timers: Disable memory pre-allocation of timer debug objects
v2 cover timers: Disable memory pre-allocation of timer debug objects
[v2] Fix a typo in patch 2 that is causing compilation problem as
reported by kernel test robot.
A circular locking dependency was reported by lockdep involving printk() called from within the memory allocator, console driver with timeout capability and timer code calling debug_object_activate() doing object pool refill by allocating memory.
One simple way to break this circular locking dependency is to disable memory allocation when timer debug objects are being handled which is what this series is about. The majority of the changes is in the debugobjects code with a one-line change in the timer code.
I have also considered updating the console driver to not hold the lock when setting up timeout, but many console drivers are using timeout in some ways. So this problem may not specific to just a single one.
Waiman Long (3):
debugobjects: Add ODEBUG_FLAG_NO_ALLOC to disable memory allocation
debugobjects: Show the state of debug_objects_enabled
timers: Disable memory pre-allocation of timer debug objects
include/linux/debugobjects.h | 6 ++++++ kernel/time/timer.c | 1 + lib/debugobjects.c | 29 +++++++++++++++++------------ 3 files changed, 24 insertions(+), 12 deletions(-)
Signature · 1 line
2.49.0
PATCH v2 1/3 debugobjects: Add ODEBUG_FLAG_NO_ALLOC to disable memory allocation
Some of the objects associated with debug_obj may be handled in a sensitive context that allowing debug_obj pre-allocation in debug_objects_fill_pool() may casue deadlock. Add a new flags parameter to the debug_obj_descr structure as well as adding the new ODEBUG_FLAG_NO_ALLOC flag to enable us to disallow memory allocation for those types of objects.
--- include/linux/debugobjects.h | 6 ++++++ lib/debugobjects.c | 10 +++++----- 2 files changed, 11 insertions(+), 5 deletions(-)
include/linux/debugobjects.h +6 -0
@@ -41,6 +41,7 @@ struct debug_obj {* struct debug_obj_descr - object type specific debug description structure * * @name: name of the object typee +* @flags: debug object flags * @debug_hint: function returning address, which have associated * kernel symbol, to allow identify the object * @is_static_object: return true if the obj is static, otherwise return false
@@ -58,6 +59,7 @@ struct debug_obj {*/ struct debug_obj_descr {const char *name; +unsigned long flags;void *(*debug_hint)(void *addr);bool (*is_static_object)(void *addr);bool (*fixup_init)(void *addr,enum debug_obj_state state);
@@ -67,6 +69,10 @@ struct debug_obj_descr {bool (*fixup_assert_init)(void *addr,enum debug_obj_state state); }; +enum debug_obj_flags { +ODEBUG_FLAG_NO_ALLOC =0x1 ,/* Disallow debug object pre-allocation */ +}; +#ifdef CONFIG_DEBUG_OBJECTS extern void debug_object_init (void *addr,const struct debug_obj_descr *descr); extern void
lib/debugobjects.c +5 -5
@@ -694,7 +694,7 @@ static struct debug_obj *lookup_object_or_alloc(void *addr, struct debug_bucketreturn NULL ; } -static void debug_objects_fill_pool (void ) +static void debug_objects_fill_pool (bool no_alloc) {if (!static_branch_likely (&obj_cache_enabled))return ;
@@ -705,7 +705,7 @@ static void debug_objects_fill_pool(void)/* Try reusing objects from obj_to_free_list */ fill_pool_from_freelist (); -if (likely (!pool_should_refill (&pool_global))) +if (likely (!pool_should_refill (&pool_global) || no_alloc))return ;/*
@@ -734,7 +734,7 @@ __debug_object_init(void *addr, const struct debug_obj_descr *descr, int onstackstruct debug_bucket *db;unsigned long flags; -debug_objects_fill_pool (); +debug_objects_fill_pool (descr->flags &ODEBUG_FLAG_NO_ALLOC ); db =get_bucket ((unsigned long ) addr);
@@ -811,7 +811,7 @@ int debug_object_activate(void *addr, const struct debug_obj_descr *descr)if (!debug_objects_enabled)return 0 ; -debug_objects_fill_pool (); +debug_objects_fill_pool (descr->flags &ODEBUG_FLAG_NO_ALLOC ); db =get_bucket ((unsigned long ) addr);
@@ -1000,7 +1000,7 @@ void debug_object_assert_init(void *addr, const struct debug_obj_descr *descr)if (!debug_objects_enabled)return ; -debug_objects_fill_pool (); +debug_objects_fill_pool (descr->flags &ODEBUG_FLAG_NO_ALLOC ); db =get_bucket ((unsigned long ) addr);
Signature · 1 line
2.49.0
PATCH v2 2/3 debugobjects: Show the state of debug_objects_enabled
In reply to Waiman LongIn the rare case that debug_objects got disabled because we are running out of free debug objects, it is not easy to figure this out. Fix that by showing the state of "debug_objects_enabled" in the stats debugfs file as well as always printing a message in the console log.
