From 0cce06ba859a515bd06224085d3addb870608b6d Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Tue, 25 Apr 2023 17:03:13 +0200 Subject: [PATCH 1/2] debugobjects,locking: Annotate debug_object_fill_pool() wait type violation There is an explicit wait-type violation in debug_object_fill_pool() for PREEMPT_RT=n kernels which allows them to more easily fill the object pool and reduce the chance of allocation failures. Lockdep's wait-type checks are designed to check the PREEMPT_RT locking rules even for PREEMPT_RT=n kernels and object to this, so create a lockdep annotation to allow this to stand. Specifically, create a 'lock' type that overrides the inner wait-type while it is held -- allowing one to temporarily raise it, such that the violation is hidden. Reported-by: Vlastimil Babka Reported-by: Qi Zheng Signed-off-by: Peter Zijlstra (Intel) Tested-by: Qi Zheng Link: https://lkml.kernel.org/r/20230429100614.GA1489784@hirez.programming.kicks-ass.net --- include/linux/lockdep.h | 14 ++++++++++++++ include/linux/lockdep_types.h | 1 + kernel/locking/lockdep.c | 28 +++++++++++++++++++++------- lib/debugobjects.c | 15 +++++++++++++-- 4 files changed, 49 insertions(+), 9 deletions(-) diff --git a/include/linux/lockdep.h b/include/linux/lockdep.h index 1023f349af71..a3329fb49b33 100644 --- a/include/linux/lockdep.h +++ b/include/linux/lockdep.h @@ -339,6 +339,16 @@ extern void lock_unpin_lock(struct lockdep_map *lock, struct pin_cookie); #define lockdep_repin_lock(l,c) lock_repin_lock(&(l)->dep_map, (c)) #define lockdep_unpin_lock(l,c) lock_unpin_lock(&(l)->dep_map, (c)) +/* + * Must use lock_map_aquire_try() with override maps to avoid + * lockdep thinking they participate in the block chain. + */ +#define DEFINE_WAIT_OVERRIDE_MAP(_name, _wait_type) \ + struct lockdep_map _name = { \ + .name = #_name "-wait-type-override", \ + .wait_type_inner = _wait_type, \ + .lock_type = LD_LOCK_WAIT_OVERRIDE, } + #else /* !CONFIG_LOCKDEP */ static inline void lockdep_init_task(struct task_struct *task) @@ -427,6 +437,9 @@ extern int lockdep_is_held(const void *); #define lockdep_repin_lock(l, c) do { (void)(l); (void)(c); } while (0) #define lockdep_unpin_lock(l, c) do { (void)(l); (void)(c); } while (0) +#define DEFINE_WAIT_OVERRIDE_MAP(_name, _wait_type) \ + struct lockdep_map __maybe_unused _name = {} + #endif /* !LOCKDEP */ enum xhlock_context_t { @@ -551,6 +564,7 @@ do { \ #define rwsem_release(l, i) lock_release(l, i) #define lock_map_acquire(l) lock_acquire_exclusive(l, 0, 0, NULL, _THIS_IP_) +#define lock_map_acquire_try(l) lock_acquire_exclusive(l, 0, 1, NULL, _THIS_IP_) #define lock_map_acquire_read(l) lock_acquire_shared_recursive(l, 0, 0, NULL, _THIS_IP_) #define lock_map_acquire_tryread(l) lock_acquire_shared_recursive(l, 0, 1, NULL, _THIS_IP_) #define lock_map_release(l) lock_release(l, _THIS_IP_) diff --git a/include/linux/lockdep_types.h b/include/linux/lockdep_types.h index d22430840b53..59f4fb1626ea 100644 --- a/include/linux/lockdep_types.h +++ b/include/linux/lockdep_types.h @@ -33,6 +33,7 @@ enum lockdep_wait_type { enum lockdep_lock_type { LD_LOCK_NORMAL = 0, /* normal, catch all */ LD_LOCK_PERCPU, /* percpu */ + LD_LOCK_WAIT_OVERRIDE, /* annotation */ LD_LOCK_MAX, }; diff --git a/kernel/locking/lockdep.c b/kernel/locking/lockdep.c index 50d4863974e7..62ef295e07e6 100644 --- a/kernel/locking/lockdep.c +++ b/kernel/locking/lockdep.c @@ -2253,6 +2253,9 @@ static inline bool usage_match(struct lock_list *entry, void *mask) static inline bool usage_skip(struct lock_list *entry, void *mask) { + if (entry->class->lock_type == LD_LOCK_NORMAL) + return false; + /* * Skip local_lock() for irq inversion detection. * @@ -2279,14 +2282,16 @@ static inline bool