summaryrefslogtreecommitdiff
path: root/kernel/printk/printk.c
diff options
context:
space:
mode:
authorLinus Torvalds <torvalds@linux-foundation.org>2024-03-12 20:54:50 -0700
committerLinus Torvalds <torvalds@linux-foundation.org>2024-03-12 20:54:50 -0700
commitb0546776ad3f332e215cebc0b063ba4351971cca (patch)
treefb1d420fc3cc7073162ff1dccfc04884dd31a9f2 /kernel/printk/printk.c
parentf88c3fb81c4badb46c2fef7d168ff138043e86bb (diff)
parent7412dc6d55eed6b76180e40ac3601412ebde29bd (diff)
Merge tag 'printk-for-6.9' of git://git.kernel.org/pub/scm/linux/kernel/git/printk/linux
Pull printk updates from Petr Mladek: "Improve the behavior during panic. The issues were found when testing the ongoing changes introducing atomic consoles and printk kthreads: - pr_flush() has to wait for the last reserved record instead of the last finalized one. Note that records are finalized in random order when generated by more CPUs in parallel. - Ignore non-finalized records during panic(). Messages printed on panic-CPU are always finalized. Messages printed by other CPUs might never be finalized when the CPUs get stopped. - Block new printk() calls on non-panic CPUs completely. Backtraces are printed before entering the panic mode. Later messages would just mess information printed by the panic CPU. - Do not take console_lock in console_flush_on_panic() at all. The original code did try_lock()/console_unlock(). The unlock part might cause a deadlock when panic() happened in a scheduler code. - Fix conversion of 64-bit sequence number for 32-bit atomic operations" * tag 'printk-for-6.9' of git://git.kernel.org/pub/scm/linux/kernel/git/printk/linux: dump_stack: Do not get cpu_sync for panic CPU panic: Flush kernel log buffer at the end printk: Avoid non-panic CPUs writing to ringbuffer printk: Disable passing console lock owner completely during panic() printk: ringbuffer: Skip non-finalized records in panic printk: Wait for all reserved records with pr_flush() printk: ringbuffer: Cleanup reader terminology printk: Add this_cpu_in_panic() printk: For @suppress_panic_printk check for other CPU in panic printk: ringbuffer: Clarify special lpos values printk: ringbuffer: Do not skip non-finalized records with prb_next_seq() printk: Use prb_first_seq() as base for 32bit seq macros printk: Adjust mapping for 32bit seq macros printk: nbcon: Relocate 32bit seq macros
Diffstat (limited to 'kernel/printk/printk.c')
-rw-r--r--kernel/printk/printk.c101
1 files changed, 59 insertions, 42 deletions
diff --git a/kernel/printk/printk.c b/kernel/printk/printk.c
index 1c6e7dfc4ba7..b06f63e276c1 100644
--- a/kernel/printk/printk.c
+++ b/kernel/printk/printk.c
@@ -347,6 +347,29 @@ static bool panic_in_progress(void)
return unlikely(atomic_read(&panic_cpu) != PANIC_CPU_INVALID);
}
+/* Return true if a panic is in progress on the current CPU. */
+bool this_cpu_in_panic(void)
+{
+ /*
+ * 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 unlikely(atomic_read(&panic_cpu) == raw_smp_processor_id());
+}
+
+/*
+ * Return true if a panic is in progress on a remote CPU.
+ *
+ * On true, the local CPU should immediately release any printing resources
+ * that may be needed by the panic CPU.
+ */
+bool other_cpu_in_panic(void)
+{
+ return (panic_in_progress() && !this_cpu_in_panic());
+}
+
/*
* This is used for debugging the mess that is the VT code by
* keeping track if we have the console semaphore held. It's
@@ -439,12 +462,6 @@ static int console_msg_format = MSG_FORMAT_DEFAULT;
static DEFINE_MUTEX(syslog_lock);
#ifdef CONFIG_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;
-
DECLARE_WAIT_QUEUE_HEAD(log_wait);
/* All 3 protected by @syslog_lock. */
/* the next printk record to read by syslog(READ) or /proc/kmsg */
@@ -1835,10 +1852,23 @@ static bool console_waiter;
*/
static void console_lock_spinning_enable(void)
{
+ /*
+ * Do not use spinning in panic(). The panic CPU wants to keep the lock.
