Merge tag 'printk-for-6.17' of git://git.kernel.org/pub/scm/linux/kernel/git/printk/linuxHEADtorvalds/mastertorvalds/HEADmaster

Pull printk updates from Petr Mladek:

 - Add new "hash_pointers=[auto|always|never]" boot parameter to force
   the hashing even with "slab_debug" enabled

 - Allow to stop CPU, after losing nbcon console ownership during
   panic(), even without proper NMI

 - Allow to use the printk kthread immediately even for the 1st
   registered nbcon

 - Compiler warning removal

* tag 'printk-for-6.17' of git://git.kernel.org/pub/scm/linux/kernel/git/printk/linux:
  printk: nbcon: Allow reacquire during panic
  printk: Allow to use the printk kthread immediately even for 1st nbcon
  slab: Decouple slab_debug and no_hash_pointers
  vsprintf: Use __diag macros to disable '-Wsuggest-attribute=format'
  compiler-gcc.h: Introduce __diag_GCC_all

3 files changed, 78 insertions, 33 deletions

diff --git a/kernel/printk/internal.h b/kernel/printk/internal.h
index bbed41ad29cf..ef282001f200 100644
--- a/kernel/printk/internal.h
+++ b/kernel/printk/internal.h
@@ -64,6 +64,7 @@ struct dev_printk_info;
 
 extern struct printk_ringbuffer *prb;
 extern bool printk_kthreads_running;
+extern bool printk_kthreads_ready;
 extern bool debug_non_panic_cpus;
 
 __printf(4, 0)
@@ -179,6 +180,7 @@ static inline void nbcon_kthread_wake(struct console *con)
 #define PRINTKRB_RECORD_MAX	0
 
 #define printk_kthreads_running (false)
+#define printk_kthreads_ready (false)
 
 /*
  * In !PRINTK builds we still export console_sem
diff --git a/kernel/printk/nbcon.c b/kernel/printk/nbcon.c
index fd12efcc4aed..646801813415 100644
--- a/kernel/printk/nbcon.c
+++ b/kernel/printk/nbcon.c
@@ -214,8 +214,9 @@ static void nbcon_seq_try_update(struct nbcon_context *ctxt, u64 new_seq)
 
 /**
  * nbcon_context_try_acquire_direct - Try to acquire directly
- * @ctxt:	The context of the caller
- * @cur:	The current console state
+ * @ctxt:		The context of the caller
+ * @cur:		The current console state
+ * @is_reacquire:	This acquire is a reacquire
  *
  * Acquire the console when it is released. Also acquire the console when
  * the current owner has a lower priority and the console is in a safe state.
@@ -225,17 +226,17 @@ static void nbcon_seq_try_update(struct nbcon_context *ctxt, u64 new_seq)
  *
  * Errors:
  *
- *	-EPERM:		A panic is in progress and this is not the panic CPU.
- *			Or the current owner or waiter has the same or higher
- *			priority. No acquire method can be successful in
- *			this case.
+ *	-EPERM:		A panic is in progress and this is neither the panic
+ *			CPU nor is this a reacquire. Or the current owner or
+ *			waiter has the same or higher priority. No acquire
+ *			method can be successful in these cases.
  *
  *	-EBUSY:		The current owner has a lower priority but the console
  *			in an unsafe state. The caller should try using
  *			the handover acquire method.
  */
 static int nbcon_context_try_acquire_direct(struct nbcon_context *ctxt,
-					    struct nbcon_state *cur)
+					    struct nbcon_state *cur, bool is_reacquire)
 {
 	unsigned int cpu = smp_processor_id();
 	struct console *con = ctxt->console;
@@ -243,14 +244,20 @@ static int nbcon_context_try_acquire_direct(struct nbcon_context *ctxt,
 
 	do {
 		/*
-		 * Panic does not imply that the console is owned. However, it
-		 * is critical that non-panic CPUs during panic are unable to
-		 * acquire ownership in order to satisfy the assumptions of
-		 * nbcon_waiter_matches(). In particular, the assumption that
-		 * lower priorities are ignored during panic.
+		 * Panic does not imply that the console is owned. However,
+		 * since all non-panic CPUs are stopped during panic(), it
+		 * is safer to have them avoid gaining console ownership.
+		 *
+		 * If this acquire is a reacquire (and an unsafe takeover
+		 * has not previously occurred) then it is allowed to attempt
+		 * a direct acquire in panic. This gives console drivers an
+		 * opportunity to perform any necessary cleanup if they were
+		 * interrupted by the panic CPU while printing.
 		 */
-		if (other_cpu_in_panic())
+		if (other_cpu_in_panic() &&
+		    (!is_reacquire || cur->unsafe_takeover)) {
 			return -EPERM;
+		}
 
