following sections describe each option for the @action argument. + * + * UTRACE_DETACH: + * + * After this, the @engine data structure is no longer accessible, + * and the thread might be reaped. The thread will start running + * again if it was stopped and no longer has any attached engines + * that want it stopped. + * + * If the @report_reap callback may already have begun, this fails + * with -%ESRCH. If the @report_death callback may already have + * begun, this fails with -%EALREADY. + * + * If @target is not already stopped, then a callback to this engine + * might be in progress or about to start on another CPU. If so, + * then this returns -%EINPROGRESS; the detach happens as soon as + * the pending callback is finished. To synchronize after an + * -%EINPROGRESS return, see utrace_barrier(). + * + * If @target is properly stopped before utrace_control() is called, + * then after successful return it's guaranteed that no more callbacks + * to the @engine->ops vector will be made. + * + * The only exception is %SIGKILL (and exec or group-exit by another + * thread in the group), which can cause asynchronous @report_death + * and/or @report_reap callbacks even when %UTRACE_STOP was used. + * (In that event, this fails with -%ESRCH or -%EALREADY, see above.) + * + * UTRACE_STOP: + * + * This asks that @target stop running. This returns 0 only if + * @target is already stopped, either for tracing or for job + * control. Then @target will remain stopped until another + * utrace_control() call is made on @engine; @target can be woken + * only by %SIGKILL (or equivalent, such as exec or termination by + * another thread in the same thread group). + * + * This returns -%EINPROGRESS if @target is not already stopped. + * Then the effect is like %UTRACE_REPORT. A @report_quiesce or + * @report_signal callback will be made soon. Your callback can + * then return %UTRACE_STOP to keep @target stopped. + * + * This does not interrupt system calls in progress, including ones + * that sleep for a long time. For that, use %UTRACE_INTERRUPT. + * To interrupt system calls and then keep @target stopped, your + * @report_signal callback can return %UTRACE_STOP. + * + * UTRACE_RESUME: + * + * Just let @target continue running normally, reversing the effect + * of a previous %UTRACE_STOP. If another engine is keeping @target + * stopped, then it remains stopped until all engines let it resume. + * If @target was not stopped, this has no effect. + * + * UTRACE_REPORT: + * + * This is like %UTRACE_RESUME, but also ensures that there will be + * a @report_quiesce or @report_signal callback made soon. If + * @target had been stopped, then there will be a callback before it + * resumes running normally. If another engine is keeping @target + * stopped, then there might be no callbacks until all engines let + * it resume. + * + * Since this is meaningless unless @report_quiesce callbacks will + * be made, it returns -%EINVAL if @engine lacks %UTRACE_EVENT(%QUIESCE). + * + * UTRACE_INTERRUPT: + * + * This is like %UTRACE_REPORT, but ensures that @target will make a + * @report_signal callback before it resumes or delivers signals. + * If @target was in a system call or about to enter one, work in + * progress will be interrupted as if by %SIGSTOP. If another + * engine is keeping @target stopped, then there might be no + * callbacks until all engines let it resume. + * + * This gives @engine an opportunity to introduce a forced signal + * disposition via its @report_signal callback. + * + * UTRACE_SINGLESTEP: + * + * It's invalid to use this unless arch_has_single_step() returned true. + * This is like %UTRACE_RESUME, but resumes for one user instruction only. + * + * Note that passing %UTRACE_SINGLESTEP or %UTRACE_BLOCKSTEP to + * utrace_control() or returning it from an event callback alone does + * not necessarily ensure that stepping will be enabled. If there are + * more callbacks made to any engine before returning to user mode, + * then the resume action is chosen only by the last set of callbacks. + * To be sure, enable %UTRACE_EVENT(%QUIESCE) and look for the + * @report_quiesce callback with a zero event mask, or the + * @report_signal callback with %UTRACE_SIGNAL_REPORT. + * + * Since this is not robust unless @report_quiesce callbacks will + * be made, it returns -%EINVAL if @engine lacks %UTRACE_EVENT(%QUIESCE). + * + * UTRACE_BLOCKSTEP: + * + * It's invalid to use this unless arch_has_block_step() returned true. + * This is like %UTRACE_SINGLESTEP, but resumes for one whole basic + * block of user instructions. + * + * Since this is not robust unless @report_quiesce callbacks will + * be made, it returns -%EINVAL if @engine lacks %UTRACE_EVENT(%QUIESCE). + * + * %UTRACE_BLOCKSTEP devolves to %UTRACE_SINGLESTEP when another + * tracing engine is using %UTRACE_SINGLESTEP at the same time. + */ +int utrace_control(struct task_struct *target, + struct utrace_engine *engine, + enum utrace_resume_action action) +{ + struct utrace *utrace; + bool reset; + int ret; + + if (unlikely(action >= UTRACE_RESUME_MAX)) { + WARN(1, "invalid action argument to utrace_control()!"); + return -EINVAL; + } + + /* + * This is a sanity check for a programming error in the caller. + * Their request can only work properly in all cases by relying on + * a follow-up callback, but they didn't set one up! This check + * doesn't do locking, but it shouldn't matter. The caller has to + * be synchronously sure the callback is set up to be operating the + * interface properly. + */ + if (action >= UTRACE_REPORT && action < UTRACE_RESUME && + unlikely(!