aboutsummaryrefslogtreecommitdiffstats
path: root/kernel
diff options
context:
space:
mode:
Diffstat (limited to 'kernel')
-rw-r--r--kernel/cgroup.c34
-rw-r--r--kernel/compat.c5
-rw-r--r--kernel/cpu.c6
-rw-r--r--kernel/cpuset.c34
-rw-r--r--kernel/dma-coherent.c42
-rw-r--r--kernel/exit.c21
-rw-r--r--kernel/fork.c17
-rw-r--r--kernel/futex.c72
-rw-r--r--kernel/kmod.c4
-rw-r--r--kernel/kprobes.c281
-rw-r--r--kernel/ksysfs.c4
-rw-r--r--kernel/module.c92
-rw-r--r--kernel/panic.c2
-rw-r--r--kernel/power/main.c6
-rw-r--r--kernel/profile.c1
-rw-r--r--kernel/rcupdate.c11
-rw-r--r--kernel/rcupreempt.c11
-rw-r--r--kernel/rcutorture.c18
-rw-r--r--kernel/rcutree.c13
-rw-r--r--kernel/signal.c3
-rw-r--r--kernel/stop_machine.c55
-rw-r--r--kernel/sys.c2
-rw-r--r--kernel/sysctl.c27
-rw-r--r--kernel/test_kprobes.c210
-rw-r--r--kernel/time.c4
-rw-r--r--kernel/tsacct.c4
26 files changed, 676 insertions, 303 deletions
diff --git a/kernel/cgroup.c b/kernel/cgroup.c
index 891a84eb9d30..f221446aa02d 100644
--- a/kernel/cgroup.c
+++ b/kernel/cgroup.c
@@ -116,7 +116,6 @@ static int root_count;
116 * be called. 116 * be called.
117 */ 117 */
118static int need_forkexit_callback __read_mostly; 118static int need_forkexit_callback __read_mostly;
119static int need_mm_owner_callback __read_mostly;
120 119
121/* convenient tests for these bits */ 120/* convenient tests for these bits */
122inline int cgroup_is_removed(const struct cgroup *cgrp) 121inline int cgroup_is_removed(const struct cgroup *cgrp)
@@ -573,7 +572,6 @@ static struct inode *cgroup_new_inode(mode_t mode, struct super_block *sb)
573 inode->i_mode = mode; 572 inode->i_mode = mode;
574 inode->i_uid = current_fsuid(); 573 inode->i_uid = current_fsuid();
575 inode->i_gid = current_fsgid(); 574 inode->i_gid = current_fsgid();
576 inode->i_blocks = 0;
577 inode->i_atime = inode->i_mtime = inode->i_ctime = CURRENT_TIME; 575 inode->i_atime = inode->i_mtime = inode->i_ctime = CURRENT_TIME;
578 inode->i_mapping->backing_dev_info = &cgroup_backing_dev_info; 576 inode->i_mapping->backing_dev_info = &cgroup_backing_dev_info;
579 } 577 }
@@ -2540,7 +2538,6 @@ static void __init cgroup_init_subsys(struct cgroup_subsys *ss)
2540 init_css_set.subsys[ss->subsys_id] = dummytop->subsys[ss->subsys_id]; 2538 init_css_set.subsys[ss->subsys_id] = dummytop->subsys[ss->subsys_id];
2541 2539
2542 need_forkexit_callback |= ss->fork || ss->exit; 2540 need_forkexit_callback |= ss->fork || ss->exit;
2543 need_mm_owner_callback |= !!ss->mm_owner_changed;
2544 2541
2545 /* At system boot, before all subsystems have been 2542 /* At system boot, before all subsystems have been
2546 * registered, no tasks have been forked, so we don't 2543 * registered, no tasks have been forked, so we don't
@@ -2790,37 +2787,6 @@ void cgroup_fork_callbacks(struct task_struct *child)
2790 } 2787 }
2791} 2788}
2792 2789
2793#ifdef CONFIG_MM_OWNER
2794/**
2795 * cgroup_mm_owner_callbacks - run callbacks when the mm->owner changes
2796 * @p: the new owner
2797 *
2798 * Called on every change to mm->owner. mm_init_owner() does not
2799 * invoke this routine, since it assigns the mm->owner the first time
2800 * and does not change it.
2801 *
2802 * The callbacks are invoked with mmap_sem held in read mode.
2803 */
2804void cgroup_mm_owner_callbacks(struct task_struct *old, struct task_struct *new)
2805{
2806 struct cgroup *oldcgrp, *newcgrp = NULL;
2807
2808 if (need_mm_owner_callback) {
2809 int i;
2810 for (i = 0; i < CGROUP_SUBSYS_COUNT; i++) {
2811 struct cgroup_subsys *ss = subsys[i];
2812 oldcgrp = task_cgroup(old, ss->subsys_id);
2813 if (new)
2814 newcgrp = task_cgroup(new, ss->subsys_id);
2815 if (oldcgrp == newcgrp)
2816 continue;
2817 if (ss->mm_owner_changed)
2818 ss->mm_owner_changed(ss, oldcgrp, newcgrp, new);
2819 }
2820 }
2821}
2822#endif /* CONFIG_MM_OWNER */
2823
2824/** 2790/**
2825 * cgroup_post_fork - called on a new task after adding it to the task list 2791 * cgroup_post_fork - called on a new task after adding it to the task list
2826 * @child: the task in question 2792 * @child: the task in question
diff --git a/kernel/compat.c b/kernel/compat.c
index d52e2ec1deb5..42d56544460f 100644
--- a/kernel/compat.c
+++ b/kernel/compat.c
@@ -24,6 +24,7 @@
24#include <linux/migrate.h> 24#include <linux/migrate.h>
25#include <linux/posix-timers.h> 25#include <linux/posix-timers.h>
26#include <linux/times.h> 26#include <linux/times.h>
27#include <linux/ptrace.h>
27 28
28#include <asm/uaccess.h> 29#include <asm/uaccess.h>
29 30
@@ -229,6 +230,7 @@ asmlinkage long compat_sys_times(struct compat_tms __user *tbuf)
229 if (copy_to_user(tbuf, &tmp, sizeof(tmp))) 230 if (copy_to_user(tbuf, &tmp, sizeof(tmp)))
230 return -EFAULT; 231 return -EFAULT;
231 } 232 }
233 force_successful_syscall_return();
232 return compat_jiffies_to_clock_t(jiffies); 234 return compat_jiffies_to_clock_t(jiffies);
233} 235}
234 236
@@ -894,8 +896,9 @@ asmlinkage long compat_sys_time(compat_time_t __user * tloc)
894 896
895 if (tloc) { 897 if (tloc) {
896 if (put_user(i,tloc)) 898 if (put_user(i,tloc))
897 i = -EFAULT; 899 return -EFAULT;
898 } 900 }
901 force_successful_syscall_return();
899 return i; 902 return i;
900} 903}
901 904
diff --git a/kernel/cpu.c b/kernel/cpu.c
index 47fff3b63cbf..30e74dd6d01b 100644
--- a/kernel/cpu.c
+++ b/kernel/cpu.c
@@ -269,8 +269,11 @@ out_release:
269 269
270int __ref cpu_down(unsigned int cpu) 270int __ref cpu_down(unsigned int cpu)
271{ 271{
272 int err = 0; 272 int err;
273 273
274 err = stop_machine_create();
275 if (err)
276 return err;
274 cpu_maps_update_begin(); 277 cpu_maps_update_begin();
275 278
276 if (cpu_hotplug_disabled) { 279 if (cpu_hotplug_disabled) {
@@ -297,6 +300,7 @@ int __ref cpu_down(unsigned int cpu)
297 300
298out: 301out:
299 cpu_maps_update_done(); 302 cpu_maps_update_done();
303 stop_machine_destroy();
300 return err; 304 return err;
301} 305}
302EXPORT_SYMBOL(cpu_down); 306EXPORT_SYMBOL(cpu_down);
diff --git a/kernel/cpuset.c b/kernel/cpuset.c
index 39c1a4c1c5a9..345ace5117de 100644
--- a/kernel/cpuset.c
+++ b/kernel/cpuset.c
@@ -240,6 +240,17 @@ static struct cpuset top_cpuset = {
240static DEFINE_MUTEX(callback_mutex); 240static DEFINE_MUTEX(callback_mutex);
241 241
242/* 242/*
243 * cpuset_buffer_lock protects both the cpuset_name and cpuset_nodelist
244 * buffers. They are statically allocated to prevent using excess stack
245 * when calling cpuset_print_task_mems_allowed().
246 */
247#define CPUSET_NAME_LEN (128)
248#define CPUSET_NODELIST_LEN (256)
249static char cpuset_name[CPUSET_NAME_LEN];
250static char cpuset_nodelist[CPUSET_NODELIST_LEN];
251static DEFINE_SPINLOCK(cpuset_buffer_lock);
252
253/*
243 * This is ugly, but preserves the userspace API for existing cpuset 254 * This is ugly, but preserves the userspace API for existing cpuset
244 * users. If someone tries to mount the "cpuset" filesystem, we 255 * users. If someone tries to mount the "cpuset" filesystem, we
245 * silently switch it to mount "cgroup" instead 256 * silently switch it to mount "cgroup" instead
@@ -2356,6 +2367,29 @@ int cpuset_mems_allowed_intersects(const struct task_struct *tsk1,
2356 return nodes_intersects(tsk1->mems_allowed, tsk2->mems_allowed); 2367 return nodes_intersects(tsk1->mems_allowed, tsk2->mems_allowed);
2357} 2368}
2358 2369
2370/**
2371 * cpuset_print_task_mems_allowed - prints task's cpuset and mems_allowed
2372 * @task: pointer to task_struct of some task.
2373 *
2374 * Description: Prints @task's name, cpuset name, and cached copy of its
2375 * mems_allowed to the kernel log. Must hold task_lock(task) to allow
2376 * dereferencing task_cs(task).
2377 */
2378void cpuset_print_task_mems_allowed(struct task_struct *tsk)
2379{
2380 struct dentry *dentry;
2381
2382 dentry = task_cs(tsk)->css.cgroup->dentry;
2383 spin_lock(&cpuset_buffer_lock);
2384 snprintf(cpuset_name, CPUSET_NAME_LEN,
2385 dentry ? (const char *)dentry->d_name.name : "/");
2386 nodelist_scnprintf(cpuset_nodelist, CPUSET_NODELIST_LEN,
2387 tsk->mems_allowed);
2388 printk(KERN_INFO "%s cpuset=%s mems_allowed=%s\n",
2389 tsk->comm, cpuset_name, cpuset_nodelist);
2390 spin_unlock(&cpuset_buffer_lock);
2391}
2392
2359/* 2393/*
2360 * Collection of memory_pressure is suppressed unless 2394 * Collection of memory_pressure is suppressed unless
2361 * this flag is enabled by writing "1" to the special 2395 * this flag is enabled by writing "1" to the special
diff --git a/kernel/dma-coherent.c b/kernel/dma-coherent.c
index f013a0c2e111..038707404b76 100644
--- a/kernel/dma-coherent.c
+++ b/kernel/dma-coherent.c
@@ -109,20 +109,40 @@ EXPORT_SYMBOL(dma_mark_declared_memory_occupied);
109int dma_alloc_from_coherent(struct device *dev, ssize_t size, 109int dma_alloc_from_coherent(struct device *dev, ssize_t size,
110 dma_addr_t *dma_handle, void **ret) 110 dma_addr_t *dma_handle, void **ret)
111{ 111{
112 struct dma_coherent_mem *mem = dev ? dev->dma_mem : NULL; 112 struct dma_coherent_mem *mem;
113 int order = get_order(size); 113 int order = get_order(size);
114 int pageno;
114 115
115 if (mem) { 116 if (!dev)
116 int page = bitmap_find_free_region(mem->bitmap, mem->size, 117 return 0;
117 order); 118 mem = dev->dma_mem;
118 if (page >= 0) { 119 if (!mem)
119 *dma_handle = mem->device_base + (page << PAGE_SHIFT); 120 return 0;
120 *ret = mem->virt_base + (page << PAGE_SHIFT); 121 if (unlikely(size > mem->size))
121 memset(*ret, 0, size); 122 return 0;
122 } else if (mem->flags & DMA_MEMORY_EXCLUSIVE) 123
123 *ret = NULL; 124 pageno = bitmap_find_free_region(mem->bitmap, mem->size, order);
125 if (pageno >= 0) {
126 /*
127 * Memory was found in the per-device arena.
128 */
129 *dma_handle = mem->device_base + (pageno << PAGE_SHIFT);
130 *ret = mem->virt_base + (pageno << PAGE_SHIFT);
131 memset(*ret, 0, size);
132 } else if (mem->flags & DMA_MEMORY_EXCLUSIVE) {
133 /*
134 * The per-device arena is exhausted and we are not
135 * permitted to fall back to generic memory.
136 */
137 *ret = NULL;
138 } else {
139 /*
140 * The per-device arena is exhausted and we are
141 * permitted to fall back to generic memory.
142 */
143 return 0;
124 } 144 }
125 return (mem != NULL); 145 return 1;
126} 146}
127EXPORT_SYMBOL(dma_alloc_from_coherent); 147EXPORT_SYMBOL(dma_alloc_from_coherent);
128 148
diff --git a/kernel/exit.c b/kernel/exit.c
index c9e5a1c14e08..c7740fa3252c 100644
--- a/kernel/exit.c
+++ b/kernel/exit.c
@@ -642,35 +642,31 @@ retry:
642 /* 642 /*
643 * We found no owner yet mm_users > 1: this implies that we are 643 * We found no owner yet mm_users > 1: this implies that we are
644 * most likely racing with swapoff (try_to_unuse()) or /proc or 644 * most likely racing with swapoff (try_to_unuse()) or /proc or
645 * ptrace or page migration (get_task_mm()). Mark owner as NULL, 645 * ptrace or page migration (get_task_mm()). Mark owner as NULL.
646 * so that subsystems can understand the callback and take action.
647 */ 646 */
648 down_write(&mm->mmap_sem);
649 cgroup_mm_owner_callbacks(mm->owner, NULL);
650 mm->owner = NULL; 647 mm->owner = NULL;
651 up_write(&mm->mmap_sem);
652 return; 648 return;
653 649
654assign_new_owner: 650assign_new_owner:
655 BUG_ON(c == p); 651 BUG_ON(c == p);
656 get_task_struct(c); 652 get_task_struct(c);
657 read_unlock(&tasklist_lock);
658 down_write(&mm->mmap_sem);
659 /* 653 /*
660 * The task_lock protects c->mm from changing. 654 * The task_lock protects c->mm from changing.
