diff options
Diffstat (limited to 'kernel')
-rw-r--r-- | kernel/cgroup.c | 8 | ||||
-rw-r--r-- | kernel/exit.c | 2 | ||||
-rw-r--r-- | kernel/extable.c | 16 | ||||
-rw-r--r-- | kernel/futex.c | 351 | ||||
-rw-r--r-- | kernel/irq/manage.c | 12 | ||||
-rw-r--r-- | kernel/lockdep.c | 60 | ||||
-rw-r--r-- | kernel/lockdep_proc.c | 28 | ||||
-rw-r--r-- | kernel/mutex.c | 10 | ||||
-rw-r--r-- | kernel/notifier.c | 8 | ||||
-rw-r--r-- | kernel/panic.c | 32 | ||||
-rw-r--r-- | kernel/posix-cpu-timers.c | 10 | ||||
-rw-r--r-- | kernel/posix-timers.c | 6 | ||||
-rw-r--r-- | kernel/printk.c | 2 | ||||
-rw-r--r-- | kernel/resource.c | 9 | ||||
-rw-r--r-- | kernel/sched.c | 3 | ||||
-rw-r--r-- | kernel/softirq.c | 14 | ||||
-rw-r--r-- | kernel/softlockup.c | 2 | ||||
-rw-r--r-- | kernel/stacktrace.c | 11 | ||||
-rw-r--r-- | kernel/sys.c | 2 |
19 files changed, 283 insertions, 303 deletions
diff --git a/kernel/cgroup.c b/kernel/cgroup.c index 8185a0f0959..2606d0fb4e5 100644 --- a/kernel/cgroup.c +++ b/kernel/cgroup.c | |||
@@ -1024,7 +1024,7 @@ static int cgroup_get_sb(struct file_system_type *fs_type, | |||
1024 | if (ret == -EBUSY) { | 1024 | if (ret == -EBUSY) { |
1025 | mutex_unlock(&cgroup_mutex); | 1025 | mutex_unlock(&cgroup_mutex); |
1026 | mutex_unlock(&inode->i_mutex); | 1026 | mutex_unlock(&inode->i_mutex); |
1027 | goto drop_new_super; | 1027 | goto free_cg_links; |
1028 | } | 1028 | } |
1029 | 1029 | ||
1030 | /* EBUSY should be the only error here */ | 1030 | /* EBUSY should be the only error here */ |
@@ -1073,10 +1073,11 @@ static int cgroup_get_sb(struct file_system_type *fs_type, | |||
1073 | 1073 | ||
1074 | return simple_set_mnt(mnt, sb); | 1074 | return simple_set_mnt(mnt, sb); |
1075 | 1075 | ||
1076 | free_cg_links: | ||
1077 | free_cg_links(&tmp_cg_links); | ||
1076 | drop_new_super: | 1078 | drop_new_super: |
1077 | up_write(&sb->s_umount); | 1079 | up_write(&sb->s_umount); |
1078 | deactivate_super(sb); | 1080 | deactivate_super(sb); |
1079 | free_cg_links(&tmp_cg_links); | ||
1080 | return ret; | 1081 | return ret; |
1081 | } | 1082 | } |
1082 | 1083 | ||
@@ -2934,9 +2935,6 @@ int cgroup_clone(struct task_struct *tsk, struct cgroup_subsys *subsys, | |||
2934 | again: | 2935 | again: |
2935 | root = subsys->root; | 2936 | root = subsys->root; |
2936 | if (root == &rootnode) { | 2937 | if (root == &rootnode) { |
2937 | printk(KERN_INFO | ||
2938 | "Not cloning cgroup for unused subsystem %s\n", | ||
2939 | subsys->name); | ||
2940 | mutex_unlock(&cgroup_mutex); | 2938 | mutex_unlock(&cgroup_mutex); |
2941 | return 0; | 2939 | return 0; |
2942 | } | 2940 | } |
diff --git a/kernel/exit.c b/kernel/exit.c index 2d8be7ebb0f..30fcdf16737 100644 --- a/kernel/exit.c +++ b/kernel/exit.c | |||
@@ -1321,10 +1321,10 @@ static int wait_task_zombie(struct task_struct *p, int options, | |||
1321 | * group, which consolidates times for all threads in the | 1321 | * group, which consolidates times for all threads in the |
1322 | * group including the group leader. | 1322 | * group including the group leader. |
1323 | */ | 1323 | */ |
1324 | thread_group_cputime(p, &cputime); | ||
1324 | spin_lock_irq(&p->parent->sighand->siglock); | 1325 | spin_lock_irq(&p->parent->sighand->siglock); |
1325 | psig = p->parent->signal; | 1326 | psig = p->parent->signal; |
1326 | sig = p->signal; | 1327 | sig = p->signal; |
1327 | thread_group_cputime(p, &cputime); | ||
1328 | psig->cutime = | 1328 | psig->cutime = |
1329 | cputime_add(psig->cutime, | 1329 | cputime_add(psig->cutime, |
1330 | cputime_add(cputime.utime, | 1330 | cputime_add(cputime.utime, |
diff --git a/kernel/extable.c b/kernel/extable.c index a26cb2e1702..adf0cc9c02d 100644 --- a/kernel/extable.c +++ b/kernel/extable.c | |||
@@ -66,3 +66,19 @@ int kernel_text_address(unsigned long addr) | |||
66 | return 1; | 66 | return 1; |
67 | return module_text_address(addr) != NULL; | 67 | return module_text_address(addr) != NULL; |
68 | } | 68 | } |
69 | |||
70 | /* | ||
71 | * On some architectures (PPC64, IA64) function pointers | ||
72 | * are actually only tokens to some data that then holds the | ||
73 | * real function address. As a result, to find if a function | ||
74 | * pointer is part of the kernel text, we need to do some | ||
75 | * special dereferencing first. | ||
76 | */ | ||
77 | int func_ptr_is_kernel_text(void *ptr) | ||
78 | { | ||
79 | unsigned long addr; | ||
80 | addr = (unsigned long) dereference_function_descriptor(ptr); | ||
81 | if (core_kernel_text(addr)) | ||
82 | return 1; | ||
83 | return module_text_address(addr) != NULL; | ||
84 | } | ||
diff --git a/kernel/futex.c b/kernel/futex.c index 8af10027514..b4f87bac91c 100644 --- a/kernel/futex.c +++ b/kernel/futex.c | |||
@@ -92,11 +92,12 @@ struct futex_pi_state { | |||
92 | * A futex_q has a woken state, just like tasks have TASK_RUNNING. | 92 | * A futex_q has a woken state, just like tasks have TASK_RUNNING. |
93 | * It is considered woken when plist_node_empty(&q->list) || q->lock_ptr == 0. | 93 | * It is considered woken when plist_node_empty(&q->list) || q->lock_ptr == 0. |
94 | * The order of wakup is always to make the first condition true, then | 94 | * The order of wakup is always to make the first condition true, then |
95 | * wake up q->waiters, then make the second condition true. | 95 | * wake up q->waiter, then make the second condition true. |
96 | */ | 96 | */ |
97 | struct futex_q { | 97 | struct futex_q { |
98 | struct plist_node list; | 98 | struct plist_node list; |
99 | wait_queue_head_t waiters; | 99 | /* There can only be a single waiter */ |
100 | wait_queue_head_t waiter; | ||
100 | 101 | ||
101 | /* Which hash list lock to use: */ | 102 | /* Which hash list lock to use: */ |
102 | spinlock_t *lock_ptr; | 103 | spinlock_t *lock_ptr; |
@@ -123,24 +124,6 @@ struct futex_hash_bucket { | |||
123 | static struct futex_hash_bucket futex_queues[1<<FUTEX_HASHBITS]; | 124 | static struct futex_hash_bucket futex_queues[1<<FUTEX_HASHBITS]; |
124 | 125 | ||
125 | /* | 126 | /* |
126 | * Take mm->mmap_sem, when futex is shared | ||
127 | */ | ||
128 | static inline void futex_lock_mm(struct rw_semaphore *fshared) | ||
129 | { | ||
130 | if (fshared) | ||
131 | down_read(fshared); | ||
132 | } | ||
133 | |||
134 | /* | ||
135 | * Release mm->mmap_sem, when the futex is shared | ||
136 | */ | ||
137 | static inline void futex_unlock_mm(struct rw_semaphore *fshared) | ||
138 | { | ||
139 | if (fshared) | ||
140 | up_read(fshared); | ||
141 | } | ||
142 | |||
143 | /* | ||
144 | * We hash on the keys returned from get_futex_key (see below). | 127 | * We hash on the keys returned from get_futex_key (see below). |
145 | */ | 128 | */ |
146 | static struct futex_hash_bucket *hash_futex(union futex_key *key) | 129 | static struct futex_hash_bucket *hash_futex(union futex_key *key) |
@@ -161,6 +144,45 @@ static inline int match_futex(union futex_key *key1, union futex_key *key2) | |||
161 | && key1->both.offset == key2->both.offset); | 144 | && key1->both.offset == key2->both.offset); |
162 | } | 145 | } |
163 | 146 | ||
147 | /* | ||
148 | * Take a reference to the resource addressed by a key. | ||
149 | * Can be called while holding spinlocks. | ||
150 | * | ||
151 | */ | ||
152 | static void get_futex_key_refs(union futex_key *key) | ||
153 | { | ||
154 | if (!key->both.ptr) | ||
155 | return; | ||
156 | |||
157 | switch (key->both.offset & (FUT_OFF_INODE|FUT_OFF_MMSHARED)) { | ||
158 | case FUT_OFF_INODE: | ||
159 | atomic_inc(&key->shared.inode->i_count); | ||
160 | break; | ||
161 | case FUT_OFF_MMSHARED: | ||
162 | atomic_inc(&key->private.mm->mm_count); | ||
163 | break; | ||
164 | } | ||
165 | } | ||
166 | |||
167 | /* | ||
168 | * Drop a reference to the resource addressed by a key. | ||
169 | * The hash bucket spinlock must not be held. | ||
170 | */ | ||
171 | static void drop_futex_key_refs(union futex_key *key) | ||
172 | { | ||
173 | if (!key->both.ptr) | ||
174 | return; | ||
175 | |||
176 | switch (key->both.offset & (FUT_OFF_INODE|FUT_OFF_MMSHARED)) { | ||
177 | case FUT_OFF_INODE: | ||
178 | iput(key->shared.inode); | ||
179 | break; | ||
180 | case FUT_OFF_MMSHARED: | ||
181 | mmdrop(key->private.mm); | ||
182 | break; | ||
183 | } | ||
184 | } | ||
185 | |||
164 | /** | 186 | /** |
165 | * get_futex_key - Get parameters which are the keys for a futex. | 187 | * get_futex_key - Get parameters which are the keys for a futex. |
166 | * @uaddr: virtual address of the futex | 188 | * @uaddr: virtual address of the futex |
@@ -179,12 +201,10 @@ static inline int match_futex(union futex_key *key1, union futex_key *key2) | |||
179 | * For other futexes, it points to ¤t->mm->mmap_sem and | 201 | * For other futexes, it points to ¤t->mm->mmap_sem and |
180 | * caller must have taken the reader lock. but NOT any spinlocks. | 202 | * caller must have taken the reader lock. but NOT any spinlocks. |
181 | */ | 203 | */ |
182 | static int get_futex_key(u32 __user *uaddr, struct rw_semaphore *fshared, | 204 | static int get_futex_key(u32 __user *uaddr, int fshared, union futex_key *key) |
183 | union futex_key *key) | ||
184 | { | 205 | { |
185 | unsigned long address = (unsigned long)uaddr; | 206 | unsigned long address = (unsigned long)uaddr; |
186 | struct mm_struct *mm = current->mm; | 207 | struct mm_struct *mm = current->mm; |
