Doug / smarc-fsl-linux-kernel

Download as
Browse Code ยป

Commit 6279a751fe096a21dc7704e918d570d3ff06e769

Authored by Oleg Nesterov
Committed by Ingo Molnar
1 parent 633fe795b8
Exists in master and in 7 other branches imx_3.0.35_4.1.0, imx_3.10.17_1.0.1_ga, imx_3.10.53_1.1.0_ga, imx_3.14.28_1.0.0_ga, smarc-imx_3.10.53_1.1.0_ga, smarc-imx_3.14.28_1.0.0_ga, smarc-l5.0.0_1.0.0-ga

posix-timers: fix RLIMIT_CPU && fork() 

See http://bugzilla.kernel.org/show_bug.cgi?id=12911

copy_signal() copies signal->rlim, but RLIMIT_CPU is "lost". Because
posix_cpu_timers_init_group() sets cputime_expires.prof_exp = 0 and thus
fastpath_timer_check() returns false unless we have other expired cpu timers.

Change copy_signal() to set cputime_expires.prof_exp if we have RLIMIT_CPU.
Also, set cputimer.running = 1 in that case. This is not strictly necessary,
but imho makes sense.

Reported-by: Peter Lojkin <ia6432@inbox.ru>
Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Peter Lojkin <ia6432@inbox.ru>
Cc: Roland McGrath <roland@redhat.com>
Cc: stable@kernel.org
LKML-Reference: <20090327000607.GA10104@redhat.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

Showing 1 changed file with 9 additions and 4 deletions Inline Diff

1 /* 1 /*
2 * linux/kernel/fork.c 2 * linux/kernel/fork.c
3 * 3 *
4 * Copyright (C) 1991, 1992 Linus Torvalds 4 * Copyright (C) 1991, 1992 Linus Torvalds
5 */ 5 */
6 6
7 /* 7 /*
8 * 'fork.c' contains the help-routines for the 'fork' system call 8 * 'fork.c' contains the help-routines for the 'fork' system call
9 * (see also entry.S and others). 9 * (see also entry.S and others).
10 * Fork is rather simple, once you get the hang of it, but the memory 10 * Fork is rather simple, once you get the hang of it, but the memory
11 * management can be a bitch. See 'mm/memory.c': 'copy_page_range()' 11 * management can be a bitch. See 'mm/memory.c': 'copy_page_range()'
12 */ 12 */
13 13
14 #include <linux/slab.h> 14 #include <linux/slab.h>
15 #include <linux/init.h> 15 #include <linux/init.h>
16 #include <linux/unistd.h> 16 #include <linux/unistd.h>
17 #include <linux/module.h> 17 #include <linux/module.h>
18 #include <linux/vmalloc.h> 18 #include <linux/vmalloc.h>
19 #include <linux/completion.h> 19 #include <linux/completion.h>
20 #include <linux/mnt_namespace.h> 20 #include <linux/mnt_namespace.h>
21 #include <linux/personality.h> 21 #include <linux/personality.h>
22 #include <linux/mempolicy.h> 22 #include <linux/mempolicy.h>
23 #include <linux/sem.h> 23 #include <linux/sem.h>
24 #include <linux/file.h> 24 #include <linux/file.h>
25 #include <linux/fdtable.h> 25 #include <linux/fdtable.h>
26 #include <linux/iocontext.h> 26 #include <linux/iocontext.h>
27 #include <linux/key.h> 27 #include <linux/key.h>
28 #include <linux/binfmts.h> 28 #include <linux/binfmts.h>
29 #include <linux/mman.h> 29 #include <linux/mman.h>
30 #include <linux/mmu_notifier.h> 30 #include <linux/mmu_notifier.h>
31 #include <linux/fs.h> 31 #include <linux/fs.h>
32 #include <linux/nsproxy.h> 32 #include <linux/nsproxy.h>
33 #include <linux/capability.h> 33 #include <linux/capability.h>
34 #include <linux/cpu.h> 34 #include <linux/cpu.h>
35 #include <linux/cgroup.h> 35 #include <linux/cgroup.h>
36 #include <linux/security.h> 36 #include <linux/security.h>
37 #include <linux/hugetlb.h> 37 #include <linux/hugetlb.h>
38 #include <linux/swap.h> 38 #include <linux/swap.h>
39 #include <linux/syscalls.h> 39 #include <linux/syscalls.h>
40 #include <linux/jiffies.h> 40 #include <linux/jiffies.h>
41 #include <linux/tracehook.h> 41 #include <linux/tracehook.h>
42 #include <linux/futex.h> 42 #include <linux/futex.h>
43 #include <linux/compat.h> 43 #include <linux/compat.h>
44 #include <linux/task_io_accounting_ops.h> 44 #include <linux/task_io_accounting_ops.h>
45 #include <linux/rcupdate.h> 45 #include <linux/rcupdate.h>
46 #include <linux/ptrace.h> 46 #include <linux/ptrace.h>
47 #include <linux/mount.h> 47 #include <linux/mount.h>
48 #include <linux/audit.h> 48 #include <linux/audit.h>
49 #include <linux/memcontrol.h> 49 #include <linux/memcontrol.h>
50 #include <linux/ftrace.h> 50 #include <linux/ftrace.h>
51 #include <linux/profile.h> 51 #include <linux/profile.h>
52 #include <linux/rmap.h> 52 #include <linux/rmap.h>
53 #include <linux/acct.h> 53 #include <linux/acct.h>
54 #include <linux/tsacct_kern.h> 54 #include <linux/tsacct_kern.h>
55 #include <linux/cn_proc.h> 55 #include <linux/cn_proc.h>
56 #include <linux/freezer.h> 56 #include <linux/freezer.h>
57 #include <linux/delayacct.h> 57 #include <linux/delayacct.h>
58 #include <linux/taskstats_kern.h> 58 #include <linux/taskstats_kern.h>
59 #include <linux/random.h> 59 #include <linux/random.h>
60 #include <linux/tty.h> 60 #include <linux/tty.h>
61 #include <linux/proc_fs.h> 61 #include <linux/proc_fs.h>
62 #include <linux/blkdev.h> 62 #include <linux/blkdev.h>
63 #include <trace/sched.h> 63 #include <trace/sched.h>
64 64
65 #include <asm/pgtable.h> 65 #include <asm/pgtable.h>
66 #include <asm/pgalloc.h> 66 #include <asm/pgalloc.h>
67 #include <asm/uaccess.h> 67 #include <asm/uaccess.h>
68 #include <asm/mmu_context.h> 68 #include <asm/mmu_context.h>
69 #include <asm/cacheflush.h> 69 #include <asm/cacheflush.h>
70 #include <asm/tlbflush.h> 70 #include <asm/tlbflush.h>
71 71
72 /* 72 /*
73 * Protected counters by write_lock_irq(&tasklist_lock) 73 * Protected counters by write_lock_irq(&tasklist_lock)
74 */ 74 */
75 unsigned long total_forks; /* Handle normal Linux uptimes. */ 75 unsigned long total_forks; /* Handle normal Linux uptimes. */
76 int nr_threads; /* The idle threads do not count.. */ 76 int nr_threads; /* The idle threads do not count.. */
77 77
78 int max_threads; /* tunable limit on nr_threads */ 78 int max_threads; /* tunable limit on nr_threads */
79 79
80 DEFINE_PER_CPU(unsigned long, process_counts) = 0; 80 DEFINE_PER_CPU(unsigned long, process_counts) = 0;
81 81
82 __cacheline_aligned DEFINE_RWLOCK(tasklist_lock); /* outer */ 82 __cacheline_aligned DEFINE_RWLOCK(tasklist_lock); /* outer */