--- lib/debugobjects.c | 19 ++++++++++++------- 1 file changed, 12 insertions(+), 7 deletions(-)
lib/debugobjects.c +12 -7
@@ -125,6 +125,12 @@ static int __init disable_object_debug(char *str) }early_param ("no_debug_objects" , disable_object_debug); +static void debug_objects_disable (const char *msg) +{ + debug_objects_enabled = false; +pr_warn ("debug_objects disabled: %s\n" , msg); +} +static const char *obj_states[ODEBUG_STATE_MAX ] = { [ODEBUG_STATE_NONE ] ="none" , [ODEBUG_STATE_INIT ] ="initialized" ,
@@ -690,7 +696,7 @@ static struct debug_obj *lookup_object_or_alloc(void *addr, struct debug_bucket }/* Out of memory. Do the cleanup outside of the locked region */ - debug_objects_enabled = false; +debug_objects_disable ("out of memory" );return NULL ; }
@@ -1161,6 +1167,8 @@ static int debug_stats_show(struct seq_file *m, void *v)seq_printf (m,"on_free_list : %u\n" ,pool_count (&pool_to_free));seq_printf (m,"objs_allocated: %d\n" , debug_objects_allocated);seq_printf (m,"objs_freed : %d\n" , debug_objects_freed); +seq_printf (m,"debug_objects : %s\n" , debug_objects_enabled ?"enabled" + :"disabled" );return 0 ; }DEFINE_SHOW_ATTRIBUTE (debug_stats);
@@ -1314,7 +1322,7 @@ check_results(void *addr, enum debug_obj_state state, int fixups, int warnings) out:raw_spin_unlock_irqrestore (&db->lock, flags);if (res) - debug_objects_enabled = false; +debug_objects_disable ("selftest" );return res; }
@@ -1486,11 +1494,8 @@ void __init debug_objects_mem_init(void) cache =kmem_cache_create ("debug_objects_cache" ,sizeof (struct debug_obj ),0 ,SLAB_DEBUG_OBJECTS |SLAB_NOLEAKTRACE ,NULL ); -if (!cache || !debug_objects_replace_static_objects (cache)) { - debug_objects_enabled = false; -pr_warn ("Out of memory.\n" ); -return ; - } +if (!cache || !debug_objects_replace_static_objects (cache)) +debug_objects_disable ("out of memory" );/* * Adjust the thresholds for allocating and freeing objects
Signature · 1 line
2.49.0
PATCH v2 3/3 timers: Disable memory pre-allocation of timer debug objects
In reply to Waiman LongA circular locking dependency lockdep splat was hit recently with a debug kernel. The dependency chain (in reverse order) is:
-> #3 (&zone->lock){-.-.}-{2:2}:
-> #2 (&base->lock){-.-.}-{2:2}:
-> #1 (&console_sch_key){-.-.}-{2:2}:
-> #0 (console_owner){..-.}-{0:0}:
The last one is from calling printk() within the rmqueue_bulk() call in mm/page_alloc.c. The "base->lock" is from lock_timer_base() and first one is due to calling add_timer_on() leading to debug_object_activate() doing actual memory allocation acquiring the zone lock.
The console_sch_key comes from a s390 console driver in driver/s390/cio. The console_sch_key -> timer dependency happens because the console driver is setting a timeout value while holding its lock. Apparently it is pretty common for a console driver to use timer for timeout or other timing purposes. So this may happen to other console drivers as well.
One way to break this circular locking dependency is to disallow any memory allocation when a timer debug object is being handled. Do this by setting the ODEBUG_FLAG_NO_ALLOC flag in the timer_debug_descr structure.
The figures below show the number of times the debug_objects_fill_pool() function has reached the statement right before and after the no_alloc check in initial bootup and after running a parallel kernel build on a 2-socket 96-threads x86-64 system.
Before After non-timer % ------ ----- -----------
Initial bootup 150,730 148,198 98.3%
Parallel kernel build 5,974,464 5,893,116 98.6%
So from object pre-allocation perspective, timer debug objects represent just a small slice of the total number of debug objects to be processed.
The allocation of debug_object to the global pool happens when its object count falls below the (256 + 16 * num_possible_cpus) threshold. Even then, there may still be free objects available in the percpu pool. So the chance that debug_objects gets disabled because it is running out of free debug_object should be minimal.
--- kernel/time/timer.c | 1 + 1 file changed, 1 insertion(+)
kernel/time/timer.c +1 -0
@@ -775,6 +775,7 @@ static bool timer_fixup_assert_init(void *addr, enum debug_obj_state state)static const struct debug_obj_descr timer_debug_descr = { .name ="timer_list" , + .flags =ODEBUG_FLAG_NO_ALLOC , .debug_hint = timer_debug_hint, .is_static_object = timer_is_static_object, .fixup_init = timer_fixup_init,
Signature · 1 line
2.49.0
kernel test robot
build botkernel test robot noticed "WARNING:possible_circular_locking_dependency_detected" on:
Hello,
kernel test robot noticed "WARNING:possible_circular_locking_dependency_detected" on:
commit: 89fd87e046372ce70f8146357324cacd365369fe ("[PATCH v2 3/3] timers: Disable memory pre-allocation of timer debug objects")
url: https://github.com/intel-lab-lkp/linux/commits/Waiman-Long/debugobjects-Add-ODEBUG_FLAG_NO_ALLOC-to-disable-memory-allocation/20250606-111718
base: https://git.kernel.org/cgit/linux/kernel/git/tip/tip.git 6c58d2791d6046727d87db50a5e46644f195dcf9
patch link: https://lore.kernel.org/all/[email protected]/
patch subject: [PATCH v2 3/3] timers: Disable memory pre-allocation of timer debug objects
in testcase: boot
config: x86_64-randconfig-003-20250608
compiler: gcc-12
test machine: qemu-system-x86_64 -enable-kvm -cpu SandyBridge -smp 2 -m 16G
(please refer to attached dmesg/kmsg for entire log/backtrace)
we observe the issue does not always happen, 21 times out of 100 runs as below. the parent keeps clean.