187 | struct vm_area_struct *vma; | ||
188 | struct page *page; | 208 | struct page *page; |
189 | int err; | 209 | int err; |
190 | 210 | ||
@@ -208,100 +228,50 @@ static int get_futex_key(u32 __user *uaddr, struct rw_semaphore *fshared, | |||
208 | return -EFAULT; | 228 | return -EFAULT; |
209 | key->private.mm = mm; | 229 | key->private.mm = mm; |
210 | key->private.address = address; | 230 | key->private.address = address; |
231 | get_futex_key_refs(key); | ||
211 | return 0; | 232 | return 0; |
212 | } | 233 | } |
213 | /* | ||
214 | * The futex is hashed differently depending on whether | ||
215 | * it's in a shared or private mapping. So check vma first. | ||
216 | */ | ||
217 | vma = find_extend_vma(mm, address); | ||
218 | if (unlikely(!vma)) | ||
219 | return -EFAULT; | ||
220 | 234 | ||
221 | /* | 235 | again: |
222 | * Permissions. | 236 | err = get_user_pages_fast(address, 1, 0, &page); |
223 | */ | 237 | if (err < 0) |
224 | if (unlikely((vma->vm_flags & (VM_IO|VM_READ)) != VM_READ)) | 238 | return err; |
225 | return (vma->vm_flags & VM_IO) ? -EPERM : -EACCES; | 239 | |
240 | lock_page(page); | ||
241 | if (!page->mapping) { | ||
242 | unlock_page(page); | ||
243 | put_page(page); | ||
244 | goto again; | ||
245 | } | ||
226 | 246 | ||
227 | /* | 247 | /* |
228 | * Private mappings are handled in a simple way. | 248 | * Private mappings are handled in a simple way. |
229 | * | 249 | * |
230 | * NOTE: When userspace waits on a MAP_SHARED mapping, even if | 250 | * NOTE: When userspace waits on a MAP_SHARED mapping, even if |
231 | * it's a read-only handle, it's expected that futexes attach to | 251 | * it's a read-only handle, it's expected that futexes attach to |
232 | * the object not the particular process. Therefore we use | 252 | * the object not the particular process. |
233 | * VM_MAYSHARE here, not VM_SHARED which is restricted to shared | ||
234 | * mappings of _writable_ handles. | ||
235 | */ | 253 | */ |
236 | if (likely(!(vma->vm_flags & VM_MAYSHARE))) { | 254 | if (PageAnon(page)) { |
237 | key->both.offset |= FUT_OFF_MMSHARED; /* reference taken on mm */ | 255 | key->both.offset |= FUT_OFF_MMSHARED; /* ref taken on mm */ |
238 | key->private.mm = mm; | 256 | key->private.mm = mm; |
239 | key->private.address = address; | 257 | key->private.address = address; |
240 | return 0; | 258 | } else { |
259 | key->both.offset |= FUT_OFF_INODE; /* inode-based key */ | ||
260 | key->shared.inode = page->mapping->host; | ||
261 | key->shared.pgoff = page->index; | ||
241 | } | 262 | } |
242 | 263 | ||
243 | /* | 264 | get_futex_key_refs(key); |
244 | * Linear file mappings are also simple. | ||
245 | */ | ||
246 | key->shared.inode = vma->vm_file->f_path.dentry->d_inode; | ||
247 | key->both.offset |= FUT_OFF_INODE; /* inode-based key. */ | ||
248 | if (likely(!(vma->vm_flags & VM_NONLINEAR))) { | ||
249 | key->shared.pgoff = (((address - vma->vm_start) >> PAGE_SHIFT) | ||
250 | + vma->vm_pgoff); | ||
251 | return 0; | ||
252 | } | ||
253 | 265 | ||
254 | /* | 266 | unlock_page(page); |
255 | * We could walk the page table to read the non-linear | 267 | put_page(page); |
256 | * pte, and get the page index without fetching the page | 268 | return 0; |
257 | * from swap. But that's a lot of code to duplicate here | ||
258 | * for a rare case, so we simply fetch the page. | ||
259 | */ | ||
260 | err = get_user_pages(current, mm, address, 1, 0, 0, &page, NULL); | ||
261 | if (err >= 0) { | ||
262 | key->shared.pgoff = | ||
263 | page->index << (PAGE_CACHE_SHIFT - PAGE_SHIFT); | ||
264 | put_page(page); | ||
265 | return 0; | ||
266 | } | ||
267 | return err; | ||
268 | } | ||
269 | |||
270 | /* | ||
271 | * Take a reference to the resource addressed by a key. | ||
272 | * Can be called while holding spinlocks. | ||
273 | * | ||
274 | */ | ||
275 | static void get_futex_key_refs(union futex_key *key) | ||
276 | { | ||
277 | if (key->both.ptr == NULL) | ||
278 | return; | ||
279 | switch (key->both.offset & (FUT_OFF_INODE|FUT_OFF_MMSHARED)) { | ||
280 | case FUT_OFF_INODE: | ||
281 | atomic_inc(&key->shared.inode->i_count); | ||
282 | break; | ||
283 | case FUT_OFF_MMSHARED: | ||
284 | atomic_inc(&key->private.mm->mm_count); | ||
285 | break; | ||
286 | } | ||
287 | } | 269 | } |
288 | 270 | ||
289 | /* | 271 | static inline |
290 | * Drop a reference to the resource addressed by a key. | 272 | void put_futex_key(int fshared, union futex_key *key) |
291 | * The hash bucket spinlock must not be held. | ||
292 | */ | ||
293 | static void drop_futex_key_refs(union futex_key *key) | ||
294 | { | 273 | { |
295 | if (!key->both.ptr) | 274 | drop_futex_key_refs(key); |
296 | return; | ||
297 | switch (key->both.offset & (FUT_OFF_INODE|FUT_OFF_MMSHARED)) { | ||
298 | case FUT_OFF_INODE: | ||
299 | iput(key->shared.inode); | ||
300 | break; | ||
301 | case FUT_OFF_MMSHARED: | ||
302 | mmdrop(key->private.mm); | ||
303 | break; | ||
304 | } | ||
305 | } | 275 | } |
306 | 276 | ||
307 | static u32 cmpxchg_futex_value_locked(u32 __user *uaddr, u32 uval, u32 newval) | 277 | static u32 cmpxchg_futex_value_locked(u32 __user *uaddr, u32 uval, u32 newval) |
@@ -328,10 +298,8 @@ static int get_futex_value_locked(u32 *dest, u32 __user *from) | |||
328 | 298 | ||
329 | /* | 299 | /* |
330 | * Fault handling. | 300 | * Fault handling. |
331 | * if fshared is non NULL, current->mm->mmap_sem is already held | ||
332 | */ | 301 | */ |
333 | static int futex_handle_fault(unsigned long address, | 302 | static int futex_handle_fault(unsigned long address, int attempt) |
334 | struct rw_semaphore *fshared, int attempt) | ||
335 | { | 303 | { |
336 | struct vm_area_struct * vma; | 304 | struct vm_area_struct * vma; |
337 | struct mm_struct *mm = current->mm; | 305 | struct mm_struct *mm = current->mm; |
@@ -340,8 +308,7 @@ static int futex_handle_fault(unsigned long address, | |||
340 | if (attempt > 2) | 308 | if (attempt > 2) |
341 | return ret; | 309 | return ret; |
342 | 310 | ||
343 | if (!fshared) | 311 | down_read(&mm->mmap_sem); |
344 | down_read(&mm->mmap_sem); | ||
345 | vma = find_vma(mm, address); | 312 | vma = find_vma(mm, address); |
346 | if (vma && address >= vma->vm_start && | 313 | if (vma && address >= vma->vm_start && |
347 | (vma->vm_flags & VM_WRITE)) { | 314 | (vma->vm_flags & VM_WRITE)) { |
@@ -361,8 +328,7 @@ static int futex_handle_fault(unsigned long address, | |||
361 | current->min_flt++; | 328 | current->min_flt++; |
362 | } | 329 | } |
363 | } | 330 | } |
364 | if (!fshared) | 331 | up_read(&mm->mmap_sem); |
365 | up_read(&mm->mmap_sem); | ||
366 | return ret; | 332 | return ret; |
367 | } | 333 | } |
368 | 334 | ||
@@ -385,6 +351,7 @@ static int refill_pi_state_cache(void) | |||
385 | /* pi_mutex gets initialized later */ | 351 | /* pi_mutex gets initialized later */ |
386 | pi_state->owner = NULL; | 352 | pi_state->owner = NULL; |
387 | atomic_set(&pi_state->refcount, 1); | 353 | atomic_set(&pi_state->refcount, 1); |
354 | pi_state->key = FUTEX_KEY_INIT; | ||
388 | 355 | ||
389 | current->pi_state_cache = pi_state; | 356 | current->pi_state_cache = pi_state; |
390 | 357 | ||
@@ -462,7 +429,7 @@ void exit_pi_state_list(struct task_struct *curr) | |||
462 | struct list_head *next, *head = &curr->pi_state_list; | 429 | struct list_head *next, *head = &curr->pi_state_list; |
463 | struct futex_pi_state *pi_state; | 430 | struct futex_pi_state *pi_state; |
464 | struct futex_hash_bucket *hb; | 431 | struct futex_hash_bucket *hb; |
465 | union futex_key key; | 432 | union futex_key key = FUTEX_KEY_INIT; |
466 | 433 | ||
467 | if (!futex_cmpxchg_enabled) | 434 | if (!futex_cmpxchg_enabled) |
468 | return; | 435 | return; |
@@ -607,7 +574,7 @@ static void wake_futex(struct futex_q *q) | |||
607 | * The lock in wake_up_all() is a crucial memory barrier after the | 574 | * The lock in wake_up_all() is a crucial memory barrier after the |
608 | * plist_del() and also before assigning to q->lock_ptr. | 575 | * plist_del() and also before assigning to q->lock_ptr. |
609 | */ | 576 | */ |
610 | wake_up_all(&q->waiters); | 577 | wake_up(&q->waiter); |
611 | /* | 578 | /* |
612 | * The waiting task can free the futex_q as soon as this is written, | 579 | * The waiting task can free the futex_q as soon as this is written, |
613 | * without taking any locks. This must come last. | 580 | * without taking any locks. This must come last. |
@@ -719,20 +686,17 @@ double_lock_hb(struct futex_hash_bucket *hb1, struct futex_hash_bucket *hb2) | |||
719 | * Wake up all waiters hashed on the physical page that is mapped | 686 | * Wake up all waiters hashed on the physical page that is mapped |
720 | * to this virtual address: | 687 | * to this virtual address: |
721 | */ | 688 | */ |
722 | static int futex_wake(u32 __user *uaddr, struct rw_semaphore *fshared, | 689 | static int futex_wake(u32 __user *uaddr, int fshared, int nr_wake, u32 bitset) |
723 | int nr_wake, u32 bitset) | ||
724 | { | 690 | { |
725 | struct futex_hash_bucket *hb; | 691 | struct futex_hash_bucket *hb; |
726 | struct futex_q *this, *next; | 692 | struct futex_q *this, *next; |
727 | struct plist_head *head; | 693 | struct plist_head *head; |
728 | union futex_key key; | 694 | union futex_key key = FUTEX_KEY_INIT; |