83 83
84 DEFINE_TRACE(sched_process_fork); 84 DEFINE_TRACE(sched_process_fork);
85 85
86 int nr_processes(void) 86 int nr_processes(void)
87 { 87 {
88 int cpu; 88 int cpu;
89 int total = 0; 89 int total = 0;
90 90
91 for_each_online_cpu(cpu) 91 for_each_online_cpu(cpu)
92 total += per_cpu(process_counts, cpu); 92 total += per_cpu(process_counts, cpu);
93 93
94 return total; 94 return total;
95 } 95 }
96 96
97 #ifndef __HAVE_ARCH_TASK_STRUCT_ALLOCATOR 97 #ifndef __HAVE_ARCH_TASK_STRUCT_ALLOCATOR
98 # define alloc_task_struct() kmem_cache_alloc(task_struct_cachep, GFP_KERNEL) 98 # define alloc_task_struct() kmem_cache_alloc(task_struct_cachep, GFP_KERNEL)
99 # define free_task_struct(tsk) kmem_cache_free(task_struct_cachep, (tsk)) 99 # define free_task_struct(tsk) kmem_cache_free(task_struct_cachep, (tsk))
100 static struct kmem_cache *task_struct_cachep; 100 static struct kmem_cache *task_struct_cachep;
101 #endif 101 #endif
102 102
103 #ifndef __HAVE_ARCH_THREAD_INFO_ALLOCATOR 103 #ifndef __HAVE_ARCH_THREAD_INFO_ALLOCATOR
104 static inline struct thread_info *alloc_thread_info(struct task_struct *tsk) 104 static inline struct thread_info *alloc_thread_info(struct task_struct *tsk)
105 { 105 {
106 #ifdef CONFIG_DEBUG_STACK_USAGE 106 #ifdef CONFIG_DEBUG_STACK_USAGE
107 gfp_t mask = GFP_KERNEL | __GFP_ZERO; 107 gfp_t mask = GFP_KERNEL | __GFP_ZERO;
108 #else 108 #else
109 gfp_t mask = GFP_KERNEL; 109 gfp_t mask = GFP_KERNEL;
110 #endif 110 #endif
111 return (struct thread_info *)__get_free_pages(mask, THREAD_SIZE_ORDER); 111 return (struct thread_info *)__get_free_pages(mask, THREAD_SIZE_ORDER);
112 } 112 }
113 113
114 static inline void free_thread_info(struct thread_info *ti) 114 static inline void free_thread_info(struct thread_info *ti)
115 { 115 {
116 free_pages((unsigned long)ti, THREAD_SIZE_ORDER); 116 free_pages((unsigned long)ti, THREAD_SIZE_ORDER);
117 } 117 }
118 #endif 118 #endif
119 119
120 /* SLAB cache for signal_struct structures (tsk->signal) */ 120 /* SLAB cache for signal_struct structures (tsk->signal) */
121 static struct kmem_cache *signal_cachep; 121 static struct kmem_cache *signal_cachep;
122 122
123 /* SLAB cache for sighand_struct structures (tsk->sighand) */ 123 /* SLAB cache for sighand_struct structures (tsk->sighand) */
124 struct kmem_cache *sighand_cachep; 124 struct kmem_cache *sighand_cachep;
125 125
126 /* SLAB cache for files_struct structures (tsk->files) */ 126 /* SLAB cache for files_struct structures (tsk->files) */
127 struct kmem_cache *files_cachep; 127 struct kmem_cache *files_cachep;
128 128
129 /* SLAB cache for fs_struct structures (tsk->fs) */ 129 /* SLAB cache for fs_struct structures (tsk->fs) */
130 struct kmem_cache *fs_cachep; 130 struct kmem_cache *fs_cachep;
131 131
132 /* SLAB cache for vm_area_struct structures */ 132 /* SLAB cache for vm_area_struct structures */
133 struct kmem_cache *vm_area_cachep; 133 struct kmem_cache *vm_area_cachep;
134 134
135 /* SLAB cache for mm_struct structures (tsk->mm) */ 135 /* SLAB cache for mm_struct structures (tsk->mm) */
136 static struct kmem_cache *mm_cachep; 136 static struct kmem_cache *mm_cachep;
137 137
138 void free_task(struct task_struct *tsk) 138 void free_task(struct task_struct *tsk)
139 { 139 {
140 prop_local_destroy_single(&tsk->dirties); 140 prop_local_destroy_single(&tsk->dirties);
141 free_thread_info(tsk->stack); 141 free_thread_info(tsk->stack);
142 rt_mutex_debug_task_free(tsk); 142 rt_mutex_debug_task_free(tsk);
143 ftrace_graph_exit_task(tsk); 143 ftrace_graph_exit_task(tsk);
144 free_task_struct(tsk); 144 free_task_struct(tsk);
145 } 145 }
146 EXPORT_SYMBOL(free_task); 146 EXPORT_SYMBOL(free_task);
147 147
148 void __put_task_struct(struct task_struct *tsk) 148 void __put_task_struct(struct task_struct *tsk)
149 { 149 {
150 WARN_ON(!tsk->exit_state); 150 WARN_ON(!tsk->exit_state);
151 WARN_ON(atomic_read(&tsk->usage)); 151 WARN_ON(atomic_read(&tsk->usage));
152 WARN_ON(tsk == current); 152 WARN_ON(tsk == current);
153 153
154 put_cred(tsk->real_cred); 154 put_cred(tsk->real_cred);
155 put_cred(tsk->cred); 155 put_cred(tsk->cred);
156 delayacct_tsk_free(tsk); 156 delayacct_tsk_free(tsk);
157 157
158 if (!profile_handoff_task(tsk)) 158 if (!profile_handoff_task(tsk))
159 free_task(tsk); 159 free_task(tsk);
160 } 160 }
161 161
162 /* 162 /*
163 * macro override instead of weak attribute alias, to workaround 163 * macro override instead of weak attribute alias, to workaround
164 * gcc 4.1.0 and 4.1.1 bugs with weak attribute and empty functions. 164 * gcc 4.1.0 and 4.1.1 bugs with weak attribute and empty functions.
165 */ 165 */
166 #ifndef arch_task_cache_init 166 #ifndef arch_task_cache_init
167 #define arch_task_cache_init() 167 #define arch_task_cache_init()
168 #endif 168 #endif
169 169
170 void __init fork_init(unsigned long mempages) 170 void __init fork_init(unsigned long mempages)
171 { 171 {
172 #ifndef __HAVE_ARCH_TASK_STRUCT_ALLOCATOR 172 #ifndef __HAVE_ARCH_TASK_STRUCT_ALLOCATOR
173 #ifndef ARCH_MIN_TASKALIGN 173 #ifndef ARCH_MIN_TASKALIGN
174 #define ARCH_MIN_TASKALIGN L1_CACHE_BYTES 174 #define ARCH_MIN_TASKALIGN L1_CACHE_BYTES
175 #endif 175 #endif
176 /* create a slab on which task_structs can be allocated */ 176 /* create a slab on which task_structs can be allocated */
177 task_struct_cachep = 177 task_struct_cachep =
178 kmem_cache_create("task_struct", sizeof(struct task_struct), 178 kmem_cache_create("task_struct", sizeof(struct task_struct),
179 ARCH_MIN_TASKALIGN, SLAB_PANIC, NULL); 179 ARCH_MIN_TASKALIGN, SLAB_PANIC, NULL);
180 #endif 180 #endif
181 181
182 /* do the arch specific task caches init */ 182 /* do the arch specific task caches init */
183 arch_task_cache_init(); 183 arch_task_cache_init();
184 184
185 /* 185 /*
186 * The default maximum number of threads is set to a safe 186 * The default maximum number of threads is set to a safe
187 * value: the thread structures can take up at most half 187 * value: the thread structures can take up at most half
188 * of memory. 188 * of memory.
