From 8ef1b9e61c04ad0aaee5b1d9eb46c04aaa0b8c4b Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Wed, 22 Feb 2012 12:03:30 +0100 Subject: [PATCH] seqlock: Prevent rt starvation If a low prio writer gets preempted while holding the seqlock write locked, a high prio reader spins forever on RT. To prevent this let the reader grab the spinlock, so it blocks and eventually boosts the writer. This way the writer can proceed and endless spinning is prevented. For seqcount writers we disable preemption over the update code path. Thaanks to Al Viro for distangling some VFS code to make that possible. Signed-off-by: Thomas Gleixner Cc: stable-rt@vger.kernel.org --- include/linux/seqlock.h | 57 ++++++++++++++++++++++++++++++++--------- include/net/dst.h | 2 +- include/net/neighbour.h | 4 +-- 3 files changed, 48 insertions(+), 15 deletions(-) diff --git a/include/linux/seqlock.h b/include/linux/seqlock.h index 8cf350325dc6..cdf56d6f8a4d 100644 --- a/include/linux/seqlock.h +++ b/include/linux/seqlock.h @@ -203,20 +203,30 @@ static inline int read_seqcount_retry(const seqcount_t *s, unsigned start) return __read_seqcount_retry(s, start); } - - -static inline void raw_write_seqcount_begin(seqcount_t *s) +static inline void __raw_write_seqcount_begin(seqcount_t *s) { s->sequence++; smp_wmb(); } -static inline void raw_write_seqcount_end(seqcount_t *s) +static inline void raw_write_seqcount_begin(seqcount_t *s) +{ + preempt_disable_rt(); + __raw_write_seqcount_begin(s); +} + +static inline void __raw_write_seqcount_end(seqcount_t *s) { smp_wmb(); s->sequence++; } +static inline void raw_write_seqcount_end(seqcount_t *s) +{ + __raw_write_seqcount_end(s); + preempt_enable_rt(); +} + /* * Sequence counter only version assumes that callers are using their * own mutexing. @@ -278,10 +288,33 @@ typedef struct { /* * Read side functions for starting and finalizing a read side section. */ +#ifndef CONFIG_PREEMPT_RT_FULL static inline unsigned read_seqbegin(const seqlock_t *sl) { return read_seqcount_begin(&sl->seqcount); } +#else +/* + * Starvation safe read side for RT + */ +static inline unsigned read_seqbegin(seqlock_t *sl) +{ + unsigned ret; + +repeat: + ret = ACCESS_ONCE(sl->seqcount.sequence); + if (unlikely(ret & 1)) { + /* + * Take the lock and let the writer proceed (i.e. evtl + * boost it), otherwise we could loop here forever. + */ + spin_lock(&sl->lock); + spin_unlock(&sl->lock); + goto repeat; + } + return ret; +} +#endif static inline unsigned read_seqretry(const seqlock_t *sl, unsigned start) { @@ -296,36 +329,36 @@ static inline unsigned read_seqretry(const seqlock_t *sl, unsigned start) static inline void write_seqlock(seqlock_t *sl) { spin_lock(&sl->lock); - write_seqcount_begin(&sl->seqcount); + __write_seqcount_begin(&sl->seqcount); } static inline void write_sequnlock(seqlock_t *sl) { - write_seqcount_end(&sl->seqcount); + __raw_write_seqcount_end(&sl->seqcount); spin_unlock(&sl->lock); } static inline void write_seqlock_bh(seqlock_t *sl) { spin_lock_bh(&sl->lock); - write_seqcount_begin(&sl->seqcount); + __write_seqcount_begin(&sl->seqcount); } static inline void write_sequnlock_bh(seqlock_t *sl) { - write_seqcount_end(&sl->seqcount); + __raw_write_seqcount_end(&sl->seqcount); spin_unlock_bh(&sl->lock); } static inline void write_seqlock_irq(seqlock_t *sl) { spin_lock_irq(&sl->lock); - write_seqcount_begin(&sl->seqcount); + __write_seqcount_begin(&sl->seqcount); } static inline void write_sequnlock_irq(seqlock_t *sl) { - write_seqcount_end(&sl->seqcount); + __raw_write_seqcount_end(&sl->seqcount); spin_unlock_irq(&sl->lock); } @@ -334,7 +367,7 @@ static inline unsigned long __write_seqlock_irqsave(seqlock_t *sl) unsigned long flags; spin_lock_irqsave(&sl->lock, flags); - write_seqcount_begin(&sl->seqcount); + __write_seqcount_begin(&sl->seqcount); return flags; } @@ -344,7 +377,7 @@ static inline unsigned long __write_seqlock_irqsave(seqlock_t *sl) static inline void write_sequnlock_irqrestore(seqlock_t *sl, unsigned long flags) { - write_seqcount_end(&sl->seqcount); + __raw_write_seqcount_end(&sl->seqcount); spin_unlock_irqrestore(&sl->lock, flags); } diff --git a/include/net/dst.h b/include/net/dst.h index 909032821c37..3a53a5c02326 100644 --- a/include/net/dst.h +++ b/include/net/dst.h @@ -393,7 +393,7 @@ static inline void dst_confirm(struct dst_entry *dst) static inline int dst_neigh_output(struct dst_entry *dst, struct neighbour *n, struct sk_buff *skb) { - const struct hh_cache *hh; + struct hh_cache *hh; if (dst->pending_confirm) { unsigned long now = jiffies; diff --git a/include/net/neighbour.h b/include/net/neighbour.h index 7277caf3743d..50e42d4dd114 100644 --- a/include/net/neighbour.h +++ b/include/net/neighbour.h @@ -388,7 +388,7 @@ static inline int neigh_hh_bridge(struct hh_cache *hh, struct sk_buff *skb) } #endif -static inline int neigh_hh_output(const struct hh_cache *hh, struct sk_buff *skb) +static inline int neigh_hh_output(struct hh_cache *hh, struct sk_buff *skb) { unsigned int seq; int hh_len; @@ -443,7 +443,7 @@ struct neighbour_cb { #define NEIGH_CB(skb) ((struct neighbour_cb *)(skb)->cb) -static inline void neigh_ha_snapshot(char *dst, const struct neighbour *n, +static inline void neigh_ha_snapshot(char *dst, struct neighbour *n, const struct net_device *dev) { unsigned int seq;