fc7e5b596de17969 89fd87e046372ce70f814635732
---------------- ---------------------------
fail:runs %reproduction fail:runs
| | |
:100 21% 21:100 dmesg.WARNING:possible_circular_locking_dependency_detected
:100 21% 21:100 dmesg.WARNING:possible_circular_locking_dependency_detected_systemd_is_trying_to_acquire_lock:at:__pm_runtime_resume_but_task_is_already_holding_lock:at:uart_write
If you fix the issue in a separate patch/commit (i.e. not just a new version of the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <[email protected]>
| Closes: https://lore.kernel.org/oe-lkp/[email protected]
[ 92.544844][ T1] WARNING: possible circular locking dependency detected
[ 92.545407][ T1] 6.15.0-rc1-00010-g89fd87e04637 #1 Tainted: G T
[ 92.546120][ T1] ------------------------------------------------------
[ 92.546721][ T1] systemd/1 is trying to acquire lock:
[ 92.547161][ T1] ffff888118c93968 (&dev->power.lock){-...}-{3:3}, at: __pm_runtime_resume (kbuild/obj/consumer/x86_64-randconfig-003-20250608/drivers/base/power/runtime.c:1180)
[ 92.548064][ T1]
[ 92.548064][ T1] but task is already holding lock:
[ 92.548680][ T1] ffffffff8806faf8 (&port_lock_key){....}-{3:3}, at: uart_write (kbuild/obj/consumer/x86_64-randconfig-003-20250608/include/linux/serial_core.h:638 kbuild/obj/consumer/x86_64-randconfig-003-20250608/include/linux/serial_core.h:672 kbuild/obj/consumer/x86_64-randconfig-003-20250608/include/linux/serial_core.h:716 kbuild/obj/consumer/x86_64-randconfig-003-20250608/drivers/tty/serial/serial_core.c:626)
[ 92.549493][ T1]
[ 92.549493][ T1] which lock already depends on the new lock.
[ 92.549493][ T1]
[ 92.550293][ T1]
[ 92.550293][ T1] the existing dependency chain (in reverse order) is:
[ 92.551049][ T1]
[ 92.551049][ T1] -> #4 (&port_lock_key){....}-{3:3}:
[ 92.551674][ T1] __lock_acquire (kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/locking/lockdep.c:5235)
[ 92.552145][ T1] lock_acquire (kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/locking/lockdep.c:472 kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/locking/lockdep.c:5868)
[ 92.552621][ T1] _raw_spin_lock_irqsave (kbuild/obj/consumer/x86_64-randconfig-003-20250608/include/linux/spinlock_api_smp.h:111 kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/locking/spinlock.c:162)
[ 92.553094][ T1] serial8250_console_write (kbuild/obj/consumer/x86_64-randconfig-003-20250608/include/linux/serial_core.h:638 kbuild/obj/consumer/x86_64-randconfig-003-20250608/include/linux/serial_core.h:672 kbuild/obj/consumer/x86_64-randconfig-003-20250608/include/linux/serial_core.h:716 kbuild/obj/consumer/x86_64-randconfig-003-20250608/drivers/tty/serial/8250/8250_port.c:3415)
[ 92.553560][ T1] univ8250_console_write (kbuild/obj/consumer/x86_64-randconfig-003-20250608/drivers/tty/serial/8250/8250_core.c:396)
[ 92.554015][ T1] console_emit_next_record (kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/printk/printk.c:3055 (discriminator 3) kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/printk/printk.c:3139 (discriminator 3))
[ 92.554517][ T1] console_flush_all (kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/printk/printk.c:3227)
[ 92.555013][ T1] console_unlock (kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/printk/printk.c:3286 kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/printk/printk.c:3325)
[ 92.555431][ T1] vprintk_emit (kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/printk/printk.c:2450)
[ 92.555844][ T1] vprintk_default (kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/printk/printk.c:2465)
[ 92.556277][ T1] vprintk (kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/printk/printk_safe.c:82)
[ 92.556641][ T1] _printk (kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/printk/printk.c:2475)
[ 92.557085][ T1] register_console (kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/printk/printk.c:4126)
[ 92.557584][ T1] univ8250_console_init (kbuild/obj/consumer/x86_64-randconfig-003-20250608/drivers/tty/serial/8250/8250_core.c:514)
[ 92.558034][ T1] console_init (kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/printk/printk.c:4323)
[ 92.558479][ T1] start_kernel (kbuild/obj/consumer/x86_64-randconfig-003-20250608/init/main.c:1035 (discriminator 3))
[ 92.558928][ T1] x86_64_start_reservations (kbuild/obj/consumer/x86_64-randconfig-003-20250608/arch/x86/kernel/head64.c:501)
[ 92.559398][ T1] x86_64_start_kernel (kbuild/obj/consumer/x86_64-randconfig-003-20250608/arch/x86/kernel/head64.c:443 (discriminator 17))
[ 92.559835][ T1] common_startup_64 (kbuild/obj/consumer/x86_64-randconfig-003-20250608/arch/x86/kernel/head_64.S:419)
[ 92.560249][ T1]
[ 92.560249][ T1] -> #3 (console_owner){....}-{0:0}:
[ 92.560869][ T1] __lock_acquire (kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/locking/lockdep.c:5235)
[ 92.561272][ T1] lock_acquire (kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/locking/lockdep.c:472 kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/locking/lockdep.c:5868)