189 */ 189 */
190 max_threads = mempages / (8 * THREAD_SIZE / PAGE_SIZE); 190 max_threads = mempages / (8 * THREAD_SIZE / PAGE_SIZE);
191 191
192 /* 192 /*
193 * we need to allow at least 20 threads to boot a system 193 * we need to allow at least 20 threads to boot a system
194 */ 194 */
195 if(max_threads < 20) 195 if(max_threads < 20)
196 max_threads = 20; 196 max_threads = 20;
197 197
198 init_task.signal->rlim[RLIMIT_NPROC].rlim_cur = max_threads/2; 198 init_task.signal->rlim[RLIMIT_NPROC].rlim_cur = max_threads/2;
199 init_task.signal->rlim[RLIMIT_NPROC].rlim_max = max_threads/2; 199 init_task.signal->rlim[RLIMIT_NPROC].rlim_max = max_threads/2;
200 init_task.signal->rlim[RLIMIT_SIGPENDING] = 200 init_task.signal->rlim[RLIMIT_SIGPENDING] =
201 init_task.signal->rlim[RLIMIT_NPROC]; 201 init_task.signal->rlim[RLIMIT_NPROC];
202 } 202 }
203 203
204 int __attribute__((weak)) arch_dup_task_struct(struct task_struct *dst, 204 int __attribute__((weak)) arch_dup_task_struct(struct task_struct *dst,
205 struct task_struct *src) 205 struct task_struct *src)
206 { 206 {
207 *dst = *src; 207 *dst = *src;
208 return 0; 208 return 0;
209 } 209 }
210 210
211 static struct task_struct *dup_task_struct(struct task_struct *orig) 211 static struct task_struct *dup_task_struct(struct task_struct *orig)
212 { 212 {
213 struct task_struct *tsk; 213 struct task_struct *tsk;
214 struct thread_info *ti; 214 struct thread_info *ti;
215 int err; 215 int err;
216 216
217 prepare_to_copy(orig); 217 prepare_to_copy(orig);
218 218
219 tsk = alloc_task_struct(); 219 tsk = alloc_task_struct();
220 if (!tsk) 220 if (!tsk)
221 return NULL; 221 return NULL;
222 222
223 ti = alloc_thread_info(tsk); 223 ti = alloc_thread_info(tsk);
224 if (!ti) { 224 if (!ti) {
225 free_task_struct(tsk); 225 free_task_struct(tsk);
226 return NULL; 226 return NULL;
227 } 227 }
228 228
229 err = arch_dup_task_struct(tsk, orig); 229 err = arch_dup_task_struct(tsk, orig);
230 if (err) 230 if (err)
231 goto out; 231 goto out;
232 232
233 tsk->stack = ti; 233 tsk->stack = ti;
234 234
235 err = prop_local_init_single(&tsk->dirties); 235 err = prop_local_init_single(&tsk->dirties);
236 if (err) 236 if (err)
237 goto out; 237 goto out;
238 238
239 setup_thread_stack(tsk, orig); 239 setup_thread_stack(tsk, orig);
240 240
241 #ifdef CONFIG_CC_STACKPROTECTOR 241 #ifdef CONFIG_CC_STACKPROTECTOR
242 tsk->stack_canary = get_random_int(); 242 tsk->stack_canary = get_random_int();
243 #endif 243 #endif
244 244
245 /* One for us, one for whoever does the "release_task()" (usually parent) */ 245 /* One for us, one for whoever does the "release_task()" (usually parent) */
246 atomic_set(&tsk->usage,2); 246 atomic_set(&tsk->usage,2);
247 atomic_set(&tsk->fs_excl, 0); 247 atomic_set(&tsk->fs_excl, 0);
248 #ifdef CONFIG_BLK_DEV_IO_TRACE 248 #ifdef CONFIG_BLK_DEV_IO_TRACE
249 tsk->btrace_seq = 0; 249 tsk->btrace_seq = 0;
250 #endif 250 #endif
251 tsk->splice_pipe = NULL; 251 tsk->splice_pipe = NULL;
252 return tsk; 252 return tsk;
253 253
254 out: 254 out:
255 free_thread_info(ti); 255 free_thread_info(ti);
256 free_task_struct(tsk); 256 free_task_struct(tsk);
257 return NULL; 257 return NULL;
258 } 258 }
259 259
260 #ifdef CONFIG_MMU 260 #ifdef CONFIG_MMU
261 static int dup_mmap(struct mm_struct *mm, struct mm_struct *oldmm) 261 static int dup_mmap(struct mm_struct *mm, struct mm_struct *oldmm)
262 { 262 {
263 struct vm_area_struct *mpnt, *tmp, **pprev; 263 struct vm_area_struct *mpnt, *tmp, **pprev;
264 struct rb_node **rb_link, *rb_parent; 264 struct rb_node **rb_link, *rb_parent;
265 int retval; 265 int retval;
266 unsigned long charge; 266 unsigned long charge;
267 struct mempolicy *pol; 267 struct mempolicy *pol;
268 268
269 down_write(&oldmm->mmap_sem); 269 down_write(&oldmm->mmap_sem);
270 flush_cache_dup_mm(oldmm); 270 flush_cache_dup_mm(oldmm);
271 /* 271 /*
272 * Not linked in yet - no deadlock potential: 272 * Not linked in yet - no deadlock potential:
273 */ 273 */
274 down_write_nested(&mm->mmap_sem, SINGLE_DEPTH_NESTING); 274 down_write_nested(&mm->mmap_sem, SINGLE_DEPTH_NESTING);
275 275
276 mm->locked_vm = 0; 276 mm->locked_vm = 0;
277 mm->mmap = NULL; 277 mm->mmap = NULL;
278 mm->mmap_cache = NULL; 278 mm->mmap_cache = NULL;
279 mm->free_area_cache = oldmm->mmap_base; 279 mm->free_area_cache = oldmm->mmap_base;
280 mm->cached_hole_size = ~0UL; 280 mm->cached_hole_size = ~0UL;
281 mm->map_count = 0; 281 mm->map_count = 0;
282 cpus_clear(mm->cpu_vm_mask); 282 cpus_clear(mm->cpu_vm_mask);
283 mm->mm_rb = RB_ROOT; 283 mm->mm_rb = RB_ROOT;
284 rb_link = &mm->mm_rb.rb_node; 284 rb_link = &mm->mm_rb.rb_node;
285 rb_parent = NULL; 285 rb_parent = NULL;
286 pprev = &mm->mmap; 286 pprev = &mm->mmap;
287 287
288 for (mpnt = oldmm->mmap; mpnt; mpnt = mpnt->vm_next) { 288 for (mpnt = oldmm->mmap; mpnt; mpnt = mpnt->vm_next) {
289 struct file *file; 289 struct file *file;
290 290
291 if (mpnt->vm_flags & VM_DONTCOPY) { 291 if (mpnt->vm_flags & VM_DONTCOPY) {
292 long pages = vma_pages(mpnt); 292 long pages = vma_pages(mpnt);
293 mm->total_vm -= pages; 293 mm->total_vm -= pages;
294 vm_stat_account(mm, mpnt->vm_flags, mpnt->vm_file, 294 vm_stat_account(mm, mpnt->vm_flags, mpnt->vm_file,
295 -pages); 295 -pages);
296 continue; 296 continue;
297 } 297 }
298 charge = 0; 298 charge = 0;
299 if (mpnt->vm_flags & VM_ACCOUNT) { 299 if (mpnt->vm_flags & VM_ACCOUNT) {
300 unsigned int len = (mpnt->vm_end - mpnt->vm_start) >> PAGE_SHIFT; 300 unsigned int len = (mpnt->vm_end - mpnt->vm_start) >> PAGE_SHIFT;
301 if (security_vm_enough_memory(len)) 301 if (security_vm_enough_memory(len))
302 goto fail_nomem; 302 goto fail_nomem;
303 charge = len; 303 charge = len;
304 } 304 }
305 tmp = kmem_cache_alloc(vm_area_cachep, GFP_KERNEL); 305 tmp = kmem_cache_alloc(vm_area_cachep, GFP_KERNEL);
306 if (!tmp) 306 if (!tmp)
307 goto fail_nomem; 307 goto fail_nomem;
308 *tmp = *mpnt; 308 *tmp = *mpnt;
309 pol = mpol_dup(vma_policy(mpnt)); 309 pol = mpol_dup(vma_policy(mpnt));
310 retval = PTR_ERR(pol); 310 retval = PTR_ERR(pol);
311 if (IS_ERR(pol)) 311 if (IS_ERR(pol))
312 goto fail_nomem_policy; 312 goto fail_nomem_policy;
313 vma_set_policy(tmp, pol); 313 vma_set_policy(tmp, pol);
314 tmp->vm_flags &= ~VM_LOCKED; 314 tmp->vm_flags &= ~VM_LOCKED;
315 tmp->vm_mm = mm; 315 tmp->vm_mm = mm;
316 tmp->vm_next = NULL; 316 tmp->vm_next = NULL;
317 anon_vma_link(tmp); 317 anon_vma_link(tmp);
318 file = tmp->vm_file; 318 file = tmp->vm_file;
319 if (file) { 319 if (file) {
320 struct inode *inode = file->f_path.dentry->d_inode; 320 struct inode *inode = file->f_path.dentry->d_inode;
321 struct address_space *mapping = file->f_mapping; 321 struct address_space *mapping = file->f_mapping;
322 322
323 get_file(file); 323 get_file(file);
324 if (tmp->vm_flags & VM_DENYWRITE) 324 if (tmp->vm_flags & VM_DENYWRITE)
325 atomic_dec(&inode->i_writecount); 325 atomic_dec(&inode->i_writecount);
326 spin_lock(&mapping->i_mmap_lock); 326 spin_lock(&mapping->i_mmap_lock);
327 if (tmp->vm_flags & VM_SHARED) 327 if (tmp->vm_flags & VM_SHARED)
328 mapping->i_mmap_writable++; 328 mapping->i_mmap_writable++;
329 tmp->vm_truncate_count = mpnt->vm_truncate_count; 329 tmp->vm_truncate_count = mpnt->vm_truncate_count;
330 flush_dcache_mmap_lock(mapping); 330 flush_dcache_mmap_lock(mapping);
331 /* insert tmp into the share list, just after mpnt */ 331 /* insert tmp into the share list, just after mpnt */
332 vma_prio_tree_add(tmp, mpnt); 332 vma_prio_tree_add(tmp, mpnt);
333 flush_dcache_mmap_unlock(mapping); 333 flush_dcache_mmap_unlock(mapping);
334 spin_unlock(&mapping->i_mmap_lock); 334 spin_unlock(&mapping->i_mmap_lock);
335 } 335 }
336 336
337 /* 337 /*
338 * Clear hugetlb-related page reserves for children. This only 338 * Clear hugetlb-related page reserves for children. This only
339 * affects MAP_PRIVATE mappings. Faults generated by the child 339 * affects MAP_PRIVATE mappings. Faults generated by the child
340 * are not guaranteed to succeed, even if read-only 340 * are not guaranteed to succeed, even if read-only
341 */ 341 */
342 if (is_vm_hugetlb_page(tmp)) 342 if (is_vm_hugetlb_page(tmp))
343 reset_vma_resv_huge_pages(tmp); 343 reset_vma_resv_huge_pages(tmp);
344 344
345 /* 345 /*
346 * Link in the new vma and copy the page table entries. 346 * Link in the new vma and copy the page table entries.
