From 6d3e0d8cc63221dec670d0ee92ac57961581e975 Mon Sep 17 00:00:00 2001 From: John Ogness Date: Mon, 17 Jul 2023 21:52:01 +0206 Subject: kdb: Do not assume write() callback available It is allowed for consoles to not provide a write() callback. For example ttynull does this. Check if a write() callback is available before using it. Signed-off-by: John Ogness Reviewed-by: Petr Mladek Reviewed-by: Douglas Anderson Reviewed-by: Daniel Thompson Acked-by: Daniel Thompson Reviewed-by: Sergey Senozhatsky Signed-off-by: Petr Mladek Link: https://lore.kernel.org/r/20230717194607.145135-2-john.ogness@linutronix.de --- kernel/debug/kdb/kdb_io.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/kernel/debug/kdb/kdb_io.c b/kernel/debug/kdb/kdb_io.c index 5c7e9ba7cd6b..e9139dfc1f0a 100644 --- a/kernel/debug/kdb/kdb_io.c +++ b/kernel/debug/kdb/kdb_io.c @@ -576,6 +576,8 @@ static void kdb_msg_write(const char *msg, int msg_len) continue; if (c == dbg_io_ops->cons) continue; + if (!c->write) + continue; /* * Set oops_in_progress to encourage the console drivers to * disregard their internal spin locks: in the current calling -- cgit v1.2.3 From 7b23a66db55ed0a55b020e913f0d6f6d52a1ad2c Mon Sep 17 00:00:00 2001 From: John Ogness Date: Mon, 17 Jul 2023 21:52:02 +0206 Subject: printk: Reduce console_unblank() usage in unsafe scenarios A semaphore is not NMI-safe, even when using down_trylock(). Both down_trylock() and up() are using internal spinlocks and up() might even call wake_up_process(). In the panic() code path it gets even worse because the internal spinlocks of the semaphore may have been taken by a CPU that has been stopped. To reduce the risk of deadlocks caused by the console semaphore in the panic path, make the following changes: - First check if any consoles have implemented the unblank() callback. If not, then there is no reason to take the console semaphore anyway. (This check is also useful for the non-panic path since the locking/unlocking of the console lock can be quite expensive due to console printing.) - If the panic path is in NMI context, bail out without attempting to take the console semaphore or calling any unblank() callbacks. Bailing out is acceptable because console_unblank() would already bail out if the console semaphore is contended. The alternative of ignoring the console semaphore and calling the unblank() callbacks anyway is a bad idea because these callbacks are also not NMI-safe. If consoles with unblank() callbacks exist and console_unblank() is called from a non-NMI panic context, it will still attempt a down_trylock(). This could still result in a deadlock if one of the stopped CPUs is holding the semaphore internal spinlock. But this is a risk that the kernel has been (and continues to be) willing to take. Signed-off-by: John Ogness Reviewed-by: Sergey Senozhatsky Reviewed-by: Petr Mladek Signed-off-by: Petr Mladek Link: https://lore.kernel.org/r/20230717194607.145135-3-john.ogness@linutronix.de --- kernel/printk/printk.c | 28 ++++++++++++++++++++++++++++ 1 file changed, 28 insertions(+) diff --git a/kernel/printk/printk.c b/kernel/printk/printk.c index 9644f6e5bf15..7aa9dbee12e8 100644 --- a/kernel/printk/printk.c +++ b/kernel/printk/printk.c @@ -3043,9 +3043,27 @@ EXPORT_SYMBOL(console_conditional_schedule); void console_unblank(void) { + bool found_unblank = false; struct console *c; int cookie; + /* + * First check if there are any consoles implementing the unblank() + * callback. If not, there is no reason to continue and take the + * console lock, which in particular can be dangerous if + * @oops_in_progress is set. + */ + cookie = console_srcu_read_lock(); + for_each_console_srcu(c) { + if ((console_srcu_read_flags(c) & CON_ENABLED) && c->unblank) { + found_unblank = true; + break; + } + } + console_srcu_read_unlock(cookie); + if (!found_unblank) + return; + /* * Stop console printing because the unblank() callback may * assume the console is not within its write() callback. @@ -3054,6 +3072,16 @@ void console_unblank(void) * In that case, attempt a trylock as best-effort. */ if (oops_in_progress) { + /* Semaphores are not NMI-safe. */ + if (in_nmi()) + return; + + /* + * Attempting to trylock the console lock can deadlock + * if another CPU was stopped while modifying the + * semaphore. "Hope and pray" that this is not the + * current situation. + */ if (down_trylock_console_sem() != 0) return; } else -- cgit v1.2.3 From 51a1d258e50e03a0216bf42b6af9ff34ec402ac1 Mon Sep 17 00:00:00 2001 From: John Ogness Date: Mon, 17 Jul 2023 21:52:03 +0206 Subject: printk: Keep non-panic-CPUs out of console lock When in a panic situation, non-panic CPUs should avoid holding the console lock so as not to contend with the panic CPU. This is already implemented with abandon_console_lock_in_panic(), which is checked after each printed line. However, non-panic CPUs should also avoid trying to acquire the console lock during a panic. Modify console_trylock() to fail and console_lock() to block() when called from a non-panic CPU during a panic. Signed-off-by: John Ogness Reviewed-by: Sergey Senozhatsky Reviewed-by: Petr Mladek Signed-off-by: Petr Mladek Link: https://lore.kernel.org/r/20230717194607.145135-4-john.ogness@linutronix.de --- kernel/printk/printk.c | 45 ++++++++++++++++++++++++++------------------- 1 file changed, 26 insertions(+), 19 deletions(-) diff --git a/kernel/printk/printk.c b/kernel/printk/printk.c index 7aa9dbee12e8..7219991885e6 100644 --- a/kernel/printk/printk.c +++ b/kernel/printk/printk.c @@ -2583,6 +2583,25 @@ static int console_cpu_notify(unsigned int cpu) return 0; } +/* + * 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(); +} + /** * console_lock - block the console subsystem from printing * @@ -2595,6 +2614,10 @@ void console_lock(void) { might_sleep(); + /* On panic, the console_lock must be left to the panic cpu. */ + while (abandon_console_lock_in_panic()) + msleep(1000); + down_console_sem(); if (console_suspended) return; @@ -2613,6 +2636,9 @@ EXPORT_SYMBOL(console_lock); */ int console_trylock(void) { + /* On panic, the console_lock must be left to the panic cpu. */ + if (abandon_console_lock_in_panic()) + return 0; if (down_trylock_console_sem()) return 0; if (console_suspended) { @@ -2631,25 +2657,6 @@ int is_console_locked(void) } EXPORT_SYMBOL(is_console_locked); -/* - * 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(); -} - /* * Check if the given console is currently capable and allowed to print * records. -- cgit v1.2.3 From eacb04ff3c5b8662a65f380ae450250698448cff Mon Sep 17 00:00:00 2001 From: John Ogness Date: Mon, 17 Jul 2023 21:52:04 +0206 Subject: printk: Do not take console lock for console_flush_on_panic() Currently console_flush_on_panic() will attempt to acquire the console lock when flushing the buffer on panic. If it fails to acquire the lock, it continues anyway because this is the last chance to get any pending records printed. The reason why the console lock was attempted at all was to prevent any other CPUs from acquiring the console lock for printing while the panic CPU was printing. But as of the previous commit, non-panic CPUs will no longer attempt to acquire the console lock in a panic situation. Therefore it is no longer strictly necessary for a panic CPU to acquire the console lock. Avoiding taking the console lock when flushing in panic has the additional