[PATCH v9] exec: Fix dead-lock in de_thread with ptrace_attach

This introduces signal->unsafe_execve_in_progress,
which is used to fix the case when at least one of the
sibling threads is traced, and therefore the trace
process may dead-lock in ptrace_attach, but de_thread
will need to wait for the tracer to continue execution.

The solution is to detect this situation and allow
ptrace_attach to continue, while de_thread() is still
waiting for traced zombies to be eventually released.
When the current thread changed the ptrace status from
non-traced to traced, we can simply abort the whole
execve and restart it by returning -ERESTARTSYS.
This needs to be done before changing the thread leader,
because the PTRACE_EVENT_EXEC needs to know the old
thread pid.

Although it is technically after the point of no return,
we just have to reset bprm->point_of_no_return here,
since at this time only the other threads have received
a fatal signal, not the current thread.

From the user's point of view the whole execve was
simply delayed until after the ptrace_attach.

Other threads die quickly since the cred_guard_mutex
is released, but a deadly signal is already pending.
In case the mutex_lock_killable misses the signal,
->unsafe_execve_in_progress makes sure they release
the mutex immediately and return with -ERESTARTNOINTR.

This means there is no API change, unlike the previous
version of this patch which was discussed here:

https://lore.kernel.org/lkml/b6537ae6-31b1-5c50-f32b-8b8332ace882@hotmail.de/

See tools/testing/selftests/ptrace/vmaccess.c
for a test case that gets fixed by this change.

Note that since the test case was originally designed to
test the ptrace_attach returning an error in this situation,
the test expectation needed to be adjusted, to allow the
API to succeed at the first attempt.

Signed-off-by: Bernd Edlinger <redacted>
---
 fs/exec.c                                 | 37 +++++++++++++++++++++++++++++--
 fs/proc/base.c                            |  6 +++++
 include/linux/sched/signal.h              | 13 +++++++++++
 kernel/ptrace.c                           |  9 ++++++++
 kernel/seccomp.c                          | 12 +++++++---
 tools/testing/selftests/ptrace/vmaccess.c | 25 ++++++++++++++-------
 6 files changed, 89 insertions(+), 13 deletions(-)

diff --git a/fs/exec.c b/fs/exec.c
index 8344fba..c7b1926 100644
--- a/fs/exec.c
+++ b/fs/exec.c

@@ -1040,6 +1040,8 @@ static int de_thread(struct task_struct *tsk)
 	struct signal_struct *sig = tsk->signal;
 	struct sighand_struct *oldsighand = tsk->sighand;
 	spinlock_t *lock = &oldsighand->siglock;
+	unsigned int prev_ptrace = tsk->ptrace;
+	struct task_struct *t = tsk;
 
 	if (thread_group_empty(tsk))
 		goto no_thread_group;

@@ -1062,6 +1064,17 @@ static int de_thread(struct task_struct *tsk)
 	if (!thread_group_leader(tsk))
 		sig->notify_count--;
 
+	while_each_thread(tsk, t) {
+		if (unlikely(t->ptrace) && t != tsk->group_leader)
+			sig->unsafe_execve_in_progress = true;
+	}
+
+	if (unlikely(sig->unsafe_execve_in_progress)) {
+		spin_unlock_irq(lock);
+		mutex_unlock(&sig->cred_guard_mutex);
+		spin_lock_irq(lock);
+	}
+
 	while (sig->notify_count) {
 		__set_current_state(TASK_KILLABLE);
 		spin_unlock_irq(lock);

@@ -1072,6 +1085,17 @@ static int de_thread(struct task_struct *tsk)
 	}
 	spin_unlock_irq(lock);
 
+	if (unlikely(sig->unsafe_execve_in_progress)) {
+		if (mutex_lock_killable(&sig->cred_guard_mutex))
+			goto killed;
+		sig->unsafe_execve_in_progress = false;
+		if (!prev_ptrace && tsk->ptrace) {
+			sig->group_exit_task = NULL;
+			sig->notify_count = 0;
+			return -ERESTARTSYS;
+		}
+	}
+
 	/*
 	 * At this point all other threads have exited, all we have to
 	 * do is to wait for the thread group leader to become inactive,