347 */ 347 */
348 *pprev = tmp; 348 *pprev = tmp;
349 pprev = &tmp->vm_next; 349 pprev = &tmp->vm_next;
350 350
351 __vma_link_rb(mm, tmp, rb_link, rb_parent); 351 __vma_link_rb(mm, tmp, rb_link, rb_parent);
352 rb_link = &tmp->vm_rb.rb_right; 352 rb_link = &tmp->vm_rb.rb_right;
353 rb_parent = &tmp->vm_rb; 353 rb_parent = &tmp->vm_rb;
354 354
355 mm->map_count++; 355 mm->map_count++;
356 retval = copy_page_range(mm, oldmm, mpnt); 356 retval = copy_page_range(mm, oldmm, mpnt);
357 357
358 if (tmp->vm_ops && tmp->vm_ops->open) 358 if (tmp->vm_ops && tmp->vm_ops->open)
359 tmp->vm_ops->open(tmp); 359 tmp->vm_ops->open(tmp);
360 360
361 if (retval) 361 if (retval)
362 goto out; 362 goto out;
363 } 363 }
364 /* a new mm has just been created */ 364 /* a new mm has just been created */
365 arch_dup_mmap(oldmm, mm); 365 arch_dup_mmap(oldmm, mm);
366 retval = 0; 366 retval = 0;
367 out: 367 out:
368 up_write(&mm->mmap_sem); 368 up_write(&mm->mmap_sem);
369 flush_tlb_mm(oldmm); 369 flush_tlb_mm(oldmm);
370 up_write(&oldmm->mmap_sem); 370 up_write(&oldmm->mmap_sem);
371 return retval; 371 return retval;
372 fail_nomem_policy: 372 fail_nomem_policy:
373 kmem_cache_free(vm_area_cachep, tmp); 373 kmem_cache_free(vm_area_cachep, tmp);
374 fail_nomem: 374 fail_nomem:
375 retval = -ENOMEM; 375 retval = -ENOMEM;
376 vm_unacct_memory(charge); 376 vm_unacct_memory(charge);
377 goto out; 377 goto out;
378 } 378 }
379 379
380 static inline int mm_alloc_pgd(struct mm_struct * mm) 380 static inline int mm_alloc_pgd(struct mm_struct * mm)
381 { 381 {
382 mm->pgd = pgd_alloc(mm); 382 mm->pgd = pgd_alloc(mm);
383 if (unlikely(!mm->pgd)) 383 if (unlikely(!mm->pgd))
384 return -ENOMEM; 384 return -ENOMEM;
385 return 0; 385 return 0;
386 } 386 }
387 387
388 static inline void mm_free_pgd(struct mm_struct * mm) 388 static inline void mm_free_pgd(struct mm_struct * mm)
389 { 389 {
390 pgd_free(mm, mm->pgd); 390 pgd_free(mm, mm->pgd);
391 } 391 }
392 #else 392 #else
393 #define dup_mmap(mm, oldmm) (0) 393 #define dup_mmap(mm, oldmm) (0)
394 #define mm_alloc_pgd(mm) (0) 394 #define mm_alloc_pgd(mm) (0)
395 #define mm_free_pgd(mm) 395 #define mm_free_pgd(mm)
396 #endif /* CONFIG_MMU */ 396 #endif /* CONFIG_MMU */
397 397
398 __cacheline_aligned_in_smp DEFINE_SPINLOCK(mmlist_lock); 398 __cacheline_aligned_in_smp DEFINE_SPINLOCK(mmlist_lock);
399 399
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
403 static unsigned long default_dump_filter = MMF_DUMP_FILTER_DEFAULT; 403 static unsigned long default_dump_filter = MMF_DUMP_FILTER_DEFAULT;
404 404
405 static int __init coredump_filter_setup(char *s) 405 static int __init coredump_filter_setup(char *s)
406 { 406 {
407 default_dump_filter = 407 default_dump_filter =
408 (simple_strtoul(s, NULL, 0) << MMF_DUMP_FILTER_SHIFT) & 408 (simple_strtoul(s, NULL, 0) << MMF_DUMP_FILTER_SHIFT) &
409 MMF_DUMP_FILTER_MASK; 409 MMF_DUMP_FILTER_MASK;
410 return 1; 410 return 1;
411 } 411 }
412 412
413 __setup("coredump_filter=", coredump_filter_setup); 413 __setup("coredump_filter=", coredump_filter_setup);
414 414
415 #include <linux/init_task.h> 415 #include <linux/init_task.h>
416 416
417 static struct mm_struct * mm_init(struct mm_struct * mm, struct task_struct *p) 417 static struct mm_struct * mm_init(struct mm_struct * mm, struct task_struct *p)
418 { 418 {
419 atomic_set(&mm->mm_users, 1); 419 atomic_set(&mm->mm_users, 1);
420 atomic_set(&mm->mm_count, 1); 420 atomic_set(&mm->mm_count, 1);
421 init_rwsem(&mm->mmap_sem); 421 init_rwsem(&mm->mmap_sem);
422 INIT_LIST_HEAD(&mm->mmlist); 422 INIT_LIST_HEAD(&mm->mmlist);
423 mm->flags = (current->mm) ? current->mm->flags : default_dump_filter; 423 mm->flags = (current->mm) ? current->mm->flags : default_dump_filter;
424 mm->core_state = NULL; 424 mm->core_state = NULL;
425 mm->nr_ptes = 0; 425 mm->nr_ptes = 0;
426 set_mm_counter(mm, file_rss, 0); 426 set_mm_counter(mm, file_rss, 0);
427 set_mm_counter(mm, anon_rss, 0); 427 set_mm_counter(mm, anon_rss, 0);
428 spin_lock_init(&mm->page_table_lock); 428 spin_lock_init(&mm->page_table_lock);
429 spin_lock_init(&mm->ioctx_lock); 429 spin_lock_init(&mm->ioctx_lock);
430 INIT_HLIST_HEAD(&mm->ioctx_list); 430 INIT_HLIST_HEAD(&mm->ioctx_list);
431 mm->free_area_cache = TASK_UNMAPPED_BASE; 431 mm->free_area_cache = TASK_UNMAPPED_BASE;
432 mm->cached_hole_size = ~0UL; 432 mm->cached_hole_size = ~0UL;
433 mm_init_owner(mm, p); 433 mm_init_owner(mm, p);
434 434
435 if (likely(!mm_alloc_pgd(mm))) { 435 if (likely(!mm_alloc_pgd(mm))) {
436 mm->def_flags = 0; 436 mm->def_flags = 0;
437 mmu_notifier_mm_init(mm); 437 mmu_notifier_mm_init(mm);
438 return mm; 438 return mm;
439 } 439 }
440 440
441 free_mm(mm); 441 free_mm(mm);
442 return NULL; 442 return NULL;
443 } 443 }
444 444
445 /* 445 /*
446 * Allocate and initialize an mm_struct. 446 * Allocate and initialize an mm_struct.