661 * We always want mm->owner->mm == mm 655 * We always want mm->owner->mm == mm
662 */ 656 */
663 task_lock(c); 657 task_lock(c);
658 /*
659 * Delay read_unlock() till we have the task_lock()
660 * to ensure that c does not slip away underneath us
661 */
662 read_unlock(&tasklist_lock);
664 if (c->mm != mm) { 663 if (c->mm != mm) {
665 task_unlock(c); 664 task_unlock(c);
666 up_write(&mm->mmap_sem);
667 put_task_struct(c); 665 put_task_struct(c);
668 goto retry; 666 goto retry;
669 } 667 }
670 cgroup_mm_owner_callbacks(mm->owner, c);
671 mm->owner = c; 668 mm->owner = c;
672 task_unlock(c); 669 task_unlock(c);
673 up_write(&mm->mmap_sem);
674 put_task_struct(c); 670 put_task_struct(c);
675} 671}
676#endif /* CONFIG_MM_OWNER */ 672#endif /* CONFIG_MM_OWNER */
@@ -1055,10 +1051,7 @@ NORET_TYPE void do_exit(long code)
1055 preempt_count()); 1051 preempt_count());
1056 1052
1057 acct_update_integrals(tsk); 1053 acct_update_integrals(tsk);
1058 if (tsk->mm) { 1054
1059 update_hiwater_rss(tsk->mm);
1060 update_hiwater_vm(tsk->mm);
1061 }
1062 group_dead = atomic_dec_and_test(&tsk->signal->live); 1055 group_dead = atomic_dec_and_test(&tsk->signal->live);
1063 if (group_dead) { 1056 if (group_dead) {
1064 hrtimer_cancel(&tsk->signal->real_timer); 1057 hrtimer_cancel(&tsk->signal->real_timer);
diff --git a/kernel/fork.c b/kernel/fork.c
index 43cbf30669e6..7b8f2a78be3d 100644
--- a/kernel/fork.c
+++ b/kernel/fork.c
@@ -400,6 +400,18 @@ __cacheline_aligned_in_smp DEFINE_SPINLOCK(mmlist_lock);
400#define allocate_mm() (kmem_cache_alloc(mm_cachep, GFP_KERNEL)) 400#define allocate_mm() (kmem_cache_alloc(mm_cachep, GFP_KERNEL))
401#define free_mm(mm) (kmem_cache_free(mm_cachep, (mm))) 401#define free_mm(mm) (kmem_cache_free(mm_cachep, (mm)))
402 402
403static unsigned long default_dump_filter = MMF_DUMP_FILTER_DEFAULT;
404
405static int __init coredump_filter_setup(char *s)
406{
407 default_dump_filter =
408 (simple_strtoul(s, NULL, 0) << MMF_DUMP_FILTER_SHIFT) &
409 MMF_DUMP_FILTER_MASK;
410 return 1;
411}
412
413__setup("coredump_filter=", coredump_filter_setup);
414
403#include <linux/init_task.h> 415#include <linux/init_task.h>
404 416
405static struct mm_struct * mm_init(struct mm_struct * mm, struct task_struct *p) 417static struct mm_struct * mm_init(struct mm_struct * mm, struct task_struct *p)
@@ -408,8 +420,7 @@ static struct mm_struct * mm_init(struct mm_struct * mm, struct task_struct *p)
408 atomic_set(&mm->mm_count, 1); 420 atomic_set(&mm->mm_count, 1);
409 init_rwsem(&mm->mmap_sem); 421 init_rwsem(&mm->mmap_sem);
410 INIT_LIST_HEAD(&mm->mmlist); 422 INIT_LIST_HEAD(&mm->mmlist);
411 mm->flags = (current->mm) ? current->mm->flags 423 mm->flags = (current->mm) ? current->mm->flags : default_dump_filter;
412 : MMF_DUMP_FILTER_DEFAULT;
413 mm->core_state = NULL; 424 mm->core_state = NULL;
414 mm->nr_ptes = 0; 425 mm->nr_ptes = 0;
415 set_mm_counter(mm, file_rss, 0); 426 set_mm_counter(mm, file_rss, 0);
@@ -758,7 +769,7 @@ static int copy_sighand(unsigned long clone_flags, struct task_struct *tsk)
758{ 769{
759 struct sighand_struct *sig; 770 struct sighand_struct *sig;
760 771
761 if (clone_flags & (CLONE_SIGHAND | CLONE_THREAD)) { 772 if (clone_flags & CLONE_SIGHAND) {
762 atomic_inc(&current->sighand->count); 773 atomic_inc(&current->sighand->count);
763 return 0; 774 return 0;
764 } 775 }
diff --git a/kernel/futex.c b/kernel/futex.c
index 7c6cbabe52b3..002aa189eb09 100644
--- a/kernel/futex.c
+++ b/kernel/futex.c
@@ -170,8 +170,11 @@ static void get_futex_key_refs(union futex_key *key)
170 */ 170 */
171static void drop_futex_key_refs(union futex_key *key) 171static void drop_futex_key_refs(union futex_key *key)
172{ 172{
173 if (!key->both.ptr) 173 if (!key->both.ptr) {
174 /* If we're here then we tried to put a key we failed to get */
175 WARN_ON_ONCE(1);
174 return; 176 return;
177 }
175 178
176 switch (key->both.offset & (FUT_OFF_INODE|FUT_OFF_MMSHARED)) { 179 switch (key->both.offset & (FUT_OFF_INODE|FUT_OFF_MMSHARED)) {
177 case FUT_OFF_INODE: 180 case FUT_OFF_INODE:
@@ -730,8 +733,8 @@ static int futex_wake(u32 __user *uaddr, int fshared, int nr_wake, u32 bitset)
730 } 733 }
731 734
732 spin_unlock(&hb->lock); 735 spin_unlock(&hb->lock);
733out:
734 put_futex_key(fshared, &key); 736 put_futex_key(fshared, &key);
737out:
735 return ret; 738 return ret;
736} 739}
737 740
@@ -755,7 +758,7 @@ retryfull:
755 goto out; 758 goto out;
756 ret = get_futex_key(uaddr2, fshared, &key2); 759 ret = get_futex_key(uaddr2, fshared, &key2);
757 if (unlikely(ret != 0)) 760 if (unlikely(ret != 0))
758 goto out; 761 goto out_put_key1;
759 762
760 hb1 = hash_futex(&key1); 763 hb1 = hash_futex(&key1);
761 hb2 = hash_futex(&key2); 764 hb2 = hash_futex(&key2);
@@ -777,12 +780,12 @@ retry:
777 * but we might get them from range checking 780 * but we might get them from range checking
778 */ 781 */
779 ret = op_ret; 782 ret = op_ret;
780 goto out; 783 goto out_put_keys;
781#endif 784#endif
782 785
783 if (unlikely(op_ret != -EFAULT)) { 786 if (unlikely(op_ret != -EFAULT)) {
784 ret = op_ret; 787 ret = op_ret;
785 goto out; 788 goto out_put_keys;
786 } 789 }
787 790
788 /* 791 /*
@@ -796,7 +799,7 @@ retry:
796 ret = futex_handle_fault((unsigned long)uaddr2, 799 ret = futex_handle_fault((unsigned long)uaddr2,
797 attempt); 800 attempt);
798 if (ret) 801 if (ret)
799 goto out; 802 goto out_put_keys;
800 goto retry; 803 goto retry;
801 } 804 }
802 805
@@ -834,10 +837,11 @@ retry:
834 spin_unlock(&hb1->lock); 837 spin_unlock(&hb1->lock);
835 if (hb1 != hb2) 838 if (hb1 != hb2)
836 spin_unlock(&hb2->lock); 839 spin_unlock(&hb2->lock);
837out: 840out_put_keys:
838 put_futex_key(fshared, &key2); 841 put_futex_key(fshared, &key2);
842out_put_key1:
839 put_futex_key(fshared, &key1); 843 put_futex_key(fshared, &key1);
840 844out:
841 return ret; 845 return ret;
842} 846}
843 847
@@ -854,13 +858,13 @@ static int futex_requeue(u32 __user *uaddr1, int fshared, u32 __user *uaddr2,
854 struct futex_q *this, *next; 858 struct futex_q *this, *next;
855 int ret, drop_count = 0; 859 int ret, drop_count = 0;
856 860
857 retry: 861retry:
858 ret = get_futex_key(uaddr1, fshared, &key1); 862 ret = get_futex_key(uaddr1, fshared, &key1);
859 if (unlikely(ret != 0)) 863 if (unlikely(ret != 0))
860 goto out; 864 goto out;
861 ret = get_futex_key(uaddr2, fshared, &key2); 865 ret = get_futex_key(uaddr2, fshared, &key2);
862 if (unlikely(ret != 0)) 866 if (unlikely(ret != 0))
863 goto out; 867 goto out_put_key1;
864 868
865 hb1 = hash_futex(&key1); 869 hb1 = hash_futex(&key1);
866 hb2 = hash_futex(&key2); 870 hb2 = hash_futex(&key2);
@@ -882,7 +886,7 @@ static int futex_requeue(u32 __user *uaddr1, int fshared, u32 __user *uaddr2,
882 if (!ret) 886 if (!ret)
883 goto retry; 887 goto retry;
884 888
885 return ret; 889 goto out_put_keys;
886 } 890 }
887 if (curval != *cmpval) { 891 if (curval != *cmpval) {
888 ret = -EAGAIN; 892 ret = -EAGAIN;
@@ -927,9 +931,11 @@ out_unlock:
927 while (--drop_count >= 0) 931 while (--drop_count >= 0)
928 drop_futex_key_refs(&key1); 932 drop_futex_key_refs(&key1);
929 933
930out: 934out_put_keys:
931 put_futex_key(fshared, &key2); 935 put_futex_key(fshared, &key2);
936out_put_key1:
932 put_futex_key(fshared, &key1); 937 put_futex_key(fshared, &key1);
938out:
933 return ret; 939 return ret;
934} 940}
935 941
@@ -990,7 +996,7 @@ static int unqueue_me(struct futex_q *q)
990 int ret = 0; 996 int ret = 0;
991 997
992 /* In the common case we don't take the spinlock, which is nice. */ 998 /* In the common case we don't take the spinlock, which is nice. */
993 retry: 999retry:
994 lock_ptr = q->lock_ptr; 1000 lock_ptr = q->lock_ptr;
995 barrier(); 1001 barrier();
996 if (lock_ptr != NULL) { 1002 if (lock_ptr != NULL) {
@@ -1172,11 +1178,11 @@ static int futex_wait(u32 __user *uaddr, int fshared,
1172 1178
1173 q.pi_state = NULL; 1179 q.pi_state = NULL;
1174 q.bitset = bitset; 1180 q.bitset = bitset;
1175 retry: 1181retry:
1176 q.key = FUTEX_KEY_INIT; 1182 q.key = FUTEX_KEY_INIT;
1177 ret = get_futex_key(uaddr, fshared, &q.key); 1183 ret = get_futex_key(uaddr, fshared, &q.key);
1178 if (unlikely(ret != 0)) 1184 if (unlikely(ret != 0))
1179 goto out_release_sem; 1185 goto out;
1180 1186
1181 hb = queue_lock(&q); 1187 hb = queue_lock(&q);
1182 1188
@@ -1204,6 +1210,7 @@ static int futex_wait(u32 __user *uaddr, int fshared,
1204 1210
1205 if (unlikely(ret)) { 1211 if (unlikely(ret)) {
1206 queue_unlock(&q, hb); 1212 queue_unlock(&q, hb);
1213 put_futex_key(fshared, &q.key);
1207 1214
1208 ret = get_user(uval, uaddr); 1215 ret = get_user(uval, uaddr);
1209 1216
@@ -1213,7 +1220,7 @@ static int futex_wait(u32 __user *uaddr, int fshared,
1213 } 1220 }
1214 ret = -EWOULDBLOCK; 1221 ret = -EWOULDBLOCK;
1215 if (uval != val) 1222 if (uval != val)
1216 goto out_unlock_release_sem; 1223 goto out_unlock_put_key;
1217 1224
1218 /* Only actually queue if *uaddr contained val. */ 1225 /* Only actually queue if *uaddr contained val. */
1219 queue_me(&q, hb); 1226 queue_me(&q, hb);
@@ -1305,11 +1312,11 @@ static int futex_wait(u32 __user *uaddr, int fshared,
1305 return -ERESTART_RESTARTBLOCK; 1312 return -ERESTART_RESTARTBLOCK;
1306 } 1313 }
1307 1314
1308 out_unlock_release_sem: 1315out_unlock_put_key:
1309 queue_unlock(&q, hb); 1316 queue_unlock(&q, hb);
1310
1311 out_release_sem:
1312 put_futex_key(fshared, &q.key); 1317 put_futex_key(fshared, &q.key);
1318
1319out:
1313 return ret; 1320 return ret;
1314} 1321}
1315 1322
@@ -1358,16 +1365,16 @@ static int futex_lock_pi(u32 __user *uaddr, int fshared,
1358 } 1365 }
1359 1366
1360 q.pi_state = NULL; 1367 q.pi_state = NULL;
1361 retry: 1368retry:
1362 q.key = FUTEX_KEY_INIT; 1369 q.key = FUTEX_KEY_INIT;
1363 ret = get_futex_key(uaddr, fshared, &q.key); 1370 ret = get_futex_key(uaddr, fshared, &q.key);
1364 if (unlikely(ret != 0)) 1371 if (unlikely(ret != 0))
1365 goto out_release_sem; 1372 goto out;
1366 1373
1367 retry_unlocked: 1374retry_unlocked:
1368 hb = queue_lock(&q); 1375 hb = queue_lock(&q);
1369 1376
1370 retry_locked: 1377retry_locked:
1371 ret = lock_taken = 0; 1378 ret = lock_taken = 0;
1372 1379
1373 /* 1380 /*
@@ -1388,14 +1395,14 @@ static int futex_lock_pi(u32 __user *uaddr, int fshared,
1388 */ 1395 */
1389 if (unlikely((curval & FUTEX_TID_MASK) == task_pid_vnr(current))) { 1396 if (unlikely((curval & FUTEX_TID_MASK) == task_pid_vnr(current))) {
1390 ret = -EDEADLK; 1397 ret = -EDEADLK;
1391 goto out_unlock_release_sem; 1398 goto out_unlock_put_key;
1392 } 1399 }
1393 1400
1394 /* 1401 /*
1395 * Surprise - we got the lock. Just return to userspace: 1402 * Surprise - we got the lock. Just return to userspace:
1396 */ 1403 */
1397 if (unlikely(!curval)) 1404 if (unlikely(!curval))
1398 goto out_unlock_release_sem; 1405 goto out_unlock_put_key;
1399 1406
1400 uval = curval; 1407 uval = curval;
1401 1408
@@ -1431,7 +1438,7 @@ static int futex_lock_pi(u32 __user *uaddr, int fshared,
1431 * We took the lock due to owner died take over. 1438 * We took the lock due to owner died take over.