@@ -1255,8 +1279,11 @@ int begin_new_exec(struct linux_binprm * bprm)
 	 * Make this the only thread in the thread group.
 	 */
 	retval = de_thread(me);
-	if (retval)
+	if (retval) {
+		if (retval == -ERESTARTSYS)
+			bprm->point_of_no_return = false;
 		goto out;
+	}
 
 	/*
 	 * Cancel any io_uring activity across execve

@@ -1466,6 +1493,11 @@ static int prepare_bprm_creds(struct linux_binprm *bprm)
 	if (mutex_lock_interruptible(&current->signal->cred_guard_mutex))
 		return -ERESTARTNOINTR;
 
+	if (unlikely(current->signal->unsafe_execve_in_progress)) {
+		mutex_unlock(&current->signal->cred_guard_mutex);
+		return -ERESTARTNOINTR;
+	}
+
 	bprm->cred = prepare_exec_creds();
 	if (likely(bprm->cred))
 		return 0;

@@ -1482,7 +1514,8 @@ static void free_bprm(struct linux_binprm *bprm)
 	}
 	free_arg_pages(bprm);
 	if (bprm->cred) {
-		mutex_unlock(&current->signal->cred_guard_mutex);
+		if (!current->signal->unsafe_execve_in_progress)
+			mutex_unlock(&current->signal->cred_guard_mutex);
 		abort_creds(bprm->cred);
 	}
 	if (bprm->file) {

diff --git a/fs/proc/base.c b/fs/proc/base.c
index 3851bfc..3b2a55c 100644
--- a/fs/proc/base.c
+++ b/fs/proc/base.c

@@ -2739,6 +2739,12 @@ static ssize_t proc_pid_attr_write(struct file * file, const char __user * buf,
 	if (rv < 0)
 		goto out_free;
 
+	if (unlikely(current->signal->unsafe_execve_in_progress)) {
+		mutex_unlock(&current->signal->cred_guard_mutex);
+		rv = -ERESTARTNOINTR;
+		goto out_free;
+	}
+
 	rv = security_setprocattr(PROC_I(inode)->op.lsm,
 				  file->f_path.dentry->d_name.name, page,
 				  count);

diff --git a/include/linux/sched/signal.h b/include/linux/sched/signal.h
index 3f6a0fc..220a083 100644
--- a/include/linux/sched/signal.h
+++ b/include/linux/sched/signal.h

@@ -214,6 +214,17 @@ struct signal_struct {
 #endif
 
 	/*
+	 * Set while execve is executing but is *not* holding
+	 * cred_guard_mutex to avoid possible dead-locks.
+	 * The cred_guard_mutex is released *after* de_thread() has
+	 * called zap_other_threads(), therefore a fatal signal is
+	 * guaranteed to be already pending in the unlikely event, that
+	 * current->signal->unsafe_execve_in_progress happens to be
+	 * true after the cred_guard_mutex was acquired.
+	 */
+	bool unsafe_execve_in_progress;
+
+	/*
 	 * Thread is the potential origin of an oom condition; kill first on
 	 * oom
 	 */

@@ -227,6 +238,8 @@ struct signal_struct {
 	struct mutex cred_guard_mutex;	/* guard against foreign influences on
 					 * credential calculations
 					 * (notably. ptrace)
+					 * Held while execve runs, except when
+					 * a sibling thread is being traced.
 					 * Deprecated do not use in new code.
 					 * Use exec_update_lock instead.
 					 */

diff --git a/kernel/ptrace.c b/kernel/ptrace.c
index 61db50f..0cbc1eb 100644
--- a/kernel/ptrace.c
+++ b/kernel/ptrace.c

@@ -468,6 +468,14 @@ static int ptrace_traceme(void)
 {
 	int ret = -EPERM;
 
+	if (mutex_lock_interruptible(&current->signal->cred_guard_mutex))
+		return -ERESTARTNOINTR;
+
+	if (unlikely(current->signal->unsafe_execve_in_progress)) {
+		mutex_unlock(&current->signal->cred_guard_mutex);
+		return -ERESTARTNOINTR;
+	}
+
 	write_lock_irq(&tasklist_lock);
 	/* Are we already being traced? */
 	if (!current->ptrace) {

@@ -483,6 +491,7 @@ static int ptrace_traceme(void)
 		}
 	}
 	write_unlock_irq(&tasklist_lock);
+	mutex_unlock(&current->signal->cred_guard_mutex);
 
 	return ret;
 }

diff --git a/kernel/seccomp.c b/kernel/seccomp.c
index 1d60fc2..b1389ee 100644
--- a/kernel/seccomp.c
+++ b/kernel/seccomp.c