(engine->flags & UTRACE_EVENT(QUIESCE)))) { + WARN(1, "utrace_control() with no QUIESCE callback in place!"); + return -EINVAL; + } + + utrace = get_utrace_lock(target, engine, true); + if (unlikely(IS_ERR(utrace))) + return PTR_ERR(utrace); + + reset = task_is_traced(target); + ret = 0; + + /* + * ->exit_state can change under us, this doesn't matter. + * We do not care about ->exit_state in fact, but we do + * care about ->reap and ->death. If either flag is set, + * we must also see ->exit_state != 0. + */ + if (unlikely(target->exit_state)) { + ret = utrace_control_dead(target, utrace, action); + if (ret) { + spin_unlock(&utrace->lock); + return ret; + } + reset = true; + } + + switch (action) { + case UTRACE_STOP: + mark_engine_wants_stop(target, engine); + if (!reset && !utrace_do_stop(target, utrace)) + ret = -EINPROGRESS; + reset = false; + break; + + case UTRACE_DETACH: + if (engine_wants_stop(engine)) + target->utrace_flags &= ~ENGINE_STOP; + mark_engine_detached(engine); + reset = reset || utrace_do_stop(target, utrace); + if (!reset) { + /* + * As in utrace_set_events(), this barrier ensures + * that our engine->flags changes have hit before we + * examine utrace->reporting, pairing with the barrier + * in start_callback(). If @target has not yet hit + * finish_callback() to clear utrace->reporting, we + * might be in the middle of a callback to @engine. + */ + smp_mb(); + if (utrace->reporting == engine) + ret = -EINPROGRESS; + } + break; + + case UTRACE_RESUME: + clear_engine_wants_stop(engine); + break; + + case UTRACE_BLOCKSTEP: + /* + * Resume from stopped, step one block. + * We fall through to treat it like UTRACE_SINGLESTEP. + */ + if (unlikely(!arch_has_block_step())) { + WARN(1, "UTRACE_BLOCKSTEP when !arch_has_block_step()"); + action = UTRACE_SINGLESTEP; + } + + case UTRACE_SINGLESTEP: + /* + * Resume from stopped, step one instruction. + * We fall through to the UTRACE_REPORT case. + */ + if (unlikely(!arch_has_single_step())) { + WARN(1, + "UTRACE_SINGLESTEP when !arch_has_single_step()"); + reset = false; + ret = -EOPNOTSUPP; + break; + } + + case UTRACE_REPORT: + /* + * Make the thread call tracehook_notify_resume() soon. + * But don't bother if it's already been interrupted. + * In that case, utrace_get_signal() will be reporting soon. + */ + clear_engine_wants_stop(engine); + if (action < utrace->resume) { + utrace->resume = action; + set_notify_resume(target); + } + break; + + case UTRACE_INTERRUPT: + /* + * Make the thread call tracehook_get_signal() soon. + */ + clear_engine_wants_stop(engine); + if (utrace->resume == UTRACE_INTERRUPT) + break; + utrace->resume = UTRACE_INTERRUPT; + + /* + * If it's not already stopped, interrupt it now. We need + * the siglock here in case it calls recalc_sigpending() + * and clears its own TIF_SIGPENDING. By taking the lock, + * we've serialized any later recalc_sigpending() after our + * setting of utrace->resume to force it on. + */ + if (reset) { + /* + * This is really just to keep the invariant that + * TIF_SIGPENDING is set with UTRACE_INTERRUPT. + * When it's stopped, we know it's always going + * through utrace_get_signal() and will recalculate. + */ + set_tsk_thread_flag(target, TIF_SIGPENDING); + } else { + struct sighand_struct *sighand; + unsigned long irqflags; + sighand = lock_task_sighand(target, &irqflags); + if (likely(sighand)) { + signal_wake_up(target, 0); + unlock_task_sighand(target, &irqflags); + } + } + break; + + default: + BUG(); /* We checked it on entry. */ + } + + /* + * Let the thread resume running. If it's not stopped now, + * there is nothing more we need to do. + */ + if (reset) + utrace_reset(target, utrace); + else + spin_unlock(&utrace->lock); + + return ret; +} +EXPORT_SYMBOL_GPL(utrace_control); + +/** + * utrace_barrier - synchronize with simultaneous tracing callbacks + * @target: thread to affect + * @engine: engine to affect (can be detached) + * + * This blocks while @target might be in the midst of making a callback to + * @engine. It can be interrupted by signals and will return -%ERESTARTSYS. + * A return value of zero means no callback from @target to @engine was + * in progress. Any effect of its return value (such as %UTRACE_STOP) has + * already been applied to @engine. + * + * It's not necessary to keep the @target pointer alive for this call. + * It's only necessary to hold a ref on @engine. This will return + * safely even if @target has been reaped and has no task refs. + * + * A successful return from utrace_barrier() guarantees its ordering + * with respect to utrace_set_events() and utrace_control() calls. If + * @target was not properly stopped, event callbacks just disabled might + * still be in progress; utrace_barrier() waits until there is no chance + * an unwanted callback can be in progress. + */ +int utrace_barrier(struct task_struct *target, struct utrace_engine *engine) +{ + struct utrace *utrace; + int ret = -ERESTARTSYS; + + if (unlikely(target == current)) + return 0; + + do { + utrace = get_utrace_lock(target, engine, false); + if (unlikely(IS_ERR(utrace))) { + ret = PTR_ERR(utrace); + if (ret != -ERESTARTSYS) + break; + } else { + /* + * All engine state changes are done while + * holding the lock, i.e. before we get here. + * Since we have the lock, we only need to + * worry about @target making a callback. + * When it has entered start_callback() but + * not yet gotten to finish_callback(), we + * will see utrace->reporting == @engine. + * When @target doesn't take the lock, it uses + * barriers to order setting utrace->reporting + * before