 		if (ctxt->prio <= cur->prio || ctxt->prio <= cur->req_prio)
 			return -EPERM;
@@ -301,8 +308,9 @@ static bool nbcon_waiter_matches(struct nbcon_state *cur, int expected_prio)
 	 * Event #1 implies this context is EMERGENCY.
 	 * Event #2 implies the new context is PANIC.
 	 * Event #3 occurs when panic() has flushed the console.
-	 * Events #4 and #5 are not possible due to the other_cpu_in_panic()
-	 * check in nbcon_context_try_acquire_direct().
+	 * Event #4 occurs when a non-panic CPU reacquires.
+	 * Event #5 is not possible due to the other_cpu_in_panic() check
+	 *          in nbcon_context_try_acquire_handover().
 	 */
 
 	return (cur->req_prio == expected_prio);
@@ -431,6 +439,16 @@ static int nbcon_context_try_acquire_handover(struct nbcon_context *ctxt,
 	WARN_ON_ONCE(ctxt->prio <= cur->prio || ctxt->prio <= cur->req_prio);
 	WARN_ON_ONCE(!cur->unsafe);
 
+	/*
+	 * Panic does not imply that the console is owned. However, it
+	 * is critical that non-panic CPUs during panic are unable to
+	 * wait for a handover in order to satisfy the assumptions of
+	 * nbcon_waiter_matches(). In particular, the assumption that
+	 * lower priorities are ignored during panic.
+	 */
+	if (other_cpu_in_panic())
+		return -EPERM;
+
 	/* Handover is not possible on the same CPU. */
 	if (cur->cpu == cpu)
 		return -EBUSY;
@@ -558,7 +576,8 @@ static struct printk_buffers panic_nbcon_pbufs;
 
 /**
  * nbcon_context_try_acquire - Try to acquire nbcon console
- * @ctxt:	The context of the caller
+ * @ctxt:		The context of the caller
+ * @is_reacquire:	This acquire is a reacquire
  *
  * Context:	Under @ctxt->con->device_lock() or local_irq_save().
  * Return:	True if the console was acquired. False otherwise.
@@ -568,7 +587,7 @@ static struct printk_buffers panic_nbcon_pbufs;
  * in an unsafe state. Otherwise, on success the caller may assume
  * the console is not in an unsafe state.
  */
-static bool nbcon_context_try_acquire(struct nbcon_context *ctxt)
+static bool nbcon_context_try_acquire(struct nbcon_context *ctxt, bool is_reacquire)
 {
 	unsigned int cpu = smp_processor_id();
 	struct console *con = ctxt->console;
@@ -577,7 +596,7 @@ static bool nbcon_context_try_acquire(struct nbcon_context *ctxt)
 
 	nbcon_state_read(con, &cur);
 try_again:
-	err = nbcon_context_try_acquire_direct(ctxt, &cur);
+	err = nbcon_context_try_acquire_direct(ctxt, &cur, is_reacquire);
 	if (err != -EBUSY)
 		goto out;
 
@@ -913,7 +932,7 @@ void nbcon_reacquire_nobuf(struct nbcon_write_context *wctxt)
 {
 	struct nbcon_context *ctxt = &ACCESS_PRIVATE(wctxt, ctxt);
 
-	while (!nbcon_context_try_acquire(ctxt))
+	while (!nbcon_context_try_acquire(ctxt, true))
 		cpu_relax();
 
 	nbcon_write_context_set_buf(wctxt, NULL, 0);
@@ -1101,7 +1120,7 @@ static bool nbcon_emit_one(struct nbcon_write_context *wctxt, bool use_atomic)
 		cant_migrate();
 	}
 
-	if (!nbcon_context_try_acquire(ctxt))
+	if (!nbcon_context_try_acquire(ctxt, false))
 		goto out;
 
 	/*
@@ -1486,7 +1505,7 @@ static int __nbcon_atomic_flush_pending_con(struct console *con, u64 stop_seq,
 	ctxt->prio			= nbcon_get_default_prio();
 	ctxt->allow_unsafe_takeover	= allow_unsafe_takeover;
 
