Merge tag 'printk-for-5.18' of git://git.kernel.org/pub/scm/linux/kernel/git/printk/linux

Pull printk updates from Petr Mladek:

- Make %pK behave the same as %p for kptr_restrict == 0 also with
no_hash_pointers parameter

- Ignore the default console in the device tree also when console=null
or console="" is used on the command line

- Document console=null and console="" behavior

- Prevent a deadlock and a livelock caused by console_lock in panic()

- Make console_lock available for panicking CPU

- Fast query for the next to-be-used sequence number

- Use the expected return values in printk.devkmsg __setup handler

- Use the correct atomic operations in wake_up_klogd() irq_work handler

- Avoid possible unaligned access when handling %4cc printing format

* tag 'printk-for-5.18' of git://git.kernel.org/pub/scm/linux/kernel/git/printk/linux:
printk: fix return value of printk.devkmsg __setup handler
vsprintf: Fix %pK with kptr_restrict == 0
printk: make suppress_panic_printk static
printk: Set console_set_on_cmdline=1 when __add_preferred_console() is called with user_specified == true
Docs: printk: add 'console=null|""' to admin/kernel-parameters
printk: use atomic updates for klogd work
printk: Drop console_sem during panic
printk: Avoid livelock with heavy printk during panic
printk: disable optimistic spin during panic
printk: Add panic_in_progress helper
vsprintf: Move space out of string literals in fourcc_string()
vsprintf: Fix potential unaligned access
printk: ringbuffer: Improve prb_next_seq() performance

Documentation/admin-guide/kernel-parameters.txt

···
 		hvc<n>	Use the hypervisor console device <n>. This is for
 			both Xen and PowerPC hypervisors.
 
+		{ null | "" }
+			Use to disable console output, i.e., to have kernel
+			console messages discarded.
+			This must be the only console= parameter used on the
+			kernel command line.
+
 		If the device connected to the port is not a TTY but a braille
 		device, prepend "brl," before the device type, for instance
 			console=brl,ttyS0
···
 			difficult since unequal pointers can no longer be
 			compared.  However, if this command-line option is
 			specified, then all normal pointers will have their true
-			value printed.  Pointers printed via %pK may still be
-			hashed.  This option should only be specified when
+			value printed. This option should only be specified when
 			debugging the kernel.  Please do not use on production
 			kernels.
 

kernel/printk/printk.c

···
  */
 int __read_mostly suppress_printk;
 
+/*
+ * During panic, heavy printk by other CPUs can delay the
+ * panic and risk deadlock on console resources.
+ */
+static int __read_mostly suppress_panic_printk;
+
 #ifdef CONFIG_LOCKDEP
 static struct lockdep_map console_lock_dep_map = {
 	.name = "console_lock"
···
 
 static int __init control_devkmsg(char *str)
 {
-	if (__control_devkmsg(str) < 0)
+	if (__control_devkmsg(str) < 0) {
+		pr_warn("printk.devkmsg: bad option string '%s'\n", str);
 		return 1;
+	}
 
 	/*
 	 * Set sysctl string accordingly:
···
 	 */
 	devkmsg_log |= DEVKMSG_LOG_MASK_LOCK;
 
-	return 0;
+	return 1;
 }
 __setup("printk.devkmsg=", control_devkmsg);
 
···
 	printk_safe_exit_irqrestore(flags);
 }
 #define up_console_sem() __up_console_sem(_RET_IP_)
+
+static bool panic_in_progress(void)
+{
+	return unlikely(atomic_read(&panic_cpu) != PANIC_CPU_INVALID);
+}
 
 /*
  * This is used for debugging the mess that is the VT code by
···
 	if (console_trylock())
 		return 1;
 
+	/*
+	 * It's unsafe to spin once a panic has begun. If we are the
+	 * panic CPU, we may have already halted the owner of the
+	 * console_sem. If we are not the panic CPU, then we should
+	 * avoid taking console_sem, so the panic CPU has a better
+	 * chance of cleanly acquiring it later.
+	 */
+	if (panic_in_progress())
+		return 0;
+
 	printk_safe_enter_irqsave(flags);
 
 	raw_spin_lock(&console_owner_lock);
···
 	if (unlikely(suppress_printk))
 		return 0;
 
