From: =?UTF-8?q?Uwe=20Kleine-K=C3=B6nig?= The patch "list_bl.h: make list head locking RT safe" introduced an unconditional __set_bit(0, (unsigned long *)b); in void hlist_bl_lock(struct hlist_bl_head *b). This clobbers the value of b->first. When the value of b->first is retrieved using hlist_bl_first the clobbering is undone using (unsigned long)h->first & ~LIST_BL_LOCKMASK and so depending on LIST_BL_LOCKMASK being one. But LIST_BL_LOCKMASK is only one if at least on of CONFIG_SMP and CONFIG_DEBUG_SPINLOCK are defined. Without these the value returned by hlist_bl_first has the zeroth bit set which likely results in a crash. So only do the clobbering in the cases where LIST_BL_LOCKMASK is one. An alternative would be to always define LIST_BL_LOCKMASK to one with CONFIG_PREEMPT_RT_BASE. Cc: stable-rt@vger.kernel.org Acked-by: Paul Gortmaker Tested-by: Paul Gortmaker Signed-off-by: Uwe Kleine-König Signed-off-by: Sebastian Andrzej Siewior --- include/linux/list_bl.h | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/include/linux/list_bl.h b/include/linux/list_bl.h index ddfd46a..becd7a6 100644 --- a/include/linux/list_bl.h +++ b/include/linux/list_bl.h @@ -131,8 +131,10 @@ static inline void hlist_bl_lock(struct hlist_bl_head *b) bit_spin_lock(0, (unsigned long *)b); #else raw_spin_lock(&b->lock); +#if defined(CONFIG_SMP) || defined(CONFIG_DEBUG_SPINLOCK) __set_bit(0, (unsigned long *)b); #endif +#endif } static inline void hlist_bl_unlock(struct hlist_bl_head *b) @@ -140,7 +142,9 @@ static inline void hlist_bl_unlock(struct hlist_bl_head *b) #ifndef CONFIG_PREEMPT_RT_BASE __bit_spin_unlock(0, (unsigned long *)b); #else +#if defined(CONFIG_SMP) || defined(CONFIG_DEBUG_SPINLOCK) __clear_bit(0, (unsigned long *)b); +#endif raw_spin_unlock(&b->lock); #endif } -- 1.7.10.4 -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/