1432 */ 1439 */
1433 if (unlikely(lock_taken)) 1440 if (unlikely(lock_taken))
1434 goto out_unlock_release_sem; 1441 goto out_unlock_put_key;
1435 1442
1436 /* 1443 /*
1437 * We dont have the lock. Look up the PI state (or create it if 1444 * We dont have the lock. Look up the PI state (or create it if
@@ -1470,7 +1477,7 @@ static int futex_lock_pi(u32 __user *uaddr, int fshared,
1470 goto retry_locked; 1477 goto retry_locked;
1471 } 1478 }
1472 default: 1479 default:
1473 goto out_unlock_release_sem; 1480 goto out_unlock_put_key;
1474 } 1481 }
1475 } 1482 }
1476 1483
@@ -1561,16 +1568,17 @@ static int futex_lock_pi(u32 __user *uaddr, int fshared,
1561 destroy_hrtimer_on_stack(&to->timer); 1568 destroy_hrtimer_on_stack(&to->timer);
1562 return ret != -EINTR ? ret : -ERESTARTNOINTR; 1569 return ret != -EINTR ? ret : -ERESTARTNOINTR;
1563 1570
1564 out_unlock_release_sem: 1571out_unlock_put_key:
1565 queue_unlock(&q, hb); 1572 queue_unlock(&q, hb);
1566 1573
1567 out_release_sem: 1574out_put_key:
1568 put_futex_key(fshared, &q.key); 1575 put_futex_key(fshared, &q.key);
1576out:
1569 if (to) 1577 if (to)
1570 destroy_hrtimer_on_stack(&to->timer); 1578 destroy_hrtimer_on_stack(&to->timer);
1571 return ret; 1579 return ret;
1572 1580
1573 uaddr_faulted: 1581uaddr_faulted:
1574 /* 1582 /*
1575 * We have to r/w *(int __user *)uaddr, and we have to modify it 1583 * We have to r/w *(int __user *)uaddr, and we have to modify it
1576 * atomically. Therefore, if we continue to fault after get_user() 1584 * atomically. Therefore, if we continue to fault after get_user()
@@ -1583,7 +1591,7 @@ static int futex_lock_pi(u32 __user *uaddr, int fshared,
1583 if (attempt++) { 1591 if (attempt++) {
1584 ret = futex_handle_fault((unsigned long)uaddr, attempt); 1592 ret = futex_handle_fault((unsigned long)uaddr, attempt);
1585 if (ret) 1593 if (ret)
1586 goto out_release_sem; 1594 goto out_put_key;
1587 goto retry_unlocked; 1595 goto retry_unlocked;
1588 } 1596 }
1589 1597
@@ -1675,9 +1683,9 @@ retry_unlocked:
1675 1683
1676out_unlock: 1684out_unlock:
1677 spin_unlock(&hb->lock); 1685 spin_unlock(&hb->lock);
1678out:
1679 put_futex_key(fshared, &key); 1686 put_futex_key(fshared, &key);
1680 1687
1688out:
1681 return ret; 1689 return ret;
1682 1690
1683pi_faulted: 1691pi_faulted:
diff --git a/kernel/kmod.c b/kernel/kmod.c
index b46dbb908669..a27a5f64443d 100644
--- a/kernel/kmod.c
+++ b/kernel/kmod.c
@@ -51,8 +51,8 @@ char modprobe_path[KMOD_PATH_LEN] = "/sbin/modprobe";
51 51
52/** 52/**
53 * request_module - try to load a kernel module 53 * request_module - try to load a kernel module
54 * @fmt: printf style format string for the name of the module 54 * @fmt: printf style format string for the name of the module
55 * @varargs: arguements as specified in the format string 55 * @...: arguments as specified in the format string
56 * 56 *
57 * Load a module using the user mode module loader. The function returns 57 * Load a module using the user mode module loader. The function returns
58 * zero on success or a negative errno code on failure. Note that a 58 * zero on success or a negative errno code on failure. Note that a
diff --git a/kernel/kprobes.c b/kernel/kprobes.c
index 9f8a3f25259a..1b9cbdc0127a 100644
--- a/kernel/kprobes.c
+++ b/kernel/kprobes.c
@@ -69,7 +69,7 @@ static struct hlist_head kretprobe_inst_table[KPROBE_TABLE_SIZE];
69/* NOTE: change this value only with kprobe_mutex held */ 69/* NOTE: change this value only with kprobe_mutex held */
70static bool kprobe_enabled; 70static bool kprobe_enabled;
71 71
72DEFINE_MUTEX(kprobe_mutex); /* Protects kprobe_table */ 72static DEFINE_MUTEX(kprobe_mutex); /* Protects kprobe_table */
73static DEFINE_PER_CPU(struct kprobe *, kprobe_instance) = NULL; 73static DEFINE_PER_CPU(struct kprobe *, kprobe_instance) = NULL;
74static struct { 74static struct {
75 spinlock_t lock ____cacheline_aligned_in_smp; 75 spinlock_t lock ____cacheline_aligned_in_smp;
@@ -115,6 +115,7 @@ enum kprobe_slot_state {
115 SLOT_USED = 2, 115 SLOT_USED = 2,
116}; 116};
117 117
118static DEFINE_MUTEX(kprobe_insn_mutex); /* Protects kprobe_insn_pages */
118static struct hlist_head kprobe_insn_pages; 119static struct hlist_head kprobe_insn_pages;
119static int kprobe_garbage_slots; 120static int kprobe_garbage_slots;
120static int collect_garbage_slots(void); 121static int collect_garbage_slots(void);
@@ -144,10 +145,10 @@ loop_end:
144} 145}
145 146
146/** 147/**
147 * get_insn_slot() - Find a slot on an executable page for an instruction. 148 * __get_insn_slot() - Find a slot on an executable page for an instruction.
148 * We allocate an executable page if there's no room on existing ones. 149 * We allocate an executable page if there's no room on existing ones.
149 */ 150 */
150kprobe_opcode_t __kprobes *get_insn_slot(void) 151static kprobe_opcode_t __kprobes *__get_insn_slot(void)
151{ 152{
152 struct kprobe_insn_page *kip; 153 struct kprobe_insn_page *kip;
153 struct hlist_node *pos; 154 struct hlist_node *pos;
@@ -196,6 +197,15 @@ kprobe_opcode_t __kprobes *get_insn_slot(void)
196 return kip->insns; 197 return kip->insns;
197} 198}
198 199
200kprobe_opcode_t __kprobes *get_insn_slot(void)
201{
202 kprobe_opcode_t *ret;
203 mutex_lock(&kprobe_insn_mutex);
204 ret = __get_insn_slot();
205 mutex_unlock(&kprobe_insn_mutex);
206 return ret;
207}
208
199/* Return 1 if all garbages are collected, otherwise 0. */ 209/* Return 1 if all garbages are collected, otherwise 0. */
200static int __kprobes collect_one_slot(struct kprobe_insn_page *kip, int idx) 210static int __kprobes collect_one_slot(struct kprobe_insn_page *kip, int idx)
201{ 211{
@@ -226,9 +236,13 @@ static int __kprobes collect_garbage_slots(void)
226{ 236{
227 struct kprobe_insn_page *kip; 237 struct kprobe_insn_page *kip;
228 struct hlist_node *pos, *next; 238 struct hlist_node *pos, *next;
239 int safety;
229 240
230 /* Ensure no-one is preepmted on the garbages */ 241 /* Ensure no-one is preepmted on the garbages */
231 if (check_safety() != 0) 242 mutex_unlock(&kprobe_insn_mutex);
243 safety = check_safety();
244 mutex_lock(&kprobe_insn_mutex);
245 if (safety != 0)
232 return -EAGAIN; 246 return -EAGAIN;
233 247
234 hlist_for_each_entry_safe(kip, pos, next, &kprobe_insn_pages, hlist) { 248 hlist_for_each_entry_safe(kip, pos, next, &kprobe_insn_pages, hlist) {
@@ -251,6 +265,7 @@ void __kprobes free_insn_slot(kprobe_opcode_t * slot, int dirty)
251 struct kprobe_insn_page *kip; 265 struct kprobe_insn_page *kip;
252 struct hlist_node *pos; 266 struct hlist_node *pos;
253 267
268 mutex_lock(&kprobe_insn_mutex);
254 hlist_for_each_entry(kip, pos, &kprobe_insn_pages, hlist) { 269 hlist_for_each_entry(kip, pos, &kprobe_insn_pages, hlist) {
255 if (kip->insns <= slot && 270 if (kip->insns <= slot &&
256 slot < kip->insns + (INSNS_PER_PAGE * MAX_INSN_SIZE)) { 271 slot < kip->insns + (INSNS_PER_PAGE * MAX_INSN_SIZE)) {
@@ -267,6 +282,8 @@ void __kprobes free_insn_slot(kprobe_opcode_t * slot, int dirty)
267 282
268 if (dirty && ++kprobe_garbage_slots > INSNS_PER_PAGE) 283 if (dirty && ++kprobe_garbage_slots > INSNS_PER_PAGE)
269 collect_garbage_slots(); 284 collect_garbage_slots();
285
286 mutex_unlock(&kprobe_insn_mutex);
270} 287}
271#endif 288#endif
272 289
@@ -310,7 +327,7 @@ static int __kprobes aggr_pre_handler(struct kprobe *p, struct pt_regs *regs)
310 struct kprobe *kp; 327 struct kprobe *kp;
311 328
312 list_for_each_entry_rcu(kp, &p->list, list) { 329 list_for_each_entry_rcu(kp, &p->list, list) {
313 if (kp->pre_handler) { 330 if (kp->pre_handler && !kprobe_gone(kp)) {
314 set_kprobe_instance(kp); 331 set_kprobe_instance(kp);
315 if (kp->pre_handler(kp, regs)) 332 if (kp->pre_handler(kp, regs))
316 return 1; 333 return 1;
@@ -326,7 +343,7 @@ static void __kprobes aggr_post_handler(struct kprobe *p, struct pt_regs *regs,
326 struct kprobe *kp; 343 struct kprobe *kp;
327 344
328 list_for_each_entry_rcu(kp, &p->list, list) { 345 list_for_each_entry_rcu(kp, &p->list, list) {
329 if (kp->post_handler) { 346 if (kp->post_handler && !kprobe_gone(kp)) {
330 set_kprobe_instance(kp); 347 set_kprobe_instance(kp);
331 kp->post_handler(kp, regs, flags); 348 kp->post_handler(kp, regs, flags);
332 reset_kprobe_instance(); 349 reset_kprobe_instance();
@@ -393,7 +410,7 @@ void __kprobes recycle_rp_inst(struct kretprobe_instance *ri,
393 hlist_add_head(&ri->hlist, head); 410 hlist_add_head(&ri->hlist, head);
394} 411}
395 412
396void kretprobe_hash_lock(struct task_struct *tsk, 413void __kprobes kretprobe_hash_lock(struct task_struct *tsk,
397 struct hlist_head **head, unsigned long *flags) 414 struct hlist_head **head, unsigned long *flags)
398{ 415{
399 unsigned long hash = hash_ptr(tsk, KPROBE_HASH_BITS); 416 unsigned long hash = hash_ptr(tsk, KPROBE_HASH_BITS);
@@ -404,13 +421,15 @@ void kretprobe_hash_lock(struct task_struct *tsk,
404 spin_lock_irqsave(hlist_lock, *flags); 421 spin_lock_irqsave(hlist_lock, *flags);
405} 422}
406 423
407static void kretprobe_table_lock(unsigned long hash, unsigned long *flags) 424static void __kprobes kretprobe_table_lock(unsigned long hash,
425 unsigned long *flags)
408{ 426{
409 spinlock_t *hlist_lock = kretprobe_table_lock_ptr(hash); 427 spinlock_t *hlist_lock = kretprobe_table_lock_ptr(hash);
410 spin_lock_irqsave(hlist_lock, *flags); 428 spin_lock_irqsave(hlist_lock, *flags);
411} 429}
412 430
413void kretprobe_hash_unlock(struct task_struct *tsk, unsigned long *flags) 431void __kprobes kretprobe_hash_unlock(struct task_struct *tsk,
432 unsigned long *flags)
414{ 433{
415 unsigned long hash = hash_ptr(tsk, KPROBE_HASH_BITS); 434 unsigned long hash = hash_ptr(tsk, KPROBE_HASH_BITS);
416 spinlock_t *hlist_lock; 435 spinlock_t *hlist_lock;
@@ -419,7 +438,7 @@ void kretprobe_hash_unlock(struct task_struct *tsk, unsigned long *flags)
419 spin_unlock_irqrestore(hlist_lock, *flags); 438 spin_unlock_irqrestore(hlist_lock, *flags);
420} 439}
421 440
422void kretprobe_table_unlock(unsigned long hash, unsigned long *flags) 441void __kprobes kretprobe_table_unlock(unsigned long hash, unsigned long *flags)
423{ 442{
424 spinlock_t *hlist_lock = kretprobe_table_lock_ptr(hash); 443 spinlock_t *hlist_lock = kretprobe_table_lock_ptr(hash);
425 spin_unlock_irqrestore(hlist_lock, *flags); 444 spin_unlock_irqrestore(hlist_lock, *flags);
@@ -526,9 +545,10 @@ static inline void add_aggr_kprobe(struct kprobe *ap, struct kprobe *p)
526 ap->addr = p->addr; 545 ap->addr = p->addr;
527 ap->pre_handler = aggr_pre_handler; 546 ap->pre_handler = aggr_pre_handler;
528 ap->fault_handler = aggr_fault_handler; 547 ap->fault_handler = aggr_fault_handler;
529 if (p->post_handler) 548 /* We don't care the kprobe which has gone. */
549 if (p->post_handler && !kprobe_gone(p))
530 ap->post_handler = aggr_post_handler; 550 ap->post_handler = aggr_post_handler;
531 if (p->break_handler) 551 if (p->break_handler && !kprobe_gone(p))
532 ap->break_handler = aggr_break_handler; 552 ap->break_handler = aggr_break_handler;
533 553
534 INIT_LIST_HEAD(&ap->list); 554 INIT_LIST_HEAD(&ap->list);
@@ -547,17 +567,41 @@ static int __kprobes register_aggr_kprobe(struct kprobe *old_p,
547 int ret = 0; 567 int ret = 0;
548 struct kprobe *ap; 568 struct kprobe *ap;
549 569
570 if (kprobe_gone(old_p)) {
571 /*
572 * Attempting to insert new probe at the same location that
573 * had a probe in the module vaddr area which already
574 * freed. So, the instruction slot has already been
575 * released. We need a new slot for the new probe.