+	if (unlikely(suppress_panic_printk) &&
+	    atomic_read(&panic_cpu) != raw_smp_processor_id())
+		return 0;
+
 	if (level == LOGLEVEL_SCHED) {
 		level = LOGLEVEL_DEFAULT;
 		in_sched = true;
···
 }
 #endif
 
+static void set_user_specified(struct console_cmdline *c, bool user_specified)
+{
+	if (!user_specified)
+		return;
+
+	/*
+	 * @c console was defined by the user on the command line.
+	 * Do not clear when added twice also by SPCR or the device tree.
+	 */
+	c->user_specified = true;
+	/* At least one console defined by the user on the command line. */
+	console_set_on_cmdline = 1;
+}
+
 static int __add_preferred_console(char *name, int idx, char *options,
 				   char *brl_options, bool user_specified)
 {
···
 		if (strcmp(c->name, name) == 0 && c->index == idx) {
 			if (!brl_options)
 				preferred_console = i;
-			if (user_specified)
-				c->user_specified = true;
+			set_user_specified(c, user_specified);
 			return 0;
 		}
 	}
···
 		preferred_console = i;
 	strlcpy(c->name, name, sizeof(c->name));
 	c->options = options;
-	c->user_specified = user_specified;
+	set_user_specified(c, user_specified);
 	braille_set_options(c, brl_options);
 
 	c->index = idx;
···
 	*s = 0;
 
 	__add_preferred_console(buf, idx, options, brl_options, true);
-	console_set_on_cmdline = 1;
 	return 1;
 }
 __setup("console=", console_setup);
···
 }
 
 /*
+ * Return true when this CPU should unlock console_sem without pushing all
+ * messages to the console. This reduces the chance that the console is
+ * locked when the panic CPU tries to use it.
+ */
+static bool abandon_console_lock_in_panic(void)
+{
+	if (!panic_in_progress())
+		return false;
+
+	/*
+	 * We can use raw_smp_processor_id() here because it is impossible for
+	 * the task to be migrated to the panic_cpu, or away from it. If
+	 * panic_cpu has already been set, and we're not currently executing on
+	 * that CPU, then we never will be.
+	 */
+	return atomic_read(&panic_cpu) != raw_smp_processor_id();
+}
+
+/*
  * Can we actually use the console at this time on this cpu?
  *
  * Console drivers may assume that per-cpu resources have been allocated. So
···
 {
 	static char ext_text[CONSOLE_EXT_LOG_MAX];
 	static char text[CONSOLE_LOG_MAX];
+	static int panic_console_dropped;
 	unsigned long flags;
 	bool do_cond_resched, retry;
 	struct printk_info info;
···
 		if (console_seq != r.info->seq) {
 			console_dropped += r.info->seq - console_seq;
 			console_seq = r.info->seq;
+			if (panic_in_progress() && panic_console_dropped++ > 10) {
+				suppress_panic_printk = 1;
+				pr_warn_once("Too many dropped messages. Suppress messages on non-panic CPUs to prevent livelock.\n");
+			}
 		}
 
 		if (suppress_message_printing(r.info->level)) {
···
 		if (handover)
 			return;
 
+		/* Allow panic_cpu to take over the consoles safely */
+		if (abandon_console_lock_in_panic())
+			break;
+
 		if (do_cond_resched)
 			cond_resched();
 	}
···
 	 * flush, no worries.
 	 */
 	retry = prb_read_valid(prb, next_seq, NULL);
-	if (retry && console_trylock())
+	if (retry && !abandon_console_lock_in_panic() && console_trylock())
 		goto again;
 }
 EXPORT_SYMBOL(console_unlock);
···
 
 static void wake_up_klogd_work_func(struct irq_work *irq_work)
 {
-	int pending = __this_cpu_xchg(printk_pending, 0);
+	int pending = this_cpu_xchg(printk_pending, 0);
 
 	if (pending & PRINTK_PENDING_OUTPUT) {
 		/* If trylock fails, someone else is doing the printing */
···
 		return;
 
 	preempt_disable();
-	__this_cpu_or(printk_pending, PRINTK_PENDING_OUTPUT);
+	this_cpu_or(printk_pending, PRINTK_PENDING_OUTPUT);
 	irq_work_queue(this_cpu_ptr(&wake_up_klogd_work));
 	preempt_enable();
 }