447 */ 447 */
448 struct mm_struct * mm_alloc(void) 448 struct mm_struct * mm_alloc(void)
449 { 449 {
450 struct mm_struct * mm; 450 struct mm_struct * mm;
451 451
452 mm = allocate_mm(); 452 mm = allocate_mm();
453 if (mm) { 453 if (mm) {
454 memset(mm, 0, sizeof(*mm)); 454 memset(mm, 0, sizeof(*mm));
455 mm = mm_init(mm, current); 455 mm = mm_init(mm, current);
456 } 456 }
457 return mm; 457 return mm;
458 } 458 }
459 459
460 /* 460 /*
461 * Called when the last reference to the mm 461 * Called when the last reference to the mm
462 * is dropped: either by a lazy thread or by 462 * is dropped: either by a lazy thread or by
463 * mmput. Free the page directory and the mm. 463 * mmput. Free the page directory and the mm.
464 */ 464 */
465 void __mmdrop(struct mm_struct *mm) 465 void __mmdrop(struct mm_struct *mm)
466 { 466 {
467 BUG_ON(mm == &init_mm); 467 BUG_ON(mm == &init_mm);
468 mm_free_pgd(mm); 468 mm_free_pgd(mm);
469 destroy_context(mm); 469 destroy_context(mm);
470 mmu_notifier_mm_destroy(mm); 470 mmu_notifier_mm_destroy(mm);
471 free_mm(mm); 471 free_mm(mm);
472 } 472 }
473 EXPORT_SYMBOL_GPL(__mmdrop); 473 EXPORT_SYMBOL_GPL(__mmdrop);
474 474
475 /* 475 /*
476 * Decrement the use count and release all resources for an mm. 476 * Decrement the use count and release all resources for an mm.
477 */ 477 */
478 void mmput(struct mm_struct *mm) 478 void mmput(struct mm_struct *mm)
479 { 479 {
480 might_sleep(); 480 might_sleep();
481 481
482 if (atomic_dec_and_test(&mm->mm_users)) { 482 if (atomic_dec_and_test(&mm->mm_users)) {
483 exit_aio(mm); 483 exit_aio(mm);
484 exit_mmap(mm); 484 exit_mmap(mm);
485 set_mm_exe_file(mm, NULL); 485 set_mm_exe_file(mm, NULL);
486 if (!list_empty(&mm->mmlist)) { 486 if (!list_empty(&mm->mmlist)) {
487 spin_lock(&mmlist_lock); 487 spin_lock(&mmlist_lock);
488 list_del(&mm->mmlist); 488 list_del(&mm->mmlist);
489 spin_unlock(&mmlist_lock); 489 spin_unlock(&mmlist_lock);
490 } 490 }
491 put_swap_token(mm); 491 put_swap_token(mm);
492 mmdrop(mm); 492 mmdrop(mm);
493 } 493 }
494 } 494 }
495 EXPORT_SYMBOL_GPL(mmput); 495 EXPORT_SYMBOL_GPL(mmput);
496 496
497 /** 497 /**
498 * get_task_mm - acquire a reference to the task's mm 498 * get_task_mm - acquire a reference to the task's mm
499 * 499 *
500 * Returns %NULL if the task has no mm. Checks PF_KTHREAD (meaning 500 * Returns %NULL if the task has no mm. Checks PF_KTHREAD (meaning
501 * this kernel workthread has transiently adopted a user mm with use_mm, 501 * this kernel workthread has transiently adopted a user mm with use_mm,
502 * to do its AIO) is not set and if so returns a reference to it, after 502 * to do its AIO) is not set and if so returns a reference to it, after
503 * bumping up the use count. User must release the mm via mmput() 503 * bumping up the use count. User must release the mm via mmput()
504 * after use. Typically used by /proc and ptrace. 504 * after use. Typically used by /proc and ptrace.
505 */ 505 */
506 struct mm_struct *get_task_mm(struct task_struct *task) 506 struct mm_struct *get_task_mm(struct task_struct *task)
507 { 507 {
508 struct mm_struct *mm; 508 struct mm_struct *mm;
509 509
510 task_lock(task); 510 task_lock(task);
511 mm = task->mm; 511 mm = task->mm;
512 if (mm) { 512 if (mm) {
513 if (task->flags & PF_KTHREAD) 513 if (task->flags & PF_KTHREAD)
514 mm = NULL; 514 mm = NULL;
515 else 515 else
516 atomic_inc(&mm->mm_users); 516 atomic_inc(&mm->mm_users);
517 } 517 }
518 task_unlock(task); 518 task_unlock(task);
519 return mm; 519 return mm;
520 } 520 }
521 EXPORT_SYMBOL_GPL(get_task_mm); 521 EXPORT_SYMBOL_GPL(get_task_mm);
522 522
523 /* Please note the differences between mmput and mm_release. 523 /* Please note the differences between mmput and mm_release.
524 * mmput is called whenever we stop holding onto a mm_struct, 524 * mmput is called whenever we stop holding onto a mm_struct,
525 * error success whatever. 525 * error success whatever.
526 * 526 *
527 * mm_release is called after a mm_struct has been removed 527 * mm_release is called after a mm_struct has been removed
528 * from the current process. 528 * from the current process.
529 * 529 *
530 * This difference is important for error handling, when we 530 * This difference is important for error handling, when we
531 * only half set up a mm_struct for a new process and need to restore 531 * only half set up a mm_struct for a new process and need to restore
532 * the old one. Because we mmput the new mm_struct before 532 * the old one. Because we mmput the new mm_struct before
533 * restoring the old one. . . 533 * restoring the old one. . .
534 * Eric Biederman 10 January 1998 534 * Eric Biederman 10 January 1998
535 */ 535 */
536 void mm_release(struct task_struct *tsk, struct mm_struct *mm) 536 void mm_release(struct task_struct *tsk, struct mm_struct *mm)
537 { 537 {
538 struct completion *vfork_done = tsk->vfork_done; 538 struct completion *vfork_done = tsk->vfork_done;
539 539
540 /* Get rid of any futexes when releasing the mm */ 540 /* Get rid of any futexes when releasing the mm */
541 #ifdef CONFIG_FUTEX 541 #ifdef CONFIG_FUTEX
542 if (unlikely(tsk->robust_list)) 542 if (unlikely(tsk->robust_list))
543 exit_robust_list(tsk); 543 exit_robust_list(tsk);
544 #ifdef CONFIG_COMPAT 544 #ifdef CONFIG_COMPAT
545 if (unlikely(tsk->compat_robust_list)) 545 if (unlikely(tsk->compat_robust_list))
546 compat_exit_robust_list(tsk); 546 compat_exit_robust_list(tsk);
547 #endif 547 #endif
548 #endif 548 #endif
549 549
550 /* Get rid of any cached register state */ 550 /* Get rid of any cached register state */
551 deactivate_mm(tsk, mm); 551 deactivate_mm(tsk, mm);
552 552
553 /* notify parent sleeping on vfork() */ 553 /* notify parent sleeping on vfork() */
554 if (vfork_done) { 554 if (vfork_done) {
555 tsk->vfork_done = NULL; 555 tsk->vfork_done = NULL;
556 complete(vfork_done); 556 complete(vfork_done);
557 } 557 }
558 558
559 /* 559 /*
560 * If we're exiting normally, clear a user-space tid field if 560 * If we're exiting normally, clear a user-space tid field if
561 * requested. We leave this alone when dying by signal, to leave 561 * requested. We leave this alone when dying by signal, to leave
562 * the value intact in a core dump, and to save the unnecessary 562 * the value intact in a core dump, and to save the unnecessary
563 * trouble otherwise. Userland only wants this done for a sys_exit. 563 * trouble otherwise. Userland only wants this done for a sys_exit.
