aboutsummaryrefslogtreecommitdiffstats
path: root/kernel/semaphore.c
diff options
context:
space:
mode:
authorLinus Torvalds <torvalds@linux-foundation.org>2008-05-10 23:43:22 -0400
committerLinus Torvalds <torvalds@linux-foundation.org>2008-05-10 23:43:22 -0400
commit00b41ec2611dc98f87f30753ee00a53db648d662 (patch)
tree431f5019ce8dfd0f9c2b0e3816128dc4a34a49ff /kernel/semaphore.c
parent39f004ba27fcd2431030a3bb3c949fa3f93fa4ca (diff)
Revert "semaphore: fix"
This reverts commit bf726eab3711cf192405d21688a4b21e07b6188a, as it has been reported to cause a regression with processes stuck in __down(), apparently because some missing wakeup. Quoth Sven Wegener: "I'm currently investigating a regression that has showed up with my last git pull yesterday. Bisecting the commits showed bf726e "semaphore: fix" to be the culprit, reverting it fixed the issue. Symptoms: During heavy filesystem usage (e.g. a kernel compile) I get several compiler processes in uninterruptible sleep, blocking all i/o on the filesystem. System is an Intel Core 2 Quad running a 64bit kernel and userspace. Filesystem is xfs on top of lvm. See below for the output of sysrq-w." See http://lkml.org/lkml/2008/5/10/45 for full report. In the meantime, we can just fix the BKL performance regression by reverting back to the good old BKL spinlock implementation instead, since any sleeping lock will generally perform badly, especially if it tries to be fair. Reported-by: Sven Wegener <sven.wegener@stealer.net> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Ingo Molnar <mingo@elte.hu> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Diffstat (limited to 'kernel/semaphore.c')
-rw-r--r--kernel/semaphore.c64
1 files changed, 34 insertions, 30 deletions
diff --git a/kernel/semaphore.c b/kernel/semaphore.c
index 5e41217239e8..5c2942e768cd 100644
--- a/kernel/semaphore.c
+++ b/kernel/semaphore.c
@@ -54,9 +54,10 @@ void down(struct semaphore *sem)
54 unsigned long flags; 54 unsigned long flags;
55 55
56 spin_lock_irqsave(&sem->lock, flags); 56 spin_lock_irqsave(&sem->lock, flags);
57 if (unlikely(!sem->count)) 57 if (likely(sem->count > 0))
58 sem->count--;
59 else
58 __down(sem); 60 __down(sem);
59 sem->count--;
60 spin_unlock_irqrestore(&sem->lock, flags); 61 spin_unlock_irqrestore(&sem->lock, flags);
61} 62}
62EXPORT_SYMBOL(down); 63EXPORT_SYMBOL(down);
@@ -76,10 +77,10 @@ int down_interruptible(struct semaphore *sem)
76 int result = 0; 77 int result = 0;
77 78
78 spin_lock_irqsave(&sem->lock, flags); 79 spin_lock_irqsave(&sem->lock, flags);
79 if (unlikely(!sem->count)) 80 if (likely(sem->count > 0))
80 result = __down_interruptible(sem);
81 if (!result)
82 sem->count--; 81 sem->count--;
82 else
83 result = __down_interruptible(sem);
83 spin_unlock_irqrestore(&sem->lock, flags); 84 spin_unlock_irqrestore(&sem->lock, flags);
84 85
85 return result; 86 return result;
@@ -102,10 +103,10 @@ int down_killable(struct semaphore *sem)
102 int result = 0; 103 int result = 0;
103 104
104 spin_lock_irqsave(&sem->lock, flags); 105 spin_lock_irqsave(&sem->lock, flags);
105 if (unlikely(!sem->count)) 106 if (likely(sem->count > 0))
106 result = __down_killable(sem);
107 if (!result)
108 sem->count--; 107 sem->count--;
108 else
109 result = __down_killable(sem);
109 spin_unlock_irqrestore(&sem->lock, flags); 110 spin_unlock_irqrestore(&sem->lock, flags);
110 111