kernel/printk/printk_ringbuffer.c

···
 	 * state has been re-checked. A memcpy() for all of @desc
 	 * cannot be used because of the atomic_t @state_var field.
 	 */
-	memcpy(&desc_out->text_blk_lpos, &desc->text_blk_lpos,
-	       sizeof(desc_out->text_blk_lpos)); /* LMM(desc_read:C) */
+	if (desc_out) {
+		memcpy(&desc_out->text_blk_lpos, &desc->text_blk_lpos,
+		       sizeof(desc_out->text_blk_lpos)); /* LMM(desc_read:C) */
+	}
 	if (seq_out)
 		*seq_out = info->seq; /* also part of desc_read:C */
 	if (caller_id_out)
···
 	state_val = atomic_long_read(state_var); /* LMM(desc_read:E) */
 	d_state = get_desc_state(id, state_val);
 out:
-	atomic_long_set(&desc_out->state_var, state_val);
+	if (desc_out)
+		atomic_long_set(&desc_out->state_var, state_val);
 	return d_state;
 }
 
···
 
 	atomic_long_cmpxchg_relaxed(&d->state_var, prev_state_val,
 			DESC_SV(id, desc_finalized)); /* LMM(desc_make_final:A) */
+
+	/* Best effort to remember the last finalized @id. */
+	atomic_long_set(&desc_ring->last_finalized_id, id);
 }
 
 /**
···
  */
 void prb_final_commit(struct prb_reserved_entry *e)
 {
+	struct prb_desc_ring *desc_ring = &e->rb->desc_ring;
+
 	_prb_commit(e, desc_finalized);
+
+	/* Best effort to remember the last finalized @id. */
+	atomic_long_set(&desc_ring->last_finalized_id, e->id);
 }
 
 /*
···
  */
 u64 prb_next_seq(struct printk_ringbuffer *rb)
 {
-	u64 seq = 0;
+	struct prb_desc_ring *desc_ring = &rb->desc_ring;
+	enum desc_state d_state;
+	unsigned long id;
+	u64 seq;
 
-	/* Search forward from the oldest descriptor. */
+	/* Check if the cached @id still points to a valid @seq. */
+	id = atomic_long_read(&desc_ring->last_finalized_id);
+	d_state = desc_read(desc_ring, id, NULL, &seq, NULL);
+
+	if (d_state == desc_finalized || d_state == desc_reusable) {
+		/*
+		 * Begin searching after the last finalized record.
+		 *
+		 * On 0, the search must begin at 0 because of hack#2
+		 * of the bootstrapping phase it is not known if a
+		 * record at index 0 exists.
+		 */
+		if (seq != 0)
+			seq++;
+	} else {
+		/*
+		 * The information about the last finalized sequence number
+		 * has gone. It should happen only when there is a flood of
+		 * new messages and the ringbuffer is rapidly recycled.
+		 * Give up and start from the beginning.
+		 */
+		seq = 0;
+	}
+
+	/*
+	 * The information about the last finalized @seq might be inaccurate.
+	 * Search forward to find the current one.
+	 */
 	while (_prb_read_valid(rb, &seq, NULL, NULL))
 		seq++;
 
···
 	rb->desc_ring.infos = infos;
 	atomic_long_set(&rb->desc_ring.head_id, DESC0_ID(descbits));
 	atomic_long_set(&rb->desc_ring.tail_id, DESC0_ID(descbits));
+	atomic_long_set(&rb->desc_ring.last_finalized_id, DESC0_ID(descbits));
 
 	rb->text_data_ring.size_bits = textbits;
 	rb->text_data_ring.data = text_buf;

kernel/printk/printk_ringbuffer.h

···
 	struct printk_info	*infos;
 	atomic_long_t		head_id;
 	atomic_long_t		tail_id;
+	atomic_long_t		last_finalized_id;
 };
 