564 */ 564 */
565 if (tsk->clear_child_tid 565 if (tsk->clear_child_tid
566 && !(tsk->flags & PF_SIGNALED) 566 && !(tsk->flags & PF_SIGNALED)
567 && atomic_read(&mm->mm_users) > 1) { 567 && atomic_read(&mm->mm_users) > 1) {
568 u32 __user * tidptr = tsk->clear_child_tid; 568 u32 __user * tidptr = tsk->clear_child_tid;
569 tsk->clear_child_tid = NULL; 569 tsk->clear_child_tid = NULL;
570 570
571 /* 571 /*
572 * We don't check the error code - if userspace has 572 * We don't check the error code - if userspace has
573 * not set up a proper pointer then tough luck. 573 * not set up a proper pointer then tough luck.
574 */ 574 */
575 put_user(0, tidptr); 575 put_user(0, tidptr);
576 sys_futex(tidptr, FUTEX_WAKE, 1, NULL, NULL, 0); 576 sys_futex(tidptr, FUTEX_WAKE, 1, NULL, NULL, 0);
577 } 577 }
578 } 578 }
579 579
580 /* 580 /*
581 * Allocate a new mm structure and copy contents from the 581 * Allocate a new mm structure and copy contents from the
582 * mm structure of the passed in task structure. 582 * mm structure of the passed in task structure.
583 */ 583 */
584 struct mm_struct *dup_mm(struct task_struct *tsk) 584 struct mm_struct *dup_mm(struct task_struct *tsk)
585 { 585 {
586 struct mm_struct *mm, *oldmm = current->mm; 586 struct mm_struct *mm, *oldmm = current->mm;
587 int err; 587 int err;
588 588
589 if (!oldmm) 589 if (!oldmm)
590 return NULL; 590 return NULL;
591 591
592 mm = allocate_mm(); 592 mm = allocate_mm();
593 if (!mm) 593 if (!mm)
594 goto fail_nomem; 594 goto fail_nomem;
595 595
596 memcpy(mm, oldmm, sizeof(*mm)); 596 memcpy(mm, oldmm, sizeof(*mm));
597 597
598 /* Initializing for Swap token stuff */ 598 /* Initializing for Swap token stuff */
599 mm->token_priority = 0; 599 mm->token_priority = 0;
600 mm->last_interval = 0; 600 mm->last_interval = 0;
601 601
602 if (!mm_init(mm, tsk)) 602 if (!mm_init(mm, tsk))
603 goto fail_nomem; 603 goto fail_nomem;
604 604
605 if (init_new_context(tsk, mm)) 605 if (init_new_context(tsk, mm))
606 goto fail_nocontext; 606 goto fail_nocontext;
607 607
608 dup_mm_exe_file(oldmm, mm); 608 dup_mm_exe_file(oldmm, mm);
609 609
610 err = dup_mmap(mm, oldmm); 610 err = dup_mmap(mm, oldmm);
611 if (err) 611 if (err)
612 goto free_pt; 612 goto free_pt;
613 613
614 mm->hiwater_rss = get_mm_rss(mm); 614 mm->hiwater_rss = get_mm_rss(mm);
615 mm->hiwater_vm = mm->total_vm; 615 mm->hiwater_vm = mm->total_vm;
616 616
617 return mm; 617 return mm;
618 618
619 free_pt: 619 free_pt:
620 mmput(mm); 620 mmput(mm);
621 621
622 fail_nomem: 622 fail_nomem:
623 return NULL; 623 return NULL;
624 624
625 fail_nocontext: 625 fail_nocontext:
626 /* 626 /*
627 * If init_new_context() failed, we cannot use mmput() to free the mm 627 * If init_new_context() failed, we cannot use mmput() to free the mm
628 * because it calls destroy_context() 628 * because it calls destroy_context()
629 */ 629 */
630 mm_free_pgd(mm); 630 mm_free_pgd(mm);
631 free_mm(mm); 631 free_mm(mm);
632 return NULL; 632 return NULL;
633 } 633 }
634 634
635 static int copy_mm(unsigned long clone_flags, struct task_struct * tsk) 635 static int copy_mm(unsigned long clone_flags, struct task_struct * tsk)
636 { 636 {
637 struct mm_struct * mm, *oldmm; 637 struct mm_struct * mm, *oldmm;
638 int retval; 638 int retval;
639 639
640 tsk->min_flt = tsk->maj_flt = 0; 640 tsk->min_flt = tsk->maj_flt = 0;
641 tsk->nvcsw = tsk->nivcsw = 0; 641 tsk->nvcsw = tsk->nivcsw = 0;
642 642
643 tsk->mm = NULL; 643 tsk->mm = NULL;
644 tsk->active_mm = NULL; 644 tsk->active_mm = NULL;
645 645
646 /* 646 /*
647 * Are we cloning a kernel thread? 647 * Are we cloning a kernel thread?
648 * 648 *
649 * We need to steal a active VM for that.. 649 * We need to steal a active VM for that..
650 */ 650 */
651 oldmm = current->mm; 651 oldmm = current->mm;
652 if (!oldmm) 652 if (!oldmm)
653 return 0; 653 return 0;
654 654
655 if (clone_flags & CLONE_VM) { 655 if (clone_flags & CLONE_VM) {
656 atomic_inc(&oldmm->mm_users); 656 atomic_inc(&oldmm->mm_users);
657 mm = oldmm; 657 mm = oldmm;
658 goto good_mm; 658 goto good_mm;
659 } 659 }
660 660
661 retval = -ENOMEM; 661 retval = -ENOMEM;
662 mm = dup_mm(tsk); 662 mm = dup_mm(tsk);
663 if (!mm) 663 if (!mm)
664 goto fail_nomem; 664 goto fail_nomem;
665 665
666 good_mm: 666 good_mm:
667 /* Initializing for Swap token stuff */ 667 /* Initializing for Swap token stuff */
668 mm->token_priority = 0; 668 mm->token_priority = 0;
669 mm->last_interval = 0; 669 mm->last_interval = 0;
670 670
671 tsk->mm = mm; 671 tsk->mm = mm;
672 tsk->active_mm = mm; 672 tsk->active_mm = mm;
673 return 0; 673 return 0;
674 674
675 fail_nomem: 675 fail_nomem:
676 return retval; 676 return retval;
677 } 677 }
678 678
679 static struct fs_struct *__copy_fs_struct(struct fs_struct *old) 679 static struct fs_struct *__copy_fs_struct(struct fs_struct *old)
680 { 680 {
681 struct fs_struct *fs = kmem_cache_alloc(fs_cachep, GFP_KERNEL); 681 struct fs_struct *fs = kmem_cache_alloc(fs_cachep, GFP_KERNEL);
682 /* We don't need to lock fs - think why ;-) */ 682 /* We don't need to lock fs - think why ;-) */
683 if (fs) { 683 if (fs) {
684 atomic_set(&fs->count, 1); 684 atomic_set(&fs->count, 1);
685 rwlock_init(&fs->lock); 685 rwlock_init(&fs->lock);
686 fs->umask = old->umask; 686 fs->umask = old->umask;
687 read_lock(&old->lock); 687 read_lock(&old->lock);
688 fs->root = old->root; 688 fs->root = old->root;
689 path_get(&old->root); 689 path_get(&old->root);
690 fs->pwd = old->pwd; 690 fs->pwd = old->pwd;
691 path_get(&old->pwd); 691 path_get(&old->pwd);
692 read_unlock(&old->lock); 692 read_unlock(&old->lock);
693 } 693 }
694 return fs; 694 return fs;
695 } 695 }
696 696
697 struct fs_struct *copy_fs_struct(struct fs_struct *old) 697 struct fs_struct *copy_fs_struct(struct fs_struct *old)
698 { 698 {
699 return __copy_fs_struct(old); 699 return __copy_fs_struct(old);
700 } 700 }
701 701
702 EXPORT_SYMBOL_GPL(copy_fs_struct); 702 EXPORT_SYMBOL_GPL(copy_fs_struct);
703 703
704 static int copy_fs(unsigned long clone_flags, struct task_struct *tsk) 704 static int copy_fs(unsigned long clone_flags, struct task_struct *tsk)
705 { 705 {
706 if (clone_flags & CLONE_FS) { 706 if (clone_flags & CLONE_FS) {
707 atomic_inc(&current->fs->count); 707 atomic_inc(&current->fs->count);
708 return 0; 708 return 0;
709 } 709 }
710 tsk->fs = __copy_fs_struct(current->fs); 710 tsk->fs = __copy_fs_struct(current->fs);
711 if (!tsk->fs) 711 if (!tsk->fs)
712 return -ENOMEM; 712 return -ENOMEM;
713 return 0; 713 return 0;
714 } 714 }
715 715
716 static int copy_files(unsigned long clone_flags, struct task_struct * tsk) 716 static int copy_files(unsigned long clone_flags, struct task_struct * tsk)
717 { 717 {
718 struct files_struct *oldf, *newf; 718 struct files_struct *oldf, *newf;
719 int error = 0; 719 int error = 0;
720 720
721 /* 721 /*
722 * A background process may not have any files ... 722 * A background process may not have any files ...