729 | int ret; | 695 | int ret; |
730 | 696 | ||
731 | if (!bitset) | 697 | if (!bitset) |
732 | return -EINVAL; | 698 | return -EINVAL; |
733 | 699 | ||
734 | futex_lock_mm(fshared); | ||
735 | |||
736 | ret = get_futex_key(uaddr, fshared, &key); | 700 | ret = get_futex_key(uaddr, fshared, &key); |
737 | if (unlikely(ret != 0)) | 701 | if (unlikely(ret != 0)) |
738 | goto out; | 702 | goto out; |
@@ -760,7 +724,7 @@ static int futex_wake(u32 __user *uaddr, struct rw_semaphore *fshared, | |||
760 | 724 | ||
761 | spin_unlock(&hb->lock); | 725 | spin_unlock(&hb->lock); |
762 | out: | 726 | out: |
763 | futex_unlock_mm(fshared); | 727 | put_futex_key(fshared, &key); |
764 | return ret; | 728 | return ret; |
765 | } | 729 | } |
766 | 730 | ||
@@ -769,19 +733,16 @@ out: | |||
769 | * to this virtual address: | 733 | * to this virtual address: |
770 | */ | 734 | */ |
771 | static int | 735 | static int |
772 | futex_wake_op(u32 __user *uaddr1, struct rw_semaphore *fshared, | 736 | futex_wake_op(u32 __user *uaddr1, int fshared, u32 __user *uaddr2, |
773 | u32 __user *uaddr2, | ||
774 | int nr_wake, int nr_wake2, int op) | 737 | int nr_wake, int nr_wake2, int op) |
775 | { | 738 | { |
776 | union futex_key key1, key2; | 739 | union futex_key key1 = FUTEX_KEY_INIT, key2 = FUTEX_KEY_INIT; |
777 | struct futex_hash_bucket *hb1, *hb2; | 740 | struct futex_hash_bucket *hb1, *hb2; |
778 | struct plist_head *head; | 741 | struct plist_head *head; |
779 | struct futex_q *this, *next; | 742 | struct futex_q *this, *next; |
780 | int ret, op_ret, attempt = 0; | 743 | int ret, op_ret, attempt = 0; |
781 | 744 | ||
782 | retryfull: | 745 | retryfull: |
783 | futex_lock_mm(fshared); | ||
784 | |||
785 | ret = get_futex_key(uaddr1, fshared, &key1); | 746 | ret = get_futex_key(uaddr1, fshared, &key1); |
786 | if (unlikely(ret != 0)) | 747 | if (unlikely(ret != 0)) |
787 | goto out; | 748 | goto out; |
@@ -826,18 +787,12 @@ retry: | |||
826 | */ | 787 | */ |
827 | if (attempt++) { | 788 | if (attempt++) { |
828 | ret = futex_handle_fault((unsigned long)uaddr2, | 789 | ret = futex_handle_fault((unsigned long)uaddr2, |
829 | fshared, attempt); | 790 | attempt); |
830 | if (ret) | 791 | if (ret) |
831 | goto out; | 792 | goto out; |
832 | goto retry; | 793 | goto retry; |
833 | } | 794 | } |
834 | 795 | ||
835 | /* | ||
836 | * If we would have faulted, release mmap_sem, | ||
837 | * fault it in and start all over again. | ||
838 | */ | ||
839 | futex_unlock_mm(fshared); | ||
840 | |||
841 | ret = get_user(dummy, uaddr2); | 796 | ret = get_user(dummy, uaddr2); |
842 | if (ret) | 797 | if (ret) |
843 | return ret; | 798 | return ret; |
@@ -873,7 +828,8 @@ retry: | |||
873 | if (hb1 != hb2) | 828 | if (hb1 != hb2) |
874 | spin_unlock(&hb2->lock); | 829 | spin_unlock(&hb2->lock); |
875 | out: | 830 | out: |
876 | futex_unlock_mm(fshared); | 831 | put_futex_key(fshared, &key2); |
832 | put_futex_key(fshared, &key1); | ||
877 | 833 | ||
878 | return ret; | 834 | return ret; |
879 | } | 835 | } |
@@ -882,19 +838,16 @@ out: | |||
882 | * Requeue all waiters hashed on one physical page to another | 838 | * Requeue all waiters hashed on one physical page to another |
883 | * physical page. | 839 | * physical page. |
884 | */ | 840 | */ |
885 | static int futex_requeue(u32 __user *uaddr1, struct rw_semaphore *fshared, | 841 | static int futex_requeue(u32 __user *uaddr1, int fshared, u32 __user *uaddr2, |
886 | u32 __user *uaddr2, | ||
887 | int nr_wake, int nr_requeue, u32 *cmpval) | 842 | int nr_wake, int nr_requeue, u32 *cmpval) |
888 | { | 843 | { |
889 | union futex_key key1, key2; | 844 | union futex_key key1 = FUTEX_KEY_INIT, key2 = FUTEX_KEY_INIT; |
890 | struct futex_hash_bucket *hb1, *hb2; | 845 | struct futex_hash_bucket *hb1, *hb2; |
891 | struct plist_head *head1; | 846 | struct plist_head *head1; |
892 | struct futex_q *this, *next; | 847 | struct futex_q *this, *next; |
893 | int ret, drop_count = 0; | 848 | int ret, drop_count = 0; |
894 | 849 | ||
895 | retry: | 850 | retry: |
896 | futex_lock_mm(fshared); | ||
897 | |||
898 | ret = get_futex_key(uaddr1, fshared, &key1); | 851 | ret = get_futex_key(uaddr1, fshared, &key1); |
899 | if (unlikely(ret != 0)) | 852 | if (unlikely(ret != 0)) |
900 | goto out; | 853 | goto out; |
@@ -917,12 +870,6 @@ static int futex_requeue(u32 __user *uaddr1, struct rw_semaphore *fshared, | |||
917 | if (hb1 != hb2) | 870 | if (hb1 != hb2) |
918 | spin_unlock(&hb2->lock); | 871 | spin_unlock(&hb2->lock); |
919 | 872 | ||
920 | /* | ||
921 | * If we would have faulted, release mmap_sem, fault | ||
922 | * it in and start all over again. | ||
923 | */ | ||
924 | futex_unlock_mm(fshared); | ||
925 | |||
926 | ret = get_user(curval, uaddr1); | 873 | ret = get_user(curval, uaddr1); |
927 | 874 | ||
928 | if (!ret) | 875 | if (!ret) |
@@ -974,7 +921,8 @@ out_unlock: | |||
974 | drop_futex_key_refs(&key1); | 921 | drop_futex_key_refs(&key1); |
975 | 922 | ||
976 | out: | 923 | out: |
977 | futex_unlock_mm(fshared); | 924 | put_futex_key(fshared, &key2); |
925 | put_futex_key(fshared, &key1); | ||
978 | return ret; | 926 | return ret; |
979 | } | 927 | } |
980 | 928 | ||
@@ -983,7 +931,7 @@ static inline struct futex_hash_bucket *queue_lock(struct futex_q *q) | |||
983 | { | 931 | { |
984 | struct futex_hash_bucket *hb; | 932 | struct futex_hash_bucket *hb; |
985 | 933 | ||
986 | init_waitqueue_head(&q->waiters); | 934 | init_waitqueue_head(&q->waiter); |
987 | 935 | ||
988 | get_futex_key_refs(&q->key); | 936 | get_futex_key_refs(&q->key); |
989 | hb = hash_futex(&q->key); | 937 | hb = hash_futex(&q->key); |
@@ -1096,8 +1044,7 @@ static void unqueue_me_pi(struct futex_q *q) | |||
1096 | * private futexes. | 1044 | * private futexes. |
1097 | */ | 1045 | */ |
1098 | static int fixup_pi_state_owner(u32 __user *uaddr, struct futex_q *q, | 1046 | static int fixup_pi_state_owner(u32 __user *uaddr, struct futex_q *q, |
1099 | struct task_struct *newowner, | 1047 | struct task_struct *newowner, int fshared) |
1100 | struct rw_semaphore *fshared) | ||
1101 | { | 1048 | { |
1102 | u32 newtid = task_pid_vnr(newowner) | FUTEX_WAITERS; | 1049 | u32 newtid = task_pid_vnr(newowner) | FUTEX_WAITERS; |
1103 | struct futex_pi_state *pi_state = q->pi_state; | 1050 | struct futex_pi_state *pi_state = q->pi_state; |
@@ -1176,7 +1123,7 @@ retry: | |||
1176 | handle_fault: | 1123 | handle_fault: |
1177 | spin_unlock(q->lock_ptr); | 1124 | spin_unlock(q->lock_ptr); |
1178 | 1125 | ||
1179 | ret = futex_handle_fault((unsigned long)uaddr, fshared, attempt++); | 1126 | ret = futex_handle_fault((unsigned long)uaddr, attempt++); |
1180 | 1127 | ||
1181 | spin_lock(q->lock_ptr); | 1128 | spin_lock(q->lock_ptr); |
1182 | 1129 | ||
@@ -1196,12 +1143,13 @@ handle_fault: | |||
1196 | * In case we must use restart_block to restart a futex_wait, | 1143 | * In case we must use restart_block to restart a futex_wait, |
1197 | * we encode in the 'flags' shared capability | 1144 | * we encode in the 'flags' shared capability |
1198 | */ | 1145 | */ |
1199 | #define FLAGS_SHARED 1 | 1146 | #define FLAGS_SHARED 0x01 |
1147 | #define FLAGS_CLOCKRT 0x02 | ||
1200 | 1148 | ||
1201 | static long futex_wait_restart(struct restart_block *restart); | 1149 | static long futex_wait_restart(struct restart_block *restart); |
1202 | 1150 | ||
1203 | static int futex_wait(u32 __user *uaddr, struct rw_semaphore *fshared, | 1151 | static int futex_wait(u32 __user *uaddr, int fshared, |
1204 | u32 val, ktime_t *abs_time, u32 bitset) | 1152 | u32 val, ktime_t *abs_time, u32 bitset, int clockrt) |
1205 | { | 1153 | { |
1206 | struct task_struct *curr = current; | 1154 | struct task_struct *curr = current; |
1207 | DECLARE_WAITQUEUE(wait, curr); | 1155 | DECLARE_WAITQUEUE(wait, curr); |
@@ -1218,8 +1166,7 @@ static int futex_wait(u32 __user *uaddr, struct rw_semaphore *fshared, | |||
1218 | q.pi_state = NULL; | 1166 | q.pi_state = NULL; |
1219 | q.bitset = bitset; | 1167 | q.bitset = bitset; |
1220 | retry: | 1168 | retry: |
1221 | futex_lock_mm(fshared); | 1169 | q.key = FUTEX_KEY_INIT; |
1222 | |||
1223 | ret = get_futex_key(uaddr, fshared, &q.key); | 1170 | ret = get_futex_key(uaddr, fshared, &q.key); |
1224 | if (unlikely(ret != 0)) | 1171 | if (unlikely(ret != 0)) |
1225 | goto out_release_sem; | 1172 | goto out_release_sem; |
@@ -1251,12 +1198,6 @@ static int futex_wait(u32 __user *uaddr, struct rw_semaphore *fshared, | |||
1251 | if (unlikely(ret)) { | 1198 | if (unlikely(ret)) { |
1252 | queue_unlock(&q, hb); | 1199 | queue_unlock(&q, hb); |
1253 | 1200 | ||
1254 | /* | ||
1255 | * If we would have faulted, release mmap_sem, fault it in and | ||
1256 | * start all over again. | ||
1257 | */ | ||
1258 | futex_unlock_mm(fshared); | ||
1259 | |||
1260 | ret = get_user(uval, uaddr); | 1201 | ret = get_user(uval, uaddr); |
1261 | 1202 | ||
1262 | if (!ret) | 1203 | if (!ret) |
@@ -1271,12 +1212,6 @@ static int futex_wait(u32 __user *uaddr, struct rw_semaphore *fshared, | |||
1271 | queue_me(&q, hb); | 1212 | queue_me(&q, hb); |
1272 | 1213 | ||
1273 | /* | 1214 | /* |
1274 | * Now the futex is queued and we have checked the data, we | ||
1275 | * don't want to hold mmap_sem while we sleep. | ||
1276 | */ | ||
1277 | futex_unlock_mm(fshared); | ||
1278 | |||
1279 | /* | ||