it examines the engine state. + */ + if (utrace->reporting != engine) + ret = 0; + spin_unlock(&utrace->lock); + if (!ret) + break; + } + schedule_timeout_interruptible(1); + } while (!signal_pending(current)); + + return ret; +} +EXPORT_SYMBOL_GPL(utrace_barrier); + +/* + * This is local state used for reporting loops, perhaps optimized away. + */ +struct utrace_report { + u32 result; + enum utrace_resume_action action; + enum utrace_resume_action resume_action; + bool detaches; + bool spurious; +}; + +#define INIT_REPORT(var) \ + struct utrace_report var = { \ + .action = UTRACE_RESUME, \ + .resume_action = UTRACE_RESUME, \ + .spurious = true \ + } + +/* + * We are now making the report, so clear the flag saying we need one. + * When there is a new attach, ->pending_attach is set just so we will + * know to do splice_attaching() here before the callback loop. + */ +static enum utrace_resume_action start_report(struct utrace *utrace) +{ + enum utrace_resume_action resume = utrace->resume; + if (utrace->pending_attach || + (resume > UTRACE_INTERRUPT && resume < UTRACE_RESUME)) { + spin_lock(&utrace->lock); + splice_attaching(utrace); + resume = utrace->resume; + if (resume > UTRACE_INTERRUPT) + utrace->resume = UTRACE_RESUME; + spin_unlock(&utrace->lock); + } + return resume; +} + +static inline void finish_report_reset(struct task_struct *task, + struct utrace *utrace, + struct utrace_report *report) +{ + if (unlikely(report->spurious || report->detaches)) { + spin_lock(&utrace->lock); + if (utrace_reset(task, utrace)) + report->action = UTRACE_RESUME; + } +} + +/* + * Complete a normal reporting pass, pairing with a start_report() call. + * This handles any UTRACE_DETACH or UTRACE_REPORT or UTRACE_INTERRUPT + * returns from engine callbacks. If @will_not_stop is true and any + * engine's last callback used UTRACE_STOP, we do UTRACE_REPORT here to + * ensure we stop before user mode. If there were no callbacks made, it + * will recompute @task->utrace_flags to avoid another false-positive. + */ +static void finish_report(struct task_struct *task, struct utrace *utrace, + struct utrace_report *report, bool will_not_stop) +{ + enum utrace_resume_action resume = report->action; + + if (resume == UTRACE_STOP) + resume = will_not_stop ? UTRACE_REPORT : UTRACE_RESUME; + + if (resume < utrace->resume) { + spin_lock(&utrace->lock); + utrace->resume = resume; + if (resume == UTRACE_INTERRUPT) + set_tsk_thread_flag(task, TIF_SIGPENDING); + else + set_tsk_thread_flag(task, TIF_NOTIFY_RESUME); + spin_unlock(&utrace->lock); + } + + finish_report_reset(task, utrace, report); +} + +static void finish_callback_report(struct task_struct *task, + struct utrace *utrace, + struct utrace_report *report, + struct utrace_engine *engine, + enum utrace_resume_action action) +{ + if (action == UTRACE_DETACH) { + /* + * By holding the lock here, we make sure that + * utrace_barrier() (really get_utrace_lock()) sees the + * effect of this detach. Otherwise utrace_barrier() could + * return 0 after this callback had returned UTRACE_DETACH. + * This way, a 0 return is an unambiguous indicator that any + * callback returning UTRACE_DETACH has indeed caused detach. + */ + spin_lock(&utrace->lock); + engine->ops = &utrace_detached_ops; + spin_unlock(&utrace->lock); + } + + /* + * If utrace_control() was used, treat that like UTRACE_DETACH here. + */ + if (engine->ops == &utrace_detached_ops) { + report->detaches = true; + return; + } + + if (action < report->action) + report->action = action; + + if (action != UTRACE_STOP) { + if (action < report->resume_action) + report->resume_action = action; + + if (engine_wants_stop(engine)) { + spin_lock(&utrace->lock); + clear_engine_wants_stop(engine); + spin_unlock(&utrace->lock); + } + + return; + } + + if (!engine_wants_stop(engine)) { + spin_lock(&utrace->lock); + /* + * If utrace_control() came in and detached us + * before we got the lock, we must not stop now. + */ + if (unlikely(engine->ops == &utrace_detached_ops)) + report->detaches = true; + else + mark_engine_wants_stop(task, engine); + spin_unlock(&utrace->lock); + } +} + +/* + * Apply the return value of one engine callback to @report. + * Returns true if @engine detached and should not get any more callbacks. + */ +static bool finish_callback(struct task_struct *task, struct utrace *utrace, + struct utrace_report *report, + struct utrace_engine *engine, + u32 ret) +{ + report->result = ret & ~UTRACE_RESUME_MASK; + finish_callback_report(task, utrace, report, engine, + utrace_resume_action(ret)); + + /* + * Now that we have applied the effect of the return value, + * clear this so that utrace_barrier() can stop waiting. + * A subsequent utrace_control() can stop or resume @engine + * and know this was ordered after its callback's action. + * + * We don't need any barriers here because utrace_barrier() + * takes utrace->lock. If we touched engine->flags above, + * the lock guaranteed this change was before utrace_barrier() + * examined utrace->reporting. + */ + utrace->reporting = NULL; + + /* + * We've just done an engine callback. These are allowed to sleep, + * though all well-behaved ones restrict that to blocking kalloc() + * or quickly-acquired mutex_lock() and the like. This is a good + * place to make sure tracing engines don't introduce too much + * latency under voluntary preemption. + */ + might_sleep(); + + return engine->ops == &utrace_detached_ops; +} + +/* + * Start the callbacks for @engine to consider @event (a bit mask). + * This makes the report_quiesce() callback first. If @engine wants + * a specific