[ 92.561827][ T1] console_lock_spinning_enable (kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/printk/printk.c:1924)
[ 92.562373][ T1] console_emit_next_record (kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/printk/printk.c:3050 (discriminator 3) kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/printk/printk.c:3137 (discriminator 3))
[ 92.562932][ T1] console_flush_all (kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/printk/printk.c:3227)
[ 92.563398][ T1] console_unlock (kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/printk/printk.c:3286 kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/printk/printk.c:3325)
[ 92.563806][ T1] vprintk_emit (kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/printk/printk.c:2450)
[ 92.564198][ T1] vprintk_default (kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/printk/printk.c:2465)
[ 92.564592][ T1] vprintk (kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/printk/printk_safe.c:82)
[ 92.564990][ T1] _printk (kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/printk/printk.c:2475)
[ 92.565340][ T1] lookup_object_or_alloc (kbuild/obj/consumer/x86_64-randconfig-003-20250608/lib/debugobjects.c:700)
[ 92.565838][ T1] __debug_object_init (kbuild/obj/consumer/x86_64-randconfig-003-20250608/lib/debugobjects.c:750)
[ 92.566274][ T1] debug_object_init (kbuild/obj/consumer/x86_64-randconfig-003-20250608/lib/debugobjects.c:785)
[ 92.566777][ T1] init_timer_key (kbuild/obj/consumer/x86_64-randconfig-003-20250608/arch/x86/include/asm/jump_label.h:36 kbuild/obj/consumer/x86_64-randconfig-003-20250608/include/trace/events/timer.h:33 kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/time/timer.c:838 kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/time/timer.c:881)
[ 92.567230][ T1] o2net_init (kbuild/obj/consumer/x86_64-randconfig-003-20250608/fs/ocfs2/cluster/tcp.c:2128 (discriminator 3))
[ 92.567629][ T1] init_o2nm (kbuild/obj/consumer/x86_64-randconfig-003-20250608/fs/ocfs2/cluster/nodemanager.c:832)
[ 92.568023][ T1] do_one_initcall (kbuild/obj/consumer/x86_64-randconfig-003-20250608/init/main.c:1257)
[ 92.568441][ T1] do_initcalls (kbuild/obj/consumer/x86_64-randconfig-003-20250608/init/main.c:1318 kbuild/obj/consumer/x86_64-randconfig-003-20250608/init/main.c:1335)
[ 92.568835][ T1] kernel_init_freeable (kbuild/obj/consumer/x86_64-randconfig-003-20250608/init/main.c:1569)
[ 92.569283][ T1] kernel_init (kbuild/obj/consumer/x86_64-randconfig-003-20250608/init/main.c:1459)
[ 92.569690][ T1] ret_from_fork (kbuild/obj/consumer/x86_64-randconfig-003-20250608/arch/x86/kernel/process.c:159)
[ 92.570193][ T1] ret_from_fork_asm (kbuild/obj/consumer/x86_64-randconfig-003-20250608/arch/x86/entry/entry_64.S:255)
[ 92.570619][ T1]
[ 92.570619][ T1] -> #2 (&obj_hash[i].lock){-.-.}-{2:2}:
[ 92.571302][ T1] __lock_acquire (kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/locking/lockdep.c:5235)
[ 92.571722][ T1] lock_acquire (kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/locking/lockdep.c:472 kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/locking/lockdep.c:5868)
[ 92.572151][ T1] _raw_spin_lock_irqsave (kbuild/obj/consumer/x86_64-randconfig-003-20250608/include/linux/spinlock_api_smp.h:111 kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/locking/spinlock.c:162)
[ 92.572704][ T1] debug_object_assert_init (kbuild/obj/consumer/x86_64-randconfig-003-20250608/lib/debugobjects.c:1014)
[ 92.573200][ T1] __mod_timer (kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/time/timer.c:1033)
[ 92.573613][ T1] mod_timer (kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/time/timer.c:1196)
[ 92.573983][ T1] worker_enter_idle (kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/workqueue.c:1053)
[ 92.574452][ T1] create_worker (kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/workqueue.c:2833)
[ 92.574850][ T1] worker_thread (kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/workqueue.c:3063 kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/workqueue.c:3115 kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/workqueue.c:3375)
[ 92.575278][ T1] kthread (kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/kthread.c:464)
[ 92.575652][ T1] ret_from_fork (kbuild/obj/consumer/x86_64-randconfig-003-20250608/arch/x86/kernel/process.c:159)
[ 92.576118][ T1] ret_from_fork_asm (kbuild/obj/consumer/x86_64-randconfig-003-20250608/arch/x86/entry/entry_64.S:255)
[ 92.576589][ T1]
[ 92.576589][ T1] -> #1 (&pool->lock){-.-.}-{2:2}:
[ 92.577245][ T1] __lock_acquire (kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/locking/lockdep.c:5235)
[ 92.577686][ T1] lock_acquire (kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/locking/lockdep.c:472 kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/locking/lockdep.c:5868)