1280 | * There might have been scheduling since the queue_me(), as we | 1215 | * There might have been scheduling since the queue_me(), as we |
1281 | * cannot hold a spinlock across the get_user() in case it | 1216 | * cannot hold a spinlock across the get_user() in case it |
1282 | * faults, and we cannot just set TASK_INTERRUPTIBLE state when | 1217 | * faults, and we cannot just set TASK_INTERRUPTIBLE state when |
@@ -1287,7 +1222,7 @@ static int futex_wait(u32 __user *uaddr, struct rw_semaphore *fshared, | |||
1287 | 1222 | ||
1288 | /* add_wait_queue is the barrier after __set_current_state. */ | 1223 | /* add_wait_queue is the barrier after __set_current_state. */ |
1289 | __set_current_state(TASK_INTERRUPTIBLE); | 1224 | __set_current_state(TASK_INTERRUPTIBLE); |
1290 | add_wait_queue(&q.waiters, &wait); | 1225 | add_wait_queue(&q.waiter, &wait); |
1291 | /* | 1226 | /* |
1292 | * !plist_node_empty() is safe here without any lock. | 1227 | * !plist_node_empty() is safe here without any lock. |
1293 | * q.lock_ptr != 0 is not safe, because of ordering against wakeup. | 1228 | * q.lock_ptr != 0 is not safe, because of ordering against wakeup. |
@@ -1300,8 +1235,10 @@ static int futex_wait(u32 __user *uaddr, struct rw_semaphore *fshared, | |||
1300 | slack = current->timer_slack_ns; | 1235 | slack = current->timer_slack_ns; |
1301 | if (rt_task(current)) | 1236 | if (rt_task(current)) |
1302 | slack = 0; | 1237 | slack = 0; |
1303 | hrtimer_init_on_stack(&t.timer, CLOCK_MONOTONIC, | 1238 | hrtimer_init_on_stack(&t.timer, |
1304 | HRTIMER_MODE_ABS); | 1239 | clockrt ? CLOCK_REALTIME : |
1240 | CLOCK_MONOTONIC, | ||
1241 | HRTIMER_MODE_ABS); | ||
1305 | hrtimer_init_sleeper(&t, current); | 1242 | hrtimer_init_sleeper(&t, current); |
1306 | hrtimer_set_expires_range_ns(&t.timer, *abs_time, slack); | 1243 | hrtimer_set_expires_range_ns(&t.timer, *abs_time, slack); |
1307 | 1244 | ||
@@ -1356,6 +1293,8 @@ static int futex_wait(u32 __user *uaddr, struct rw_semaphore *fshared, | |||
1356 | 1293 | ||
1357 | if (fshared) | 1294 | if (fshared) |
1358 | restart->futex.flags |= FLAGS_SHARED; | 1295 | restart->futex.flags |= FLAGS_SHARED; |
1296 | if (clockrt) | ||
1297 | restart->futex.flags |= FLAGS_CLOCKRT; | ||
1359 | return -ERESTART_RESTARTBLOCK; | 1298 | return -ERESTART_RESTARTBLOCK; |
1360 | } | 1299 | } |
1361 | 1300 | ||
@@ -1363,7 +1302,7 @@ static int futex_wait(u32 __user *uaddr, struct rw_semaphore *fshared, | |||
1363 | queue_unlock(&q, hb); | 1302 | queue_unlock(&q, hb); |
1364 | 1303 | ||
1365 | out_release_sem: | 1304 | out_release_sem: |
1366 | futex_unlock_mm(fshared); | 1305 | put_futex_key(fshared, &q.key); |
1367 | return ret; | 1306 | return ret; |
1368 | } | 1307 | } |
1369 | 1308 | ||
@@ -1371,15 +1310,16 @@ static int futex_wait(u32 __user *uaddr, struct rw_semaphore *fshared, | |||
1371 | static long futex_wait_restart(struct restart_block *restart) | 1310 | static long futex_wait_restart(struct restart_block *restart) |
1372 | { | 1311 | { |
1373 | u32 __user *uaddr = (u32 __user *)restart->futex.uaddr; | 1312 | u32 __user *uaddr = (u32 __user *)restart->futex.uaddr; |
1374 | struct rw_semaphore *fshared = NULL; | 1313 | int fshared = 0; |
1375 | ktime_t t; | 1314 | ktime_t t; |
1376 | 1315 | ||
1377 | t.tv64 = restart->futex.time; | 1316 | t.tv64 = restart->futex.time; |
1378 | restart->fn = do_no_restart_syscall; | 1317 | restart->fn = do_no_restart_syscall; |
1379 | if (restart->futex.flags & FLAGS_SHARED) | 1318 | if (restart->futex.flags & FLAGS_SHARED) |
1380 | fshared = ¤t->mm->mmap_sem; | 1319 | fshared = 1; |
1381 | return (long)futex_wait(uaddr, fshared, restart->futex.val, &t, | 1320 | return (long)futex_wait(uaddr, fshared, restart->futex.val, &t, |
1382 | restart->futex.bitset); | 1321 | restart->futex.bitset, |
1322 | restart->futex.flags & FLAGS_CLOCKRT); | ||
1383 | } | 1323 | } |
1384 | 1324 | ||
1385 | 1325 | ||
@@ -1389,7 +1329,7 @@ static long futex_wait_restart(struct restart_block *restart) | |||
1389 | * if there are waiters then it will block, it does PI, etc. (Due to | 1329 | * if there are waiters then it will block, it does PI, etc. (Due to |
1390 | * races the kernel might see a 0 value of the futex too.) | 1330 | * races the kernel might see a 0 value of the futex too.) |
1391 | */ | 1331 | */ |
1392 | static int futex_lock_pi(u32 __user *uaddr, struct rw_semaphore *fshared, | 1332 | static int futex_lock_pi(u32 __user *uaddr, int fshared, |
1393 | int detect, ktime_t *time, int trylock) | 1333 | int detect, ktime_t *time, int trylock) |
1394 | { | 1334 | { |
1395 | struct hrtimer_sleeper timeout, *to = NULL; | 1335 | struct hrtimer_sleeper timeout, *to = NULL; |
@@ -1412,8 +1352,7 @@ static int futex_lock_pi(u32 __user *uaddr, struct rw_semaphore *fshared, | |||
1412 | 1352 | ||
1413 | q.pi_state = NULL; | 1353 | q.pi_state = NULL; |
1414 | retry: | 1354 | retry: |
1415 | futex_lock_mm(fshared); | 1355 | q.key = FUTEX_KEY_INIT; |
1416 | |||
1417 | ret = get_futex_key(uaddr, fshared, &q.key); | 1356 | ret = get_futex_key(uaddr, fshared, &q.key); |
1418 | if (unlikely(ret != 0)) | 1357 | if (unlikely(ret != 0)) |
1419 | goto out_release_sem; | 1358 | goto out_release_sem; |
@@ -1502,7 +1441,6 @@ static int futex_lock_pi(u32 __user *uaddr, struct rw_semaphore *fshared, | |||
1502 | * exit to complete. | 1441 | * exit to complete. |
1503 | */ | 1442 | */ |
1504 | queue_unlock(&q, hb); | 1443 | queue_unlock(&q, hb); |
1505 | futex_unlock_mm(fshared); | ||
1506 | cond_resched(); | 1444 | cond_resched(); |
1507 | goto retry; | 1445 | goto retry; |
1508 | 1446 | ||
@@ -1534,12 +1472,6 @@ static int futex_lock_pi(u32 __user *uaddr, struct rw_semaphore *fshared, | |||
1534 | */ | 1472 | */ |
1535 | queue_me(&q, hb); | 1473 | queue_me(&q, hb); |
1536 | 1474 | ||
1537 | /* | ||
1538 | * Now the futex is queued and we have checked the data, we | ||
1539 | * don't want to hold mmap_sem while we sleep. | ||
1540 | */ | ||
1541 | futex_unlock_mm(fshared); | ||
1542 | |||
1543 | WARN_ON(!q.pi_state); | 1475 | WARN_ON(!q.pi_state); |
1544 | /* | 1476 | /* |
1545 | * Block on the PI mutex: | 1477 | * Block on the PI mutex: |
@@ -1552,7 +1484,6 @@ static int futex_lock_pi(u32 __user *uaddr, struct rw_semaphore *fshared, | |||
1552 | ret = ret ? 0 : -EWOULDBLOCK; | 1484 | ret = ret ? 0 : -EWOULDBLOCK; |
1553 | } | 1485 | } |
1554 | 1486 | ||
1555 | futex_lock_mm(fshared); | ||
1556 | spin_lock(q.lock_ptr); | 1487 | spin_lock(q.lock_ptr); |
1557 | 1488 | ||
1558 | if (!ret) { | 1489 | if (!ret) { |
@@ -1618,7 +1549,6 @@ static int futex_lock_pi(u32 __user *uaddr, struct rw_semaphore *fshared, | |||
1618 | 1549 | ||
1619 | /* Unqueue and drop the lock */ | 1550 | /* Unqueue and drop the lock */ |
1620 | unqueue_me_pi(&q); | 1551 | unqueue_me_pi(&q); |
1621 | futex_unlock_mm(fshared); | ||
1622 | 1552 | ||
1623 | if (to) | 1553 | if (to) |
1624 | destroy_hrtimer_on_stack(&to->timer); | 1554 | destroy_hrtimer_on_stack(&to->timer); |
@@ -1628,34 +1558,30 @@ static int futex_lock_pi(u32 __user *uaddr, struct rw_semaphore *fshared, | |||
1628 | queue_unlock(&q, hb); | 1558 | queue_unlock(&q, hb); |
1629 | 1559 | ||
1630 | out_release_sem: | 1560 | out_release_sem: |
1631 | futex_unlock_mm(fshared); | 1561 | put_futex_key(fshared, &q.key); |
1632 | if (to) | 1562 | if (to) |
1633 | destroy_hrtimer_on_stack(&to->timer); | 1563 | destroy_hrtimer_on_stack(&to->timer); |
1634 | return ret; | 1564 | return ret; |
1635 | 1565 | ||
1636 | uaddr_faulted: | 1566 | uaddr_faulted: |
1637 | /* | 1567 | /* |
1638 | * We have to r/w *(int __user *)uaddr, but we can't modify it | 1568 | * We have to r/w *(int __user *)uaddr, and we have to modify it |
1639 | * non-atomically. Therefore, if get_user below is not | 1569 | * atomically. Therefore, if we continue to fault after get_user() |
1640 | * enough, we need to handle the fault ourselves, while | 1570 | * below, we need to handle the fault ourselves, while still holding |
1641 | * still holding the mmap_sem. | 1571 | * the mmap_sem. This can occur if the uaddr is under contention as |
1642 | * | 1572 | * we have to drop the mmap_sem in order to call get_user(). |
1643 | * ... and hb->lock. :-) --ANK | ||
1644 | */ | 1573 | */ |
1645 | queue_unlock(&q, hb); | 1574 | queue_unlock(&q, hb); |
1646 | 1575 | ||
1647 | if (attempt++) { | 1576 | if (attempt++) { |
1648 | ret = futex_handle_fault((unsigned long)uaddr, fshared, | 1577 | ret = futex_handle_fault((unsigned long)uaddr, attempt); |
1649 | attempt); | ||
1650 | if (ret) | 1578 | if (ret) |
1651 | goto out_release_sem; | 1579 | goto out_release_sem; |
1652 | goto retry_unlocked; | 1580 | goto retry_unlocked; |
1653 | } | 1581 | } |
1654 | 1582 | ||
1655 | futex_unlock_mm(fshared); | ||
1656 | |||
1657 | ret = get_user(uval, uaddr); | 1583 | ret = get_user(uval, uaddr); |
1658 | if (!ret && (uval != -EFAULT)) | 1584 | if (!ret) |
1659 | goto retry; | 1585 | goto retry; |
1660 | 1586 | ||
1661 | if (to) | 1587 | if (to) |
@@ -1668,13 +1594,13 @@ static int futex_lock_pi(u32 __user *uaddr, struct rw_semaphore *fshared, | |||
1668 | * This is the in-kernel slowpath: we look up the PI state (if any), | 1594 | * This is the in-kernel slowpath: we look up the PI state (if any), |
1669 | * and do the rt-mutex unlock. | 1595 | * and do the rt-mutex unlock. |