@@ -1824,9 +1824,15 @@ static long seccomp_set_mode_filter(unsigned int flags,
 	 * Make sure we cannot change seccomp or nnp state via TSYNC
 	 * while another thread is in the middle of calling exec.
 	 */
-	if (flags & SECCOMP_FILTER_FLAG_TSYNC &&
-	    mutex_lock_killable(&current->signal->cred_guard_mutex))
-		goto out_put_fd;
+	if (flags & SECCOMP_FILTER_FLAG_TSYNC) {
+		if (mutex_lock_killable(&current->signal->cred_guard_mutex))
+			goto out_put_fd;
+
+		if (unlikely(current->signal->unsafe_execve_in_progress)) {
+			mutex_unlock(&current->signal->cred_guard_mutex);
+			goto out_put_fd;
+		}
+	}
 
 	spin_lock_irq(&current->sighand->siglock);
 

diff --git a/tools/testing/selftests/ptrace/vmaccess.c b/tools/testing/selftests/ptrace/vmaccess.c
index 4db327b..c7c2242 100644
--- a/tools/testing/selftests/ptrace/vmaccess.c
+++ b/tools/testing/selftests/ptrace/vmaccess.c

@@ -39,8 +39,15 @@ static void *thread(void *arg)
 	f = open(mm, O_RDONLY);
 	ASSERT_GE(f, 0);
 	close(f);
-	f = kill(pid, SIGCONT);
-	ASSERT_EQ(f, 0);
+	f = waitpid(-1, NULL, 0);
+	ASSERT_NE(f, -1);
+	ASSERT_NE(f, 0);
+	ASSERT_NE(f, pid);
+	f = waitpid(-1, NULL, 0);
+	ASSERT_EQ(f, pid);
+	f = waitpid(-1, NULL, 0);
+	ASSERT_EQ(f, -1);
+	ASSERT_EQ(errno, ECHILD);
 }
 
 TEST(attach)

@@ -57,22 +64,24 @@ static void *thread(void *arg)
 
 	sleep(1);
 	k = ptrace(PTRACE_ATTACH, pid, 0L, 0L);
-	ASSERT_EQ(errno, EAGAIN);
-	ASSERT_EQ(k, -1);
+	ASSERT_EQ(k, 0);
 	k = waitpid(-1, &s, WNOHANG);
 	ASSERT_NE(k, -1);
 	ASSERT_NE(k, 0);
 	ASSERT_NE(k, pid);
 	ASSERT_EQ(WIFEXITED(s), 1);
 	ASSERT_EQ(WEXITSTATUS(s), 0);
-	sleep(1);
-	k = ptrace(PTRACE_ATTACH, pid, 0L, 0L);
-	ASSERT_EQ(k, 0);
 	k = waitpid(-1, &s, 0);
 	ASSERT_EQ(k, pid);
 	ASSERT_EQ(WIFSTOPPED(s), 1);
 	ASSERT_EQ(WSTOPSIG(s), SIGSTOP);
-	k = ptrace(PTRACE_DETACH, pid, 0L, 0L);
+	k = ptrace(PTRACE_CONT, pid, 0L, 0L);
+	ASSERT_EQ(k, 0);
+	k = waitpid(-1, &s, 0);
+	ASSERT_EQ(k, pid);
+	ASSERT_EQ(WIFSTOPPED(s), 1);
+	ASSERT_EQ(WSTOPSIG(s), SIGTRAP);
+	k = ptrace(PTRACE_CONT, pid, 0L, 0L);
 	ASSERT_EQ(k, 0);
 	k = waitpid(-1, &s, 0);
 	ASSERT_EQ(k, pid);

-- 
1.9.1

On Fri, 11 Jun 2021 17:55:09 +0200 Bernd Edlinger [off-list ref] wrote:

This introduces signal->unsafe_execve_in_progress,
which is used to fix the case when at least one of the
sibling threads is traced, and therefore the trace
process may dead-lock in ptrace_attach, but de_thread
will need to wait for the tracer to continue execution.