[ 92.578132][ T1] _raw_spin_lock (kbuild/obj/consumer/x86_64-randconfig-003-20250608/include/linux/spinlock_api_smp.h:134 kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/locking/spinlock.c:154)
[ 92.578524][ T1] __queue_work (kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/workqueue.c:2311)
[ 92.578929][ T1] queue_work_on (kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/workqueue.c:2393)
[ 92.579474][ T1] rpm_suspend (kbuild/obj/consumer/x86_64-randconfig-003-20250608/include/linux/workqueue.h:662 kbuild/obj/consumer/x86_64-randconfig-003-20250608/drivers/base/power/runtime.c:673)
[ 92.579939][ T1] rpm_idle (kbuild/obj/consumer/x86_64-randconfig-003-20250608/drivers/base/power/runtime.c:547 (discriminator 1))
[ 92.580379][ T1] __pm_runtime_idle (kbuild/obj/consumer/x86_64-randconfig-003-20250608/include/linux/spinlock.h:406 (discriminator 5) kbuild/obj/consumer/x86_64-randconfig-003-20250608/drivers/base/power/runtime.c:1113 (discriminator 5))
[ 92.580896][ T1] __device_attach (kbuild/obj/consumer/x86_64-randconfig-003-20250608/include/linux/device.h:937 kbuild/obj/consumer/x86_64-randconfig-003-20250608/drivers/base/dd.c:1051)
[ 92.581512][ T1] device_initial_probe (kbuild/obj/consumer/x86_64-randconfig-003-20250608/drivers/base/dd.c:1079)
[ 92.581983][ T1] bus_probe_device (kbuild/obj/consumer/x86_64-randconfig-003-20250608/drivers/base/bus.c:537)
[ 92.582481][ T1] device_add (kbuild/obj/consumer/x86_64-randconfig-003-20250608/drivers/base/core.c:3673)
[ 92.582934][ T1] serial_base_port_add (kbuild/obj/consumer/x86_64-randconfig-003-20250608/drivers/tty/serial/serial_base_bus.c:179)
[ 92.583403][ T1] serial_core_register_port (kbuild/obj/consumer/x86_64-randconfig-003-20250608/drivers/tty/serial/serial_core.c:3342 kbuild/obj/consumer/x86_64-randconfig-003-20250608/drivers/tty/serial/serial_core.c:3381)
[ 92.583950][ T1] serial_ctrl_register_port (kbuild/obj/consumer/x86_64-randconfig-003-20250608/drivers/tty/serial/serial_ctrl.c:41)
[ 92.584518][ T1] uart_add_one_port (kbuild/obj/consumer/x86_64-randconfig-003-20250608/drivers/tty/serial/serial_port.c:143)
[ 92.585031][ T1] serial8250_register_8250_port (kbuild/obj/consumer/x86_64-randconfig-003-20250608/drivers/tty/serial/8250/8250_core.c:822)
[ 92.585578][ T1] serial_pnp_probe (kbuild/obj/consumer/x86_64-randconfig-003-20250608/drivers/tty/serial/8250/8250_pnp.c:480)
[ 92.586119][ T1] pnp_device_probe (kbuild/obj/consumer/x86_64-randconfig-003-20250608/drivers/pnp/driver.c:111)
[ 92.586574][ T1] really_probe (kbuild/obj/consumer/x86_64-randconfig-003-20250608/drivers/base/dd.c:579 kbuild/obj/consumer/x86_64-randconfig-003-20250608/drivers/base/dd.c:658)
[ 92.586973][ T1] __driver_probe_device (kbuild/obj/consumer/x86_64-randconfig-003-20250608/drivers/base/dd.c:800)
[ 92.587468][ T1] driver_probe_device (kbuild/obj/consumer/x86_64-randconfig-003-20250608/drivers/base/dd.c:830)
[ 92.587949][ T1] __driver_attach (kbuild/obj/consumer/x86_64-randconfig-003-20250608/drivers/base/dd.c:1217)
[ 92.588438][ T1] bus_for_each_dev (kbuild/obj/consumer/x86_64-randconfig-003-20250608/drivers/base/bus.c:370)
[ 92.588872][ T1] driver_attach (kbuild/obj/consumer/x86_64-randconfig-003-20250608/drivers/base/dd.c:1234)
[ 92.589342][ T1] bus_add_driver (kbuild/obj/consumer/x86_64-randconfig-003-20250608/drivers/base/bus.c:678)
[ 92.589759][ T1] driver_register (kbuild/obj/consumer/x86_64-randconfig-003-20250608/drivers/base/driver.c:249)
[ 92.590215][ T1] pnp_register_driver (kbuild/obj/consumer/x86_64-randconfig-003-20250608/drivers/pnp/driver.c:280)
[ 92.590609][ T1] serial8250_pnp_init (kbuild/obj/consumer/x86_64-randconfig-003-20250608/drivers/tty/serial/8250/8250_pnp.c:530)
[ 92.591096][ T1] serial8250_init (kbuild/obj/consumer/x86_64-randconfig-003-20250608/drivers/tty/serial/8250/8250_platform.c:315)
[ 92.591488][ T1] do_one_initcall (kbuild/obj/consumer/x86_64-randconfig-003-20250608/init/main.c:1257)
[ 92.591939][ T1] do_initcalls (kbuild/obj/consumer/x86_64-randconfig-003-20250608/init/main.c:1318 kbuild/obj/consumer/x86_64-randconfig-003-20250608/init/main.c:1335)
[ 92.592334][ T1] kernel_init_freeable (kbuild/obj/consumer/x86_64-randconfig-003-20250608/init/main.c:1569)
[ 92.592879][ T1] kernel_init (kbuild/obj/consumer/x86_64-randconfig-003-20250608/init/main.c:1459)
[ 92.593289][ T1] ret_from_fork (kbuild/obj/consumer/x86_64-randconfig-003-20250608/arch/x86/kernel/process.c:159)
[ 92.593701][ T1] ret_from_fork_asm (kbuild/obj/consumer/x86_64-randconfig-003-20250608/arch/x86/entry/entry_64.S:255)
[ 92.594139][ T1]
[ 92.594139][ T1] -> #0 (&dev->power.lock){-...}-{3:3}:
[ 92.594791][ T1] check_prev_add (kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/locking/lockdep.c:3167)
[ 92.595300][ T1] validate_chain (kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/locking/lockdep.c:3286 kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/locking/lockdep.c:3909)
[ 92.595732][ T1] __lock_acquire (kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/locking/lockdep.c:5235)