callback for @event, we return the ops vector to use. + * If not, we return NULL. The return value from the ops->callback + * function called should be passed to finish_callback(). + */ +static const struct utrace_engine_ops *start_callback( + struct utrace *utrace, struct utrace_report *report, + struct utrace_engine *engine, struct task_struct *task, + unsigned long event) +{ + const struct utrace_engine_ops *ops; + unsigned long want; + + /* + * This barrier ensures that we've set utrace->reporting before + * we examine engine->flags or engine->ops. utrace_barrier() + * relies on this ordering to indicate that the effect of any + * utrace_control() and utrace_set_events() calls is in place + * by the time utrace->reporting can be seen to be NULL. + */ + utrace->reporting = engine; + smp_mb(); + + /* + * This pairs with the barrier in mark_engine_detached(). + * It makes sure that we never see the old ops vector with + * the new flags, in case the original vector had no report_quiesce. + */ + want = engine->flags; + smp_rmb(); + ops = engine->ops; + + if ((want & UTRACE_EVENT(QUIESCE)) || ops == &utrace_detached_ops) { + if (finish_callback(task, utrace, report, engine, + (*ops->report_quiesce)(report->action, + engine, event))) + return NULL; + + if (!event) { + /* We only got here to report QUIESCE */ + report->spurious = false; + return NULL; + } + + /* + * finish_callback() reset utrace->reporting after the + * quiesce callback. Now we set it again (as above) + * before re-examining engine->flags, which could have + * been changed synchronously by ->report_quiesce or + * asynchronously by utrace_control() or utrace_set_events(). + */ + utrace->reporting = engine; + smp_mb(); + want = engine->flags; + } + + if (want & ENGINE_STOP) + report->action = UTRACE_STOP; + + if (want & event) { + report->spurious = false; + return ops; + } + + utrace->reporting = NULL; + return NULL; +} + +/* + * Do a normal reporting pass for engines interested in @event. + * @callback is the name of the member in the ops vector, and remaining + * args are the extras it takes after the standard three args. + */ +#define REPORT_CALLBACKS(rev, task, utrace, report, event, callback, ...) \ + do { \ + struct utrace_engine *engine; \ + const struct utrace_engine_ops *ops; \ + list_for_each_entry##rev(engine, &utrace->attached, entry) { \ + ops = start_callback(utrace, report, engine, task, \ + event); \ + if (!ops) \ + continue; \ + finish_callback(task, utrace, report, engine, \ + (*ops->callback)(__VA_ARGS__)); \ + } \ + } while (0) +#define REPORT(task, utrace, report, event, callback, ...) \ + do { \ + start_report(utrace); \ + REPORT_CALLBACKS(, task, utrace, report, event, callback, \ + (report)->action, engine, ## __VA_ARGS__); \ + finish_report(task, utrace, report, true); \ + } while (0) + +/* + * Called iff UTRACE_EVENT(EXEC) flag is set. + */ +void utrace_report_exec(struct linux_binfmt *fmt, struct linux_binprm *bprm, + struct pt_regs *regs) +{ + struct task_struct *task = current; + struct utrace *utrace = task_utrace_struct(task); + INIT_REPORT(report); + + REPORT(task, utrace, &report, UTRACE_EVENT(EXEC), + report_exec, fmt, bprm, regs); +} + +static u32 do_report_syscall_entry(struct pt_regs *regs, + struct task_struct *task, + struct utrace *utrace, + struct utrace_report *report, + u32 resume_report) +{ + start_report(utrace); + REPORT_CALLBACKS(_reverse, task, utrace, report, + UTRACE_EVENT(SYSCALL_ENTRY), report_syscall_entry, + resume_report | report->result | report->action, + engine, regs); + finish_report(task, utrace, report, false); + + if (report->action != UTRACE_STOP) + return 0; + + utrace_stop(task, utrace, report->resume_action); + + if (fatal_signal_pending(task)) { + /* + * We are continuing despite UTRACE_STOP because of a + * SIGKILL. Don't let the system call actually proceed. + */ + report->result = UTRACE_SYSCALL_ABORT; + } else if (utrace->resume <= UTRACE_REPORT) { + /* + * If we've been asked for another report after our stop, + * go back to report (and maybe stop) again before we run + * the system call. The second (and later) reports are + * marked with the UTRACE_SYSCALL_RESUMED flag so that + * engines know this is a second report at the same + * entry. This gives them the chance to examine the + * registers anew after they might have been changed + * while we were stopped. + */ + report->detaches = false; + report->spurious = true; + report->action = report->resume_action = UTRACE_RESUME; + return UTRACE_SYSCALL_RESUMED; + } + + return 0; +} + +/* + * Called iff UTRACE_EVENT(SYSCALL_ENTRY) flag is set. + * Return true to prevent the system call. + */ +bool utrace_report_syscall_entry(struct pt_regs *regs) +{ + struct task_struct *task = current; + struct utrace *utrace = task_utrace_struct(task); + INIT_REPORT(report); + u32 resume_report = 0; + + do { + resume_report = do_report_syscall_entry(regs, task, utrace, + &report, resume_report); + } while (resume_report); + + return utrace_syscall_action(report.result) == UTRACE_SYSCALL_ABORT; +} + +/* + * Called iff UTRACE_EVENT(SYSCALL_EXIT) flag is set. + */ +void utrace_report_syscall_exit(struct pt_regs *regs) +{ + struct task_struct *task = current; + struct utrace *utrace = task_utrace_struct(task); + INIT_REPORT(report); + + REPORT(task, utrace, &report, UTRACE_EVENT(SYSCALL_EXIT), + report_syscall_exit, regs); +} + +/* + * Called iff UTRACE_EVENT(CLONE) flag is set. + * This notification call blocks the wake_up_new_task call on the child. + * So we must not quiesce here. tracehook_report_clone_complete will do + * a quiescence check momentarily. + */ +void utrace_report_clone(unsigned long clone_flags, struct task_struct *child) +{ + struct task_struct *task = current; + struct utrace *utrace = task_utrace_struct(task); + INIT_REPORT(report); + + /* + * We don't use the REPORT() macro here, because we need + * to clear utrace->cloning before finish_report(). + * After finish_report(), utrace can be a stale pointer + * in cases when report.action is still UTRACE_RESUME. + */ + start_report(utrace); + utrace->cloning = child; + + REPORT_CALLBACKS(, task, utrace, &report, + UTRACE_EVENT(CLONE), report_clone, + report.action, engine, clone_flags, child); + + utrace->cloning = NULL; + finish_report(task, utrace, &report, !