-	if (!nbcon_context_try_acquire(ctxt))
+	if (!nbcon_context_try_acquire(ctxt, false))
 		return -EPERM;
 
 	while (nbcon_seq_read(con) < stop_seq) {
@@ -1671,6 +1690,9 @@ bool nbcon_alloc(struct console *con)
 {
 	struct nbcon_state state = { };
 
+	/* Synchronize the kthread start. */
+	lockdep_assert_console_list_lock_held();
+
 	/* The write_thread() callback is mandatory. */
 	if (WARN_ON(!con->write_thread))
 		return false;
@@ -1701,12 +1723,15 @@ bool nbcon_alloc(struct console *con)
 			return false;
 		}
 
-		if (printk_kthreads_running) {
+		if (printk_kthreads_ready && !have_boot_console) {
 			if (!nbcon_kthread_create(con)) {
 				kfree(con->pbufs);
 				con->pbufs = NULL;
 				return false;
 			}
+
+			/* Might be the first kthread. */
+			printk_kthreads_running = true;
 		}
 	}
 
@@ -1716,14 +1741,30 @@ bool nbcon_alloc(struct console *con)
 /**
  * nbcon_free - Free and cleanup the nbcon console specific data
  * @con:	Console to free/cleanup nbcon data
+ *
+ * Important: @have_nbcon_console must be updated before calling
+ *	this function. In particular, it can be set only when there
+ *	is still another nbcon console registered.
  */
 void nbcon_free(struct console *con)
 {
 	struct nbcon_state state = { };
 
-	if (printk_kthreads_running)
+	/* Synchronize the kthread stop. */
+	lockdep_assert_console_list_lock_held();
+
+	if (printk_kthreads_running) {
 		nbcon_kthread_stop(con);
 
+		/* Might be the last nbcon console.
+		 *
+		 * Do not rely on printk_kthreads_check_locked(). It is not
+		 * called in some code paths, see nbcon_free() callers.
+		 */
+		if (!have_nbcon_console)
+			printk_kthreads_running = false;
+	}
+
 	nbcon_state_set(con, &state);
 
 	/* Boot consoles share global printk buffers. */
@@ -1762,7 +1803,7 @@ bool nbcon_device_try_acquire(struct console *con)
 	ctxt->console	= con;
 	ctxt->prio	= NBCON_PRIO_NORMAL;
 
-	if (!nbcon_context_try_acquire(ctxt))
+	if (!nbcon_context_try_acquire(ctxt, false))
 		return false;
 
 	if (!nbcon_context_enter_unsafe(ctxt))
diff --git a/kernel/printk/printk.c b/kernel/printk/printk.c
index 1eea80d0648e..0efbcdda9aab 100644
--- a/kernel/printk/printk.c
+++ b/kernel/printk/printk.c
@@ -3574,7 +3574,7 @@ EXPORT_SYMBOL(console_resume);
 static int unregister_console_locked(struct console *console);
 
 /* True when system boot is far enough to create printer threads. */
-static bool printk_kthreads_ready __ro_after_init;
+bool printk_kthreads_ready __ro_after_init;
 
 static struct task_struct *printk_legacy_kthread;
 
@@ -3713,6 +3713,7 @@ static void printk_kthreads_check_locked(void)
 	if (!printk_kthreads_ready)
 		return;
 
+	/* Start or stop the legacy kthread when needed. */
 	if (have_legacy_console || have_boot_console) {
 		if (!printk_legacy_kthread &&
 		    force_legacy_kthread() &&
@@ -4204,14 +4205,6 @@ static int unregister_console_locked(struct console *console)
 	 */
 	synchronize_srcu(&console_srcu);
 
-	if (console->flags & CON_NBCON)
-		nbcon_free(console);
-
-	console_sysfs_notify();
-
-	if (console->exit)
-		res = console->exit(console);
-
 	/*
 	 * With this console gone, the global flags tracking registered
 	 * console types may have changed. Update them.
@@ -4232,6 +4225,15 @@ static int unregister_console_locked(struct console *console)
 	if (!found_nbcon_con)
 		have_nbcon_console = found_nbcon_con;
 
+	/* @have_nbcon_console must be updated before calling nbcon_free(). */
+	if (console->flags & CON_NBCON)
+		nbcon_free(console);
+
+	console_sysfs_notify();
+
+	if (console->exit)
+		res = console->exit(console);
+
 	/* Changed console list, may require printer threads to start/stop. */
 	printk_kthreads_check_locked();