benefit of avoiding possible deadlocks due to semaphore usage in NMI context (semaphores are not NMI-safe) and avoiding possible deadlocks if another CPU accesses the semaphore and is stopped while holding one of the semaphore's internal spinlocks. Signed-off-by: John Ogness Reviewed-by: Sergey Senozhatsky Reviewed-by: Petr Mladek Signed-off-by: Petr Mladek Link: https://lore.kernel.org/r/20230717194607.145135-5-john.ogness@linutronix.de --- kernel/printk/printk.c | 28 +++++++++++++++++++--------- 1 file changed, 19 insertions(+), 9 deletions(-) diff --git a/kernel/printk/printk.c b/kernel/printk/printk.c index 7219991885e6..51445e8ea730 100644 --- a/kernel/printk/printk.c +++ b/kernel/printk/printk.c @@ -3118,14 +3118,24 @@ void console_unblank(void) */ void console_flush_on_panic(enum con_flush_mode mode) { + bool handover; + u64 next_seq; + /* - * If someone else is holding the console lock, trylock will fail - * and may_schedule may be set. Ignore and proceed to unlock so - * that messages are flushed out. As this can be called from any - * context and we don't want to get preempted while flushing, - * ensure may_schedule is cleared. + * Ignore the console lock and flush out the messages. Attempting a + * trylock would not be useful because: + * + * - if it is contended, it must be ignored anyway + * - console_lock() and console_trylock() block and fail + * respectively in panic for non-panic CPUs + * - semaphores are not NMI-safe + */ + + /* + * If another context is holding the console lock, + * @console_may_schedule might be set. Clear it so that + * this context does not call cond_resched() while flushing. */ - console_trylock(); console_may_schedule = 0; if (mode == CONSOLE_REPLAY_ALL) { @@ -3138,15 +3148,15 @@ void console_flush_on_panic(enum con_flush_mode mode) cookie = console_srcu_read_lock(); for_each_console_srcu(c) { /* - * If the above console_trylock() failed, this is an - * unsynchronized assignment. But in that case, the + * This is an unsynchronized assignment, but the * kernel is in "hope and pray" mode anyway. */ c->seq = seq; } console_srcu_read_unlock(cookie); } - console_unlock(); + + console_flush_all(false, &next_seq, &handover); } /* -- cgit v1.2.3 From 696ffaf50e1f8dbc66223ff614473f945f5fb8d8 Mon Sep 17 00:00:00 2001 From: John Ogness Date: Mon, 17 Jul 2023 21:52:05 +0206 Subject: printk: Consolidate console deferred printing Printing to consoles can be deferred for several reasons: - explicitly with printk_deferred() - printk() in NMI context - recursive printk() calls The current implementation is not consistent. For printk_deferred(), irq work is scheduled twice. For NMI und recursive, panic CPU suppression and caller delays are not properly enforced. Correct these inconsistencies by consolidating the deferred printing code so that vprintk_deferred() is the top-level function for deferred printing and vprintk_emit() will perform whichever irq_work queueing is appropriate. Also add kerneldoc for wake_up_klogd() and defer_console_output() to clarify their differences and appropriate usage. Signed-off-by: John Ogness Reviewed-by: Sergey Senozhatsky Reviewed-by: Petr Mladek Signed-off-by: Petr Mladek Link: https://lore.kernel.org/r/20230717194607.145135-6-john.ogness@linutronix.de --- kernel/printk/printk.c | 35 ++++++++++++++++++++++++++++------- kernel/printk/printk_safe.c | 9 ++------- 2 files changed, 30 insertions(+), 14 deletions(-) diff --git a/kernel/printk/printk.c b/kernel/printk/printk.c index 51445e8ea730..6e853a1441a7 100644 --- a/kernel/printk/printk.c +++ b/kernel/printk/printk.c @@ -2306,7 +2306,11 @@ asmlinkage int vprintk_emit(int facility, int level, preempt_enable(); } - wake_up_klogd(); + if (in_sched) + defer_console_output(); + else + wake_up_klogd(); + return printed_len; } EXPORT_SYMBOL(vprintk_emit); @@ -3841,11 +3845,33 @@ static void __wake_up_klogd(int val) preempt_enable(); } +/** + * wake_up_klogd - Wake kernel logging daemon + * + * Use this function when new records have been added to the ringbuffer + * and the console printing of those records has already occurred or is + * known to be handled by some other context. This function will only + * wake the logging daemon. + * + * Context: Any context. + */ void wake_up_klogd(void) { __wake_up_klogd(PRINTK_PENDING_WAKEUP); } +/** + * defer_console_output - Wake kernel logging daemon and trigger + * console printing in a deferred context + * + * Use this function when new records have been added to the ringbuffer, + * this context is responsible for console printing those records, but + * the current context is not allowed to perform the console printing. + * Trigger an irq_work context to perform the console printing. This + * function also wakes the logging daemon. + * + * Context: Any context. + */ void defer_console_output(void) { /* @@ -3862,12 +3888,7 @@ void printk_trigger_flush(void) int vprintk_deferred(const char *fmt, va_list args) { - int r; - - r = vprintk_emit(0, LOGLEVEL_SCHED, NULL, fmt, args); - defer_console_output(); - - return r; + return vprintk_emit(0, LOGLEVEL_SCHED, NULL, fmt, args); } int _printk_deferred(const char *fmt, ...) diff --git a/kernel/printk/printk_safe.c b/kernel/printk/printk_safe.c index ef0f9a2044da..6d10927a07d8 100644 --- a/kernel/printk/printk_safe.c +++ b/kernel/printk/printk_safe.c @@ -38,13 +38,8 @@ asmlinkage int vprintk(const char *fmt, va_list args) * Use the main logbuf even in NMI. But avoid calling console * drivers that might have their own locks. */ - if (this_cpu_read(printk_context) || in_nmi()) { - int len; - - len = vprintk_store(0, LOGLEVEL_DEFAULT, NULL, fmt, args); - defer_console_output(); - return len; - } + if (this_cpu_read(printk_context) || in_nmi()) + return vprintk_deferred(fmt, args); /* No obstacles. */ return vprintk_default(fmt, args); -- cgit v1.2.3 From 9e70a5e109a4a23367810de09be826c52d27ee2f Mon Sep 17 00:00:00 2001 From: John Ogness Date: Mon, 17 Jul 2023 21:52:06 +0206 Subject: printk: Add per-console suspended state Currently the global @console_suspended is used to determine if consoles are in a suspended state. Its primary purpose is to allow usage of the console_lock when suspended without causing console printing. It is synchronized by the console_lock. Rather than relying on the console_lock to determine suspended state, make it an official per-console state that is set within console->flags. This allows the state to be queried via SRCU. Remove @console_suspended. Console printing will still be avoided when suspended because console_is_usable() returns false when the new suspended flag is set for that console. Signed-off-by: John Ogness Reviewed-by: Sergey Senozhatsky Reviewed-by: Petr Mladek Signed-off-by: Petr Mladek Link: https://lore.kernel.org/r/20230717194607.145135-7-john.ogness@linutronix.de --- include/linux/console.h | 3 ++ kernel/printk/printk.c | 74 +++++++++++++++++++++++++++++-------------------- 2 files changed, 47 insertions(+), 30 deletions(-) diff --git a/include/linux/console.h b/include/linux/console.h index d3195664baa5..7de11c763eb3 100644 --- a/include/linux/console.h +++ b/include/linux/console.h @@ -154,6 +154,8 @@ static inline int con_debug_leave(void) * receiving the printk spam for obvious reasons. * @CON_EXTENDED: The console supports the extended output format of * /dev/kmesg which requires a larger output buffer. + * @CON_SUSPENDED: Indicates if a console is suspended. If true, the + * printing callbacks must not be called. */ enum cons_flags { CON_PRINTBUFFER = BIT(0), @@ -163,6 +165,7 @@ enum cons_flags { CON_ANYTIME = BIT(4), CON_BRL = BIT(5), CON_EXTENDED = BIT(6), + CON_SUSPENDED = BIT(7), }; /** diff --git a/kernel/printk/printk.c b/kernel/printk/printk.c index 6e853a1441a7..efe577477913 100644 --- a/kernel/printk/printk.c +++ b/kernel/printk/printk.c @@ -86,7 +86,7 @@ EXPORT_SYMBOL(oops_in_progress); static DEFINE_MUTEX(console_mutex); /* - * console_sem protects updates to console->seq and console_suspended, + * console_sem protects updates to console->seq * and also provides serialization for console printing. */ static DEFINE_SEMAPHORE(console_sem); @@ -359,7 +359,7 @@ static bool panic_in_progress(void) * paths in the console code where we end up in places I want * locked without the console semaphore held). */ -static int console_locked, console_suspended; +static int console_locked; /* * Array of consoles built from command line options (console=) @@ -2549,22 +2549,46 @@ MODULE_PARM_DESC(console_no_auto_verbose, "Disable console loglevel raise to hig */ void suspend_console(void) { + struct console *con; + if (!console_suspend_enabled) return; pr_info("Suspending console(s) (use no_console_suspend to debug)\n"); pr_flush(1000, true); - console_lock(); - console_suspended = 1; - up_console_sem(); + + console_list_lock(); + for_each_console(con) + console_srcu_write_flags(con, con->flags | CON_SUSPENDED); + console_list_unlock(); + + /* + * Ensure that all SRCU list walks have completed. All printing + * contexts must be able to see that they are suspended so that it + * is guaranteed that all printing has stopped when this function + * completes. + */ + synchronize_srcu(&console_srcu); } void resume_console(void) { + struct console *con; + if (!console_suspend_enabled) return; - down_console_sem(); - console_suspended = 0; - console_unlock(); + + console_list_lock(); + for_each_console(con) + console_srcu_write_flags(con, con->flags & ~CON_SUSPENDED); + console_list_unlock(); + + /* + * Ensure that all SRCU list walks have completed. All printing + * contexts must be able to see they are no longer suspended so + * that they are guaranteed to wake up and resume printing. + */ + synchronize_srcu(&console_srcu); + pr_flush(1000, true); } @@ -2623,8 +2647,6 @@ void console_lock(void) msleep(1000); down_console_sem(); - if (console_suspended) - return; console_locked = 1; console_may_schedule = 1; } @@ -2645,10 +2667,6 @@ int console_trylock(void) return 0; if (down_trylock_console_sem()) return 0; - if (console_suspended) { - up_console_sem(); - return 0; - } console_locked = 1; console_may_schedule = 0; return 1; @@ -2674,6 +2692,9 @@ static inline bool console_is_usable(struct console *con) if (!(flags & CON_ENABLED)) return false; + if ((flags & CON_SUSPENDED)) + return false; + if (!con->write) return false; @@ -2992,11 +3013,6 @@ void console_unlock(void) bool flushed; u64 next_seq; - if (console_suspended) { - up_console_sem(); - return; - } - /* * Console drivers are called with interrupts disabled, so * @console_may_schedule should be cleared before; however, we may @@ -3726,8 +3742,7 @@ static bool __pr_flush(struct console *con, int timeout_ms, bool reset_on_progre /* * Hold the console_lock to guarantee safe access to - * console->seq and to prevent changes to @console_suspended - * until all consoles have been processed. + * console->seq. */ console_lock(); @@ -3735,6 +3750,11 @@ static bool __pr_flush(struct console *con, int timeout_ms, bool reset_on_progre for_each_console_srcu(c) { if (con && con != c) continue; + /* + * If consoles