(clone_flags & CLONE_VFORK)); + + /* + * For a vfork, we will go into an uninterruptible block waiting + * for the child. We need UTRACE_STOP to happen before this, not + * after. For CLONE_VFORK, utrace_finish_vfork() will be called. + */ + if (report.action == UTRACE_STOP && (clone_flags & CLONE_VFORK)) { + spin_lock(&utrace->lock); + utrace->vfork_stop = 1; + spin_unlock(&utrace->lock); + } +} + +/* + * We're called after utrace_report_clone() for a CLONE_VFORK. + * If UTRACE_STOP was left from the clone report, we stop here. + * After this, we'll enter the uninterruptible wait_for_completion() + * waiting for the child. + */ +void utrace_finish_vfork(struct task_struct *task) +{ + struct utrace *utrace = task_utrace_struct(task); + + if (utrace->vfork_stop) { + spin_lock(&utrace->lock); + utrace->vfork_stop = 0; + spin_unlock(&utrace->lock); + utrace_stop(task, utrace, UTRACE_RESUME); /* XXX */ + } +} + +/* + * Called iff UTRACE_EVENT(JCTL) flag is set. + * + * Called with siglock held. + */ +void utrace_report_jctl(int notify, int what) +{ + struct task_struct *task = current; + struct utrace *utrace = task_utrace_struct(task); + INIT_REPORT(report); + + spin_unlock_irq(&task->sighand->siglock); + + REPORT(task, utrace, &report, UTRACE_EVENT(JCTL), + report_jctl, what, notify); + + spin_lock_irq(&task->sighand->siglock); +} + +/* + * Called iff UTRACE_EVENT(EXIT) flag is set. + */ +void utrace_report_exit(long *exit_code) +{ + struct task_struct *task = current; + struct utrace *utrace = task_utrace_struct(task); + INIT_REPORT(report); + long orig_code = *exit_code; + + REPORT(task, utrace, &report, UTRACE_EVENT(EXIT), + report_exit, orig_code, exit_code); + + if (report.action == UTRACE_STOP) + utrace_stop(task, utrace, report.resume_action); +} + +/* + * Called iff UTRACE_EVENT(DEATH) or UTRACE_EVENT(QUIESCE) flag is set. + * + * It is always possible that we are racing with utrace_release_task here. + * For this reason, utrace_release_task checks for the event bits that get + * us here, and delays its cleanup for us to do. + */ +void utrace_report_death(struct task_struct *task, struct utrace *utrace, + bool group_dead, int signal) +{ + INIT_REPORT(report); + + BUG_ON(!task->exit_state); + + /* + * We are presently considered "quiescent"--which is accurate + * inasmuch as we won't run any more user instructions ever again. + * But for utrace_control and utrace_set_events to be robust, they + * must be sure whether or not we will run any more callbacks. If + * a call comes in before we do, taking the lock here synchronizes + * us so we don't run any callbacks just disabled. Calls that come + * in while we're running the callbacks will see the exit.death + * flag and know that we are not yet fully quiescent for purposes + * of detach bookkeeping. + */ + spin_lock(&utrace->lock); + BUG_ON(utrace->death); + utrace->death = 1; + utrace->resume = UTRACE_RESUME; + splice_attaching(utrace); + spin_unlock(&utrace->lock); + + REPORT_CALLBACKS(, task, utrace, &report, UTRACE_EVENT(DEATH), + report_death, engine, group_dead, signal); + + utrace_maybe_reap(task, utrace, false); +} + +/* + * Finish the last reporting pass before returning to user mode. + */ +static void finish_resume_report(struct task_struct *task, + struct utrace *utrace, + struct utrace_report *report) +{ + finish_report_reset(task, utrace, report); + + switch (report->action) { + case UTRACE_STOP: + utrace_stop(task, utrace, report->resume_action); + break; + + case UTRACE_INTERRUPT: + if (!signal_pending(task)) + set_tsk_thread_flag(task, TIF_SIGPENDING); + break; + + case UTRACE_BLOCKSTEP: + if (likely(arch_has_block_step())) { + user_enable_block_step(task); + break; + } + + /* + * This means some callback is to blame for failing + * to check arch_has_block_step() itself. Warn and + * then fall through to treat it as SINGLESTEP. + */ + WARN(1, "UTRACE_BLOCKSTEP when !arch_has_block_step()"); + + case UTRACE_SINGLESTEP: + if (likely(arch_has_single_step())) { + user_enable_single_step(task); + } else { + /* + * This means some callback is to blame for failing + * to check arch_has_single_step() itself. Spew + * about it so the loser will fix his module. + */ + WARN(1, + "UTRACE_SINGLESTEP when !arch_has_single_step()"); + } + break; + + case UTRACE_REPORT: + case UTRACE_RESUME: + default: + user_disable_single_step(task); + break; + } +} + +/* + * This is called when TIF_NOTIFY_RESUME had been set (and is now clear). + * We are close to user mode, and this is the place to report or stop. + * When we return, we're going to user mode or into the signals code. + */ +void utrace_resume(struct task_struct *task, struct pt_regs *regs) +{ + struct utrace *utrace = task_utrace_struct(task); + INIT_REPORT(report); + struct utrace_engine *engine; + + /* + * Some machines get here with interrupts disabled. The same arch + * code path leads to calling