[ 92.596198][ T1] lock_acquire (kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/locking/lockdep.c:472 kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/locking/lockdep.c:5868)
[ 92.596626][ T1] _raw_spin_lock_irqsave (kbuild/obj/consumer/x86_64-randconfig-003-20250608/include/linux/spinlock_api_smp.h:111 kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/locking/spinlock.c:162)
[ 92.597149][ T1] __pm_runtime_resume (kbuild/obj/consumer/x86_64-randconfig-003-20250608/drivers/base/power/runtime.c:1180)
[ 92.597624][ T1] __uart_start (kbuild/obj/consumer/x86_64-randconfig-003-20250608/include/linux/pm_runtime.h:414 kbuild/obj/consumer/x86_64-randconfig-003-20250608/drivers/tty/serial/serial_core.c:148)
[ 92.598037][ T1] uart_write (kbuild/obj/consumer/x86_64-randconfig-003-20250608/include/linux/serial_core.h:638 kbuild/obj/consumer/x86_64-randconfig-003-20250608/include/linux/serial_core.h:682 kbuild/obj/consumer/x86_64-randconfig-003-20250608/include/linux/serial_core.h:785 kbuild/obj/consumer/x86_64-randconfig-003-20250608/drivers/tty/serial/serial_core.c:636)
[ 92.598483][ T1] process_output_block (kbuild/obj/consumer/x86_64-randconfig-003-20250608/drivers/tty/n_tty.c:561)
[ 92.598979][ T1] n_tty_write (kbuild/obj/consumer/x86_64-randconfig-003-20250608/drivers/tty/n_tty.c:2377)
[ 92.599384][ T1] iterate_tty_write (kbuild/obj/consumer/x86_64-randconfig-003-20250608/drivers/tty/tty_io.c:1015)
[ 92.599892][ T1] file_tty_write+0x13f/0x240
[ 92.600397][ T1] redirected_tty_write (kbuild/obj/consumer/x86_64-randconfig-003-20250608/drivers/tty/tty_io.c:1111 kbuild/obj/consumer/x86_64-randconfig-003-20250608/drivers/tty/tty_io.c:1134)
[ 92.600918][ T1] do_iter_readv_writev (kbuild/obj/consumer/x86_64-randconfig-003-20250608/fs/read_write.c:825)
[ 92.601380][ T1] vfs_writev (kbuild/obj/consumer/x86_64-randconfig-003-20250608/fs/read_write.c:1055)
[ 92.601825][ T1] do_writev (kbuild/obj/consumer/x86_64-randconfig-003-20250608/fs/read_write.c:1101)
[ 92.602226][ T1] __ia32_sys_writev (kbuild/obj/consumer/x86_64-randconfig-003-20250608/fs/read_write.c:1169 kbuild/obj/consumer/x86_64-randconfig-003-20250608/fs/read_write.c:1166 kbuild/obj/consumer/x86_64-randconfig-003-20250608/fs/read_write.c:1166)
[ 92.602662][ T1] ia32_sys_call (kbuild/obj/consumer/x86_64-randconfig-003-20250608/./arch/x86/include/generated/asm/syscalls_32.h:147)
[ 92.603141][ T1] do_int80_emulation (kbuild/obj/consumer/x86_64-randconfig-003-20250608/arch/x86/entry/syscall_32.c:83 kbuild/obj/consumer/x86_64-randconfig-003-20250608/arch/x86/entry/syscall_32.c:172)
[ 92.603655][ T1] asm_int80_emulation (kbuild/obj/consumer/x86_64-randconfig-003-20250608/arch/x86/include/asm/idtentry.h:626)
[ 92.604247][ T1]
[ 92.604247][ T1] other info that might help us debug this:
[ 92.604247][ T1]
[ 92.605052][ T1] Chain exists of:
[ 92.605052][ T1] &dev->power.lock --> console_owner --> &port_lock_key
[ 92.605052][ T1]
[ 92.606202][ T1] Possible unsafe locking scenario:
[ 92.606202][ T1]
[ 92.606763][ T1] CPU0 CPU1
[ 92.607130][ T1] ---- ----
[ 92.607543][ T1] lock(&port_lock_key);
[ 92.607947][ T1] lock(console_owner);
[ 92.608624][ T1] lock(&port_lock_key);
[ 92.609191][ T1] lock(&dev->power.lock);
[ 92.609615][ T1]
[ 92.609615][ T1] *** DEADLOCK ***
[ 92.609615][ T1]
[ 92.610237][ T1] 5 locks held by systemd/1:
[ 92.610655][ T1] #0: ffff8881193a3bb8 (&tty->ldisc_sem){++++}-{0:0}, at: ldsem_down_read (kbuild/obj/consumer/x86_64-randconfig-003-20250608/drivers/tty/tty_ldsem.c:340) The kernel config and materials to reproduce are available at:
https://download.01.org/0day-ci/archive/20250612/[email protected]
Signature · 2 lines
0-DAY CI Kernel Test Service https://github.com/intel/lkp-tests/wiki
Quoted 2 lines · kernel test robot
kernel test robot noticed "WARNING:possible_circular_locking_dependency_detected" on:
The lockdep issue comes from patch 2/3 which prints with the hash bucket lock held.
Quoted 1 line
commit:
89fd87e046372ce70f8146357324cacd365369fe("[PATCH v2 3/3] timers: Disable memory pre-allocation of timer debug objects")
This one triggers it because the preallocation disable makes debug object go out of memory.
Quoted 2 lines
[ 92.564990][ T1] _printk (kbuild/obj/consumer/x86_64-randconfig-003-20250608/kernel/printk/printk.c:2475) [ 92.565340][ T1] lookup_object_or_alloc (kbuild/obj/consumer/x86_64-randconfig-003-20250608/lib/debugobjects.c:700)
That's the:
debug_objects_disable("out of memory");introduced by patch 2/3. The printk() in debug_objects_disable() needs to be deferred.
Quoted 2 lines
[ 92.565838][ T1] __debug_object_init (kbuild/obj/consumer/x86_64-randconfig-003-20250608/lib/debugobjects.c:750) [ 92.566274][ T1] debug_object_init (kbuild/obj/consumer/x86_64-randconfig-003-20250608/lib/debugobjects.c:785)
Which means that the changelog saying:
"So the chance that debug_objects gets disabled because it is running out
of free debug_object should be minimal."
is slightly off ....