are not usable, it cannot be expected + * that they make forward progress, so only increment + * @diff for usable consoles. + */ if (!console_is_usable(c)) continue; printk_seq = c->seq; @@ -3743,18 +3763,12 @@ static bool __pr_flush(struct console *con, int timeout_ms, bool reset_on_progre } console_srcu_read_unlock(cookie); - /* - * If consoles are suspended, it cannot be expected that they - * make forward progress, so timeout immediately. @diff is - * still used to return a valid flush status. - */ - if (console_suspended) - remaining = 0; - else if (diff != last_diff && reset_on_progress) + if (diff != last_diff && reset_on_progress) remaining = timeout_ms; console_unlock(); + /* Note: @diff is 0 if there are no usable consoles. */ if (diff == 0 || remaining == 0) break; @@ -3788,7 +3802,7 @@ static bool __pr_flush(struct console *con, int timeout_ms, bool reset_on_progre * printer has been seen to make some forward progress. * * Context: Process context. May sleep while acquiring console lock. - * Return: true if all enabled printers are caught up. + * Return: true if all usable printers are caught up. */ static bool pr_flush(int timeout_ms, bool reset_on_progress) { -- cgit v1.2.3 From 132a90d1527fedba2d95085c951ccf00dbbebe41 Mon Sep 17 00:00:00 2001 From: John Ogness Date: Mon, 17 Jul 2023 21:52:07 +0206 Subject: printk: Rename abandon_console_lock_in_panic() to other_cpu_in_panic() Currently abandon_console_lock_in_panic() is only used to determine if the current CPU should immediately release the console lock because another CPU is in panic. However, later this function will be used by the CPU to immediately release other resources in this situation. Rename the function to other_cpu_in_panic(), which is a better description and does not assume it is related to the console lock. Signed-off-by: John Ogness Reviewed-by: Sergey Senozhatsky Reviewed-by: Petr Mladek Signed-off-by: Petr Mladek Link: https://lore.kernel.org/r/20230717194607.145135-8-john.ogness@linutronix.de --- kernel/printk/internal.h | 2 ++ kernel/printk/printk.c | 15 ++++++++------- 2 files changed, 10 insertions(+), 7 deletions(-) diff --git a/kernel/printk/internal.h b/kernel/printk/internal.h index 2a17704136f1..7d4979d5c3ce 100644 --- a/kernel/printk/internal.h +++ b/kernel/printk/internal.h @@ -103,3 +103,5 @@ struct printk_message { u64 seq; unsigned long dropped; }; + +bool other_cpu_in_panic(void); diff --git a/kernel/printk/printk.c b/kernel/printk/printk.c index efe577477913..8787d3a72114 100644 --- a/kernel/printk/printk.c +++ b/kernel/printk/printk.c @@ -2612,11 +2612,12 @@ static int console_cpu_notify(unsigned int cpu) } /* - * 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. + * 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. */ -static bool abandon_console_lock_in_panic(void) +bool other_cpu_in_panic(void) { if (!panic_in_progress()) return false; @@ -2643,7 +2644,7 @@ void console_lock(void) might_sleep(); /* On panic, the console_lock must be left to the panic cpu. */ - while (abandon_console_lock_in_panic()) + while (other_cpu_in_panic()) msleep(1000); down_console_sem(); @@ -2663,7 +2664,7 @@ EXPORT_SYMBOL(console_lock); int console_trylock(void) { /* On panic, the console_lock must be left to the panic cpu. */ - if (abandon_console_lock_in_panic()) + if (other_cpu_in_panic()) return 0; if (down_trylock_console_sem()) return 0; @@ -2978,7 +2979,7 @@ static bool console_flush_all(bool do_cond_resched, u64 *next_seq, bool *handove any_progress = true; /* Allow panic_cpu to take over the consoles safely. */ - if (abandon_console_lock_in_panic()) + if (other_cpu_in_panic()) goto abandon; if (do_cond_resched) -- cgit v1.2.3