into get_signal_to_deliver(), which + * implicitly reenables them by virtue of spin_unlock_irq. + */ + local_irq_enable(); + + /* + * If this flag is still set it's because there was a signal + * handler setup done but no report_signal following it. Clear + * the flag before we get to user so it doesn't confuse us later. + */ + if (unlikely(utrace->signal_handler)) { + spin_lock(&utrace->lock); + utrace->signal_handler = 0; + spin_unlock(&utrace->lock); + } + + /* + * Update our bookkeeping even if there are no callbacks made here. + */ + report.action = start_report(utrace); + + switch (report.action) { + case UTRACE_RESUME: + /* + * Anything we might have done was already handled by + * utrace_get_signal(), or this is an entirely spurious + * call. (The arch might use TIF_NOTIFY_RESUME for other + * purposes as well as calling us.) + */ + return; + case UTRACE_REPORT: + if (unlikely(!(task->utrace_flags & UTRACE_EVENT(QUIESCE)))) + break; + /* + * Do a simple reporting pass, with no specific + * callback after report_quiesce. + */ + report.action = UTRACE_RESUME; + list_for_each_entry(engine, &utrace->attached, entry) + start_callback(utrace, &report, engine, task, 0); + break; + default: + /* + * Even if this report was truly spurious, there is no need + * for utrace_reset() now. TIF_NOTIFY_RESUME was already + * cleared--it doesn't stay spuriously set. + */ + report.spurious = false; + break; + } + + /* + * Finish the report and either stop or get ready to resume. + * If utrace->resume was not UTRACE_REPORT, this applies its + * effect now (i.e. step or interrupt). + */ + finish_resume_report(task, utrace, &report); +} + +/* + * Return true if current has forced signal_pending(). + * + * This is called only when current->utrace_flags is nonzero, so we know + * that current->utrace must be set. It's not inlined in tracehook.h + * just so that struct utrace can stay opaque outside this file. + */ +bool utrace_interrupt_pending(void) +{ + return task_utrace_struct(current)->resume == UTRACE_INTERRUPT; +} + +/* + * Take the siglock and push @info back on our queue. + * Returns with @task->sighand->siglock held. + */ +static void push_back_signal(struct task_struct *task, siginfo_t *info) + __acquires(task->sighand->siglock) +{ + struct sigqueue *q; + + if (unlikely(!info->si_signo)) { /* Oh, a wise guy! */ + spin_lock_irq(&task->sighand->siglock); + return; + } + + q = sigqueue_alloc(); + if (likely(q)) { + q->flags = 0; + copy_siginfo(&q->info, info); + } + + spin_lock_irq(&task->sighand->siglock); + + sigaddset(&task->pending.signal, info->si_signo); + if (likely(q)) + list_add(&q->list, &task->pending.list); + + set_tsk_thread_flag(task, TIF_SIGPENDING); +} + +/* + * This is the hook from the signals code, called with the siglock held. + * Here is the ideal place to stop. We also dequeue and intercept signals. + */ +int utrace_get_signal(struct task_struct *task, struct pt_regs *regs, + siginfo_t *info, struct k_sigaction *return_ka) + __releases(task->sighand->siglock) + __acquires(task->sighand->siglock) +{ + struct utrace *utrace; + struct k_sigaction *ka; + INIT_REPORT(report); + struct utrace_engine *engine; + const struct utrace_engine_ops *ops; + unsigned long event, want; + u32 ret; + int signr; + + utrace = task_utrace_struct(task); + if (utrace->resume < UTRACE_RESUME || + utrace->pending_attach || utrace->signal_handler) { + enum utrace_resume_action resume; + + /* + * We've been asked for an explicit report before we + * even check for pending signals. + */ + + spin_unlock_irq(&task->sighand->siglock); + + spin_lock(&utrace->lock); + + splice_attaching(utrace); + + report.result = utrace->signal_handler ? + UTRACE_SIGNAL_HANDLER : UTRACE_SIGNAL_REPORT; + utrace->signal_handler = 0; + + resume = utrace->resume; + utrace->resume = UTRACE_RESUME; + + spin_unlock(&utrace->lock); + + /* + * Make sure signal_pending() only returns true + * if there are real signals pending. + */ + if (signal_pending(task)) { + spin_lock_irq(&task->sighand->siglock); + recalc_sigpending(); + spin_unlock_irq(&task->sighand->siglock); + } + + if (resume > UTRACE_REPORT) { + /* + * We only got here to process utrace->resume. + * Despite no callbacks, this report is not spurious. + */ + report.action = resume; + report.spurious = false; + finish_resume_report(task, utrace, &report); + return -1; + } else if (!(task->utrace_flags & UTRACE_EVENT(QUIESCE))) { + /* + * We only got here to clear utrace->signal_handler. + */ + return -1; + } + + /* + * Do a reporting pass for no signal, just for EVENT(QUIESCE). + * The engine callbacks can fill in *info and *return_ka. + * We'll pass NULL for the @orig_ka argument to indicate + * that there was no original signal. + */ + event = 0; + ka = NULL; + memset(return_ka, 0, sizeof *return_ka); + } else if (!