111 return result; 112 return result;
@@ -156,10 +157,10 @@ int down_timeout(struct semaphore *sem, long jiffies)
156 int result = 0; 157 int result = 0;
157 158
158 spin_lock_irqsave(&sem->lock, flags); 159 spin_lock_irqsave(&sem->lock, flags);
159 if (unlikely(!sem->count)) 160 if (likely(sem->count > 0))
160 result = __down_timeout(sem, jiffies);
161 if (!result)
162 sem->count--; 161 sem->count--;
162 else
163 result = __down_timeout(sem, jiffies);
163 spin_unlock_irqrestore(&sem->lock, flags); 164 spin_unlock_irqrestore(&sem->lock, flags);
164 165
165 return result; 166 return result;
@@ -178,8 +179,9 @@ void up(struct semaphore *sem)
178 unsigned long flags; 179 unsigned long flags;
179 180
180 spin_lock_irqsave(&sem->lock, flags); 181 spin_lock_irqsave(&sem->lock, flags);
181 sem->count++; 182 if (likely(list_empty(&sem->wait_list)))
182 if (unlikely(!list_empty(&sem->wait_list))) 183 sem->count++;
184 else
183 __up(sem); 185 __up(sem);
184 spin_unlock_irqrestore(&sem->lock, flags); 186 spin_unlock_irqrestore(&sem->lock, flags);
185} 187}
@@ -190,6 +192,7 @@ EXPORT_SYMBOL(up);
190struct semaphore_waiter { 192struct semaphore_waiter {
191 struct list_head list; 193 struct list_head list;
192 struct task_struct *task; 194 struct task_struct *task;
195 int up;
193}; 196};
194 197
195/* 198/*
@@ -202,34 +205,33 @@ static inline int __sched __down_common(struct semaphore *sem, long state,
202{ 205{
203 struct task_struct *task = current; 206 struct task_struct *task = current;
204 struct semaphore_waiter waiter; 207 struct semaphore_waiter waiter;
205 int ret = 0;
206 208
207 waiter.task = task;
208 list_add_tail(&waiter.list, &sem->wait_list); 209 list_add_tail(&waiter.list, &sem->wait_list);
210 waiter.task = task;
211 waiter.up = 0;
209 212
210 for (;;) { 213 for (;;) {
211 if (state == TASK_INTERRUPTIBLE && signal_pending(task)) { 214 if (state == TASK_INTERRUPTIBLE && signal_pending(task))
212 ret = -EINTR; 215 goto interrupted;
213 break; 216 if (state == TASK_KILLABLE && fatal_signal_pending(task))
214 } 217 goto interrupted;
215 if (state == TASK_KILLABLE && fatal_signal_pending(task)) { 218 if (timeout <= 0)
216 ret = -EINTR; 219 goto timed_out;
217 break;
218 }
219 if (timeout <= 0) {
220 ret = -ETIME;
221 break;
222 }
223 __set_task_state(task, state); 220 __set_task_state(task, state);
224 spin_unlock_irq(&sem->lock); 221 spin_unlock_irq(&sem->lock);
225 timeout = schedule_timeout(timeout); 222 timeout = schedule_timeout(timeout);
226 spin_lock_irq(&sem->lock); 223 spin_lock_irq(&sem->lock);
227 if (sem->count > 0) 224 if (waiter.up)
228 break; 225 return 0;
229 } 226 }
230 227
228 timed_out:
229 list_del(&waiter.list);
230 return -ETIME;
231
232 interrupted:
231 list_del(&waiter.list); 233 list_del(&waiter.list);
232 return ret; 234 return -EINTR;
233} 235}
234 236
235static noinline void __sched __down(struct semaphore *sem) 237static noinline void __sched __down(struct semaphore *sem)
@@ -256,5 +258,7 @@ static noinline void __sched __up(struct semaphore *sem)
256{ 258{
257 struct semaphore_waiter *waiter = list_first_entry(&sem->wait_list, 259 struct semaphore_waiter *waiter = list_first_entry(&sem->wait_list,
258 struct semaphore_waiter, list); 260 struct semaphore_waiter, list);
261 list_del(&waiter->list);
262 waiter->up = 1;
259 wake_up_process(waiter->task); 263 wake_up_process(waiter->task);
260} 264}