1670 | */ | 1596 | */ |
1671 | static int futex_unlock_pi(u32 __user *uaddr, struct rw_semaphore *fshared) | 1597 | static int futex_unlock_pi(u32 __user *uaddr, int fshared) |
1672 | { | 1598 | { |
1673 | struct futex_hash_bucket *hb; | 1599 | struct futex_hash_bucket *hb; |
1674 | struct futex_q *this, *next; | 1600 | struct futex_q *this, *next; |
1675 | u32 uval; | 1601 | u32 uval; |
1676 | struct plist_head *head; | 1602 | struct plist_head *head; |
1677 | union futex_key key; | 1603 | union futex_key key = FUTEX_KEY_INIT; |
1678 | int ret, attempt = 0; | 1604 | int ret, attempt = 0; |
1679 | 1605 | ||
1680 | retry: | 1606 | retry: |
@@ -1685,10 +1611,6 @@ retry: | |||
1685 | */ | 1611 | */ |
1686 | if ((uval & FUTEX_TID_MASK) != task_pid_vnr(current)) | 1612 | if ((uval & FUTEX_TID_MASK) != task_pid_vnr(current)) |
1687 | return -EPERM; | 1613 | return -EPERM; |
1688 | /* | ||
1689 | * First take all the futex related locks: | ||
1690 | */ | ||
1691 | futex_lock_mm(fshared); | ||
1692 | 1614 | ||
1693 | ret = get_futex_key(uaddr, fshared, &key); | 1615 | ret = get_futex_key(uaddr, fshared, &key); |
1694 | if (unlikely(ret != 0)) | 1616 | if (unlikely(ret != 0)) |
@@ -1747,34 +1669,30 @@ retry_unlocked: | |||
1747 | out_unlock: | 1669 | out_unlock: |
1748 | spin_unlock(&hb->lock); | 1670 | spin_unlock(&hb->lock); |
1749 | out: | 1671 | out: |
1750 | futex_unlock_mm(fshared); | 1672 | put_futex_key(fshared, &key); |
1751 | 1673 | ||
1752 | return ret; | 1674 | return ret; |
1753 | 1675 | ||
1754 | pi_faulted: | 1676 | pi_faulted: |
1755 | /* | 1677 | /* |
1756 | * We have to r/w *(int __user *)uaddr, but we can't modify it | 1678 | * We have to r/w *(int __user *)uaddr, and we have to modify it |
1757 | * non-atomically. Therefore, if get_user below is not | 1679 | * atomically. Therefore, if we continue to fault after get_user() |
1758 | * enough, we need to handle the fault ourselves, while | 1680 | * below, we need to handle the fault ourselves, while still holding |
1759 | * still holding the mmap_sem. | 1681 | * the mmap_sem. This can occur if the uaddr is under contention as |
1760 | * | 1682 | * we have to drop the mmap_sem in order to call get_user(). |
1761 | * ... and hb->lock. --ANK | ||
1762 | */ | 1683 | */ |
1763 | spin_unlock(&hb->lock); | 1684 | spin_unlock(&hb->lock); |
1764 | 1685 | ||
1765 | if (attempt++) { | 1686 | if (attempt++) { |
1766 | ret = futex_handle_fault((unsigned long)uaddr, fshared, | 1687 | ret = futex_handle_fault((unsigned long)uaddr, attempt); |
1767 | attempt); | ||
1768 | if (ret) | 1688 | if (ret) |
1769 | goto out; | 1689 | goto out; |
1770 | uval = 0; | 1690 | uval = 0; |
1771 | goto retry_unlocked; | 1691 | goto retry_unlocked; |
1772 | } | 1692 | } |
1773 | 1693 | ||
1774 | futex_unlock_mm(fshared); | ||
1775 | |||
1776 | ret = get_user(uval, uaddr); | 1694 | ret = get_user(uval, uaddr); |
1777 | if (!ret && (uval != -EFAULT)) | 1695 | if (!ret) |
1778 | goto retry; | 1696 | goto retry; |
1779 | 1697 | ||
1780 | return ret; | 1698 | return ret; |
@@ -1898,8 +1816,7 @@ retry: | |||
1898 | * PI futexes happens in exit_pi_state(): | 1816 | * PI futexes happens in exit_pi_state(): |
1899 | */ | 1817 | */ |
1900 | if (!pi && (uval & FUTEX_WAITERS)) | 1818 | if (!pi && (uval & FUTEX_WAITERS)) |
1901 | futex_wake(uaddr, &curr->mm->mmap_sem, 1, | 1819 | futex_wake(uaddr, 1, 1, FUTEX_BITSET_MATCH_ANY); |
1902 | FUTEX_BITSET_MATCH_ANY); | ||
1903 | } | 1820 | } |
1904 | return 0; | 1821 | return 0; |
1905 | } | 1822 | } |
@@ -1993,18 +1910,22 @@ void exit_robust_list(struct task_struct *curr) | |||
1993 | long do_futex(u32 __user *uaddr, int op, u32 val, ktime_t *timeout, | 1910 | long do_futex(u32 __user *uaddr, int op, u32 val, ktime_t *timeout, |
1994 | u32 __user *uaddr2, u32 val2, u32 val3) | 1911 | u32 __user *uaddr2, u32 val2, u32 val3) |
1995 | { | 1912 | { |
1996 | int ret = -ENOSYS; | 1913 | int clockrt, ret = -ENOSYS; |
1997 | int cmd = op & FUTEX_CMD_MASK; | 1914 | int cmd = op & FUTEX_CMD_MASK; |
1998 | struct rw_semaphore *fshared = NULL; | 1915 | int fshared = 0; |
1999 | 1916 | ||
2000 | if (!(op & FUTEX_PRIVATE_FLAG)) | 1917 | if (!(op & FUTEX_PRIVATE_FLAG)) |
2001 | fshared = ¤t->mm->mmap_sem; | 1918 | fshared = 1; |
1919 | |||
1920 | clockrt = op & FUTEX_CLOCK_REALTIME; | ||
1921 | if (clockrt && cmd != FUTEX_WAIT_BITSET) | ||
1922 | return -ENOSYS; | ||
2002 | 1923 | ||
2003 | switch (cmd) { | 1924 | switch (cmd) { |
2004 | case FUTEX_WAIT: | 1925 | case FUTEX_WAIT: |
2005 | val3 = FUTEX_BITSET_MATCH_ANY; | 1926 | val3 = FUTEX_BITSET_MATCH_ANY; |
2006 | case FUTEX_WAIT_BITSET: | 1927 | case FUTEX_WAIT_BITSET: |
2007 | ret = futex_wait(uaddr, fshared, val, timeout, val3); | 1928 | ret = futex_wait(uaddr, fshared, val, timeout, val3, clockrt); |
2008 | break; | 1929 | break; |
2009 | case FUTEX_WAKE: | 1930 | case FUTEX_WAKE: |
2010 | val3 = FUTEX_BITSET_MATCH_ANY; | 1931 | val3 = FUTEX_BITSET_MATCH_ANY; |
diff --git a/kernel/irq/manage.c b/kernel/irq/manage.c index 801addda3c4..e9d1c8205a3 100644 --- a/kernel/irq/manage.c +++ b/kernel/irq/manage.c | |||
@@ -673,6 +673,18 @@ int request_irq(unsigned int irq, irq_handler_t handler, | |||
673 | struct irq_desc *desc; | 673 | struct irq_desc *desc; |
674 | int retval; | 674 | int retval; |
675 | 675 | ||
676 | /* | ||
677 | * handle_IRQ_event() always ignores IRQF_DISABLED except for | ||
678 | * the _first_ irqaction (sigh). That can cause oopsing, but | ||
679 | * the behavior is classified as "will not fix" so we need to | ||
680 | * start nudging drivers away from using that idiom. | ||
681 | */ | ||
682 | if ((irqflags & (IRQF_SHARED|IRQF_DISABLED)) | ||
683 | == (IRQF_SHARED|IRQF_DISABLED)) | ||
684 | pr_warning("IRQ %d/%s: IRQF_DISABLED is not " | ||
685 | "guaranteed on shared IRQs\n", | ||
686 | irq, devname); | ||
687 | |||
676 | #ifdef CONFIG_LOCKDEP | 688 | #ifdef CONFIG_LOCKDEP |
677 | /* | 689 | /* |
678 | * Lockdep wants atomic interrupt handlers: | 690 | * Lockdep wants atomic interrupt handlers: |
diff --git a/kernel/lockdep.c b/kernel/lockdep.c index 46a404173db..4fa6eeb4e8a 100644 --- a/kernel/lockdep.c +++ b/kernel/lockdep.c | |||
@@ -136,16 +136,16 @@ static inline struct lock_class *hlock_class(struct held_lock *hlock) | |||
136 | #ifdef CONFIG_LOCK_STAT | 136 | #ifdef CONFIG_LOCK_STAT |
137 | static DEFINE_PER_CPU(struct lock_class_stats[MAX_LOCKDEP_KEYS], lock_stats); | 137 | static DEFINE_PER_CPU(struct lock_class_stats[MAX_LOCKDEP_KEYS], lock_stats); |
138 | 138 | ||
139 | static int lock_contention_point(struct lock_class *class, unsigned long ip) | 139 | static int lock_point(unsigned long points[], unsigned long ip) |
140 | { | 140 | { |
141 | int i; | 141 | int i; |
142 | 142 | ||
143 | for (i = 0; i < ARRAY_SIZE(class->contention_point); i++) { | 143 | for (i = 0; i < LOCKSTAT_POINTS; i++) { |
144 | if (class->contention_point[i] == 0) { | 144 | if (points[i] == 0) { |
145 | class->contention_point[i] = ip; | 145 | points[i] = ip; |
146 | break; | 146 | break; |
147 | } | 147 | } |
148 | if (class->contention_point[i] == ip) | 148 | if (points[i] == ip) |
149 | break; | 149 | break; |
150 | } | 150 | } |
151 | 151 | ||
@@ -185,6 +185,9 @@ struct lock_class_stats lock_stats(struct lock_class *class) | |||
185 | for (i = 0; i < ARRAY_SIZE(stats.contention_point); i++) | 185 | for (i = 0; i < ARRAY_SIZE(stats.contention_point); i++) |
186 | stats.contention_point[i] += pcs->contention_point[i]; | 186 | stats.contention_point[i] += pcs->contention_point[i]; |
187 | 187 | ||
188 | for (i = 0; i < ARRAY_SIZE(stats.contending_point); i++) | ||
189 | stats.contending_point[i] += pcs->contending_point[i]; | ||
190 | |||
188 | lock_time_add(&pcs->read_waittime, &stats.read_waittime); | 191 | lock_time_add(&pcs->read_waittime, &stats.read_waittime); |
189 | lock_time_add(&pcs->write_waittime, &stats.write_waittime); | 192 | lock_time_add(&pcs->write_waittime, &stats.write_waittime); |
190 | 193 | ||
@@ -209,6 +212,7 @@ void clear_lock_stats(struct lock_class *class) | |||
209 | memset(cpu_stats, 0, sizeof(struct lock_class_stats)); | 212 | memset(cpu_stats, 0, sizeof(struct lock_class_stats)); |
210 | } | 213 | } |
211 | memset(class->contention_point, 0, sizeof(class->contention_point)); | 214 | memset(class->contention_point, 0, sizeof(class->contention_point)); |
215 | memset(class->contending_point, 0, sizeof(class->contending_point)); | ||
212 | } | 216 | } |
213 | 217 | ||
214 | static struct lock_class_stats *get_lock_stats(struct lock_class *class) | 218 | static struct lock_class_stats *get_lock_stats(struct lock_class *class) |
@@ -287,14 +291,12 @@ void lockdep_off(void) | |||
287 | { | 291 | { |
288 | current->lockdep_recursion++; | 292 | current->lockdep_recursion++; |
289 | } | 293 | } |
290 | |||
291 | EXPORT_SYMBOL(lockdep_off); | 294 | EXPORT_SYMBOL(lockdep_off); |
292 | 295 | ||
293 | void lockdep_on(void) | 296 | void lockdep_on(void) |
294 | { | 297 | { |
295 | current->lockdep_recursion--; | 298 | current->lockdep_recursion--; |
296 | } | 299 | } |
297 | |||
298 | EXPORT_SYMBOL(lockdep_on); | 300 | EXPORT_SYMBOL(lockdep_on); |
299 | 301 | ||
300 | /* | 302 | /* |
@@ -576,7 +578,8 @@ static void print_lock_class_header(struct lock_class *class, int depth) | |||
