linux-kernel - [PATCH 07/17] signal: Always set JOBCTL_WILL

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date: Tue, 18 Jun 2024 23:08:14 -0500
From: "Eric W. Biederman" <ebiederm@...ssion.com>
To: Oleg Nesterov <oleg@...hat.com>
Cc: Andrew Morton <akpm@...ux-foundation.org>,  Tejun Heo <tj@...nel.org>,
  linux-kernel@...r.kernel.org
Subject: [PATCH 07/17] signal: Always set JOBCTL_WILL_EXIT for exiting tasks


This is a semantic change.  Previously JOBCTL_WILL_EXIT replaced the
use of sigaddset(&t->pending.signal, SIGKILL) to mark tasks for
termination but was otherwise identical.  Now JOBCTL_WILL_EXIT once
set remains set (except possibly for the thread performing a
coredump).

Keeping JOBCTL_WILL_EXIT set makes it possible to detect and skip
unnecessary work when the decision has been made for a task to exit.

Getting there involves not clearing JOBCTL_WILL_EXIT when short
circuited signals are being processed, and setting JOBCTL_WILL_EXIT on
all tasks when SIGNAL_GROUP_EXIT is set.

Never clearing JOBCTL_WILL_EXIT requires updating ptrace_stop
so it will continue to stop in PTRACE_EVENT_EXIT.

Set JOBCTL_WILL_EXIT in coredump_finish so that it winds up on the
coredumping thread.  It was initially set in zap_process when all of
the tasks of the processed were asked to exit, and then cleared in
zap_threads so that the coredump thread can be killed.

Signed-off-by: "Eric W. Biederman" <ebiederm@...ssion.com>
---
 fs/coredump.c   |  1 +
 kernel/exit.c   |  2 ++
 kernel/signal.c | 20 ++++++++++++--------
 3 files changed, 15 insertions(+), 8 deletions(-)

diff --git a/fs/coredump.c b/fs/coredump.c
index bec9e290802a..c8b057724bf6 100644
--- a/fs/coredump.c
+++ b/fs/coredump.c
@@ -443,6 +443,7 @@ static void coredump_finish(bool core_dumped)
 		current->signal->group_exit_code |= 0x80;
 	next = current->signal->core_state->dumper.next;
 	current->signal->core_state = NULL;
+	current->jobctl |= JOBCTL_WILL_EXIT;
 	spin_unlock_irq(&current->sighand->siglock);
 
 	while ((curr = next) != NULL) {
diff --git a/kernel/exit.c b/kernel/exit.c
index 08de33740b9c..0059c60946a3 100644
--- a/kernel/exit.c
+++ b/kernel/exit.c
@@ -801,6 +801,7 @@ static void synchronize_group_exit(struct task_struct *tsk, long code)
 	struct signal_struct *signal = tsk->signal;
 
 	spin_lock_irq(&sighand->siglock);
+	tsk->jobctl |= JOBCTL_WILL_EXIT;
 	signal->quick_threads--;
 	if ((signal->quick_threads == 0) &&
 	    !(signal->flags & SIGNAL_GROUP_EXIT)) {
@@ -1012,6 +1013,7 @@ do_group_exit(int exit_code)
 			sig->group_exit_code = exit_code;
 			sig->flags = SIGNAL_GROUP_EXIT;
 			zap_other_threads(current);
+			current->jobctl |= JOBCTL_WILL_EXIT;
 		}
 		spin_unlock_irq(&sighand->siglock);
 	}
diff --git a/kernel/signal.c b/kernel/signal.c
index ea7753b31be7..d169b47775b0 100644
--- a/kernel/signal.c
+++ b/kernel/signal.c
@@ -2253,15 +2253,19 @@ static int ptrace_stop(int exit_code, int why, unsigned long message,
 		spin_lock_irq(&current->sighand->siglock);
 	}
 
-	/*
-	 * After this point ptrace_signal_wake_up or signal_wake_up
-	 * will clear TASK_TRACED if ptrace_unlink happens or a fatal
-	 * signal comes in.  Handle previous ptrace_unlinks and fatal
-	 * signals here to prevent ptrace_stop sleeping in schedule.
-	 */
-	if (!current->ptrace || __fatal_signal_pending(current))
+	/* Do not stop if ptrace_unlink has happened. */
+	if (!current->ptrace)
+		return exit_code;
+
+	/* Do not stop in a killed task except for PTRACE_EVENT_EXIT */
+	if (task_exit_pending(current) &&
+	    ((exit_code >> 8) != PTRACE_EVENT_EXIT))
 		return exit_code;
 
+	/*
+	 * After this point ptrace_unlink or a fatal signal will clear
+	 * TASK_TRACED preventing ptrace_stop from sleeping.
+	 */
 	set_special_state(TASK_TRACED);
 	current->jobctl |= JOBCTL_TRACED;
 
@@ -2746,7 +2750,6 @@ bool get_signal(struct ksignal *ksig)
 		if ((signal->flags & SIGNAL_GROUP_EXIT) ||
 		     signal->group_exec_task) {
 			signr = SIGKILL;
-			current->jobctl &= ~JOBCTL_WILL_EXIT;
 			trace_signal_deliver(SIGKILL, SEND_SIG_NOINFO,
 					     &sighand->action[SIGKILL-1]);
 			recalc_sigpending();
@@ -2889,6 +2892,7 @@ bool get_signal(struct ksignal *ksig)
 			signal->group_exit_code = exit_code;
 			signal->flags = SIGNAL_GROUP_EXIT;
 			zap_other_threads(current);
+			current->jobctl |= JOBCTL_WILL_EXIT;
 		}
 	fatal:
 		spin_unlock_irq(&sighand->siglock);
-- 
2.41.0