usage_skip(struct lock_list *entry, void *mask) * As a result, we will skip local_lock(), when we search for irq * inversion bugs. */ - if (entry->class->lock_type == LD_LOCK_PERCPU) { - if (DEBUG_LOCKS_WARN_ON(entry->class->wait_type_inner < LD_WAIT_CONFIG)) - return false; + if (entry->class->lock_type == LD_LOCK_PERCPU && + DEBUG_LOCKS_WARN_ON(entry->class->wait_type_inner < LD_WAIT_CONFIG)) + return false; - return true; - } + /* + * Skip WAIT_OVERRIDE for irq inversion detection -- it's not actually + * a lock and only used to override the wait_type. + */ - return false; + return true; } /* @@ -4752,7 +4757,8 @@ static int check_wait_context(struct task_struct *curr, struct held_lock *next) for (; depth < curr->lockdep_depth; depth++) { struct held_lock *prev = curr->held_locks + depth; - u8 prev_inner = hlock_class(prev)->wait_type_inner; + struct lock_class *class = hlock_class(prev); + u8 prev_inner = class->wait_type_inner; if (prev_inner) { /* @@ -4762,6 +4768,14 @@ static int check_wait_context(struct task_struct *curr, struct held_lock *next) * Also due to trylocks. */ curr_inner = min(curr_inner, prev_inner); + + /* + * Allow override for annotations -- this is typically + * only valid/needed for code that only exists when + * CONFIG_PREEMPT_RT=n. + */ + if (unlikely(class->lock_type == LD_LOCK_WAIT_OVERRIDE)) + curr_inner = prev_inner; } } diff --git a/lib/debugobjects.c b/lib/debugobjects.c index 003edc5ebd67..826c617b10a7 100644 --- a/lib/debugobjects.c +++ b/lib/debugobjects.c @@ -591,10 +591,21 @@ static void debug_objects_fill_pool(void) { /* * On RT enabled kernels the pool refill must happen in preemptible - * context: + * context -- for !RT kernels we rely on the fact that spinlock_t and + * raw_spinlock_t are basically the same type and this lock-type + * inversion works just fine. */ - if (!IS_ENABLED(CONFIG_PREEMPT_RT) || preemptible()) + if (!IS_ENABLED(CONFIG_PREEMPT_RT) || preemptible()) { + /* + * Annotate away the spinlock_t inside raw_spinlock_t warning + * by temporarily raising the wait-type to WAIT_SLEEP, matching + * the preemptible() condition above. + */ + static DEFINE_WAIT_OVERRIDE_MAP(fill_pool_map, LD_WAIT_SLEEP); + lock_map_acquire_try(&fill_pool_map); fill_pool(); + lock_map_release(&fill_pool_map); + } } static void From eb799279fb1f9c63c520fe8c1c41cb9154252db6 Mon Sep 17 00:00:00 2001 From: Tetsuo Handa Date: Thu, 11 May 2023 22:47:32 +0900 Subject: [PATCH 2/2] debugobjects: Don't wake up kswapd from fill_pool() syzbot is reporting a lockdep warning in fill_pool() because the allocation from debugobjects is using GFP_ATOMIC, which is (__GFP_HIGH | __GFP_KSWAPD_RECLAIM) and therefore tries to wake up kswapd, which acquires kswapd_wait::lock. Since fill_pool() might be called with arbitrary locks held, fill_pool() should not assume that acquiring kswapd_wait::lock is safe. Use __GFP_HIGH instead and remove __GFP_NORETRY as it is pointless for !__GFP_DIRECT_RECLAIM allocation. Fixes: 3ac7fe5a4aab ("infrastructure to debug (dynamic) objects") Reported-by: syzbot Signed-off-by: Tetsuo Handa Signed-off-by: Thomas Gleixner Link: https://lore.kernel.org/r/6577e1fa-b6ee-f2be-2414-a2b51b1c5e30@I-love.SAKURA.ne.jp Closes: https://syzkaller.appspot.com/bug?extid=fe0c72f0ccbb93786380 --- lib/debugobjects.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/lib/debugobjects.c b/lib/debugobjects.c index 826c617b10a7..984985c39c9b 100644 --- a/lib/debugobjects.c +++ b/lib/debugobjects.c @@ -126,7 +126,7 @@ static const char *obj_states[ODEBUG_STATE_MAX] = { static void fill_pool(void) { - gfp_t gfp = GFP_ATOMIC | __GFP_NORETRY | __GFP_NOWARN; + gfp_t gfp = __GFP_HIGH | __GFP_NOWARN; struct debug_obj *obj; unsigned long flags;