576 | /* | 578 | /* |
577 | * printk all lock dependencies starting at <entry>: | 579 | * printk all lock dependencies starting at <entry>: |
578 | */ | 580 | */ |
579 | static void print_lock_dependencies(struct lock_class *class, int depth) | 581 | static void __used |
582 | print_lock_dependencies(struct lock_class *class, int depth) | ||
580 | { | 583 | { |
581 | struct lock_list *entry; | 584 | struct lock_list *entry; |
582 | 585 | ||
@@ -2508,7 +2511,6 @@ void lockdep_init_map(struct lockdep_map *lock, const char *name, | |||
2508 | if (subclass) | 2511 | if (subclass) |
2509 | register_lock_class(lock, subclass, 1); | 2512 | register_lock_class(lock, subclass, 1); |
2510 | } | 2513 | } |
2511 | |||
2512 | EXPORT_SYMBOL_GPL(lockdep_init_map); | 2514 | EXPORT_SYMBOL_GPL(lockdep_init_map); |
2513 | 2515 | ||
2514 | /* | 2516 | /* |
@@ -2689,8 +2691,9 @@ static int check_unlock(struct task_struct *curr, struct lockdep_map *lock, | |||
2689 | } | 2691 | } |
2690 | 2692 | ||
2691 | static int | 2693 | static int |
2692 | __lock_set_subclass(struct lockdep_map *lock, | 2694 | __lock_set_class(struct lockdep_map *lock, const char *name, |
2693 | unsigned int subclass, unsigned long ip) | 2695 | struct lock_class_key *key, unsigned int subclass, |
2696 | unsigned long ip) | ||
2694 | { | 2697 | { |
2695 | struct task_struct *curr = current; | 2698 | struct task_struct *curr = current; |
2696 | struct held_lock *hlock, *prev_hlock; | 2699 | struct held_lock *hlock, *prev_hlock; |
@@ -2717,6 +2720,7 @@ __lock_set_subclass(struct lockdep_map *lock, | |||
2717 | return print_unlock_inbalance_bug(curr, lock, ip); | 2720 | return print_unlock_inbalance_bug(curr, lock, ip); |
2718 | 2721 | ||
2719 | found_it: | 2722 | found_it: |
2723 | lockdep_init_map(lock, name, key, 0); | ||
2720 | class = register_lock_class(lock, subclass, 0); | 2724 | class = register_lock_class(lock, subclass, 0); |
2721 | hlock->class_idx = class - lock_classes + 1; | 2725 | hlock->class_idx = class - lock_classes + 1; |
2722 | 2726 | ||
@@ -2901,9 +2905,9 @@ static void check_flags(unsigned long flags) | |||
2901 | #endif | 2905 | #endif |
2902 | } | 2906 | } |
2903 | 2907 | ||
2904 | void | 2908 | void lock_set_class(struct lockdep_map *lock, const char *name, |
2905 | lock_set_subclass(struct lockdep_map *lock, | 2909 | struct lock_class_key *key, unsigned int subclass, |
2906 | unsigned int subclass, unsigned long ip) | 2910 | unsigned long ip) |
2907 | { | 2911 | { |
2908 | unsigned long flags; | 2912 | unsigned long flags; |
2909 | 2913 | ||
@@ -2913,13 +2917,12 @@ lock_set_subclass(struct lockdep_map *lock, | |||
2913 | raw_local_irq_save(flags); | 2917 | raw_local_irq_save(flags); |
2914 | current->lockdep_recursion = 1; | 2918 | current->lockdep_recursion = 1; |
2915 | check_flags(flags); | 2919 | check_flags(flags); |
2916 | if (__lock_set_subclass(lock, subclass, ip)) | 2920 | if (__lock_set_class(lock, name, key, subclass, ip)) |
2917 | check_chain_key(current); | 2921 | check_chain_key(current); |
2918 | current->lockdep_recursion = 0; | 2922 | current->lockdep_recursion = 0; |
2919 | raw_local_irq_restore(flags); | 2923 | raw_local_irq_restore(flags); |
2920 | } | 2924 | } |
2921 | 2925 | EXPORT_SYMBOL_GPL(lock_set_class); | |
2922 | EXPORT_SYMBOL_GPL(lock_set_subclass); | ||
2923 | 2926 | ||
2924 | /* | 2927 | /* |
2925 | * We are not always called with irqs disabled - do that here, | 2928 | * We are not always called with irqs disabled - do that here, |
@@ -2943,7 +2946,6 @@ void lock_acquire(struct lockdep_map *lock, unsigned int subclass, | |||
2943 | current->lockdep_recursion = 0; | 2946 | current->lockdep_recursion = 0; |
2944 | raw_local_irq_restore(flags); | 2947 | raw_local_irq_restore(flags); |
2945 | } | 2948 | } |
2946 | |||
2947 | EXPORT_SYMBOL_GPL(lock_acquire); | 2949 | EXPORT_SYMBOL_GPL(lock_acquire); |
2948 | 2950 | ||
2949 | void lock_release(struct lockdep_map *lock, int nested, | 2951 | void lock_release(struct lockdep_map *lock, int nested, |
@@ -2961,7 +2963,6 @@ void lock_release(struct lockdep_map *lock, int nested, | |||
2961 | current->lockdep_recursion = 0; | 2963 | current->lockdep_recursion = 0; |
2962 | raw_local_irq_restore(flags); | 2964 | raw_local_irq_restore(flags); |
2963 | } | 2965 | } |
2964 | |||
2965 | EXPORT_SYMBOL_GPL(lock_release); | 2966 | EXPORT_SYMBOL_GPL(lock_release); |
2966 | 2967 | ||
2967 | #ifdef CONFIG_LOCK_STAT | 2968 | #ifdef CONFIG_LOCK_STAT |
@@ -2999,7 +3000,7 @@ __lock_contended(struct lockdep_map *lock, unsigned long ip) | |||
2999 | struct held_lock *hlock, *prev_hlock; | 3000 | struct held_lock *hlock, *prev_hlock; |
3000 | struct lock_class_stats *stats; | 3001 | struct lock_class_stats *stats; |
3001 | unsigned int depth; | 3002 | unsigned int depth; |
3002 | int i, point; | 3003 | int i, contention_point, contending_point; |
3003 | 3004 | ||
3004 | depth = curr->lockdep_depth; | 3005 | depth = curr->lockdep_depth; |
3005 | if (DEBUG_LOCKS_WARN_ON(!depth)) | 3006 | if (DEBUG_LOCKS_WARN_ON(!depth)) |
@@ -3023,18 +3024,22 @@ __lock_contended(struct lockdep_map *lock, unsigned long ip) | |||
3023 | found_it: | 3024 | found_it: |
3024 | hlock->waittime_stamp = sched_clock(); | 3025 | hlock->waittime_stamp = sched_clock(); |
3025 | 3026 | ||
3026 | point = lock_contention_point(hlock_class(hlock), ip); | 3027 | contention_point = lock_point(hlock_class(hlock)->contention_point, ip); |
3028 | contending_point = lock_point(hlock_class(hlock)->contending_point, | ||
3029 | lock->ip); | ||
3027 | 3030 | ||
3028 | stats = get_lock_stats(hlock_class(hlock)); | 3031 | stats = get_lock_stats(hlock_class(hlock)); |
3029 | if (point < ARRAY_SIZE(stats->contention_point)) | 3032 | if (contention_point < LOCKSTAT_POINTS) |
3030 | stats->contention_point[point]++; | 3033 | stats->contention_point[contention_point]++; |
3034 | if (contending_point < LOCKSTAT_POINTS) | ||
3035 | stats->contending_point[contending_point]++; | ||
3031 | if (lock->cpu != smp_processor_id()) | 3036 | if (lock->cpu != smp_processor_id()) |
3032 | stats->bounces[bounce_contended + !!hlock->read]++; | 3037 | stats->bounces[bounce_contended + !!hlock->read]++; |
3033 | put_lock_stats(stats); | 3038 | put_lock_stats(stats); |
3034 | } | 3039 | } |
3035 | 3040 | ||
3036 | static void | 3041 | static void |
3037 | __lock_acquired(struct lockdep_map *lock) | 3042 | __lock_acquired(struct lockdep_map *lock, unsigned long ip) |
3038 | { | 3043 | { |
3039 | struct task_struct *curr = current; | 3044 | struct task_struct *curr = current; |
3040 | struct held_lock *hlock, *prev_hlock; | 3045 | struct held_lock *hlock, *prev_hlock; |
@@ -3083,6 +3088,7 @@ found_it: | |||
3083 | put_lock_stats(stats); | 3088 | put_lock_stats(stats); |
3084 | 3089 | ||
3085 | lock->cpu = cpu; | 3090 | lock->cpu = cpu; |
3091 | lock->ip = ip; | ||
3086 | } | 3092 | } |
3087 | 3093 | ||
3088 | void lock_contended(struct lockdep_map *lock, unsigned long ip) | 3094 | void lock_contended(struct lockdep_map *lock, unsigned long ip) |
@@ -3104,7 +3110,7 @@ void lock_contended(struct lockdep_map *lock, unsigned long ip) | |||
3104 | } | 3110 | } |
3105 | EXPORT_SYMBOL_GPL(lock_contended); | 3111 | EXPORT_SYMBOL_GPL(lock_contended); |
3106 | 3112 | ||
3107 | void lock_acquired(struct lockdep_map *lock) | 3113 | void lock_acquired(struct lockdep_map *lock, unsigned long ip) |
3108 | { | 3114 | { |
3109 | unsigned long flags; | 3115 | unsigned long flags; |
3110 | 3116 | ||
@@ -3117,7 +3123,7 @@ void lock_acquired(struct lockdep_map *lock) | |||
3117 | raw_local_irq_save(flags); | 3123 | raw_local_irq_save(flags); |
3118 | check_flags(flags); | 3124 | check_flags(flags); |
3119 | current->lockdep_recursion = 1; | 3125 | current->lockdep_recursion = 1; |
3120 | __lock_acquired(lock); | 3126 | __lock_acquired(lock, ip); |
3121 | current->lockdep_recursion = 0; | 3127 | current->lockdep_recursion = 0; |
3122 | raw_local_irq_restore(flags); | 3128 | raw_local_irq_restore(flags); |
3123 | } | 3129 | } |
@@ -3441,7 +3447,6 @@ retry: | |||
3441 | if (unlock) | 3447 | if (unlock) |
3442 | read_unlock(&tasklist_lock); | 3448 | read_unlock(&tasklist_lock); |
3443 | } | 3449 | } |
3444 | |||
3445 | EXPORT_SYMBOL_GPL(debug_show_all_locks); | 3450 | EXPORT_SYMBOL_GPL(debug_show_all_locks); |
3446 | 3451 | ||
3447 | /* | 3452 | /* |
@@ -3462,7 +3467,6 @@ void debug_show_held_locks(struct task_struct *task) | |||
3462 | { | 3467 | { |
3463 | __debug_show_held_locks(task); | 3468 | __debug_show_held_locks(task); |
3464 | } | 3469 | } |
3465 | |||
3466 | EXPORT_SYMBOL_GPL(debug_show_held_locks); | 3470 | EXPORT_SYMBOL_GPL(debug_show_held_locks); |
3467 | 3471 | ||
3468 | void lockdep_sys_exit(void) | 3472 | void lockdep_sys_exit(void) |
diff --git a/kernel/lockdep_proc.c b/kernel/lockdep_proc.c index 20dbcbf9c7d..13716b81389 100644 --- a/kernel/lockdep_proc.c +++ b/kernel/lockdep_proc.c | |||
@@ -470,11 +470,12 @@ static void seq_line(struct seq_file *m, char c, int offset, int length) | |||
470 | 470 | ||
471 | static void snprint_time(char *buf, size_t bufsiz, s64 nr) | 471 | static void snprint_time(char *buf, size_t bufsiz, s64 nr) |
472 | { | 472 | { |
473 | unsigned long rem; | 473 | s64 div; |
474 | s32 rem; | ||
474 | 475 | ||