723 */ 723 */
724 oldf = current->files; 724 oldf = current->files;
725 if (!oldf) 725 if (!oldf)
726 goto out; 726 goto out;
727 727
728 if (clone_flags & CLONE_FILES) { 728 if (clone_flags & CLONE_FILES) {
729 atomic_inc(&oldf->count); 729 atomic_inc(&oldf->count);
730 goto out; 730 goto out;
731 } 731 }
732 732
733 newf = dup_fd(oldf, &error); 733 newf = dup_fd(oldf, &error);
734 if (!newf) 734 if (!newf)
735 goto out; 735 goto out;
736 736
737 tsk->files = newf; 737 tsk->files = newf;
738 error = 0; 738 error = 0;
739 out: 739 out:
740 return error; 740 return error;
741 } 741 }
742 742
743 static int copy_io(unsigned long clone_flags, struct task_struct *tsk) 743 static int copy_io(unsigned long clone_flags, struct task_struct *tsk)
744 { 744 {
745 #ifdef CONFIG_BLOCK 745 #ifdef CONFIG_BLOCK
746 struct io_context *ioc = current->io_context; 746 struct io_context *ioc = current->io_context;
747 747
748 if (!ioc) 748 if (!ioc)
749 return 0; 749 return 0;
750 /* 750 /*
751 * Share io context with parent, if CLONE_IO is set 751 * Share io context with parent, if CLONE_IO is set
752 */ 752 */
753 if (clone_flags & CLONE_IO) { 753 if (clone_flags & CLONE_IO) {
754 tsk->io_context = ioc_task_link(ioc); 754 tsk->io_context = ioc_task_link(ioc);
755 if (unlikely(!tsk->io_context)) 755 if (unlikely(!tsk->io_context))
756 return -ENOMEM; 756 return -ENOMEM;
757 } else if (ioprio_valid(ioc->ioprio)) { 757 } else if (ioprio_valid(ioc->ioprio)) {
758 tsk->io_context = alloc_io_context(GFP_KERNEL, -1); 758 tsk->io_context = alloc_io_context(GFP_KERNEL, -1);
759 if (unlikely(!tsk->io_context)) 759 if (unlikely(!tsk->io_context))
760 return -ENOMEM; 760 return -ENOMEM;
761 761
762 tsk->io_context->ioprio = ioc->ioprio; 762 tsk->io_context->ioprio = ioc->ioprio;
763 } 763 }
764 #endif 764 #endif
765 return 0; 765 return 0;
766 } 766 }
767 767
768 static int copy_sighand(unsigned long clone_flags, struct task_struct *tsk) 768 static int copy_sighand(unsigned long clone_flags, struct task_struct *tsk)
769 { 769 {
770 struct sighand_struct *sig; 770 struct sighand_struct *sig;
771 771
772 if (clone_flags & CLONE_SIGHAND) { 772 if (clone_flags & CLONE_SIGHAND) {
773 atomic_inc(&current->sighand->count); 773 atomic_inc(&current->sighand->count);
774 return 0; 774 return 0;
775 } 775 }
776 sig = kmem_cache_alloc(sighand_cachep, GFP_KERNEL); 776 sig = kmem_cache_alloc(sighand_cachep, GFP_KERNEL);
777 rcu_assign_pointer(tsk->sighand, sig); 777 rcu_assign_pointer(tsk->sighand, sig);
778 if (!sig) 778 if (!sig)
779 return -ENOMEM; 779 return -ENOMEM;
780 atomic_set(&sig->count, 1); 780 atomic_set(&sig->count, 1);
781 memcpy(sig->action, current->sighand->action, sizeof(sig->action)); 781 memcpy(sig->action, current->sighand->action, sizeof(sig->action));
782 return 0; 782 return 0;
783 } 783 }
784 784
785 void __cleanup_sighand(struct sighand_struct *sighand) 785 void __cleanup_sighand(struct sighand_struct *sighand)
786 { 786 {
787 if (atomic_dec_and_test(&sighand->count)) 787 if (atomic_dec_and_test(&sighand->count))
788 kmem_cache_free(sighand_cachep, sighand); 788 kmem_cache_free(sighand_cachep, sighand);
789 } 789 }
790 790
791 791
792 /* 792 /*
793 * Initialize POSIX timer handling for a thread group. 793 * Initialize POSIX timer handling for a thread group.