576 */
577 ret = arch_prepare_kprobe(old_p);
578 if (ret)
579 return ret;
580 }
550 if (old_p->pre_handler == aggr_pre_handler) { 581 if (old_p->pre_handler == aggr_pre_handler) {
551 copy_kprobe(old_p, p); 582 copy_kprobe(old_p, p);
552 ret = add_new_kprobe(old_p, p); 583 ret = add_new_kprobe(old_p, p);
584 ap = old_p;
553 } else { 585 } else {
554 ap = kzalloc(sizeof(struct kprobe), GFP_KERNEL); 586 ap = kzalloc(sizeof(struct kprobe), GFP_KERNEL);
555 if (!ap) 587 if (!ap) {
588 if (kprobe_gone(old_p))
589 arch_remove_kprobe(old_p);
556 return -ENOMEM; 590 return -ENOMEM;
591 }
557 add_aggr_kprobe(ap, old_p); 592 add_aggr_kprobe(ap, old_p);
558 copy_kprobe(ap, p); 593 copy_kprobe(ap, p);
559 ret = add_new_kprobe(ap, p); 594 ret = add_new_kprobe(ap, p);
560 } 595 }
596 if (kprobe_gone(old_p)) {
597 /*
598 * If the old_p has gone, its breakpoint has been disarmed.
599 * We have to arm it again after preparing real kprobes.
600 */
601 ap->flags &= ~KPROBE_FLAG_GONE;
602 if (kprobe_enabled)
603 arch_arm_kprobe(ap);
604 }
561 return ret; 605 return ret;
562} 606}
563 607
@@ -600,8 +644,7 @@ static kprobe_opcode_t __kprobes *kprobe_addr(struct kprobe *p)
600 return (kprobe_opcode_t *)(((char *)addr) + p->offset); 644 return (kprobe_opcode_t *)(((char *)addr) + p->offset);
601} 645}
602 646
603static int __kprobes __register_kprobe(struct kprobe *p, 647int __kprobes register_kprobe(struct kprobe *p)
604 unsigned long called_from)
605{ 648{
606 int ret = 0; 649 int ret = 0;
607 struct kprobe *old_p; 650 struct kprobe *old_p;
@@ -620,28 +663,30 @@ static int __kprobes __register_kprobe(struct kprobe *p,
620 return -EINVAL; 663 return -EINVAL;
621 } 664 }
622 665
623 p->mod_refcounted = 0; 666 p->flags = 0;
624
625 /* 667 /*
626 * Check if are we probing a module. 668 * Check if are we probing a module.
627 */ 669 */
628 probed_mod = __module_text_address((unsigned long) p->addr); 670 probed_mod = __module_text_address((unsigned long) p->addr);
629 if (probed_mod) { 671 if (probed_mod) {
630 struct module *calling_mod;
631 calling_mod = __module_text_address(called_from);
632 /* 672 /*
633 * We must allow modules to probe themself and in this case 673 * We must hold a refcount of the probed module while updating
634 * avoid incrementing the module refcount, so as to allow 674 * its code to prohibit unexpected unloading.
635 * unloading of self probing modules.
636 */ 675 */
637 if (calling_mod && calling_mod != probed_mod) { 676 if (unlikely(!try_module_get(probed_mod))) {
638 if (unlikely(!try_module_get(probed_mod))) { 677 preempt_enable();
639 preempt_enable(); 678 return -EINVAL;
640 return -EINVAL; 679 }
641 } 680 /*
642 p->mod_refcounted = 1; 681 * If the module freed .init.text, we couldn't insert
643 } else 682 * kprobes in there.
644 probed_mod = NULL; 683 */
684 if (within_module_init((unsigned long)p->addr, probed_mod) &&
685 probed_mod->state != MODULE_STATE_COMING) {
686 module_put(probed_mod);
687 preempt_enable();
688 return -EINVAL;
689 }
645 } 690 }
646 preempt_enable(); 691 preempt_enable();
647 692
@@ -668,8 +713,9 @@ static int __kprobes __register_kprobe(struct kprobe *p,
668out: 713out:
669 mutex_unlock(&kprobe_mutex); 714 mutex_unlock(&kprobe_mutex);
670 715
671 if (ret && probed_mod) 716 if (probed_mod)
672 module_put(probed_mod); 717 module_put(probed_mod);
718
673 return ret; 719 return ret;
674} 720}
675 721
@@ -697,16 +743,16 @@ valid_p:
697 list_is_singular(&old_p->list))) { 743 list_is_singular(&old_p->list))) {
698 /* 744 /*
699 * Only probe on the hash list. Disarm only if kprobes are 745 * Only probe on the hash list. Disarm only if kprobes are
700 * enabled - otherwise, the breakpoint would already have 746 * enabled and not gone - otherwise, the breakpoint would
701 * been removed. We save on flushing icache. 747 * already have been removed. We save on flushing icache.
702 */ 748 */
703 if (kprobe_enabled) 749 if (kprobe_enabled && !kprobe_gone(old_p))
704 arch_disarm_kprobe(p); 750 arch_disarm_kprobe(p);
705 hlist_del_rcu(&old_p->hlist); 751 hlist_del_rcu(&old_p->hlist);
706 } else { 752 } else {
707 if (p->break_handler) 753 if (p->break_handler && !kprobe_gone(p))
708 old_p->break_handler = NULL; 754 old_p->break_handler = NULL;
709 if (p->post_handler) { 755 if (p->post_handler && !kprobe_gone(p)) {
710 list_for_each_entry_rcu(list_p, &old_p->list, list) { 756 list_for_each_entry_rcu(list_p, &old_p->list, list) {
711 if ((list_p != p) && (list_p->post_handler)) 757 if ((list_p != p) && (list_p->post_handler))
712 goto noclean; 758 goto noclean;
@@ -721,39 +767,27 @@ noclean:
721 767
722static void __kprobes __unregister_kprobe_bottom(struct kprobe *p) 768static void __kprobes __unregister_kprobe_bottom(struct kprobe *p)
723{ 769{
724 struct module *mod;
725 struct kprobe *old_p; 770 struct kprobe *old_p;
726 771
727 if (p->mod_refcounted) { 772 if (list_empty(&p->list))
728 /*
729 * Since we've already incremented refcount,
730 * we don't need to disable preemption.
731 */
732 mod = module_text_address((unsigned long)p->addr);
733 if (mod)
734 module_put(mod);
735 }
736
737 if (list_empty(&p->list) || list_is_singular(&p->list)) {
738 if (!list_empty(&p->list)) {
739 /* "p" is the last child of an aggr_kprobe */
740 old_p = list_entry(p->list.next, struct kprobe, list);
741 list_del(&p->list);
742 kfree(old_p);
743 }
744 arch_remove_kprobe(p); 773 arch_remove_kprobe(p);
774 else if (list_is_singular(&p->list)) {
775 /* "p" is the last child of an aggr_kprobe */
776 old_p = list_entry(p->list.next, struct kprobe, list);
777 list_del(&p->list);
778 arch_remove_kprobe(old_p);
779 kfree(old_p);
745 } 780 }
746} 781}
747 782
748static int __register_kprobes(struct kprobe **kps, int num, 783int __kprobes register_kprobes(struct kprobe **kps, int num)
749 unsigned long called_from)
750{ 784{
751 int i, ret = 0; 785 int i, ret = 0;
752 786
753 if (num <= 0) 787 if (num <= 0)
754 return -EINVAL; 788 return -EINVAL;
755 for (i = 0; i < num; i++) { 789 for (i = 0; i < num; i++) {
756 ret = __register_kprobe(kps[i], called_from); 790 ret = register_kprobe(kps[i]);
757 if (ret < 0) { 791 if (ret < 0) {
758 if (i > 0) 792 if (i > 0)
759 unregister_kprobes(kps, i); 793 unregister_kprobes(kps, i);
@@ -763,26 +797,11 @@ static int __register_kprobes(struct kprobe **kps, int num,
763 return ret; 797 return ret;
764} 798}
765 799
766/*
767 * Registration and unregistration functions for kprobe.
768 */
769int __kprobes register_kprobe(struct kprobe *p)
770{
771 return __register_kprobes(&p, 1,
772 (unsigned long)__builtin_return_address(0));
773}
774
775void __kprobes unregister_kprobe(struct kprobe *p) 800void __kprobes unregister_kprobe(struct kprobe *p)
776{ 801{
777 unregister_kprobes(&p, 1); 802 unregister_kprobes(&p, 1);
778} 803}
779 804
780int __kprobes register_kprobes(struct kprobe **kps, int num)
781{
782 return __register_kprobes(kps, num,
783 (unsigned long)__builtin_return_address(0));
784}
785
786void __kprobes unregister_kprobes(struct kprobe **kps, int num) 805void __kprobes unregister_kprobes(struct kprobe **kps, int num)
787{ 806{
788 int i; 807 int i;
@@ -811,8 +830,7 @@ unsigned long __weak arch_deref_entry_point(void *entry)
811 return (unsigned long)entry; 830 return (unsigned long)entry;
812} 831}
813 832
814static int __register_jprobes(struct jprobe **jps, int num, 833int __kprobes register_jprobes(struct jprobe **jps, int num)
815 unsigned long called_from)
816{ 834{
817 struct jprobe *jp; 835 struct jprobe *jp;
818 int ret = 0, i; 836 int ret = 0, i;
@@ -830,7 +848,7 @@ static int __register_jprobes(struct jprobe **jps, int num,
830 /* Todo: Verify probepoint is a function entry point */ 848 /* Todo: Verify probepoint is a function entry point */
831 jp->kp.pre_handler = setjmp_pre_handler; 849 jp->kp.pre_handler = setjmp_pre_handler;
832 jp->kp.break_handler = longjmp_break_handler; 850 jp->kp.break_handler = longjmp_break_handler;
833 ret = __register_kprobe(&jp->kp, called_from); 851 ret = register_kprobe(&jp->kp);
834 } 852 }
835 if (ret < 0) { 853 if (ret < 0) {
836 if (i > 0) 854 if (i > 0)
@@ -843,8 +861,7 @@ static int __register_jprobes(struct jprobe **jps, int num,
843 861
844int __kprobes register_jprobe(struct jprobe *jp) 862int __kprobes register_jprobe(struct jprobe *jp)
845{ 863{
846 return __register_jprobes(&jp, 1, 864 return register_jprobes(&jp, 1);
847 (unsigned long)__builtin_return_address(0));
848} 865}
849 866
850void __kprobes unregister_jprobe(struct jprobe *jp) 867void __kprobes unregister_jprobe(struct jprobe *jp)
@@ -852,12 +869,6 @@ void __kprobes unregister_jprobe(struct jprobe *jp)
852 unregister_jprobes(&jp, 1); 869 unregister_jprobes(&jp, 1);
853} 870}
854 871
855int __kprobes register_jprobes(struct jprobe **jps, int num)
856{
857 return __register_jprobes(jps, num,
858 (unsigned long)__builtin_return_address(0));
859}
860
861void __kprobes unregister_jprobes(struct jprobe **jps, int num) 872void __kprobes unregister_jprobes(struct jprobe **jps, int num)
862{ 873{
863 int i; 874 int i;
@@ -920,8 +931,7 @@ static int __kprobes pre_handler_kretprobe(struct kprobe *p,
920 return 0; 931 return 0;
921} 932}
922 933
923static int __kprobes __register_kretprobe(struct kretprobe *rp, 934int __kprobes register_kretprobe(struct kretprobe *rp)
924 unsigned long called_from)
925{ 935{
926 int ret = 0; 936 int ret = 0;
927 struct kretprobe_instance *inst; 937 struct kretprobe_instance *inst;
@@ -967,21 +977,20 @@ static int __kprobes __register_kretprobe(struct kretprobe *rp,
967 977
968 rp->nmissed = 0; 978 rp->nmissed = 0;
969 /* Establish function entry probe point */ 979 /* Establish function entry probe point */
970 ret = __register_kprobe(&rp->kp, called_from); 980 ret = register_kprobe(&rp->kp);
971 if (ret != 0) 981 if (ret != 0)
972 free_rp_inst(rp); 982 free_rp_inst(rp);
973 return ret; 983 return ret;
974} 984}
975 985
976static int __register_kretprobes(struct kretprobe **rps, int num, 986int __kprobes register_kretprobes(struct kretprobe **rps, int num)
977 unsigned long called_from)
978{ 987{
979 int ret = 0, i; 988 int ret = 0, i;
980 989
981 if (num <= 0) 990 if (num <= 0)
982 return -EINVAL; 991 return -EINVAL;
983 for (i = 0; i < num; i++) { 992 for (i = 0; i < num; i++) {
984 ret = __register_kretprobe(rps[i], called_from); 993 ret = register_kretprobe(rps[i]);
985 if (ret < 0) { 994 if (ret < 0) {
986 if (i > 0) 995 if (i > 0)
987 unregister_kretprobes(rps, i); 996 unregister_kretprobes(rps, i);
@@ -991,23 +1000,11 @@ static int __register_kretprobes(struct kretprobe **rps, int num,
991 return ret; 1000 return ret;
992} 1001}
993 1002
994int __kprobes register_kretprobe(struct kretprobe *rp)
995{
996 return __register_kretprobes(&rp, 1,
997 (unsigned long)__builtin_return_address(0));
998}
999
1000void __kprobes unregister_kretprobe(struct kretprobe *rp) 1003void __kprobes unregister_kretprobe(struct kretprobe *rp)
1001{ 1004{
1002 unregister_kretprobes(&rp, 1); 1005 unregister_kretprobes(&rp, 1);
1003} 1006}
1004 1007
1005int __kprobes register_kretprobes(struct kretprobe **rps, int num)
1006{
1007 return __register_kretprobes(rps, num,
1008 (unsigned long)__builtin_return_address(0));
1009}
1010
1011void __kprobes unregister_kretprobes(struct kretprobe **rps, int num) 1008void __kprobes unregister_kretprobes(struct kretprobe **rps, int num)
1012{ 1009{
1013 int i; 1010 int i;
@@ -1055,6 +1052,72 @@ static int __kprobes pre_handler_kretprobe(struct kprobe *p,
1055 1052
1056#endif /* CONFIG_KRETPROBES */ 1053#endif /* CONFIG_KRETPROBES */
1057 1054
1055/* Set the kprobe gone and remove its instruction buffer. */
1056static void __kprobes kill_kprobe(struct kprobe *p)
1057{
1058 struct kprobe *kp;
1059 p->flags |= KPROBE_FLAG_GONE;
1060 if (p->pre_handler == aggr_pre_handler) {
1061 /*
1062 * If this is an aggr_kprobe, we have to list all the
1063 * chained probes and mark them GONE.