(task->utrace_flags & UTRACE_EVENT_SIGNAL_ALL) || + unlikely(task->signal->group_stop_count)) { + /* + * If no engine is interested in intercepting signals or + * we must stop, let the caller just dequeue them normally + * or participate in group-stop. + */ + return 0; + } else { + /* + * Steal the next signal so we can let tracing engines + * examine it. From the signal number and sigaction, + * determine what normal delivery would do. If no + * engine perturbs it, we'll do that by returning the + * signal number after setting *return_ka. + */ + signr = dequeue_signal(task, &task->blocked, info); + if (signr == 0) + return signr; + BUG_ON(signr != info->si_signo); + + ka = &task->sighand->action[signr - 1]; + *return_ka = *ka; + + /* + * We are never allowed to interfere with SIGKILL. + * Just punt after filling in *return_ka for our caller. + */ + if (signr == SIGKILL) + return signr; + + if (ka->sa.sa_handler == SIG_IGN) { + event = UTRACE_EVENT(SIGNAL_IGN); + report.result = UTRACE_SIGNAL_IGN; + } else if (ka->sa.sa_handler != SIG_DFL) { + event = UTRACE_EVENT(SIGNAL); + report.result = UTRACE_SIGNAL_DELIVER; + } else if (sig_kernel_coredump(signr)) { + event = UTRACE_EVENT(SIGNAL_CORE); + report.result = UTRACE_SIGNAL_CORE; + } else if (sig_kernel_ignore(signr)) { + event = UTRACE_EVENT(SIGNAL_IGN); + report.result = UTRACE_SIGNAL_IGN; + } else if (signr == SIGSTOP) { + event = UTRACE_EVENT(SIGNAL_STOP); + report.result = UTRACE_SIGNAL_STOP; + } else if (sig_kernel_stop(signr)) { + event = UTRACE_EVENT(SIGNAL_STOP); + report.result = UTRACE_SIGNAL_TSTP; + } else { + event = UTRACE_EVENT(SIGNAL_TERM); + report.result = UTRACE_SIGNAL_TERM; + } + + /* + * Now that we know what event type this signal is, we + * can short-circuit if no engines care about those. + */ + if ((task->utrace_flags & (event | UTRACE_EVENT(QUIESCE))) == 0) + return signr; + + /* + * We have some interested engines, so tell them about + * the signal and let them change its disposition. + */ + spin_unlock_irq(&task->sighand->siglock); + } + + /* + * This reporting pass chooses what signal disposition we'll act on. + */ + list_for_each_entry(engine, &utrace->attached, entry) { + /* + * See start_callback() comment about this barrier. + */ + utrace->reporting = engine; + smp_mb(); + + /* + * This pairs with the barrier in mark_engine_detached(), + * see start_callback() comments. + */ + want = engine->flags; + smp_rmb(); + ops = engine->ops; + + if ((want & (event | UTRACE_EVENT(QUIESCE))) == 0) { + utrace->reporting = NULL; + continue; + } + + if (ops->report_signal) + ret = (*ops->report_signal)( + report.result | report.action, engine, + regs, info, ka, return_ka); + else + ret = (report.result | (*ops->report_quiesce)( + report.action, engine, event)); + + /* + * Avoid a tight loop reporting again and again if some + * engine is too stupid. + */ + switch (utrace_resume_action(ret)) { + default: + break; + case UTRACE_INTERRUPT: + case UTRACE_REPORT: + ret = (ret & ~UTRACE_RESUME_MASK) | UTRACE_RESUME; + break; + } + + finish_callback(task, utrace, &report, engine, ret); + } + + /* + * We express the chosen action to the signals code in terms + * of a representative signal whose default action does it. + * Our caller uses our return value (signr) to decide what to + * do, but uses info->si_signo as the signal number to report. + */ + switch (utrace_signal_action(report.result)) { + case UTRACE_SIGNAL_TERM: + signr = SIGTERM; + break; + + case UTRACE_SIGNAL_CORE: + signr = SIGQUIT; + break; + + case UTRACE_SIGNAL_STOP: + signr = SIGSTOP; + break; + + case UTRACE_SIGNAL_TSTP: + signr = SIGTSTP; + break; + + case UTRACE_SIGNAL_DELIVER: + signr = info->si_signo; + + if (return_ka->sa.sa_handler == SIG_DFL) { + /* + * We'll do signr's normal default action. + * For ignore, we'll fall through below. + * For stop/death, break locks and returns it. + */ + if (likely(signr) && !sig_kernel_ignore(signr)) + break; + } else if (return_ka->sa.sa_handler != SIG_IGN && + likely(signr)) { + /* + * Complete the bookkeeping after the report. + * The handler will run. If an engine wanted to + * stop or step, then make sure we do another + * report after signal handler setup. + */ + if (report.action != UTRACE_RESUME) + report.action = UTRACE_INTERRUPT; + finish_report(task, utrace, &report, true); + + if (unlikely(report.result & UTRACE_SIGNAL_HOLD)) + push_back_signal(task, info); + else + spin_lock_irq(&task->sighand->siglock); + + /* + * We do the SA_ONESHOT work here since the + * normal path will only touch *return_ka now. + */ + if (unlikely(return_ka->sa.sa_flags & SA_ONESHOT)) { + return_ka->sa.sa_flags &= ~SA_ONESHOT; + if (likely(valid_signal(signr))) { + ka = &task->sighand->action[signr - 1]; + ka->sa.sa_handler = SIG_DFL; + } + } + + return signr; + } + + /* Fall through for an ignored signal. */ + + case UTRACE_SIGNAL_IGN: + case UTRACE_SIGNAL_REPORT: + default: + /* + * If the signal is being ignored, then we are on the way + * directly back to user mode. We can stop here, or step, + * as in utrace_resume(), above. After we've dealt with that, + * our caller will relock and come back through here. + */ + finish_resume_report(task, utrace, &report); + + if (unlikely(fatal_signal_pending(task))) { + /* + * The only reason we woke up now was because of a + * SIGKILL. Don't do normal dequeuing in case it + * might get a signal other than SIGKILL. That would + * perturb the death state so it might differ from + * what the debugger would have allowed to happen. + * Instead, pluck out just the SIGKILL to be sure + * we'll die immediately with nothing else different + * from the quiescent state the debugger wanted us in. + */ + sigset_t sigkill_only; + siginitsetinv(&sigkill_only, sigmask(SIGKILL)); + spin_lock_irq(&task->sighand->siglock); + signr = dequeue_signal(task, &sigkill_only, info); + BUG_ON(signr != SIGKILL); + *return_ka = task->sighand->action[SIGKILL - 1]; + return signr; + } + + if (unlikely(report.result & UTRACE_SIGNAL_HOLD)) { + push_back_signal(task, info); + spin_unlock_irq(&task->sighand->siglock); + } + + return -1; + } + + /* + * Complete the bookkeeping after the report. + * This sets utrace->resume if UTRACE_STOP was used. + */ + finish_report(task, utrace, &report, true); + + return_ka->sa.sa_handler = SIG_DFL; + + /* + * If this signal is fatal, si_signo gets through as