Thanks,
tglx
Quoted 19 lines · Waiman Long
A circular locking dependency lockdep splat was hit recently with a debug kernel. The dependency chain (in reverse order) is:
-> #3 (&zone->lock){-.-.}-{2:2}:
-> #2 (&base->lock){-.-.}-{2:2}:
-> #1 (&console_sch_key){-.-.}-{2:2}:
-> #0 (console_owner){..-.}-{0:0}:The last one is from calling
printk()within thermqueue_bulk()call inmm/page_alloc.c. The "base->lock" is fromlock_timer_base()and first one is due to callingadd_timer_on()leading todebug_object_activate()doing actual memory allocation acquiring the zone lock.The
console_sch_keycomes from a s390 console driver in driver/s390/cio. Theconsole_sch_key-> timer dependency happens because the console driver is setting a timeout value while holding its lock. Apparently it is pretty common for a console driver to use timer for timeout or other timing purposes. So this may happen to other console drivers as well.
Quoted 4 lines
One way to break this circular locking dependency is to disallow any memory allocation when a timer debug object is being handled. Do this by setting the
ODEBUG_FLAG_NO_ALLOCflag in thetimer_debug_descrstructure.
Well. I'm absolutely not convinced that this is the right approach.
I have a hard time to find the printk() in rmqueue_bulk(). But if there is one then it has to go or has to be converted to a deferred printk() simply because that code can be called from so many contexts, which all can legitimately create a lock dependency chain into the console drivers in some way or the other. That's like invoking printk() from the guts of the scheduler or locking code.
Quoted 5 lines
The figures below show the number of times the
debug_objects_fill_pool()function has reached the statement right before and after theno_alloccheck in initial bootup and after running a parallel kernel build on a 2-socket 96-threads x86-64 system.
Quoted 7 lines
Before After non-timer % ------ ----- -----------Initial bootup 150,730 148,198 98.3%
Parallel kernel build 5,974,464 5,893,116 98.6%So from object pre-allocation perspective, timer debug objects represent just a small slice of the total number of debug objects to be processed.
That math is skewed due to the way how debugobjects handles the allocations for the global pool.
The initial decision to attempt a refill is:
count < min_cnt
where min_cnt = 256 + 16 * num_possible_cpus()
That makes _one_ context go into the allocation path unless
count < min_cnt / 2
which forces all contexts to try allocating in order not to deplete the pool.
So let's assume we have 16 CPUs, then min_cnt = 512 and therefore min_cnt / 2 = 256.
As the initial context which allocates when the count goes below 512 might be preempted, the rest of the callers can lower the pool count to 256 easily.
In the 0-day splat the debug_objects OOM happens from o2net_init():
[ 92.566274][ T1] debug_object_init (kbuild/obj/consumer/x86_64-randconfig-003-20250608/lib/debugobjects.c:785) [ 92.566777][ T1] init_timer_key (kbuild/obj/consumer/x86_64-randconfig-003-20250608/arch/x86/include/asm/jump_label.h:36 [ 92.567230][ T1] o2net_init (kbuild/obj/consumer/x86_64-randconfig-003-20250608/fs/ocfs2/cluster/tcp.c:2128 (discriminator 3)) [ 92.567629][ T1] init_o2nm (kbuild/obj/consumer/x86_64-randconfig-003-20250608/fs/ocfs2/cluster/nodemanager.c:832) [ 92.568023][ T1] do_one_initcall (kbuild/obj/consumer/x86_64-randconfig-003-20250608/init/main.c:1257)
o2net_init() initializes 255 nodes and each node has three delayed works. Each delayed work contains a timer for which debugobjects needs to create a new tracking object. So with your brute force approach of disabling allocations for timers blindy o2net_init() can trivially deplete the pool.
For the o2net case this requires that workqueue debugobjects are disabled, but you can't argue that this is silly because there are other code paths which do bulk initialization of timers w/o having a work involved.
So using the percentage of timer operations for evaluating how this change can cause a debug object OOM is just bogus and wishful thinking.
Let's take a step back and ask the obvious question, when there is actually consumption of debug objects happening:
1) For all dynamically initialized objects it happens in debug_object_init()
2) For statically initialized objects it happens in debug_object_activate()
#2 is arguably irrelevant as there are not gazillions of statically
initialized objects, which are trackable by debugobjects
#1 is the vast majority and the good news is that the initialization of
such objects (after allocation) happens mostly in preemptible context with no locks held.
So the obvious thing to try is not adding some random flag to timers (and tomorrow to RCU and work), but to restrict the allocation requirement to debug_object_init().
Something like the untested below should make all of these headaches go away. Except for the general observation that debugobjects is not the only way to create nasty dependency chains into console driver locks, but that's a headache the MM folks have to sort out.
Thanks,
tglx
---
lib/debugobjects.c +0 -4
@@ -811,8 +811,6 @@ int debug_object_activate(void *addr, coif (!debug_objects_enabled)return 0 ; -debug_objects_fill_pool (); - db =get_bucket ((unsigned long ) addr);raw_spin_lock_irqsave (&db->lock, flags);
@@ -1000,8 +998,6 @@ void debug_object_assert_init(void *addrif (!debug_objects_enabled)return ; -debug_objects_fill_pool (); - db =get_bucket ((unsigned long ) addr);raw_spin_lock_irqsave (&db->lock, flags);
Quoted 24 lines · Thomas Gleixner
Quoted 19 lines · Waiman Long
A circular locking dependency lockdep splat was hit recently with a debug kernel. The dependency chain (in reverse order) is:
-> #3 (&zone->lock){-.-.}-{2:2}: -> #2 (&base->lock){-.-.}-{2:2}: -> #1 (&console_sch_key){-.-.}-{2:2}: -> #0 (console_owner){..-.}-{0:0}:The last one is from calling
printk()within thermqueue_bulk()call inmm/page_alloc.c. The "base->lock" is fromlock_timer_base()and first one is due to callingadd_timer_on()leading todebug_object_activate()doing actual memory allocation acquiring the zone lock.The
console_sch_keycomes from a s390 console driver in driver/s390/cio. Theconsole_sch_key-> timer dependency happens because the console driver is setting a timeout value while holding its lock. Apparently it is pretty common for a console driver to use timer for timeout or other timing purposes. So this may happen to other console drivers as well.Quoted 4 lines
One way to break this circular locking dependency is to disallow any memory allocation when a timer debug object is being handled. Do this by setting the
ODEBUG_FLAG_NO_ALLOCflag in thetimer_debug_descrstructure.