1064 */
1065 list_for_each_entry_rcu(kp, &p->list, list)
1066 kp->flags |= KPROBE_FLAG_GONE;
1067 p->post_handler = NULL;
1068 p->break_handler = NULL;
1069 }
1070 /*
1071 * Here, we can remove insn_slot safely, because no thread calls
1072 * the original probed function (which will be freed soon) any more.
1073 */
1074 arch_remove_kprobe(p);
1075}
1076
1077/* Module notifier call back, checking kprobes on the module */
1078static int __kprobes kprobes_module_callback(struct notifier_block *nb,
1079 unsigned long val, void *data)
1080{
1081 struct module *mod = data;
1082 struct hlist_head *head;
1083 struct hlist_node *node;
1084 struct kprobe *p;
1085 unsigned int i;
1086 int checkcore = (val == MODULE_STATE_GOING);
1087
1088 if (val != MODULE_STATE_GOING && val != MODULE_STATE_LIVE)
1089 return NOTIFY_DONE;
1090
1091 /*
1092 * When MODULE_STATE_GOING was notified, both of module .text and
1093 * .init.text sections would be freed. When MODULE_STATE_LIVE was
1094 * notified, only .init.text section would be freed. We need to
1095 * disable kprobes which have been inserted in the sections.
1096 */
1097 mutex_lock(&kprobe_mutex);
1098 for (i = 0; i < KPROBE_TABLE_SIZE; i++) {
1099 head = &kprobe_table[i];
1100 hlist_for_each_entry_rcu(p, node, head, hlist)
1101 if (within_module_init((unsigned long)p->addr, mod) ||
1102 (checkcore &&
1103 within_module_core((unsigned long)p->addr, mod))) {
1104 /*
1105 * The vaddr this probe is installed will soon
1106 * be vfreed buy not synced to disk. Hence,
1107 * disarming the breakpoint isn't needed.
1108 */
1109 kill_kprobe(p);
1110 }
1111 }
1112 mutex_unlock(&kprobe_mutex);
1113 return NOTIFY_DONE;
1114}
1115
1116static struct notifier_block kprobe_module_nb = {
1117 .notifier_call = kprobes_module_callback,
1118 .priority = 0
1119};
1120
1058static int __init init_kprobes(void) 1121static int __init init_kprobes(void)
1059{ 1122{
1060 int i, err = 0; 1123 int i, err = 0;
@@ -1111,6 +1174,9 @@ static int __init init_kprobes(void)
1111 err = arch_init_kprobes(); 1174 err = arch_init_kprobes();
1112 if (!err) 1175 if (!err)
1113 err = register_die_notifier(&kprobe_exceptions_nb); 1176 err = register_die_notifier(&kprobe_exceptions_nb);
1177 if (!err)
1178 err = register_module_notifier(&kprobe_module_nb);
1179
1114 kprobes_initialized = (err == 0); 1180 kprobes_initialized = (err == 0);
1115 1181
1116 if (!err) 1182 if (!err)
@@ -1131,10 +1197,12 @@ static void __kprobes report_probe(struct seq_file *pi, struct kprobe *p,
1131 else 1197 else
1132 kprobe_type = "k"; 1198 kprobe_type = "k";
1133 if (sym) 1199 if (sym)
1134 seq_printf(pi, "%p %s %s+0x%x %s\n", p->addr, kprobe_type, 1200 seq_printf(pi, "%p %s %s+0x%x %s %s\n", p->addr, kprobe_type,
1135 sym, offset, (modname ? modname : " ")); 1201 sym, offset, (modname ? modname : " "),
1202 (kprobe_gone(p) ? "[GONE]" : ""));
1136 else 1203 else
1137 seq_printf(pi, "%p %s %p\n", p->addr, kprobe_type, p->addr); 1204 seq_printf(pi, "%p %s %p %s\n", p->addr, kprobe_type, p->addr,
1205 (kprobe_gone(p) ? "[GONE]" : ""));
1138} 1206}
1139 1207
1140static void __kprobes *kprobe_seq_start(struct seq_file *f, loff_t *pos) 1208static void __kprobes *kprobe_seq_start(struct seq_file *f, loff_t *pos)
@@ -1215,7 +1283,8 @@ static void __kprobes enable_all_kprobes(void)
1215 for (i = 0; i < KPROBE_TABLE_SIZE; i++) { 1283 for (i = 0; i < KPROBE_TABLE_SIZE; i++) {
1216 head = &kprobe_table[i]; 1284 head = &kprobe_table[i];
1217 hlist_for_each_entry_rcu(p, node, head, hlist) 1285 hlist_for_each_entry_rcu(p, node, head, hlist)
1218 arch_arm_kprobe(p); 1286 if (!kprobe_gone(p))
1287 arch_arm_kprobe(p);
1219 } 1288 }
1220 1289
1221 kprobe_enabled = true; 1290 kprobe_enabled = true;
@@ -1244,7 +1313,7 @@ static void __kprobes disable_all_kprobes(void)
1244 for (i = 0; i < KPROBE_TABLE_SIZE; i++) { 1313 for (i = 0; i < KPROBE_TABLE_SIZE; i++) {
1245 head = &kprobe_table[i]; 1314 head = &kprobe_table[i];
1246 hlist_for_each_entry_rcu(p, node, head, hlist) { 1315 hlist_for_each_entry_rcu(p, node, head, hlist) {
1247 if (!arch_trampoline_kprobe(p)) 1316 if (!arch_trampoline_kprobe(p) && !kprobe_gone(p))
1248 arch_disarm_kprobe(p); 1317 arch_disarm_kprobe(p);
1249 } 1318 }
1250 } 1319 }
diff --git a/kernel/ksysfs.c b/kernel/ksysfs.c
index 08dd8ed86c77..528dd78e7e7e 100644
--- a/kernel/ksysfs.c
+++ b/kernel/ksysfs.c
@@ -24,7 +24,7 @@ static struct kobj_attribute _name##_attr = __ATTR_RO(_name)
24static struct kobj_attribute _name##_attr = \ 24static struct kobj_attribute _name##_attr = \
25 __ATTR(_name, 0644, _name##_show, _name##_store) 25 __ATTR(_name, 0644, _name##_show, _name##_store)
26 26
27#if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET) 27#if defined(CONFIG_HOTPLUG)
28/* current uevent sequence number */ 28/* current uevent sequence number */
29static ssize_t uevent_seqnum_show(struct kobject *kobj, 29static ssize_t uevent_seqnum_show(struct kobject *kobj,
30 struct kobj_attribute *attr, char *buf) 30 struct kobj_attribute *attr, char *buf)
@@ -137,7 +137,7 @@ struct kobject *kernel_kobj;
137EXPORT_SYMBOL_GPL(kernel_kobj); 137EXPORT_SYMBOL_GPL(kernel_kobj);
138 138
139static struct attribute * kernel_attrs[] = { 139static struct attribute * kernel_attrs[] = {
140#if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET) 140#if defined(CONFIG_HOTPLUG)
141 &uevent_seqnum_attr.attr, 141 &uevent_seqnum_attr.attr,
142 &uevent_helper_attr.attr, 142 &uevent_helper_attr.attr,
143#endif 143#endif
diff --git a/kernel/module.c b/kernel/module.c
index dd2a54155b54..496dcb57b608 100644
--- a/kernel/module.c
+++ b/kernel/module.c
@@ -43,7 +43,6 @@
43#include <linux/device.h> 43#include <linux/device.h>
44#include <linux/string.h> 44#include <linux/string.h>
45#include <linux/mutex.h> 45#include <linux/mutex.h>
46#include <linux/unwind.h>
47#include <linux/rculist.h> 46#include <linux/rculist.h>
48#include <asm/uaccess.h> 47#include <asm/uaccess.h>
49#include <asm/cacheflush.h> 48#include <asm/cacheflush.h>
@@ -757,8 +756,16 @@ sys_delete_module(const char __user *name_user, unsigned int flags)
757 return -EFAULT; 756 return -EFAULT;
758 name[MODULE_NAME_LEN-1] = '\0'; 757 name[MODULE_NAME_LEN-1] = '\0';
759 758
760 if (mutex_lock_interruptible(&module_mutex) != 0) 759 /* Create stop_machine threads since free_module relies on
761 return -EINTR; 760 * a non-failing stop_machine call. */
761 ret = stop_machine_create();
762 if (ret)
763 return ret;
764
765 if (mutex_lock_interruptible(&module_mutex) != 0) {
766 ret = -EINTR;
767 goto out_stop;
768 }
762 769
763 mod = find_module(name); 770 mod = find_module(name);
764 if (!mod) { 771 if (!mod) {
@@ -817,10 +824,12 @@ sys_delete_module(const char __user *name_user, unsigned int flags)
817 824
818 out: 825 out:
819 mutex_unlock(&module_mutex); 826 mutex_unlock(&module_mutex);
827out_stop:
828 stop_machine_destroy();
820 return ret; 829 return ret;
821} 830}
822 831
823static void print_unload_info(struct seq_file *m, struct module *mod) 832static inline void print_unload_info(struct seq_file *m, struct module *mod)
824{ 833{
825 struct module_use *use; 834 struct module_use *use;
826 int printed_something = 0; 835 int printed_something = 0;
@@ -893,7 +902,7 @@ void module_put(struct module *module)
893EXPORT_SYMBOL(module_put); 902EXPORT_SYMBOL(module_put);
894 903
895#else /* !CONFIG_MODULE_UNLOAD */ 904#else /* !CONFIG_MODULE_UNLOAD */
896static void print_unload_info(struct seq_file *m, struct module *mod) 905static inline void print_unload_info(struct seq_file *m, struct module *mod)
897{ 906{
898 /* We don't know the usage count, or what modules are using. */ 907 /* We don't know the usage count, or what modules are using. */
899 seq_printf(m, " - -"); 908 seq_printf(m, " - -");
@@ -1439,8 +1448,6 @@ static void free_module(struct module *mod)
1439 remove_sect_attrs(mod); 1448 remove_sect_attrs(mod);
1440 mod_kobject_remove(mod); 1449 mod_kobject_remove(mod);
1441 1450
1442 unwind_remove_table(mod->unwind_info, 0);
1443
1444 /* Arch-specific cleanup. */ 1451 /* Arch-specific cleanup. */
1445 module_arch_cleanup(mod); 1452 module_arch_cleanup(mod);
1446 1453
@@ -1578,11 +1585,21 @@ static int simplify_symbols(Elf_Shdr *sechdrs,
1578 return ret; 1585 return ret;
1579} 1586}
1580 1587
1588/* Additional bytes needed by arch in front of individual sections */
1589unsigned int __weak arch_mod_section_prepend(struct module *mod,
1590 unsigned int section)
1591{
1592 /* default implementation just returns zero */
1593 return 0;
1594}
1595
1581/* Update size with this section: return offset. */ 1596/* Update size with this section: return offset. */
1582static long get_offset(unsigned int *size, Elf_Shdr *sechdr) 1597static long get_offset(struct module *mod, unsigned int *size,
1598 Elf_Shdr *sechdr, unsigned int section)
1583{ 1599{
1584 long ret; 1600 long ret;
1585 1601
1602 *size += arch_mod_section_prepend(mod, section);
1586 ret = ALIGN(*size, sechdr->sh_addralign ?: 1); 1603 ret = ALIGN(*size, sechdr->sh_addralign ?: 1);
1587 *size = ret + sechdr->sh_size; 1604 *size = ret + sechdr->sh_size;
1588 return ret; 1605 return ret;
@@ -1622,7 +1639,7 @@ static void layout_sections(struct module *mod,
1622 || strncmp(secstrings + s->sh_name, 1639 || strncmp(secstrings + s->sh_name,
1623 ".init", 5) == 0) 1640 ".init", 5) == 0)
1624 continue; 1641 continue;
1625 s->sh_entsize = get_offset(&mod->core_size, s); 1642 s->sh_entsize = get_offset(mod, &mod->core_size, s, i);
1626 DEBUGP("\t%s\n", secstrings + s->sh_name); 1643 DEBUGP("\t%s\n", secstrings + s->sh_name);
1627 } 1644 }
1628 if (m == 0) 1645 if (m == 0)
@@ -1640,7 +1657,7 @@ static void layout_sections(struct module *mod,
1640 || strncmp(secstrings + s->sh_name, 1657 || strncmp(secstrings + s->sh_name,
1641 ".init", 5) != 0) 1658 ".init", 5) != 0)
1642 continue; 1659 continue;
1643 s->sh_entsize = (get_offset(&mod->init_size, s) 1660 s->sh_entsize = (get_offset(mod, &mod->init_size, s, i)
1644 | INIT_OFFSET_MASK); 1661 | INIT_OFFSET_MASK);
1645 DEBUGP("\t%s\n", secstrings + s->sh_name); 1662 DEBUGP("\t%s\n", secstrings + s->sh_name);
1646 } 1663 }
@@ -1725,15 +1742,15 @@ static const struct kernel_symbol *lookup_symbol(const char *name,
1725 return NULL; 1742 return NULL;
1726} 1743}
1727 1744
1728static int is_exported(const char *name, const struct module *mod) 1745static int is_exported(const char *name, unsigned long value,
1746 const struct module *mod)
1729{ 1747{
1730 if (!mod && lookup_symbol(name, __start___ksymtab, __stop___ksymtab)) 1748 const struct kernel_symbol *ks;
1731 return 1; 1749 if (!mod)
1750 ks = lookup_symbol(name, __start___ksymtab, __stop___ksymtab);
1732 else 1751 else
1733 if (mod && lookup_symbol(name, mod->syms, mod->syms + mod->num_syms)) 1752 ks = lookup_symbol(name, mod->syms, mod->syms + mod->num_syms);
1734 return 1; 1753 return ks != NULL && ks->value == value;
1735 else
1736 return 0;
1737} 1754}
1738 1755
1739/* As per nm */ 1756/* As per nm */
@@ -1847,7 +1864,6 @@ static noinline struct module *load_module(void __user *umod,
1847 unsigned int symindex = 0; 1864 unsigned int symindex = 0;
1848 unsigned int strindex = 0; 1865 unsigned int strindex = 0;
1849 unsigned int modindex, versindex, infoindex, pcpuindex; 1866 unsigned int modindex, versindex, infoindex, pcpuindex;
1850 unsigned int unwindex = 0;
1851 unsigned int num_kp, num_mcount; 1867 unsigned int num_kp, num_mcount;
1852 struct kernel_param *kp; 1868 struct kernel_param *kp;