exit_code. + * We can't allow a completely bogus value there or else core + * kernel code can freak out. (If an engine wants to control + * the exit_code value exactly, it can do so in report_exit.) + * We'll produce a big complaint in dmesg, but won't crash. + * That's nicer for debugging your utrace engine. + */ + if (unlikely(info->si_signo & 0x80)) { + WARN(1, "utrace engine left bogus si_signo value!"); + info->si_signo = SIGTRAP; + } + + if (unlikely(report.result & UTRACE_SIGNAL_HOLD)) + push_back_signal(task, info); + else + spin_lock_irq(&task->sighand->siglock); + + if (sig_kernel_stop(signr)) + task->group_stop |= GROUP_STOP_DEQUEUED; + + return signr; +} + +/* + * This gets called after a signal handler has been set up. + * We set a flag so the next report knows it happened. + * If we're already stepping, make sure we do a report_signal. + * If not, make sure we get into utrace_resume() where we can + * clear the signal_handler flag before resuming. + */ +void utrace_signal_handler(struct task_struct *task, int stepping) +{ + struct utrace *utrace = task_utrace_struct(task); + + spin_lock(&utrace->lock); + + utrace->signal_handler = 1; + if (utrace->resume > UTRACE_INTERRUPT) { + if (stepping) { + utrace->resume = UTRACE_INTERRUPT; + set_tsk_thread_flag(task, TIF_SIGPENDING); + } else if (utrace->resume == UTRACE_RESUME) { + set_tsk_thread_flag(task, TIF_NOTIFY_RESUME); + } + } + + spin_unlock(&utrace->lock); +} + +/** + * utrace_prepare_examine - prepare to examine thread state + * @target: thread of interest, a &struct task_struct pointer + * @engine: engine pointer returned by utrace_attach_task() + * @exam: temporary state, a &struct utrace_examiner pointer + * + * This call prepares to safely examine the thread @target using + * &struct user_regset calls, or direct access to thread-synchronous fields. + * + * When @target is current, this call is superfluous. When @target is + * another thread, it must be held stopped via %UTRACE_STOP by @engine. + * + * This call may block the caller until @target stays stopped, so it must + * be called only after the caller is sure @target is about to unschedule. + * This means a zero return from a utrace_control() call on @engine giving + * %UTRACE_STOP, or a report_quiesce() or report_signal() callback to + * @engine that used %UTRACE_STOP in its return value. + * + * Returns -%ESRCH if @target is dead or -%EINVAL if %UTRACE_STOP was + * not used. If @target has started running again despite %UTRACE_STOP + * (for %SIGKILL or a spurious wakeup), this call returns -%EAGAIN. + * + * When this call returns zero, it's safe to use &struct user_regset + * calls and task_user_regset_view() on @target and to examine some of + * its fields directly. When the examination is complete, a + * utrace_finish_examine() call must follow to check whether it was + * completed safely. + */ +int utrace_prepare_examine(struct task_struct *target, + struct utrace_engine *engine, + struct utrace_examiner *exam) +{ + int ret = 0; + + if (unlikely(target == current)) + return 0; + + rcu_read_lock(); + if (unlikely(!engine_wants_stop(engine))) + ret = -EINVAL; + else if (unlikely(target->exit_state)) + ret = -ESRCH; + else { + exam->state = target->state; + if (unlikely(exam->state == TASK_RUNNING)) + ret = -EAGAIN; + else + get_task_struct(target); + } + rcu_read_unlock(); + + if (likely(!ret)) { + exam->ncsw = wait_task_inactive(target, exam->state); + put_task_struct(target); + if (unlikely(!exam->ncsw)) + ret = -EAGAIN; + } + + return ret; +} +EXPORT_SYMBOL_GPL(utrace_prepare_examine); + +/** + * utrace_finish_examine - complete an examination of thread state + * @target: thread of interest, a &struct task_struct pointer + * @engine: engine pointer returned by utrace_attach_task() + * @exam: pointer passed to utrace_prepare_examine() call + * + * This call completes an examination on the thread @target begun by a + * paired utrace_prepare_examine() call with the same arguments that + * returned success (zero). + * + * When @target is current, this call is superfluous. When @target is + * another thread, this returns zero if @target has remained unscheduled + * since the paired utrace_prepare_examine() call returned zero. + * + * When this returns an error, any examination done since the paired + * utrace_prepare_examine() call is unreliable and the data extracted + * should be discarded. The error is -%EINVAL if @engine is not + * keeping @target stopped, or -%EAGAIN if @target woke up unexpectedly. + */ +int utrace_finish_examine(struct task_struct *target, + struct utrace_engine *engine, + struct utrace_examiner *exam) +{ + int ret = 0; + + if (unlikely(target == current)) + return 0; + + rcu_read_lock(); + if (unlikely(!engine_wants_stop(engine))) + ret = -EINVAL; + else if (unlikely(target->state != exam->state)) + ret = -EAGAIN; + else + get_task_struct(target); + rcu_read_unlock(); + + if (likely(!ret)) { + unsigned long ncsw = wait_task_inactive(target, exam->state); + if (unlikely(ncsw != exam->ncsw)) + ret = -EAGAIN; + put_task_struct(target); + } + + return ret; +} +EXPORT_SYMBOL_GPL(utrace_finish_examine); + +/* + * This is declared in linux/regset.h and defined in machine-dependent + * code. We put the export here to ensure no machine forgets it. + */ +EXPORT_SYMBOL_GPL(task_user_regset_view); + +/* + * Called with rcu_read_lock() held. + */ +void task_utrace_proc_status(struct seq_file *m, struct task_struct *p) +{ + seq_printf(m, "Utrace:\t%lx\n", p->utrace_flags); +} -- 1.5.5.1

1 0

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

kernel July 2011