475 | nr += 5; /* for display rounding */ | 476 | nr += 5; /* for display rounding */ |
476 | rem = do_div(nr, 1000); /* XXX: do_div_signed */ | 477 | div = div_s64_rem(nr, 1000, &rem); |
477 | snprintf(buf, bufsiz, "%lld.%02d", (long long)nr, (int)rem/10); | 478 | snprintf(buf, bufsiz, "%lld.%02d", (long long)div, (int)rem/10); |
478 | } | 479 | } |
479 | 480 | ||
480 | static void seq_time(struct seq_file *m, s64 time) | 481 | static void seq_time(struct seq_file *m, s64 time) |
@@ -556,7 +557,7 @@ static void seq_stats(struct seq_file *m, struct lock_stat_data *data) | |||
556 | if (stats->read_holdtime.nr) | 557 | if (stats->read_holdtime.nr) |
557 | namelen += 2; | 558 | namelen += 2; |
558 | 559 | ||
559 | for (i = 0; i < ARRAY_SIZE(class->contention_point); i++) { | 560 | for (i = 0; i < LOCKSTAT_POINTS; i++) { |
560 | char sym[KSYM_SYMBOL_LEN]; | 561 | char sym[KSYM_SYMBOL_LEN]; |
561 | char ip[32]; | 562 | char ip[32]; |
562 | 563 | ||
@@ -573,6 +574,23 @@ static void seq_stats(struct seq_file *m, struct lock_stat_data *data) | |||
573 | stats->contention_point[i], | 574 | stats->contention_point[i], |
574 | ip, sym); | 575 | ip, sym); |
575 | } | 576 | } |
577 | for (i = 0; i < LOCKSTAT_POINTS; i++) { | ||
578 | char sym[KSYM_SYMBOL_LEN]; | ||
579 | char ip[32]; | ||
580 | |||
581 | if (class->contending_point[i] == 0) | ||
582 | break; | ||
583 | |||
584 | if (!i) | ||
585 | seq_line(m, '-', 40-namelen, namelen); | ||
586 | |||
587 | sprint_symbol(sym, class->contending_point[i]); | ||
588 | snprintf(ip, sizeof(ip), "[<%p>]", | ||
589 | (void *)class->contending_point[i]); | ||
590 | seq_printf(m, "%40s %14lu %29s %s\n", name, | ||
591 | stats->contending_point[i], | ||
592 | ip, sym); | ||
593 | } | ||
576 | if (i) { | 594 | if (i) { |
577 | seq_puts(m, "\n"); | 595 | seq_puts(m, "\n"); |
578 | seq_line(m, '.', 0, 40 + 1 + 10 * (14 + 1)); | 596 | seq_line(m, '.', 0, 40 + 1 + 10 * (14 + 1)); |
@@ -582,7 +600,7 @@ static void seq_stats(struct seq_file *m, struct lock_stat_data *data) | |||
582 | 600 | ||
583 | static void seq_header(struct seq_file *m) | 601 | static void seq_header(struct seq_file *m) |
584 | { | 602 | { |
585 | seq_printf(m, "lock_stat version 0.2\n"); | 603 | seq_printf(m, "lock_stat version 0.3\n"); |
586 | seq_line(m, '-', 0, 40 + 1 + 10 * (14 + 1)); | 604 | seq_line(m, '-', 0, 40 + 1 + 10 * (14 + 1)); |
587 | seq_printf(m, "%40s %14s %14s %14s %14s %14s %14s %14s %14s " | 605 | seq_printf(m, "%40s %14s %14s %14s %14s %14s %14s %14s %14s " |
588 | "%14s %14s\n", | 606 | "%14s %14s\n", |
diff --git a/kernel/mutex.c b/kernel/mutex.c index 12c779dc65d..4f45d4b658e 100644 --- a/kernel/mutex.c +++ b/kernel/mutex.c | |||
@@ -59,7 +59,7 @@ EXPORT_SYMBOL(__mutex_init); | |||
59 | * We also put the fastpath first in the kernel image, to make sure the | 59 | * We also put the fastpath first in the kernel image, to make sure the |
60 | * branch is predicted by the CPU as default-untaken. | 60 | * branch is predicted by the CPU as default-untaken. |
61 | */ | 61 | */ |
62 | static void noinline __sched | 62 | static __used noinline void __sched |
63 | __mutex_lock_slowpath(atomic_t *lock_count); | 63 | __mutex_lock_slowpath(atomic_t *lock_count); |
64 | 64 | ||
65 | /*** | 65 | /*** |
@@ -96,7 +96,7 @@ void inline __sched mutex_lock(struct mutex *lock) | |||
96 | EXPORT_SYMBOL(mutex_lock); | 96 | EXPORT_SYMBOL(mutex_lock); |
97 | #endif | 97 | #endif |
98 | 98 | ||
99 | static noinline void __sched __mutex_unlock_slowpath(atomic_t *lock_count); | 99 | static __used noinline void __sched __mutex_unlock_slowpath(atomic_t *lock_count); |
100 | 100 | ||
101 | /*** | 101 | /*** |
102 | * mutex_unlock - release the mutex | 102 | * mutex_unlock - release the mutex |
@@ -184,7 +184,7 @@ __mutex_lock_common(struct mutex *lock, long state, unsigned int subclass, | |||
184 | } | 184 | } |
185 | 185 | ||
186 | done: | 186 | done: |
187 | lock_acquired(&lock->dep_map); | 187 | lock_acquired(&lock->dep_map, ip); |
188 | /* got the lock - rejoice! */ | 188 | /* got the lock - rejoice! */ |
189 | mutex_remove_waiter(lock, &waiter, task_thread_info(task)); | 189 | mutex_remove_waiter(lock, &waiter, task_thread_info(task)); |
190 | debug_mutex_set_owner(lock, task_thread_info(task)); | 190 | debug_mutex_set_owner(lock, task_thread_info(task)); |
@@ -268,7 +268,7 @@ __mutex_unlock_common_slowpath(atomic_t *lock_count, int nested) | |||
268 | /* | 268 | /* |
269 | * Release the lock, slowpath: | 269 | * Release the lock, slowpath: |
270 | */ | 270 | */ |
271 | static noinline void | 271 | static __used noinline void |
272 | __mutex_unlock_slowpath(atomic_t *lock_count) | 272 | __mutex_unlock_slowpath(atomic_t *lock_count) |
273 | { | 273 | { |
274 | __mutex_unlock_common_slowpath(lock_count, 1); | 274 | __mutex_unlock_common_slowpath(lock_count, 1); |
@@ -313,7 +313,7 @@ int __sched mutex_lock_killable(struct mutex *lock) | |||
313 | } | 313 | } |
314 | EXPORT_SYMBOL(mutex_lock_killable); | 314 | EXPORT_SYMBOL(mutex_lock_killable); |
315 | 315 | ||
316 | static noinline void __sched | 316 | static __used noinline void __sched |
317 | __mutex_lock_slowpath(atomic_t *lock_count) | 317 | __mutex_lock_slowpath(atomic_t *lock_count) |
318 | { | 318 | { |
319 | struct mutex *lock = container_of(lock_count, struct mutex, count); | 319 | struct mutex *lock = container_of(lock_count, struct mutex, count); |
diff --git a/kernel/notifier.c b/kernel/notifier.c index 4282c0a40a5..61d5aa5eced 100644 --- a/kernel/notifier.c +++ b/kernel/notifier.c | |||
@@ -82,6 +82,14 @@ static int __kprobes notifier_call_chain(struct notifier_block **nl, | |||
82 | 82 | ||
83 | while (nb && nr_to_call) { | 83 | while (nb && nr_to_call) { |
84 | next_nb = rcu_dereference(nb->next); | 84 | next_nb = rcu_dereference(nb->next); |
85 | |||
86 | #ifdef CONFIG_DEBUG_NOTIFIERS | ||
87 | if (unlikely(!func_ptr_is_kernel_text(nb->notifier_call))) { | ||
88 | WARN(1, "Invalid notifier called!"); | ||
89 | nb = next_nb; | ||
90 | continue; | ||
91 | } | ||
92 | #endif | ||
85 | ret = nb->notifier_call(nb, val, v); | 93 | ret = nb->notifier_call(nb, val, v); |
86 | 94 | ||
87 | if (nr_calls) | 95 | if (nr_calls) |
diff --git a/kernel/panic.c b/kernel/panic.c index 4d5088355bf..13f06349a78 100644 --- a/kernel/panic.c +++ b/kernel/panic.c | |||
@@ -21,6 +21,7 @@ | |||
21 | #include <linux/debug_locks.h> | 21 | #include <linux/debug_locks.h> |
22 | #include <linux/random.h> | 22 | #include <linux/random.h> |
23 | #include <linux/kallsyms.h> | 23 | #include <linux/kallsyms.h> |
24 | #include <linux/dmi.h> | ||
24 | 25 | ||
25 | int panic_on_oops; | 26 | int panic_on_oops; |
26 | static unsigned long tainted_mask; | 27 | static unsigned long tainted_mask; |
@@ -321,36 +322,27 @@ void oops_exit(void) | |||
321 | } | 322 | } |
322 | 323 | ||
323 | #ifdef WANT_WARN_ON_SLOWPATH | 324 | #ifdef WANT_WARN_ON_SLOWPATH |
324 | void warn_on_slowpath(const char *file, int line) | ||
325 | { | ||
326 | char function[KSYM_SYMBOL_LEN]; | ||
327 | unsigned long caller = (unsigned long) __builtin_return_address(0); | ||
328 | sprint_symbol(function, caller); | ||
329 | |||
330 | printk(KERN_WARNING "------------[ cut here ]------------\n"); | ||
331 | printk(KERN_WARNING "WARNING: at %s:%d %s()\n", file, | ||
332 | line, function); | ||
333 | print_modules(); | ||
334 | dump_stack(); | ||
335 | print_oops_end_marker(); | ||
336 | add_taint(TAINT_WARN); | ||
337 | } | ||
338 | EXPORT_SYMBOL(warn_on_slowpath); | ||
339 | |||
340 | |||
341 | void warn_slowpath(const char *file, int line, const char *fmt, ...) | 325 | void warn_slowpath(const char *file, int line, const char *fmt, ...) |
342 | { | 326 | { |
343 | va_list args; | 327 | va_list args; |
344 | char function[KSYM_SYMBOL_LEN]; | 328 | char function[KSYM_SYMBOL_LEN]; |
345 | unsigned long caller = (unsigned long)__builtin_return_address(0); | 329 | unsigned long caller = (unsigned long)__builtin_return_address(0); |
330 | const char *board; | ||
331 | |||
346 | sprint_symbol(function, caller); | 332 | sprint_symbol(function, caller); |
347 | 333 | ||
348 | printk(KERN_WARNING "------------[ cut here ]------------\n"); | 334 | printk(KERN_WARNING "------------[ cut here ]------------\n"); |
349 | printk(KERN_WARNING "WARNING: at %s:%d %s()\n", file, | 335 | printk(KERN_WARNING "WARNING: at %s:%d %s()\n", file, |
350 | line, function); | 336 | line, function); |
351 | va_start(args, fmt); | 337 | board = dmi_get_system_info(DMI_PRODUCT_NAME); |
352 | vprintk(fmt, args); | 338 | if (board) |
353 | va_end(args); | 339 | printk(KERN_WARNING "Hardware name: %s\n", board); |
340 | |||
341 | if (fmt) { | ||
342 | va_start(args, fmt); | ||
343 | vprintk(fmt, args); | ||
344 | va_end(args); | ||
345 | } | ||
354 | 346 | ||
355 | print_modules(); | 347 | print_modules(); |
356 | dump_stack(); | 348 | dump_stack(); |
diff --git a/kernel/posix-cpu-timers.c b/kernel/posix-cpu-timers.c index 4e5288a831d..157de3a4783 100644 --- a/kernel/posix-cpu-timers.c +++ b/kernel/posix-cpu-timers.c | |||
@@ -58,21 +58,21 @@ void thread_group_cputime( | |||
58 | struct task_struct *tsk, | 58 | struct task_struct *tsk, |
59 | struct task_cputime *times) | 59 | struct task_cputime *times) |
60 | { | 60 | { |
61 | struct signal_struct *sig; | 61 | struct task_cputime *totals, *tot; |