+ * Non-panic CPUs abandon the flush anyway.
+ *
+ * Just keep the lockdep annotation. The panic-CPU should avoid
+ * taking console_owner_lock because it might cause a deadlock.
+ * This looks like the easiest way how to prevent false lockdep
+ * reports without handling races a lockless way.
+ */
+ if (panic_in_progress())
+ goto lockdep;
+
raw_spin_lock(&console_owner_lock);
console_owner = current;
raw_spin_unlock(&console_owner_lock);
+lockdep:
/* The waiter may spin on us after setting console_owner */
spin_acquire(&console_owner_dep_map, 0, 0, _THIS_IP_);
}
@@ -1863,6 +1893,22 @@ static int console_lock_spinning_disable_and_check(int cookie)
{
int waiter;
+ /*
+ * Ignore spinning waiters during panic() because they might get stopped
+ * or blocked at any time,
+ *
+ * It is safe because nobody is allowed to start spinning during panic
+ * in the first place. If there has been a waiter then non panic CPUs
+ * might stay spinning. They would get stopped anyway. The panic context
+ * will never start spinning and an interrupted spin on panic CPU will
+ * never continue.
+ */
+ if (panic_in_progress()) {
+ /* Keep lockdep happy. */
+ spin_release(&console_owner_dep_map, _THIS_IP_);
+ return 0;
+ }
+
raw_spin_lock(&console_owner_lock);
waiter = READ_ONCE(console_waiter);
console_owner = NULL;
@@ -2259,8 +2305,12 @@ asmlinkage int vprintk_emit(int facility, int level,
if (unlikely(suppress_printk))
return 0;
- if (unlikely(suppress_panic_printk) &&
- atomic_read(&panic_cpu) != raw_smp_processor_id())
+ /*
+ * The messages on the panic CPU are the most important. If
+ * non-panic CPUs are generating any messages, they will be
+ * silently dropped.
+ */
+ if (other_cpu_in_panic())
return 0;
if (level == LOGLEVEL_SCHED) {
@@ -2590,26 +2640,6 @@ static int console_cpu_notify(unsigned int cpu)
return 0;
}
-/*
- * Return true if a panic is in progress on a remote CPU.
- *
- * On true, the local CPU should immediately release any printing resources
- * that may be needed by the panic CPU.
- */
-bool other_cpu_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();
-}
-
/**
* console_lock - block the console subsystem from printing
*
@@ -2765,8 +2795,6 @@ void console_prepend_dropped(struct printk_message *pmsg, unsigned long dropped)
bool printk_get_next_message(struct printk_message *pmsg, u64 seq,
bool is_extended, bool may_suppress)
{
- static int panic_console_dropped;
-
struct printk_buffers *pbufs = pmsg->pbufs;
const size_t scratchbuf_sz = sizeof(pbufs->scratchbuf);
const size_t outbuf_sz = sizeof(pbufs->outbuf);
@@ -2794,17 +2822,6 @@ bool printk_get_next_message(struct printk_message *pmsg, u64 seq,
pmsg->seq = r.info->seq;
pmsg->dropped = r.info->seq - seq;
- /*
- * Check for dropped messages in panic here so that printk
- * suppression can occur as early as possible if necessary.
- */
- if (pmsg->dropped &&
- 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");
- }
-
/* Skip record that has level above the console loglevel. */
if (may_suppress && suppress_message_printing(r.info->level))
goto out;
@@ -3750,7 +3767,7 @@ static bool __pr_flush(struct console *con, int timeout_ms, bool reset_on_progre
might_sleep();
- seq = prb_next_seq(prb);
+ seq = prb_next_reserve_seq(prb);
/* Flush the consoles so that records up to @seq are printed. */
console_lock();