1853 struct module *mod; 1869 struct module *mod;
@@ -1865,6 +1881,13 @@ static noinline struct module *load_module(void __user *umod,
1865 /* vmalloc barfs on "unusual" numbers. Check here */ 1881 /* vmalloc barfs on "unusual" numbers. Check here */
1866 if (len > 64 * 1024 * 1024 || (hdr = vmalloc(len)) == NULL) 1882 if (len > 64 * 1024 * 1024 || (hdr = vmalloc(len)) == NULL)
1867 return ERR_PTR(-ENOMEM); 1883 return ERR_PTR(-ENOMEM);
1884
1885 /* Create stop_machine threads since the error path relies on
1886 * a non-failing stop_machine call. */
1887 err = stop_machine_create();
1888 if (err)
1889 goto free_hdr;
1890
1868 if (copy_from_user(hdr, umod, len) != 0) { 1891 if (copy_from_user(hdr, umod, len) != 0) {
1869 err = -EFAULT; 1892 err = -EFAULT;
1870 goto free_hdr; 1893 goto free_hdr;
@@ -1930,9 +1953,6 @@ static noinline struct module *load_module(void __user *umod,
1930 versindex = find_sec(hdr, sechdrs, secstrings, "__versions"); 1953 versindex = find_sec(hdr, sechdrs, secstrings, "__versions");
1931 infoindex = find_sec(hdr, sechdrs, secstrings, ".modinfo"); 1954 infoindex = find_sec(hdr, sechdrs, secstrings, ".modinfo");
1932 pcpuindex = find_pcpusec(hdr, sechdrs, secstrings); 1955 pcpuindex = find_pcpusec(hdr, sechdrs, secstrings);
1933#ifdef ARCH_UNWIND_SECTION_NAME
1934 unwindex = find_sec(hdr, sechdrs, secstrings, ARCH_UNWIND_SECTION_NAME);
1935#endif
1936 1956
1937 /* Don't keep modinfo and version sections. */ 1957 /* Don't keep modinfo and version sections. */
1938 sechdrs[infoindex].sh_flags &= ~(unsigned long)SHF_ALLOC; 1958 sechdrs[infoindex].sh_flags &= ~(unsigned long)SHF_ALLOC;
@@ -1942,8 +1962,6 @@ static noinline struct module *load_module(void __user *umod,
1942 sechdrs[symindex].sh_flags |= SHF_ALLOC; 1962 sechdrs[symindex].sh_flags |= SHF_ALLOC;
1943 sechdrs[strindex].sh_flags |= SHF_ALLOC; 1963 sechdrs[strindex].sh_flags |= SHF_ALLOC;
1944#endif 1964#endif
1945 if (unwindex)
1946 sechdrs[unwindex].sh_flags |= SHF_ALLOC;
1947 1965
1948 /* Check module struct version now, before we try to use module. */ 1966 /* Check module struct version now, before we try to use module. */
1949 if (!check_modstruct_version(sechdrs, versindex, mod)) { 1967 if (!check_modstruct_version(sechdrs, versindex, mod)) {
@@ -2240,14 +2258,10 @@ static noinline struct module *load_module(void __user *umod,
2240 add_sect_attrs(mod, hdr->e_shnum, secstrings, sechdrs); 2258 add_sect_attrs(mod, hdr->e_shnum, secstrings, sechdrs);
2241 add_notes_attrs(mod, hdr->e_shnum, secstrings, sechdrs); 2259 add_notes_attrs(mod, hdr->e_shnum, secstrings, sechdrs);
2242 2260
2243 /* Size of section 0 is 0, so this works well if no unwind info. */
2244 mod->unwind_info = unwind_add_table(mod,
2245 (void *)sechdrs[unwindex].sh_addr,
2246 sechdrs[unwindex].sh_size);
2247
2248 /* Get rid of temporary copy */ 2261 /* Get rid of temporary copy */
2249 vfree(hdr); 2262 vfree(hdr);
2250 2263
2264 stop_machine_destroy();
2251 /* Done! */ 2265 /* Done! */
2252 return mod; 2266 return mod;
2253 2267
@@ -2270,6 +2284,7 @@ static noinline struct module *load_module(void __user *umod,
2270 kfree(args); 2284 kfree(args);
2271 free_hdr: 2285 free_hdr:
2272 vfree(hdr); 2286 vfree(hdr);
2287 stop_machine_destroy();
2273 return ERR_PTR(err); 2288 return ERR_PTR(err);
2274 2289
2275 truncated: 2290 truncated:
@@ -2337,11 +2352,12 @@ sys_init_module(void __user *umod,
2337 /* Now it's a first class citizen! Wake up anyone waiting for it. */ 2352 /* Now it's a first class citizen! Wake up anyone waiting for it. */
2338 mod->state = MODULE_STATE_LIVE; 2353 mod->state = MODULE_STATE_LIVE;
2339 wake_up(&module_wq); 2354 wake_up(&module_wq);
2355 blocking_notifier_call_chain(&module_notify_list,
2356 MODULE_STATE_LIVE, mod);
2340 2357
2341 mutex_lock(&module_mutex); 2358 mutex_lock(&module_mutex);
2342 /* Drop initial reference. */ 2359 /* Drop initial reference. */
2343 module_put(mod); 2360 module_put(mod);
2344 unwind_remove_table(mod->unwind_info, 1);
2345 module_free(mod, mod->module_init); 2361 module_free(mod, mod->module_init);
2346 mod->module_init = NULL; 2362 mod->module_init = NULL;
2347 mod->init_size = 0; 2363 mod->init_size = 0;
@@ -2376,7 +2392,7 @@ static const char *get_ksymbol(struct module *mod,
2376 unsigned long nextval; 2392 unsigned long nextval;
2377 2393
2378 /* At worse, next value is at end of module */ 2394 /* At worse, next value is at end of module */
2379 if (within(addr, mod->module_init, mod->init_size)) 2395 if (within_module_init(addr, mod))
2380 nextval = (unsigned long)mod->module_init+mod->init_text_size; 2396 nextval = (unsigned long)mod->module_init+mod->init_text_size;
2381 else 2397 else
2382 nextval = (unsigned long)mod->module_core+mod->core_text_size; 2398 nextval = (unsigned long)mod->module_core+mod->core_text_size;
@@ -2424,8 +2440,8 @@ const char *module_address_lookup(unsigned long addr,
2424 2440
2425 preempt_disable(); 2441 preempt_disable();
2426 list_for_each_entry_rcu(mod, &modules, list) { 2442 list_for_each_entry_rcu(mod, &modules, list) {
2427 if (within(addr, mod->module_init, mod->init_size) 2443 if (within_module_init(addr, mod) ||
2428 || within(addr, mod->module_core, mod->core_size)) { 2444 within_module_core(addr, mod)) {
2429 if (modname) 2445 if (modname)
2430 *modname = mod->name; 2446 *modname = mod->name;
2431 ret = get_ksymbol(mod, addr, size, offset); 2447 ret = get_ksymbol(mod, addr, size, offset);
@@ -2447,8 +2463,8 @@ int lookup_module_symbol_name(unsigned long addr, char *symname)
2447 2463
2448 preempt_disable(); 2464 preempt_disable();
2449 list_for_each_entry_rcu(mod, &modules, list) { 2465 list_for_each_entry_rcu(mod, &modules, list) {
2450 if (within(addr, mod->module_init, mod->init_size) || 2466 if (within_module_init(addr, mod) ||
2451 within(addr, mod->module_core, mod->core_size)) { 2467 within_module_core(addr, mod)) {
2452 const char *sym; 2468 const char *sym;
2453 2469
2454 sym = get_ksymbol(mod, addr, NULL, NULL); 2470 sym = get_ksymbol(mod, addr, NULL, NULL);
@@ -2471,8 +2487,8 @@ int lookup_module_symbol_attrs(unsigned long addr, unsigned long *size,
2471 2487
2472 preempt_disable(); 2488 preempt_disable();
2473 list_for_each_entry_rcu(mod, &modules, list) { 2489 list_for_each_entry_rcu(mod, &modules, list) {
2474 if (within(addr, mod->module_init, mod->init_size) || 2490 if (within_module_init(addr, mod) ||
2475 within(addr, mod->module_core, mod->core_size)) { 2491 within_module_core(addr, mod)) {
2476 const char *sym; 2492 const char *sym;
2477 2493
2478 sym = get_ksymbol(mod, addr, size, offset); 2494 sym = get_ksymbol(mod, addr, size, offset);
@@ -2504,7 +2520,7 @@ int module_get_kallsym(unsigned int symnum, unsigned long *value, char *type,
2504 strlcpy(name, mod->strtab + mod->symtab[symnum].st_name, 2520 strlcpy(name, mod->strtab + mod->symtab[symnum].st_name,
2505 KSYM_NAME_LEN); 2521 KSYM_NAME_LEN);
2506 strlcpy(module_name, mod->name, MODULE_NAME_LEN); 2522 strlcpy(module_name, mod->name, MODULE_NAME_LEN);
2507 *exported = is_exported(name, mod); 2523 *exported = is_exported(name, *value, mod);
2508 preempt_enable(); 2524 preempt_enable();
2509 return 0; 2525 return 0;
2510 } 2526 }
@@ -2691,7 +2707,7 @@ int is_module_address(unsigned long addr)
2691 preempt_disable(); 2707 preempt_disable();
2692 2708
2693 list_for_each_entry_rcu(mod, &modules, list) { 2709 list_for_each_entry_rcu(mod, &modules, list) {
2694 if (within(addr, mod->module_core, mod->core_size)) { 2710 if (within_module_core(addr, mod)) {
2695 preempt_enable(); 2711 preempt_enable();
2696 return 1; 2712 return 1;
2697 } 2713 }
diff --git a/kernel/panic.c b/kernel/panic.c
index 13f06349a786..2a2ff36ff44d 100644
--- a/kernel/panic.c
+++ b/kernel/panic.c
@@ -299,6 +299,8 @@ static int init_oops_id(void)
299{ 299{
300 if (!oops_id) 300 if (!oops_id)
301 get_random_bytes(&oops_id, sizeof(oops_id)); 301 get_random_bytes(&oops_id, sizeof(oops_id));
302 else
303 oops_id++;
302 304
303 return 0; 305 return 0;
304} 306}
diff --git a/kernel/power/main.c b/kernel/power/main.c
index 613f16941b85..239988873971 100644
--- a/kernel/power/main.c
+++ b/kernel/power/main.c
@@ -615,7 +615,7 @@ static void __init test_wakealarm(struct rtc_device *rtc, suspend_state_t state)
615 /* this may fail if the RTC hasn't been initialized */ 615 /* this may fail if the RTC hasn't been initialized */
616 status = rtc_read_time(rtc, &alm.time); 616 status = rtc_read_time(rtc, &alm.time);
617 if (status < 0) { 617 if (status < 0) {
618 printk(err_readtime, rtc->dev.bus_id, status); 618 printk(err_readtime, dev_name(&rtc->dev), status);
619 return; 619 return;
620 } 620 }
621 rtc_tm_to_time(&alm.time, &now); 621 rtc_tm_to_time(&alm.time, &now);
@@ -626,7 +626,7 @@ static void __init test_wakealarm(struct rtc_device *rtc, suspend_state_t state)
626 626
627 status = rtc_set_alarm(rtc, &alm); 627 status = rtc_set_alarm(rtc, &alm);
628 if (status < 0) { 628 if (status < 0) {
629 printk(err_wakealarm, rtc->dev.bus_id, status); 629 printk(err_wakealarm, dev_name(&rtc->dev), status);
630 return; 630 return;
631 } 631 }
632 632
@@ -660,7 +660,7 @@ static int __init has_wakealarm(struct device *dev, void *name_ptr)
660 if (!device_may_wakeup(candidate->dev.parent)) 660 if (!device_may_wakeup(candidate->dev.parent))
661 return 0; 661 return 0;
662 662
663 *(char **)name_ptr = dev->bus_id; 663 *(const char **)name_ptr = dev_name(dev);
664 return 1; 664 return 1;
665} 665}
666 666
diff --git a/kernel/profile.c b/kernel/profile.c
index d18e2d2654f2..784933acf5b8 100644
--- a/kernel/profile.c
+++ b/kernel/profile.c
@@ -445,7 +445,6 @@ void profile_tick(int type)
445#ifdef CONFIG_PROC_FS 445#ifdef CONFIG_PROC_FS
446#include <linux/proc_fs.h> 446#include <linux/proc_fs.h>
447#include <asm/uaccess.h> 447#include <asm/uaccess.h>
448#include <asm/ptrace.h>
449 448
450static int prof_cpu_mask_read_proc(char *page, char **start, off_t off, 449static int prof_cpu_mask_read_proc(char *page, char **start, off_t off,
451 int count, int *eof, void *data) 450 int count, int *eof, void *data)
diff --git a/kernel/rcupdate.c b/kernel/rcupdate.c
index ad63af8b2521..d92a76a881aa 100644
--- a/kernel/rcupdate.c
+++ b/kernel/rcupdate.c
@@ -77,8 +77,15 @@ void wakeme_after_rcu(struct rcu_head *head)
77 * sections are delimited by rcu_read_lock() and rcu_read_unlock(), 77 * sections are delimited by rcu_read_lock() and rcu_read_unlock(),
78 * and may be nested. 78 * and may be nested.
79 */ 79 */
80void synchronize_rcu(void); /* Makes kernel-doc tools happy */ 80void synchronize_rcu(void)
81synchronize_rcu_xxx(synchronize_rcu, call_rcu) 81{
82 struct rcu_synchronize rcu;
83 init_completion(&rcu.completion);
84 /* Will wake me after RCU finished. */
85 call_rcu(&rcu.head, wakeme_after_rcu);
86 /* Wait for it. */
87 wait_for_completion(&rcu.completion);
88}
82EXPORT_SYMBOL_GPL(synchronize_rcu); 89EXPORT_SYMBOL_GPL(synchronize_rcu);
83 90
84static void rcu_barrier_callback(struct rcu_head *notused) 91static void rcu_barrier_callback(struct rcu_head *notused)
diff --git a/kernel/rcupreempt.c b/kernel/rcupreempt.c
index f9dc8f3720f6..33cfc50781f9 100644
--- a/kernel/rcupreempt.c
+++ b/kernel/rcupreempt.c
@@ -1177,7 +1177,16 @@ EXPORT_SYMBOL_GPL(call_rcu_sched);
1177 * in -rt this does -not- necessarily result in all currently executing 1177 * in -rt this does -not- necessarily result in all currently executing
1178 * interrupt -handlers- having completed. 1178 * interrupt -handlers- having completed.