62 | int i; | 62 | int i; |
63 | struct task_cputime *tot; | ||
64 | 63 | ||
65 | sig = tsk->signal; | 64 | totals = tsk->signal->cputime.totals; |
66 | if (unlikely(!sig) || !sig->cputime.totals) { | 65 | if (!totals) { |
67 | times->utime = tsk->utime; | 66 | times->utime = tsk->utime; |
68 | times->stime = tsk->stime; | 67 | times->stime = tsk->stime; |
69 | times->sum_exec_runtime = tsk->se.sum_exec_runtime; | 68 | times->sum_exec_runtime = tsk->se.sum_exec_runtime; |
70 | return; | 69 | return; |
71 | } | 70 | } |
71 | |||
72 | times->stime = times->utime = cputime_zero; | 72 | times->stime = times->utime = cputime_zero; |
73 | times->sum_exec_runtime = 0; | 73 | times->sum_exec_runtime = 0; |
74 | for_each_possible_cpu(i) { | 74 | for_each_possible_cpu(i) { |
75 | tot = per_cpu_ptr(tsk->signal->cputime.totals, i); | 75 | tot = per_cpu_ptr(totals, i); |
76 | times->utime = cputime_add(times->utime, tot->utime); | 76 | times->utime = cputime_add(times->utime, tot->utime); |
77 | times->stime = cputime_add(times->stime, tot->stime); | 77 | times->stime = cputime_add(times->stime, tot->stime); |
78 | times->sum_exec_runtime += tot->sum_exec_runtime; | 78 | times->sum_exec_runtime += tot->sum_exec_runtime; |
diff --git a/kernel/posix-timers.c b/kernel/posix-timers.c index 5e79c662294..a140e44eebb 100644 --- a/kernel/posix-timers.c +++ b/kernel/posix-timers.c | |||
@@ -197,6 +197,11 @@ static int common_timer_create(struct k_itimer *new_timer) | |||
197 | return 0; | 197 | return 0; |
198 | } | 198 | } |
199 | 199 | ||
200 | static int no_timer_create(struct k_itimer *new_timer) | ||
201 | { | ||
202 | return -EOPNOTSUPP; | ||
203 | } | ||
204 | |||
200 | /* | 205 | /* |
201 | * Return nonzero if we know a priori this clockid_t value is bogus. | 206 | * Return nonzero if we know a priori this clockid_t value is bogus. |
202 | */ | 207 | */ |
@@ -248,6 +253,7 @@ static __init int init_posix_timers(void) | |||
248 | .clock_getres = hrtimer_get_res, | 253 | .clock_getres = hrtimer_get_res, |
249 | .clock_get = posix_get_monotonic_raw, | 254 | .clock_get = posix_get_monotonic_raw, |
250 | .clock_set = do_posix_clock_nosettime, | 255 | .clock_set = do_posix_clock_nosettime, |
256 | .timer_create = no_timer_create, | ||
251 | }; | 257 | }; |
252 | 258 | ||
253 | register_posix_clock(CLOCK_REALTIME, &clock_realtime); | 259 | register_posix_clock(CLOCK_REALTIME, &clock_realtime); |
diff --git a/kernel/printk.c b/kernel/printk.c index f492f1583d7..e651ab05655 100644 --- a/kernel/printk.c +++ b/kernel/printk.c | |||
@@ -662,7 +662,7 @@ asmlinkage int vprintk(const char *fmt, va_list args) | |||
662 | if (recursion_bug) { | 662 | if (recursion_bug) { |
663 | recursion_bug = 0; | 663 | recursion_bug = 0; |
664 | strcpy(printk_buf, recursion_bug_msg); | 664 | strcpy(printk_buf, recursion_bug_msg); |
665 | printed_len = sizeof(recursion_bug_msg); | 665 | printed_len = strlen(recursion_bug_msg); |
666 | } | 666 | } |
667 | /* Emit the output into the temporary buffer */ | 667 | /* Emit the output into the temporary buffer */ |
668 | printed_len += vscnprintf(printk_buf + printed_len, | 668 | printed_len += vscnprintf(printk_buf + printed_len, |
diff --git a/kernel/resource.c b/kernel/resource.c index 4337063663e..e633106b12f 100644 --- a/kernel/resource.c +++ b/kernel/resource.c | |||
@@ -853,6 +853,15 @@ int iomem_map_sanity_check(resource_size_t addr, unsigned long size) | |||
853 | if (PFN_DOWN(p->start) <= PFN_DOWN(addr) && | 853 | if (PFN_DOWN(p->start) <= PFN_DOWN(addr) && |
854 | PFN_DOWN(p->end) >= PFN_DOWN(addr + size - 1)) | 854 | PFN_DOWN(p->end) >= PFN_DOWN(addr + size - 1)) |
855 | continue; | 855 | continue; |
856 | /* | ||
857 | * if a resource is "BUSY", it's not a hardware resource | ||
858 | * but a driver mapping of such a resource; we don't want | ||
859 | * to warn for those; some drivers legitimately map only | ||
860 | * partial hardware resources. (example: vesafb) | ||
861 | */ | ||
862 | if (p->flags & IORESOURCE_BUSY) | ||
863 | continue; | ||
864 | |||
856 | printk(KERN_WARNING "resource map sanity check conflict: " | 865 | printk(KERN_WARNING "resource map sanity check conflict: " |
857 | "0x%llx 0x%llx 0x%llx 0x%llx %s\n", | 866 | "0x%llx 0x%llx 0x%llx 0x%llx %s\n", |
858 | (unsigned long long)addr, | 867 | (unsigned long long)addr, |
diff --git a/kernel/sched.c b/kernel/sched.c index e4bb1dd7b30..3e70963120a 100644 --- a/kernel/sched.c +++ b/kernel/sched.c | |||
@@ -4203,7 +4203,6 @@ void account_steal_time(struct task_struct *p, cputime_t steal) | |||
4203 | 4203 | ||
4204 | if (p == rq->idle) { | 4204 | if (p == rq->idle) { |
4205 | p->stime = cputime_add(p->stime, steal); | 4205 | p->stime = cputime_add(p->stime, steal); |
4206 | account_group_system_time(p, steal); | ||
4207 | if (atomic_read(&rq->nr_iowait) > 0) | 4206 | if (atomic_read(&rq->nr_iowait) > 0) |
4208 | cpustat->iowait = cputime64_add(cpustat->iowait, tmp); | 4207 | cpustat->iowait = cputime64_add(cpustat->iowait, tmp); |
4209 | else | 4208 | else |
@@ -4339,7 +4338,7 @@ void __kprobes sub_preempt_count(int val) | |||
4339 | /* | 4338 | /* |
4340 | * Underflow? | 4339 | * Underflow? |
4341 | */ | 4340 | */ |
4342 | if (DEBUG_LOCKS_WARN_ON(val > preempt_count())) | 4341 | if (DEBUG_LOCKS_WARN_ON(val > preempt_count() - (!!kernel_locked()))) |
4343 | return; | 4342 | return; |
4344 | /* | 4343 | /* |
4345 | * Is the spinlock portion underflowing? | 4344 | * Is the spinlock portion underflowing? |
diff --git a/kernel/softirq.c b/kernel/softirq.c index 80d323e6f61..466e75ce271 100644 --- a/kernel/softirq.c +++ b/kernel/softirq.c | |||
@@ -102,20 +102,6 @@ void local_bh_disable(void) | |||
102 | 102 | ||
103 | EXPORT_SYMBOL(local_bh_disable); | 103 | EXPORT_SYMBOL(local_bh_disable); |
104 | 104 | ||
105 | void __local_bh_enable(void) | ||
106 | { | ||
107 | WARN_ON_ONCE(in_irq()); | ||
108 | |||
109 | /* | ||
110 | * softirqs should never be enabled by __local_bh_enable(), | ||
111 | * it always nests inside local_bh_enable() sections: | ||
112 | */ | ||
113 | WARN_ON_ONCE(softirq_count() == SOFTIRQ_OFFSET); | ||
114 | |||
115 | sub_preempt_count(SOFTIRQ_OFFSET); | ||
116 | } | ||
117 | EXPORT_SYMBOL_GPL(__local_bh_enable); | ||
118 | |||
119 | /* | 105 | /* |
120 | * Special-case - softirqs can safely be enabled in | 106 | * Special-case - softirqs can safely be enabled in |
121 | * cond_resched_softirq(), or by __do_softirq(), | 107 | * cond_resched_softirq(), or by __do_softirq(), |
diff --git a/kernel/softlockup.c b/kernel/softlockup.c index dc0b3be6b7d..1ab790c67b1 100644 --- a/kernel/softlockup.c +++ b/kernel/softlockup.c | |||
@@ -164,7 +164,7 @@ unsigned long __read_mostly sysctl_hung_task_check_count = 1024; | |||
164 | /* | 164 | /* |
165 | * Zero means infinite timeout - no checking done: | 165 | * Zero means infinite timeout - no checking done: |
166 | */ | 166 | */ |
167 | unsigned long __read_mostly sysctl_hung_task_timeout_secs = 120; | 167 | unsigned long __read_mostly sysctl_hung_task_timeout_secs = 480; |
168 | 168 | ||
169 | unsigned long __read_mostly sysctl_hung_task_warnings = 10; | 169 | unsigned long __read_mostly sysctl_hung_task_warnings = 10; |
170 | 170 | ||
diff --git a/kernel/stacktrace.c b/kernel/stacktrace.c index 94b527ef1d1..eb212f8f8bc 100644 --- a/kernel/stacktrace.c +++ b/kernel/stacktrace.c | |||
@@ -6,6 +6,7 @@ | |||
6 | * Copyright (C) 2006 Red Hat, Inc., Ingo Molnar <mingo@redhat.com> | 6 | * Copyright (C) 2006 Red Hat, Inc., Ingo Molnar <mingo@redhat.com> |
7 | */ | 7 | */ |
8 | #include <linux/sched.h> | 8 | #include <linux/sched.h> |
9 | #include <linux/kernel.h> | ||
9 | #include <linux/module.h> | 10 | #include <linux/module.h> |
10 | #include <linux/kallsyms.h> | 11 | #include <linux/kallsyms.h> |
11 | #include <linux/stacktrace.h> | 12 | #include <linux/stacktrace.h> |
@@ -24,3 +25,13 @@ void print_stack_trace(struct stack_trace *trace, int spaces) | |||
24 | } | 25 | } |
25 | EXPORT_SYMBOL_GPL(print_stack_trace); | 26 | EXPORT_SYMBOL_GPL(print_stack_trace); |
26 | 27 | ||
28 | /* | ||
29 | * Architectures that do not implement save_stack_trace_tsk get this | ||
30 | * weak alias and a once-per-bootup warning (whenever this facility | ||
31 | * is utilized - for example by procfs): | ||
32 | */ | ||
33 | __weak void | ||
34 | save_stack_trace_tsk(struct task_struct *tsk, struct stack_trace *trace) | ||
35 | { | ||
36 | WARN_ONCE(1, KERN_INFO "save_stack_trace_tsk() not implemented yet.\n"); | ||
37 | } | ||
diff --git a/kernel/sys.c b/kernel/sys.c index 31deba8f7d1..5fc3a0cfb99 100644 --- a/kernel/sys.c +++ b/kernel/sys.c | |||
@@ -858,8 +858,8 @@ void do_sys_times(struct tms *tms) | |||
858 | struct task_cputime cputime; | 858 | struct task_cputime cputime; |
859 | cputime_t cutime, cstime; | 859 | cputime_t cutime, cstime; |
860 | 860 | ||
861 | spin_lock_irq(¤t->sighand->siglock); | ||
862 | thread_group_cputime(current, &cputime); | 861 | thread_group_cputime(current, &cputime); |
862 | spin_lock_irq(¤t->sighand->siglock); | ||
863 | cutime = current->signal->cutime; | 863 | cutime = current->signal->cutime; |
864 | cstime = current->signal->cstime; | 864 | cstime = current->signal->cstime; |
865 | spin_unlock_irq(¤t->sighand->siglock); | 865 | spin_unlock_irq(¤t->sighand->siglock); |