 /*
···
 		.infos		= &_##name##_infos[0],						\
 		.head_id	= ATOMIC_INIT(DESC0_ID(descbits)),				\
 		.tail_id	= ATOMIC_INIT(DESC0_ID(descbits)),				\
+		.last_finalized_id = ATOMIC_INIT(DESC0_ID(descbits)),				\
 	},											\
 	.text_data_ring = {									\
 		.size_bits	= (avgtextbits) + (descbits),					\

lib/vsprintf.c

···
 
 #include <asm/page.h>		/* for PAGE_SIZE */
 #include <asm/byteorder.h>	/* cpu_to_le16 */
+#include <asm/unaligned.h>
 
 #include <linux/string_helpers.h>
 #include "kstrtox.h"
+
+/* Disable pointer hashing if requested */
+bool no_hash_pointers __ro_after_init;
+EXPORT_SYMBOL_GPL(no_hash_pointers);
 
 static noinline unsigned long long simple_strntoull(const char *startp, size_t max_chars, char **endp, unsigned int base)
 {
···
 	return pointer_string(buf, end, (const void *)hashval, spec);
 }
 
+static char *default_pointer(char *buf, char *end, const void *ptr,
+			     struct printf_spec spec)
+{
+	/*
+	 * default is to _not_ leak addresses, so hash before printing,
+	 * unless no_hash_pointers is specified on the command line.
+	 */
+	if (unlikely(no_hash_pointers))
+		return pointer_string(buf, end, ptr, spec);
+
+	return ptr_to_id(buf, end, ptr, spec);
+}
+
 int kptr_restrict __read_mostly;
 
 static noinline_for_stack
···
 	switch (kptr_restrict) {
 	case 0:
 		/* Handle as %p, hash and do _not_ leak addresses. */
-		return ptr_to_id(buf, end, ptr, spec);
+		return default_pointer(buf, end, ptr, spec);
 	case 1: {
 		const struct cred *cred;
 
···
 	char output[sizeof("0123 little-endian (0x01234567)")];
 	char *p = output;
 	unsigned int i;
-	u32 val;
+	u32 orig, val;
 
 	if (fmt[1] != 'c' || fmt[2] != 'c')
 		return error_string(buf, end, "(%p4?)", spec);
···
 	if (check_pointer(&buf, end, fourcc, spec))
 		return buf;
 
-	val = *fourcc & ~BIT(31);
+	orig = get_unaligned(fourcc);
+	val = orig & ~BIT(31);
 
-	for (i = 0; i < sizeof(*fourcc); i++) {
+	for (i = 0; i < sizeof(u32); i++) {
 		unsigned char c = val >> (i * 8);
 
 		/* Print non-control ASCII characters as-is, dot otherwise */
 		*p++ = isascii(c) && isprint(c) ? c : '.';
 	}
 
-	strcpy(p, *fourcc & BIT(31) ? " big-endian" : " little-endian");
+	*p++ = ' ';
+	strcpy(p, orig & BIT(31) ? "big-endian" : "little-endian");
 	p += strlen(p);
 
 	*p++ = ' ';
 	*p++ = '(';
-	p = special_hex_number(p, output + sizeof(output) - 2, *fourcc, sizeof(u32));
+	p = special_hex_number(p, output + sizeof(output) - 2, orig, sizeof(u32));
 	*p++ = ')';
 	*p = '\0';
 
···
 	return widen_string(buf, buf - buf_start, end, spec);
 }
 
-/* Disable pointer hashing if requested */
-bool no_hash_pointers __ro_after_init;
-EXPORT_SYMBOL_GPL(no_hash_pointers);
-
 int __init no_hash_pointers_enable(char *str)
 {
 	if (no_hash_pointers)
···
 	case 'e':
 		/* %pe with a non-ERR_PTR gets treated as plain %p */
 		if (!IS_ERR(ptr))
-			break;
+			return default_pointer(buf, end, ptr, spec);
 		return err_ptr(buf, end, ptr, spec);
 	case 'u':
 	case 'k':
···
 		default:
 			return error_string(buf, end, "(einval)", spec);
 		}
+	default:
+		return default_pointer(buf, end, ptr, spec);
 	}
-
-	/*
-	 * default is to _not_ leak addresses, so hash before printing,
-	 * unless no_hash_pointers is specified on the command line.
-	 */
-	if (unlikely(no_hash_pointers))
-		return pointer_string(buf, end, ptr, spec);
-	else
-		return ptr_to_id(buf, end, ptr, spec);
 }
 
 /*