| Message ID | 20211123170134.y6xb7pmpgdn4m3bn@linutronix.de (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | [v2] locking: Fixup write_lock_nested() implementation. | expand |
Hi all, On Tue, 23 Nov 2021 18:01:34 +0100 Sebastian Andrzej Siewior <bigeasy@linutronix.de> wrote: > > Andrew, please merge it into: > locking/rwlocks: introduce write_lock_nested > locking-rwlocks-introduce-write_lock_nested.patch > > And if someone could test it, I get sh4 defconfig built with and without > lockdep. x86 seems still to build, too. So it can't be that bad. > > v1…v2: I noticed a typo in _raw_write_lock_nested() and decided that it > is no needed so now it is removed for !CONFIG_INLINE_WRITE_LOCK. > > Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> > --- > include/linux/rwlock_api_smp.h | 1 - > kernel/locking/spinlock.c | 4 ++++ > 2 files changed, 4 insertions(+), 1 deletion(-) > > diff --git a/include/linux/rwlock_api_smp.h b/include/linux/rwlock_api_smp.h > index f0c535ec4e654..dceb0a59b6927 100644 > --- a/include/linux/rwlock_api_smp.h > +++ b/include/linux/rwlock_api_smp.h > @@ -47,7 +47,6 @@ _raw_write_unlock_irqrestore(rwlock_t *lock, unsigned long flags) > > #ifdef CONFIG_INLINE_WRITE_LOCK > #define _raw_write_lock(lock) __raw_write_lock(lock) > -#define _raw_write_lock_nested(lock, subclass) __raw_write_lock_nested(lock, subclass) > #endif > > #ifdef CONFIG_INLINE_READ_LOCK_BH > diff --git a/kernel/locking/spinlock.c b/kernel/locking/spinlock.c > index 996811efa6d6e..7f49baaa49793 100644 > --- a/kernel/locking/spinlock.c > +++ b/kernel/locking/spinlock.c > @@ -301,6 +301,10 @@ void __lockfunc _raw_write_lock(rwlock_t *lock) > } > EXPORT_SYMBOL(_raw_write_lock); > > +#ifndef CONFIG_DEBUG_LOCK_ALLOC > +#define __raw_write_lock_nested(lock, subclass) __raw_write_lock(((void)(subclass), (lock))) > +#endif > + > void __lockfunc _raw_write_lock_nested(rwlock_t *lock, int subclass) > { > __raw_write_lock_nested(lock, subclass); > -- > 2.34.0 > I have added that patch to iinux-next today.
diff --git a/include/linux/rwlock_api_smp.h b/include/linux/rwlock_api_smp.h index f0c535ec4e654..dceb0a59b6927 100644 --- a/include/linux/rwlock_api_smp.h +++ b/include/linux/rwlock_api_smp.h @@ -47,7 +47,6 @@ _raw_write_unlock_irqrestore(rwlock_t *lock, unsigned long flags) #ifdef CONFIG_INLINE_WRITE_LOCK #define _raw_write_lock(lock) __raw_write_lock(lock) -#define _raw_write_lock_nested(lock, subclass) __raw_write_lock_nested(lock, subclass) #endif #ifdef CONFIG_INLINE_READ_LOCK_BH diff --git a/kernel/locking/spinlock.c b/kernel/locking/spinlock.c index 996811efa6d6e..7f49baaa49793 100644 --- a/kernel/locking/spinlock.c +++ b/kernel/locking/spinlock.c @@ -301,6 +301,10 @@ void __lockfunc _raw_write_lock(rwlock_t *lock) } EXPORT_SYMBOL(_raw_write_lock); +#ifndef CONFIG_DEBUG_LOCK_ALLOC +#define __raw_write_lock_nested(lock, subclass) __raw_write_lock(((void)(subclass), (lock))) +#endif + void __lockfunc _raw_write_lock_nested(rwlock_t *lock, int subclass) { __raw_write_lock_nested(lock, subclass);
Andrew, please merge it into: locking/rwlocks: introduce write_lock_nested locking-rwlocks-introduce-write_lock_nested.patch And if someone could test it, I get sh4 defconfig built with and without lockdep. x86 seems still to build, too. So it can't be that bad. v1…v2: I noticed a typo in _raw_write_lock_nested() and decided that it is no needed so now it is removed for !CONFIG_INLINE_WRITE_LOCK. Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> --- include/linux/rwlock_api_smp.h | 1 - kernel/locking/spinlock.c | 4 ++++ 2 files changed, 4 insertions(+), 1 deletion(-)