1179 */ 1179 */
1180synchronize_rcu_xxx(__synchronize_sched, call_rcu_sched) 1180void __synchronize_sched(void)
1181{
1182 struct rcu_synchronize rcu;
1183
1184 init_completion(&rcu.completion);
1185 /* Will wake me after RCU finished. */
1186 call_rcu_sched(&rcu.head, wakeme_after_rcu);
1187 /* Wait for it. */
1188 wait_for_completion(&rcu.completion);
1189}
1181EXPORT_SYMBOL_GPL(__synchronize_sched); 1190EXPORT_SYMBOL_GPL(__synchronize_sched);
1182 1191
1183/* 1192/*
diff --git a/kernel/rcutorture.c b/kernel/rcutorture.c
index 3245b40952c6..1cff28db56b6 100644
--- a/kernel/rcutorture.c
+++ b/kernel/rcutorture.c
@@ -136,7 +136,7 @@ static int stutter_pause_test = 0;
136#endif 136#endif
137int rcutorture_runnable = RCUTORTURE_RUNNABLE_INIT; 137int rcutorture_runnable = RCUTORTURE_RUNNABLE_INIT;
138 138
139#define FULLSTOP_SIGNALED 1 /* Bail due to signal. */ 139#define FULLSTOP_SHUTDOWN 1 /* Bail due to system shutdown/panic. */
140#define FULLSTOP_CLEANUP 2 /* Orderly shutdown. */ 140#define FULLSTOP_CLEANUP 2 /* Orderly shutdown. */
141static int fullstop; /* stop generating callbacks at test end. */ 141static int fullstop; /* stop generating callbacks at test end. */
142DEFINE_MUTEX(fullstop_mutex); /* protect fullstop transitions and */ 142DEFINE_MUTEX(fullstop_mutex); /* protect fullstop transitions and */
@@ -151,12 +151,10 @@ rcutorture_shutdown_notify(struct notifier_block *unused1,
151{ 151{
152 if (fullstop) 152 if (fullstop)
153 return NOTIFY_DONE; 153 return NOTIFY_DONE;
154 if (signal_pending(current)) { 154 mutex_lock(&fullstop_mutex);
155 mutex_lock(&fullstop_mutex); 155 if (!fullstop)
156 if (!ACCESS_ONCE(fullstop)) 156 fullstop = FULLSTOP_SHUTDOWN;
157 fullstop = FULLSTOP_SIGNALED; 157 mutex_unlock(&fullstop_mutex);
158 mutex_unlock(&fullstop_mutex);
159 }
160 return NOTIFY_DONE; 158 return NOTIFY_DONE;
161} 159}
162 160
@@ -624,7 +622,7 @@ rcu_torture_writer(void *arg)
624 rcu_stutter_wait(); 622 rcu_stutter_wait();
625 } while (!kthread_should_stop() && !fullstop); 623 } while (!kthread_should_stop() && !fullstop);
626 VERBOSE_PRINTK_STRING("rcu_torture_writer task stopping"); 624 VERBOSE_PRINTK_STRING("rcu_torture_writer task stopping");
627 while (!kthread_should_stop() && fullstop != FULLSTOP_SIGNALED) 625 while (!kthread_should_stop() && fullstop != FULLSTOP_SHUTDOWN)
628 schedule_timeout_uninterruptible(1); 626 schedule_timeout_uninterruptible(1);
629 return 0; 627 return 0;
630} 628}
@@ -649,7 +647,7 @@ rcu_torture_fakewriter(void *arg)
649 } while (!kthread_should_stop() && !fullstop); 647 } while (!kthread_should_stop() && !fullstop);
650 648
651 VERBOSE_PRINTK_STRING("rcu_torture_fakewriter task stopping"); 649 VERBOSE_PRINTK_STRING("rcu_torture_fakewriter task stopping");
652 while (!kthread_should_stop() && fullstop != FULLSTOP_SIGNALED) 650 while (!kthread_should_stop() && fullstop != FULLSTOP_SHUTDOWN)
653 schedule_timeout_uninterruptible(1); 651 schedule_timeout_uninterruptible(1);
654 return 0; 652 return 0;
655} 653}
@@ -759,7 +757,7 @@ rcu_torture_reader(void *arg)
759 VERBOSE_PRINTK_STRING("rcu_torture_reader task stopping"); 757 VERBOSE_PRINTK_STRING("rcu_torture_reader task stopping");
760 if (irqreader && cur_ops->irqcapable) 758 if (irqreader && cur_ops->irqcapable)
761 del_timer_sync(&t); 759 del_timer_sync(&t);
762 while (!kthread_should_stop() && fullstop != FULLSTOP_SIGNALED) 760 while (!kthread_should_stop() && fullstop != FULLSTOP_SHUTDOWN)
763 schedule_timeout_uninterruptible(1); 761 schedule_timeout_uninterruptible(1);
764 return 0; 762 return 0;
765} 763}
diff --git a/kernel/rcutree.c b/kernel/rcutree.c
index a342b032112c..f2d8638e6c60 100644
--- a/kernel/rcutree.c
+++ b/kernel/rcutree.c
@@ -79,7 +79,10 @@ struct rcu_state rcu_bh_state = RCU_STATE_INITIALIZER(rcu_bh_state);
79DEFINE_PER_CPU(struct rcu_data, rcu_bh_data); 79DEFINE_PER_CPU(struct rcu_data, rcu_bh_data);
80 80
81#ifdef CONFIG_NO_HZ 81#ifdef CONFIG_NO_HZ
82DEFINE_PER_CPU(struct rcu_dynticks, rcu_dynticks); 82DEFINE_PER_CPU(struct rcu_dynticks, rcu_dynticks) = {
83 .dynticks_nesting = 1,
84 .dynticks = 1,
85};
83#endif /* #ifdef CONFIG_NO_HZ */ 86#endif /* #ifdef CONFIG_NO_HZ */
84 87
85static int blimit = 10; /* Maximum callbacks per softirq. */ 88static int blimit = 10; /* Maximum callbacks per softirq. */
@@ -572,6 +575,7 @@ rcu_start_gp(struct rcu_state *rsp, unsigned long flags)
572 /* Special-case the common single-level case. */ 575 /* Special-case the common single-level case. */
573 if (NUM_RCU_NODES == 1) { 576 if (NUM_RCU_NODES == 1) {
574 rnp->qsmask = rnp->qsmaskinit; 577 rnp->qsmask = rnp->qsmaskinit;
578 rsp->signaled = RCU_SIGNAL_INIT; /* force_quiescent_state OK. */
575 spin_unlock_irqrestore(&rnp->lock, flags); 579 spin_unlock_irqrestore(&rnp->lock, flags);
576 return; 580 return;
577 } 581 }
@@ -1379,13 +1383,6 @@ rcu_init_percpu_data(int cpu, struct rcu_state *rsp)
1379 1383
1380static void __cpuinit rcu_online_cpu(int cpu) 1384static void __cpuinit rcu_online_cpu(int cpu)
1381{ 1385{
1382#ifdef CONFIG_NO_HZ
1383 struct rcu_dynticks *rdtp = &per_cpu(rcu_dynticks, cpu);
1384
1385 rdtp->dynticks_nesting = 1;
1386 rdtp->dynticks |= 1; /* need consecutive #s even for hotplug. */
1387 rdtp->dynticks_nmi = (rdtp->dynticks_nmi + 1) & ~0x1;
1388#endif /* #ifdef CONFIG_NO_HZ */
1389 rcu_init_percpu_data(cpu, &rcu_state); 1386 rcu_init_percpu_data(cpu, &rcu_state);
1390 rcu_init_percpu_data(cpu, &rcu_bh_state); 1387 rcu_init_percpu_data(cpu, &rcu_bh_state);
1391 open_softirq(RCU_SOFTIRQ, rcu_process_callbacks); 1388 open_softirq(RCU_SOFTIRQ, rcu_process_callbacks);
diff --git a/kernel/signal.c b/kernel/signal.c
index 8e95855ff3cf..3152ac3b62e2 100644
--- a/kernel/signal.c
+++ b/kernel/signal.c
@@ -858,7 +858,8 @@ static int send_signal(int sig, struct siginfo *info, struct task_struct *t,
858 q->info.si_signo = sig; 858 q->info.si_signo = sig;
859 q->info.si_errno = 0; 859 q->info.si_errno = 0;
860 q->info.si_code = SI_USER; 860 q->info.si_code = SI_USER;
861 q->info.si_pid = task_pid_vnr(current); 861 q->info.si_pid = task_tgid_nr_ns(current,
862 task_active_pid_ns(t));
862 q->info.si_uid = current_uid(); 863 q->info.si_uid = current_uid();
863 break; 864 break;
864 case (unsigned long) SEND_SIG_PRIV: 865 case (unsigned long) SEND_SIG_PRIV:
diff --git a/kernel/stop_machine.c b/kernel/stop_machine.c
index 286c41722e8c..0cd415ee62a2 100644
--- a/kernel/stop_machine.c
+++ b/kernel/stop_machine.c
@@ -38,7 +38,10 @@ struct stop_machine_data {
38static unsigned int num_threads; 38static unsigned int num_threads;
39static atomic_t thread_ack; 39static atomic_t thread_ack;
40static DEFINE_MUTEX(lock); 40static DEFINE_MUTEX(lock);
41 41/* setup_lock protects refcount, stop_machine_wq and stop_machine_work. */
42static DEFINE_MUTEX(setup_lock);
43/* Users of stop_machine. */
44static int refcount;
42static struct workqueue_struct *stop_machine_wq; 45static struct workqueue_struct *stop_machine_wq;
43static struct stop_machine_data active, idle; 46static struct stop_machine_data active, idle;
44static const cpumask_t *active_cpus; 47static const cpumask_t *active_cpus;
@@ -109,6 +112,43 @@ static int chill(void *unused)
109 return 0; 112 return 0;
110} 113}
111 114
115int stop_machine_create(void)
116{
117 mutex_lock(&setup_lock);
118 if (refcount)
119 goto done;
120 stop_machine_wq = create_rt_workqueue("kstop");
121 if (!stop_machine_wq)
122 goto err_out;
123 stop_machine_work = alloc_percpu(struct work_struct);
124 if (!stop_machine_work)
125 goto err_out;
126done:
127 refcount++;
128 mutex_unlock(&setup_lock);
129 return 0;
130
131err_out:
132 if (stop_machine_wq)
133 destroy_workqueue(stop_machine_wq);
134 mutex_unlock(&setup_lock);
135 return -ENOMEM;
136}
137EXPORT_SYMBOL_GPL(stop_machine_create);
138
139void stop_machine_destroy(void)
140{
141 mutex_lock(&setup_lock);
142 refcount--;
143 if (refcount)
144 goto done;
145 destroy_workqueue(stop_machine_wq);
146 free_percpu(stop_machine_work);
147done:
148 mutex_unlock(&setup_lock);
149}
150EXPORT_SYMBOL_GPL(stop_machine_destroy);
151
112int __stop_machine(int (*fn)(void *), void *data, const struct cpumask *cpus) 152int __stop_machine(int (*fn)(void *), void *data, const struct cpumask *cpus)
113{ 153{
114 struct work_struct *sm_work; 154 struct work_struct *sm_work;
@@ -146,19 +186,14 @@ int stop_machine(int (*fn)(void *), void *data, const struct cpumask *cpus)
146{ 186{
147 int ret; 187 int ret;
148 188
189 ret = stop_machine_create();
190 if (ret)
191 return ret;
149 /* No CPUs can come up or down during this. */ 192 /* No CPUs can come up or down during this. */
150 get_online_cpus(); 193 get_online_cpus();
151 ret = __stop_machine(fn, data, cpus); 194 ret = __stop_machine(fn, data, cpus);
152 put_online_cpus(); 195 put_online_cpus();
153 196 stop_machine_destroy();
154 return ret; 197 return ret;
155} 198}
156EXPORT_SYMBOL_GPL(stop_machine); 199EXPORT_SYMBOL_GPL(stop_machine);
157
158static int __init stop_machine_init(void)
159{
160 stop_machine_wq = create_rt_workqueue("kstop");
161 stop_machine_work = alloc_percpu(struct work_struct);
162 return 0;
163}
164core_initcall(stop_machine_init);
diff --git a/kernel/sys.c b/kernel/sys.c
index d356d79e84ac..4a43617cd565 100644
--- a/kernel/sys.c
+++ b/kernel/sys.c
@@ -33,6 +33,7 @@
33#include <linux/task_io_accounting_ops.h> 33#include <linux/task_io_accounting_ops.h>
34#include <linux/seccomp.h> 34#include <linux/seccomp.h>
35#include <linux/cpu.h> 35#include <linux/cpu.h>
36#include <linux/ptrace.h>
36 37
37#include <linux/compat.h> 38#include <linux/compat.h>
38#include <linux/syscalls.h> 39#include <linux/syscalls.h>
@@ -927,6 +928,7 @@ asmlinkage long sys_times(struct tms __user * tbuf)
927 if (copy_to_user(tbuf, &tmp, sizeof(struct tms))) 928 if (copy_to_user(tbuf, &tmp, sizeof(struct tms)))
928 return -EFAULT; 929 return -EFAULT;
929 } 930 }
931 force_successful_syscall_return();
930 return (long) jiffies_64_to_clock_t(get_jiffies_64()); 932 return (long) jiffies_64_to_clock_t(get_jiffies_64());
931} 933}
932 934
diff --git a/kernel/sysctl.c b/kernel/sysctl.c
index ff6d45c7626f..92f6e5bc3c24 100644
--- a/kernel/sysctl.c
+++ b/kernel/sysctl.c
@@ -87,10 +87,6 @@ extern int rcutorture_runnable;
87#endif /* #ifdef CONFIG_RCU_TORTURE_TEST */ 87#endif /* #ifdef CONFIG_RCU_TORTURE_TEST */
88 88
89/* Constants used for minimum and maximum */ 89/* Constants used for minimum and maximum */
90#if defined(CONFIG_HIGHMEM) || defined(CONFIG_DETECT_SOFTLOCKUP)
91static int one = 1;
92#endif
93
94#ifdef CONFIG_DETECT_SOFTLOCKUP 90#ifdef CONFIG_DETECT_SOFTLOCKUP
95static int sixty = 60; 91static int sixty = 60;
96static int neg_one = -1; 92static int neg_one = -1;
@@ -101,6 +97,7 @@ static int two = 2;
101#endif 97#endif
102 98
103static int zero; 99static int zero;
100static int one = 1;
104static int one_hundred = 100; 101static int one_hundred = 100;
105 102
106/* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */ 103/* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
@@ -952,12 +949,22 @@ static struct ctl_table vm_table[] = {
952 .data = &dirty_background_ratio, 949 .data = &dirty_background_ratio,
953 .maxlen = sizeof(dirty_background_ratio), 950 .maxlen = sizeof(dirty_background_ratio),
954 .mode = 0644, 951 .mode = 0644,
955 .proc_handler = &proc_dointvec_minmax, 952 .proc_handler = &dirty_background_ratio_handler,
956 .strategy = &sysctl_intvec, 953 .strategy = &sysctl_intvec,
957 .extra1 = &zero, 954 .extra1 = &zero,
958 .extra2 = &one_hundred, 955 .extra2 = &one_hundred,
959 }, 956 },
960 { 957 {
958 .ctl_name = CTL_UNNUMBERED,
959 .procname = "dirty_background_bytes",
960 .data = &dirty_background_bytes,
961 .maxlen = sizeof(dirty_background_bytes),
962 .mode = 0644,
963 .proc_handler = &dirty_background_bytes_handler,
964 .strategy = &sysctl_intvec,
965 .extra1 = &one,
966 },
967 {
961 .ctl_name = VM_DIRTY_RATIO, 968 .ctl_name = VM_DIRTY_RATIO,
962 .procname = "dirty_ratio", 969 .procname = "dirty_ratio",
963 .data = &vm_dirty_ratio, 970 .data = &vm_dirty_ratio,
@@ -969,6 +976,16 @@ static struct ctl_table vm_table[] = {
969 .extra2 = &one_hundred, 976 .extra2 = &one_hundred,
970 }, 977 },
971 { 978 {
979 .ctl_name = CTL_UNNUMBERED,
980 .procname = "dirty_bytes",
981 .data = &vm_dirty_bytes,
982 .maxlen = sizeof(vm_dirty_bytes),
983 .mode = 0644,
984 .proc_handler = &dirty_bytes_handler,
985 .strategy = &sysctl_intvec,
986 .extra1 = &one,
987 },
988 {
972 .procname = "dirty_writeback_centisecs", 989 .procname = "dirty_writeback_centisecs",
973 .data = &dirty_writeback_interval, 990 .data = &dirty_writeback_interval,
974 .maxlen = sizeof(dirty_writeback_interval), 991 .maxlen = sizeof(dirty_writeback_interval),
diff --git a/kernel/test_kprobes.c b/kernel/test_kprobes.c
index 06b6395b45b2..4f104515a19b 100644
--- a/kernel/test_kprobes.c
+++ b/kernel/test_kprobes.c
@@ -22,21 +22,11 @@
22 22
23static u32 rand1, preh_val, posth_val, jph_val; 23static u32 rand1, preh_val, posth_val, jph_val;
24static int errors, handler_errors, num_tests; 24static int errors, handler_errors, num_tests;
25static u32 (*target)(u32 value);
26static u32 (*target2)(u32 value);
25 27
26static noinline u32 kprobe_target(u32 value) 28static noinline u32 kprobe_target(u32 value)
27{ 29{
28 /*
29 * gcc ignores noinline on some architectures unless we stuff
30 * sufficient lard into the function. The get_kprobe() here is
31 * just for that.