The solution is to detect this situation and allow
ptrace_attach to continue, while de_thread() is still
waiting for traced zombies to be eventually released.
When the current thread changed the ptrace status from
non-traced to traced, we can simply abort the whole
execve and restart it by returning -ERESTARTSYS.
This needs to be done before changing the thread leader,
because the PTRACE_EVENT_EXEC needs to know the old
thread pid.

Although it is technically after the point of no return,
we just have to reset bprm->point_of_no_return here,
since at this time only the other threads have received
a fatal signal, not the current thread.

quoted

From the user's point of view the whole execve was

simply delayed until after the ptrace_attach.

Other threads die quickly since the cred_guard_mutex
is released, but a deadly signal is already pending.
In case the mutex_lock_killable misses the signal,
->unsafe_execve_in_progress makes sure they release
the mutex immediately and return with -ERESTARTNOINTR.

This means there is no API change, unlike the previous
version of this patch which was discussed here:

https://lore.kernel.org/lkml/b6537ae6-31b1-5c50-f32b-8b8332ace882@hotmail.de/

See tools/testing/selftests/ptrace/vmaccess.c
for a test case that gets fixed by this change.

Note that since the test case was originally designed to
test the ptrace_attach returning an error in this situation,
the test expectation needed to be adjusted, to allow the
API to succeed at the first attempt.

err, sorry.  I replied to the v8 patch, not to v9.

--- a/fs/exec.c~exec-fix-dead-lock-in-de_thread-with-ptrace_attach-v9
+++ a/fs/exec.c

@@ -1056,29 +1056,31 @@ static int de_thread(struct task_struct
 		return -EAGAIN;
 	}
 
-	while_each_thread(tsk, t) {
-		if (unlikely(t->ptrace) && t != tsk->group_leader)
-			sig->unsafe_execve_in_progress = true;
-	}
-
 	sig->group_exit_task = tsk;
 	sig->notify_count = zap_other_threads(tsk);
 	if (!thread_group_leader(tsk))
 		sig->notify_count--;
-	spin_unlock_irq(lock);
 
-	if (unlikely(sig->unsafe_execve_in_progress))
+	while_each_thread(tsk, t) {
+		if (unlikely(t->ptrace) && t != tsk->group_leader)
+			sig->unsafe_execve_in_progress = true;
+	}
+
+	if (unlikely(sig->unsafe_execve_in_progress)) {
+		spin_unlock_irq(lock);
 		mutex_unlock(&sig->cred_guard_mutex);
+		spin_lock_irq(lock);
+	}
 
-	for (;;) {
-		set_current_state(TASK_KILLABLE);
-		if (!sig->notify_count)
-			break;
+	while (sig->notify_count) {
+		__set_current_state(TASK_KILLABLE);
+		spin_unlock_irq(lock);
 		schedule();
 		if (__fatal_signal_pending(tsk))
 			goto killed;
+		spin_lock_irq(lock);
 	}
-	__set_current_state(TASK_RUNNING);
+	spin_unlock_irq(lock);
 
 	if (unlikely(sig->unsafe_execve_in_progress)) {
 		if (mutex_lock_killable(&sig->cred_guard_mutex))

_

On 6/12/21 1:16 AM, Andrew Morton wrote:

On Fri, 11 Jun 2021 17:55:09 +0200 Bernd Edlinger [off-list ref] wrote:

quoted

This introduces signal->unsafe_execve_in_progress,
which is used to fix the case when at least one of the
sibling threads is traced, and therefore the trace
process may dead-lock in ptrace_attach, but de_thread
will need to wait for the tracer to continue execution.

The solution is to detect this situation and allow
ptrace_attach to continue, while de_thread() is still
waiting for traced zombies to be eventually released.
When the current thread changed the ptrace status from
non-traced to traced, we can simply abort the whole
execve and restart it by returning -ERESTARTSYS.
This needs to be done before changing the thread leader,
because the PTRACE_EVENT_EXEC needs to know the old
thread pid.

Although it is technically after the point of no return,
we just have to reset bprm->point_of_no_return here,
since at this time only the other threads have received
a fatal signal, not the current thread.

quoted

From the user's point of view the whole execve was

simply delayed until after the ptrace_attach.

Other threads die quickly since the cred_guard_mutex
is released, but a deadly signal is already pending.
In case the mutex_lock_killable misses the signal,
->unsafe_execve_in_progress makes sure they release
the mutex immediately and return with -ERESTARTNOINTR.

