Messages in this thread | | | Date | Mon, 16 Jan 2023 17:21:09 -0500 | Subject | Re: [PATCH 4/3] locking/lockdep: Improve the deadlock scenario print for sync and read lock | From | Waiman Long <> |
| |
On 1/13/23 18:57, Boqun Feng wrote: > Lock scenario print is always a weak spot of lockdep splats. Improvement > can be made if we rework the dependency search and the error printing. > > However without touching the graph search, we can improve a little for > the circular deadlock case, since we have the to-be-added lock > dependency, and know whether these two locks are read/write/sync. > > In order to know whether a held_lock is sync or not, a bit was > "stolen" from ->references, which reduce our limit for the same lock > class nesting from 2^12 to 2^11, and it should still be good enough. > > Besides, since we now have bit in held_lock for sync, we don't need the > "hardirqoffs being 1" trick, and also we can avoid the __lock_release() > if we jump out of __lock_acquire() before the held_lock stored. > > With these changes, a deadlock case evolved with read lock and sync gets > a better print-out from: > > [...] Possible unsafe locking scenario: > [...] > [...] CPU0 CPU1 > [...] ---- ---- > [...] lock(srcuA); > [...] lock(srcuB); > [...] lock(srcuA); > [...] lock(srcuB); > > to > > [...] Possible unsafe locking scenario: > [...] > [...] CPU0 CPU1 > [...] ---- ---- > [...] rlock(srcuA); > [...] lock(srcuB); > [...] lock(srcuA); > [...] sync(srcuB); > > Signed-off-by: Boqun Feng <boqun.feng@gmail.com> > --- > include/linux/lockdep.h | 3 ++- > kernel/locking/lockdep.c | 48 ++++++++++++++++++++++++++-------------- > 2 files changed, 34 insertions(+), 17 deletions(-) > > diff --git a/include/linux/lockdep.h b/include/linux/lockdep.h > index ba09df6a0872..febd7ecc225c 100644 > --- a/include/linux/lockdep.h > +++ b/include/linux/lockdep.h > @@ -134,7 +134,8 @@ struct held_lock { > unsigned int read:2; /* see lock_acquire() comment */ > unsigned int check:1; /* see lock_acquire() comment */ > unsigned int hardirqs_off:1; > - unsigned int references:12; /* 32 bits */ > + unsigned int sync:1; > + unsigned int references:11; /* 32 bits */ > unsigned int pin_count; > }; > > diff --git a/kernel/locking/lockdep.c b/kernel/locking/lockdep.c > index cffa026a765f..4031d87f6829 100644 > --- a/kernel/locking/lockdep.c > +++ b/kernel/locking/lockdep.c > @@ -1880,6 +1880,8 @@ print_circular_lock_scenario(struct held_lock *src, > struct lock_class *source = hlock_class(src); > struct lock_class *target = hlock_class(tgt); > struct lock_class *parent = prt->class; > + int src_read = src->read; > + int tgt_read = tgt->read; > > /* > * A direct locking problem where unsafe_class lock is taken > @@ -1907,7 +1909,10 @@ print_circular_lock_scenario(struct held_lock *src, > printk(" Possible unsafe locking scenario:\n\n"); > printk(" CPU0 CPU1\n"); > printk(" ---- ----\n"); > - printk(" lock("); > + if (tgt_read != 0) > + printk(" rlock("); > + else > + printk(" lock("); > __print_lock_name(target); > printk(KERN_CONT ");\n"); > printk(" lock("); > @@ -1916,7 +1921,12 @@ print_circular_lock_scenario(struct held_lock *src, > printk(" lock("); > __print_lock_name(target); > printk(KERN_CONT ");\n"); > - printk(" lock("); > + if (src_read != 0) > + printk(" rlock("); > + else if (src->sync) > + printk(" sync("); > + else > + printk(" lock("); > __print_lock_name(source); > printk(KERN_CONT ");\n"); > printk("\n *** DEADLOCK ***\n\n");
src can be sync() but not the target. Is there a reason why that is the case?
> @@ -4530,7 +4540,13 @@ mark_usage(struct task_struct *curr, struct held_lock *hlock, int check) > return 0; > } > } > - if (!hlock->hardirqs_off) { > + > + /* > + * For lock_sync(), don't mark the ENABLED usage, since lock_sync() > + * creates no critical section and no extra dependency can be introduced > + * by interrupts > + */ > + if (!hlock->hardirqs_off && !hlock->sync) { > if (hlock->read) { > if (!mark_lock(curr, hlock, > LOCK_ENABLED_HARDIRQ_READ)) > @@ -4909,7 +4925,7 @@ static int __lock_is_held(const struct lockdep_map *lock, int read); > static int __lock_acquire(struct lockdep_map *lock, unsigned int subclass, > int trylock, int read, int check, int hardirqs_off, > struct lockdep_map *nest_lock, unsigned long ip, > - int references, int pin_count) > + int references, int pin_count, int sync) > { > struct task_struct *curr = current; > struct lock_class *class = NULL; > @@ -4960,7 +4976,8 @@ static int __lock_acquire(struct lockdep_map *lock, unsigned int subclass, > > class_idx = class - lock_classes; > > - if (depth) { /* we're holding locks */ > + if (depth && !sync) { > + /* we're holding locks and the new held lock is not a sync */ > hlock = curr->held_locks + depth - 1; > if (hlock->class_idx == class_idx && nest_lock) { > if (!references) > @@ -4994,6 +5011,7 @@ static int __lock_acquire(struct lockdep_map *lock, unsigned int subclass, > hlock->trylock = trylock; > hlock->read = read; > hlock->check = check; > + hlock->sync = !!sync; > hlock->hardirqs_off = !!hardirqs_off; > hlock->references = references; > #ifdef CONFIG_LOCK_STAT > @@ -5055,6 +5073,10 @@ static int __lock_acquire(struct lockdep_map *lock, unsigned int subclass, > if (!validate_chain(curr, hlock, chain_head, chain_key)) > return 0; > > + /* For lock_sync(), we are done here since no actual critical section */ > + if (hlock->sync) > + return 1; > + > curr->curr_chain_key = chain_key; > curr->lockdep_depth++; > check_chain_key(curr);
Even with sync, there is still a corresponding lock_acquire() and lock_release(), you can't exit here without increasing lockdep_depth. That can cause underflow.
Cheers, Longman
| |