32 *
33 * NOTE: We aren't concerned about the correctness of get_kprobe()
34 * here; hence, this call is neither under !preempt nor with the
35 * kprobe_mutex held. This is fine(tm)
36 */
37 if (get_kprobe((void *)0xdeadbeef))
38 printk(KERN_INFO "Kprobe smoke test: probe on 0xdeadbeef!\n");
39
40 return (value / div_factor); 30 return (value / div_factor);
41} 31}
42 32
@@ -74,7 +64,7 @@ static int test_kprobe(void)
74 return ret; 64 return ret;
75 } 65 }
76 66
77 ret = kprobe_target(rand1); 67 ret = target(rand1);
78 unregister_kprobe(&kp); 68 unregister_kprobe(&kp);
79 69
80 if (preh_val == 0) { 70 if (preh_val == 0) {
@@ -92,6 +82,84 @@ static int test_kprobe(void)
92 return 0; 82 return 0;
93} 83}
94 84
85static noinline u32 kprobe_target2(u32 value)
86{
87 return (value / div_factor) + 1;
88}
89
90static int kp_pre_handler2(struct kprobe *p, struct pt_regs *regs)
91{
92 preh_val = (rand1 / div_factor) + 1;
93 return 0;
94}
95
96static void kp_post_handler2(struct kprobe *p, struct pt_regs *regs,
97 unsigned long flags)
98{
99 if (preh_val != (rand1 / div_factor) + 1) {
100 handler_errors++;
101 printk(KERN_ERR "Kprobe smoke test failed: "
102 "incorrect value in post_handler2\n");
103 }
104 posth_val = preh_val + div_factor;
105}
106
107static struct kprobe kp2 = {
108 .symbol_name = "kprobe_target2",
109 .pre_handler = kp_pre_handler2,
110 .post_handler = kp_post_handler2
111};
112
113static int test_kprobes(void)
114{
115 int ret;
116 struct kprobe *kps[2] = {&kp, &kp2};
117
118 kp.addr = 0; /* addr should be cleard for reusing kprobe. */
119 ret = register_kprobes(kps, 2);
120 if (ret < 0) {
121 printk(KERN_ERR "Kprobe smoke test failed: "
122 "register_kprobes returned %d\n", ret);
123 return ret;
124 }
125
126 preh_val = 0;
127 posth_val = 0;
128 ret = target(rand1);
129
130 if (preh_val == 0) {
131 printk(KERN_ERR "Kprobe smoke test failed: "
132 "kprobe pre_handler not called\n");
133 handler_errors++;
134 }
135
136 if (posth_val == 0) {
137 printk(KERN_ERR "Kprobe smoke test failed: "
138 "kprobe post_handler not called\n");
139 handler_errors++;
140 }
141
142 preh_val = 0;
143 posth_val = 0;
144 ret = target2(rand1);
145
146 if (preh_val == 0) {
147 printk(KERN_ERR "Kprobe smoke test failed: "
148 "kprobe pre_handler2 not called\n");
149 handler_errors++;
150 }
151
152 if (posth_val == 0) {
153 printk(KERN_ERR "Kprobe smoke test failed: "
154 "kprobe post_handler2 not called\n");
155 handler_errors++;
156 }
157
158 unregister_kprobes(kps, 2);
159 return 0;
160
161}
162
95static u32 j_kprobe_target(u32 value) 163static u32 j_kprobe_target(u32 value)
96{ 164{
97 if (value != rand1) { 165 if (value != rand1) {
@@ -121,7 +189,7 @@ static int test_jprobe(void)
121 return ret; 189 return ret;
122 } 190 }
123 191
124 ret = kprobe_target(rand1); 192 ret = target(rand1);
125 unregister_jprobe(&jp); 193 unregister_jprobe(&jp);
126 if (jph_val == 0) { 194 if (jph_val == 0) {
127 printk(KERN_ERR "Kprobe smoke test failed: " 195 printk(KERN_ERR "Kprobe smoke test failed: "
@@ -132,6 +200,43 @@ static int test_jprobe(void)
132 return 0; 200 return 0;
133} 201}
134 202
203static struct jprobe jp2 = {
204 .entry = j_kprobe_target,
205 .kp.symbol_name = "kprobe_target2"
206};
207
208static int test_jprobes(void)
209{
210 int ret;
211 struct jprobe *jps[2] = {&jp, &jp2};
212
213 jp.kp.addr = 0; /* addr should be cleard for reusing kprobe. */
214 ret = register_jprobes(jps, 2);
215 if (ret < 0) {
216 printk(KERN_ERR "Kprobe smoke test failed: "
217 "register_jprobes returned %d\n", ret);
218 return ret;
219 }
220
221 jph_val = 0;
222 ret = target(rand1);
223 if (jph_val == 0) {
224 printk(KERN_ERR "Kprobe smoke test failed: "
225 "jprobe handler not called\n");
226 handler_errors++;
227 }
228
229 jph_val = 0;
230 ret = target2(rand1);
231 if (jph_val == 0) {
232 printk(KERN_ERR "Kprobe smoke test failed: "
233 "jprobe handler2 not called\n");
234 handler_errors++;
235 }
236 unregister_jprobes(jps, 2);
237
238 return 0;
239}
135#ifdef CONFIG_KRETPROBES 240#ifdef CONFIG_KRETPROBES
136static u32 krph_val; 241static u32 krph_val;
137 242
@@ -177,7 +282,7 @@ static int test_kretprobe(void)
177 return ret; 282 return ret;
178 } 283 }
179 284
180 ret = kprobe_target(rand1); 285 ret = target(rand1);
181 unregister_kretprobe(&rp); 286 unregister_kretprobe(&rp);
182 if (krph_val != rand1) { 287 if (krph_val != rand1) {
183 printk(KERN_ERR "Kprobe smoke test failed: " 288 printk(KERN_ERR "Kprobe smoke test failed: "
@@ -187,12 +292,72 @@ static int test_kretprobe(void)
187 292
188 return 0; 293 return 0;
189} 294}
295
296static int return_handler2(struct kretprobe_instance *ri, struct pt_regs *regs)
297{
298 unsigned long ret = regs_return_value(regs);
299
300 if (ret != (rand1 / div_factor) + 1) {
301 handler_errors++;
302 printk(KERN_ERR "Kprobe smoke test failed: "
303 "incorrect value in kretprobe handler2\n");
304 }
305 if (krph_val == 0) {
306 handler_errors++;
307 printk(KERN_ERR "Kprobe smoke test failed: "
308 "call to kretprobe entry handler failed\n");
309 }
310
311 krph_val = rand1;
312 return 0;
313}
314
315static struct kretprobe rp2 = {
316 .handler = return_handler2,
317 .entry_handler = entry_handler,
318 .kp.symbol_name = "kprobe_target2"
319};
320
321static int test_kretprobes(void)
322{
323 int ret;
324 struct kretprobe *rps[2] = {&rp, &rp2};
325
326 rp.kp.addr = 0; /* addr should be cleard for reusing kprobe. */
327 ret = register_kretprobes(rps, 2);
328 if (ret < 0) {
329 printk(KERN_ERR "Kprobe smoke test failed: "
330 "register_kretprobe returned %d\n", ret);
331 return ret;
332 }
333
334 krph_val = 0;
335 ret = target(rand1);
336 if (krph_val != rand1) {
337 printk(KERN_ERR "Kprobe smoke test failed: "
338 "kretprobe handler not called\n");
339 handler_errors++;
340 }
341
342 krph_val = 0;
343 ret = target2(rand1);
344 if (krph_val != rand1) {
345 printk(KERN_ERR "Kprobe smoke test failed: "
346 "kretprobe handler2 not called\n");
347 handler_errors++;
348 }
349 unregister_kretprobes(rps, 2);
350 return 0;
351}
190#endif /* CONFIG_KRETPROBES */ 352#endif /* CONFIG_KRETPROBES */
191 353
192int init_test_probes(void) 354int init_test_probes(void)
193{ 355{
194 int ret; 356 int ret;
195 357
358 target = kprobe_target;
359 target2 = kprobe_target2;
360
196 do { 361 do {
197 rand1 = random32(); 362 rand1 = random32();
198 } while (rand1 <= div_factor); 363 } while (rand1 <= div_factor);
@@ -204,15 +369,30 @@ int init_test_probes(void)
204 errors++; 369 errors++;
205 370
206 num_tests++; 371 num_tests++;
372 ret = test_kprobes();
373 if (ret < 0)
374 errors++;
375
376 num_tests++;
207 ret = test_jprobe(); 377 ret = test_jprobe();
208 if (ret < 0) 378 if (ret < 0)
209 errors++; 379 errors++;
210 380
381 num_tests++;
382 ret = test_jprobes();
383 if (ret < 0)
384 errors++;
385
211#ifdef CONFIG_KRETPROBES 386#ifdef CONFIG_KRETPROBES
212 num_tests++; 387 num_tests++;
213 ret = test_kretprobe(); 388 ret = test_kretprobe();
214 if (ret < 0) 389 if (ret < 0)
215 errors++; 390 errors++;
391
392 num_tests++;
393 ret = test_kretprobes();
394 if (ret < 0)
395 errors++;
216#endif /* CONFIG_KRETPROBES */ 396#endif /* CONFIG_KRETPROBES */
217 397
218 if (errors) 398 if (errors)
diff --git a/kernel/time.c b/kernel/time.c
index d63a4336fad6..4886e3ce83a4 100644
--- a/kernel/time.c
+++ b/kernel/time.c
@@ -37,6 +37,7 @@
37#include <linux/fs.h> 37#include <linux/fs.h>
38#include <linux/slab.h> 38#include <linux/slab.h>
39#include <linux/math64.h> 39#include <linux/math64.h>
40#include <linux/ptrace.h>
40 41
41#include <asm/uaccess.h> 42#include <asm/uaccess.h>
42#include <asm/unistd.h> 43#include <asm/unistd.h>
@@ -65,8 +66,9 @@ asmlinkage long sys_time(time_t __user * tloc)
65 66
66 if (tloc) { 67 if (tloc) {
67 if (put_user(i,tloc)) 68 if (put_user(i,tloc))
68 i = -EFAULT; 69 return -EFAULT;
69 } 70 }
71 force_successful_syscall_return();
70 return i; 72 return i;
71} 73}
72 74
diff --git a/kernel/tsacct.c b/kernel/tsacct.c
index 2dc06ab35716..43f891b05a4b 100644
--- a/kernel/tsacct.c
+++ b/kernel/tsacct.c
@@ -92,8 +92,8 @@ void xacct_add_tsk(struct taskstats *stats, struct task_struct *p)
92 mm = get_task_mm(p); 92 mm = get_task_mm(p);
93 if (mm) { 93 if (mm) {
94 /* adjust to KB unit */ 94 /* adjust to KB unit */
95 stats->hiwater_rss = mm->hiwater_rss * PAGE_SIZE / KB; 95 stats->hiwater_rss = get_mm_hiwater_rss(mm) * PAGE_SIZE / KB;
96 stats->hiwater_vm = mm->hiwater_vm * PAGE_SIZE / KB; 96 stats->hiwater_vm = get_mm_hiwater_vm(mm) * PAGE_SIZE / KB;
97 mmput(mm); 97 mmput(mm);
98 } 98 }
99 stats->read_char = p->ioac.rchar; 99 stats->read_char = p->ioac.rchar;
generated by cgit v1.2.2 (git 2.25.0) at 2025-10-30 08:48:51 -0400