This means there is no API change, unlike the previous
version of this patch which was discussed here:

https://lore.kernel.org/lkml/b6537ae6-31b1-5c50-f32b-8b8332ace882@hotmail.de/

See tools/testing/selftests/ptrace/vmaccess.c
for a test case that gets fixed by this change.

Note that since the test case was originally designed to
test the ptrace_attach returning an error in this situation,
the test expectation needed to be adjusted, to allow the
API to succeed at the first attempt.

err, sorry.  I replied to the v8 patch, not to v9.

Sorry for the confusion.

Originally the loop here looked was entered with
sighand locked and was like this:

	while (sig->notify_count) {
		__set_current_state(TASK_KILLABLE);
		if (!sig->notify_count)
			break;
		spin_unlock_irq(lock);
		schedule();
		if (__fatal_signal_pending(tsk))
			goto killed;
	}
	spin_unlock_irq(lock);

v8 did this (tried avoid lots of spin-lock/unlocks):

	sig->group_exit_task = tsk;
	sig->notify_count = zap_other_threads(tsk);
	if (!thread_group_leader(tsk))
		sig->notify_count--;
	spin_unlock_irq(lock);

	if (unlikely(sig->unsafe_execve_in_progress))
		mutex_unlock(&sig->cred_guard_mutex);

	for (;;) {
		set_current_state(TASK_KILLABLE);
		if (!sig->notify_count)
			break;
		schedule();
		if (__fatal_signal_pending(tsk))
			goto killed;
	}

but here I overlooked that there is an execution path without
any spin-lock where sig->group_exit_task is set to NULL, which
could create a race with __signal_exit.

So v9 keeps the loop as it was, and instead does this:

	if (unlikely(sig->unsafe_execve_in_progress)) {
		spin_unlock_irq(lock);
		mutex_unlock(&sig->cred_guard_mutex);
		spin_lock_irq(lock);
	}

because I would not like to release the mutex while an
interrupt spin-lock is held.


Bernd.

On Fri, Jun 11, 2021 at 05:55:09PM +0200, Bernd Edlinger wrote:

This introduces signal->unsafe_execve_in_progress,
which is used to fix the case when at least one of the
sibling threads is traced, and therefore the trace
process may dead-lock in ptrace_attach, but de_thread
will need to wait for the tracer to continue execution.

The solution is to detect this situation and allow
ptrace_attach to continue, while de_thread() is still
waiting for traced zombies to be eventually released.
When the current thread changed the ptrace status from
non-traced to traced, we can simply abort the whole
execve and restart it by returning -ERESTARTSYS.
This needs to be done before changing the thread leader,
because the PTRACE_EVENT_EXEC needs to know the old
thread pid.

Although it is technically after the point of no return,
we just have to reset bprm->point_of_no_return here,
since at this time only the other threads have received
a fatal signal, not the current thread.

quoted

From the user's point of view the whole execve was

simply delayed until after the ptrace_attach.

Other threads die quickly since the cred_guard_mutex
is released, but a deadly signal is already pending.
In case the mutex_lock_killable misses the signal,
->unsafe_execve_in_progress makes sure they release
the mutex immediately and return with -ERESTARTNOINTR.

This means there is no API change, unlike the previous
version of this patch which was discussed here:

https://lore.kernel.org/lkml/b6537ae6-31b1-5c50-f32b-8b8332ace882@hotmail.de/

See tools/testing/selftests/ptrace/vmaccess.c
for a test case that gets fixed by this change.

Note that since the test case was originally designed to
test the ptrace_attach returning an error in this situation,
the test expectation needed to be adjusted, to allow the
API to succeed at the first attempt.

Signed-off-by: Bernd Edlinger <redacted>
---
 fs/exec.c                                 | 37 +++++++++++++++++++++++++++++--
 fs/proc/base.c                            |  6 +++++
 include/linux/sched/signal.h              | 13 +++++++++++
 kernel/ptrace.c                           |  9 ++++++++
 kernel/seccomp.c                          | 12 +++++++---
 tools/testing/selftests/ptrace/vmaccess.c | 25 ++++++++++++++-------
 6 files changed, 89 insertions(+), 13 deletions(-)

<formletter>

This is not the correct way to submit patches for inclusion in the
stable kernel tree.  Please read:
    https://www.kernel.org/doc/html/latest/process/stable-kernel-rules.html
for how to do this properly.

</formletter>