Quoted 8 lines
Well. I'm absolutely not convinced that this is the right approach.
I have a hard time to find the
printk()inrmqueue_bulk(). But if there is one then it has to go or has to be converted to a deferredprintk()simply because that code can be called from so many contexts, which all can legitimately create a lock dependency chain into the console drivers in some way or the other. That's like invokingprintk()from the guts of the scheduler or locking code.
Actually, rmqueue_bulk() calls expand() which, in turn, calls __add_to_free_list() and the printk() comes from the VM_WARN_ONCE() macro there.
In a sense, printk() is called because of some other issues in the mm code.
Quoted 101 lines
Quoted 6 lines
The figures below show the number of times the
debug_objects_fill_pool()function has reached the statement right before and after theno_alloccheck in initial bootup and after running a parallel kernel build on a 2-socket 96-threads x86-64 system.Quoted 7 lines
Before After non-timer % ------ ----- -----------Initial bootup 150,730 148,198 98.3%
Parallel kernel build 5,974,464 5,893,116 98.6%So from object pre-allocation perspective, timer debug objects represent just a small slice of the total number of debug objects to be processed.
That math is skewed due to the way how debugobjects handles the allocations for the global pool.
The initial decision to attempt a refill is:
count <
min_cntwhere
min_cnt= 256 + 16 *num_possible_cpus()That makes _one_ context go into the allocation path unless
count <
min_cnt/ 2which forces all contexts to try allocating in order not to deplete the pool.
So let's assume we have 16 CPUs, then
min_cnt= 512 and thereforemin_cnt/ 2 = 256.As the initial context which allocates when the count goes below 512 might be preempted, the rest of the callers can lower the pool count to 256 easily.
In the 0-day splat the
debug_objectsOOM happens fromo2net_init():[ 92.566274][ T1] debug_object_init (kbuild/obj/consumer/x86_64-randconfig-003-20250608/lib/debugobjects.c:785) [ 92.566777][ T1] init_timer_key (kbuild/obj/consumer/x86_64-randconfig-003-20250608/arch/x86/include/asm/jump_label.h:36 [ 92.567230][ T1] o2net_init (kbuild/obj/consumer/x86_64-randconfig-003-20250608/fs/ocfs2/cluster/tcp.c:2128 (discriminator 3)) [ 92.567629][ T1] init_o2nm (kbuild/obj/consumer/x86_64-randconfig-003-20250608/fs/ocfs2/cluster/nodemanager.c:832) [ 92.568023][ T1] do_one_initcall (kbuild/obj/consumer/x86_64-randconfig-003-20250608/init/main.c:1257)
o2net_init()initializes 255 nodes and each node has three delayed works. Each delayed work contains a timer for which debugobjects needs to create a new tracking object. So with your brute force approach of disabling allocations for timers blindyo2net_init()can trivially deplete the pool.For the o2net case this requires that workqueue debugobjects are disabled, but you can't argue that this is silly because there are other code paths which do bulk initialization of timers w/o having a work involved.
So using the percentage of timer operations for evaluating how this change can cause a debug object OOM is just bogus and wishful thinking.
Let's take a step back and ask the obvious question, when there is actually consumption of debug objects happening:
1) For all dynamically initialized objects it happens in
debug_object_init()2) For statically initialized objects it happens in
debug_object_activate()#2 is arguably irrelevant as there are not gazillions of statically
initialized objects, which are trackable by debugobjects#1 is the vast majority and the good news is that the initialization of
such objects (after allocation) happens mostly in preemptible context with no locks held.So the obvious thing to try is not adding some random flag to timers (and tomorrow to RCU and work), but to restrict the allocation requirement to
debug_object_init().Something like the untested below should make all of these headaches go away. Except for the general observation that debugobjects is not the only way to create nasty dependency chains into console driver locks, but that's a headache the MM folks have to sort out.
Thanks,
tglx
---lib/debugobjects.c +0
-2@@ -811,8 +811,6 @@ int debug_object_activate(void *addr, coif (!debug_objects_enabled)return 0 ; -debug_objects_fill_pool (); - db =get_bucket ((unsigned long ) addr);raw_spin_lock_irqsave (&db->lock, flags);
lib/debugobjects.c void debug_object_assert_init(void *addr · line 1003
Quoted 8 lines
lib/debugobjects.c +0
-2@@ -1000,8 +998,6 @@ void debug_object_assert_init(void *addrif (!debug_objects_enabled)return ; -debug_objects_fill_pool (); - db =get_bucket ((unsigned long ) addr);raw_spin_lock_irqsave (&db->lock, flags);
Thanks for the suggestion. I will take further look at this problem.
Cheers,
Longman
Quoted 11 lines · Waiman Long
Quoted 6 lines · Thomas Gleixner
I have a hard time to find the
printk()inrmqueue_bulk(). But if there is one then it has to go or has to be converted to a deferredprintk()simply because that code can be called from so many contexts, which all can legitimately create a lock dependency chain into the console drivers in some way or the other. That's like invokingprintk()from the guts of the scheduler or locking code.Actually,
rmqueue_bulk()callsexpand()which, in turn, calls__add_to_free_list()and theprintk()comes from theVM_WARN_ONCE()macro there.
Duh. I missed that one ....
Quoted 1 line
In a sense,
printk()is called because of some other issues in the mm code.
:)
Thanks,
tglx