794 */ 794 */
795 static void posix_cpu_timers_init_group(struct signal_struct *sig) 795 static void posix_cpu_timers_init_group(struct signal_struct *sig)
796 { 796 {
797 /* Thread group counters. */ 797 /* Thread group counters. */
798 thread_group_cputime_init(sig); 798 thread_group_cputime_init(sig);
799 799
800 /* Expiration times and increments. */ 800 /* Expiration times and increments. */
801 sig->it_virt_expires = cputime_zero; 801 sig->it_virt_expires = cputime_zero;
802 sig->it_virt_incr = cputime_zero; 802 sig->it_virt_incr = cputime_zero;
803 sig->it_prof_expires = cputime_zero; 803 sig->it_prof_expires = cputime_zero;
804 sig->it_prof_incr = cputime_zero; 804 sig->it_prof_incr = cputime_zero;
805 805
806 /* Cached expiration times. */ 806 /* Cached expiration times. */
807 sig->cputime_expires.prof_exp = cputime_zero; 807 sig->cputime_expires.prof_exp = cputime_zero;
808 sig->cputime_expires.virt_exp = cputime_zero; 808 sig->cputime_expires.virt_exp = cputime_zero;
809 sig->cputime_expires.sched_exp = 0; 809 sig->cputime_expires.sched_exp = 0;
810 810
811 if (sig->rlim[RLIMIT_CPU].rlim_cur != RLIM_INFINITY) {
812 sig->cputime_expires.prof_exp =
813 secs_to_cputime(sig->rlim[RLIMIT_CPU].rlim_cur);
814 sig->cputimer.running = 1;
815 }
816
811 /* The timer lists. */ 817 /* The timer lists. */
812 INIT_LIST_HEAD(&sig->cpu_timers[0]); 818 INIT_LIST_HEAD(&sig->cpu_timers[0]);
813 INIT_LIST_HEAD(&sig->cpu_timers[1]); 819 INIT_LIST_HEAD(&sig->cpu_timers[1]);
814 INIT_LIST_HEAD(&sig->cpu_timers[2]); 820 INIT_LIST_HEAD(&sig->cpu_timers[2]);
815 } 821 }
816 822
817 static int copy_signal(unsigned long clone_flags, struct task_struct *tsk) 823 static int copy_signal(unsigned long clone_flags, struct task_struct *tsk)
818 { 824 {
819 struct signal_struct *sig; 825 struct signal_struct *sig;
820 826
821 if (clone_flags & CLONE_THREAD) { 827 if (clone_flags & CLONE_THREAD) {
822 atomic_inc(&current->signal->count); 828 atomic_inc(&current->signal->count);
823 atomic_inc(&current->signal->live); 829 atomic_inc(&current->signal->live);
824 return 0; 830 return 0;
825 } 831 }
826 sig = kmem_cache_alloc(signal_cachep, GFP_KERNEL);
827 832
828 if (sig) 833 sig = kmem_cache_alloc(signal_cachep, GFP_KERNEL);
829 posix_cpu_timers_init_group(sig);
830
831 tsk->signal = sig; 834 tsk->signal = sig;
832 if (!sig) 835 if (!sig)
833 return -ENOMEM; 836 return -ENOMEM;
834 837
835 atomic_set(&sig->count, 1); 838 atomic_set(&sig->count, 1);
836 atomic_set(&sig->live, 1); 839 atomic_set(&sig->live, 1);
837 init_waitqueue_head(&sig->wait_chldexit); 840 init_waitqueue_head(&sig->wait_chldexit);
838 sig->flags = 0; 841 sig->flags = 0;
839 sig->group_exit_code = 0; 842 sig->group_exit_code = 0;
840 sig->group_exit_task = NULL; 843 sig->group_exit_task = NULL;
841 sig->group_stop_count = 0; 844 sig->group_stop_count = 0;
842 sig->curr_target = tsk; 845 sig->curr_target = tsk;
843 init_sigpending(&sig->shared_pending); 846 init_sigpending(&sig->shared_pending);
844 INIT_LIST_HEAD(&sig->posix_timers); 847 INIT_LIST_HEAD(&sig->posix_timers);
845 848
846 hrtimer_init(&sig->real_timer, CLOCK_MONOTONIC, HRTIMER_MODE_REL); 849 hrtimer_init(&sig->real_timer, CLOCK_MONOTONIC, HRTIMER_MODE_REL);
847 sig->it_real_incr.tv64 = 0; 850 sig->it_real_incr.tv64 = 0;
848 sig->real_timer.function = it_real_fn; 851 sig->real_timer.function = it_real_fn;
849 852
850 sig->leader = 0; /* session leadership doesn't inherit */ 853 sig->leader = 0; /* session leadership doesn't inherit */
851 sig->tty_old_pgrp = NULL; 854 sig->tty_old_pgrp = NULL;
852 sig->tty = NULL; 855 sig->tty = NULL;
853 856
854 sig->utime = sig->stime = sig->cutime = sig->cstime = cputime_zero; 857 sig->utime = sig->stime = sig->cutime = sig->cstime = cputime_zero;
855 sig->gtime = cputime_zero; 858 sig->gtime = cputime_zero;
856 sig->cgtime = cputime_zero; 859 sig->cgtime = cputime_zero;
857 sig->nvcsw = sig->nivcsw = sig->cnvcsw = sig->cnivcsw = 0; 860 sig->nvcsw = sig->nivcsw = sig->cnvcsw = sig->cnivcsw = 0;
858 sig->min_flt = sig->maj_flt = sig->cmin_flt = sig->cmaj_flt = 0; 861 sig->min_flt = sig->maj_flt = sig->cmin_flt = sig->cmaj_flt = 0;
859 sig->inblock = sig->oublock = sig->cinblock = sig->coublock = 0; 862 sig->inblock = sig->oublock = sig->cinblock = sig->coublock = 0;
860 task_io_accounting_init(&sig->ioac); 863 task_io_accounting_init(&sig->ioac);
861 sig->sum_sched_runtime = 0; 864 sig->sum_sched_runtime = 0;
862 taskstats_tgid_init(sig); 865 taskstats_tgid_init(sig);
863 866
864 task_lock(current->group_leader); 867 task_lock(current->group_leader);
865 memcpy(sig->rlim, current->signal->rlim, sizeof sig->rlim); 868 memcpy(sig->rlim, current->signal->rlim, sizeof sig->rlim);
866 task_unlock(current->group_leader); 869 task_unlock(current->group_leader);
870
871 posix_cpu_timers_init_group(sig);
867 872
868 acct_init_pacct(&sig->pacct); 873 acct_init_pacct(&sig->pacct);
869 874
870 tty_audit_fork(sig); 875 tty_audit_fork(sig);
871 876
872 return 0; 877 return 0;
873 } 878 }
874 879
875 void __cleanup_signal(struct signal_struct *sig) 880 void __cleanup_signal(struct signal_struct *sig)
876 { 881 {
877 thread_group_cputime_free(sig); 882 thread_group_cputime_free(sig);
878 tty_kref_put(sig->tty); 883 tty_kref_put(sig->tty);
879 kmem_cache_free(signal_cachep, sig); 884 kmem_cache_free(signal_cachep, sig);
880 } 885 }
881 886
882 static void cleanup_signal(struct task_struct *tsk) 887 static void cleanup_signal(struct task_struct *tsk)
883 { 888 {
884 struct signal_struct *sig = tsk->signal; 889 struct signal_struct *sig = tsk->signal;
885 890
886 atomic_dec(&sig->live); 891 atomic_dec(&sig->live);
887 892
888 if (atomic_dec_and_test(&sig->count)) 893 if (atomic_dec_and_test(&sig->count))
889 __cleanup_signal(sig); 894 __cleanup_signal(sig);
890 } 895 }
891 896
892 static void copy_flags(unsigned long clone_flags, struct task_struct *p) 897 static void copy_flags(unsigned long clone_flags, struct task_struct *p)
893 { 898 {
894 unsigned long new_flags = p->flags; 899 unsigned long new_flags = p->flags;
895 900
896 new_flags &= ~PF_SUPERPRIV; 901 new_flags &= ~PF_SUPERPRIV;
897 new_flags |= PF_FORKNOEXEC; 902 new_flags |= PF_FORKNOEXEC;
898 new_flags |= PF_STARTING; 903 new_flags |= PF_STARTING;
899 p->flags = new_flags; 904 p->flags = new_flags;
900 clear_freeze_flag(p); 905 clear_freeze_flag(p);
901 } 906 }
902 907
903 SYSCALL_DEFINE1(set_tid_address, int __user *, tidptr) 908 SYSCALL_DEFINE1(set_tid_address, int __user *, tidptr)
904 { 909 {
905 current->clear_child_tid = tidptr; 910 current->clear_child_tid = tidptr;
906 911
907 return task_pid_vnr(current); 912 return task_pid_vnr(current);
908 } 913 }
909 914
910 static void rt_mutex_init_task(struct task_struct *p) 915 static void rt_mutex_init_task(struct task_struct *p)
911 { 916 {
912 spin_lock_init(&p->pi_lock); 917 spin_lock_init(&p->pi_lock);
913 #ifdef CONFIG_RT_MUTEXES 918 #ifdef CONFIG_RT_MUTEXES
914 plist_head_init(&p->pi_waiters, &p->pi_lock); 919 plist_head_init(&p->pi_waiters, &p->pi_lock);
915 p->pi_blocked_on = NULL; 920 p->pi_blocked_on = NULL;
916 #endif 921 #endif
917 } 922 }
918 923
919 #ifdef CONFIG_MM_OWNER 924 #ifdef CONFIG_MM_OWNER
920 void mm_init_owner(struct mm_struct *mm, struct task_struct *p) 925 void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
921 { 926 {
922 mm->owner = p; 927 mm->owner = p;
923 } 928 }
924 #endif /* CONFIG_MM_OWNER */ 929 #endif /* CONFIG_MM_OWNER */
925 930
926 /* 931 /*
927 * Initialize POSIX timer handling for a single task. 932 * Initialize POSIX timer handling for a single task.
928 */ 933 */
929 static void posix_cpu_timers_init(struct task_struct *tsk) 934 static void posix_cpu_timers_init(struct task_struct *tsk)
930 { 935 {
931 tsk->cputime_expires.prof_exp = cputime_zero; 936 tsk->cputime_expires.prof_exp = cputime_zero;
932 tsk->cputime_expires.virt_exp = cputime_zero; 937 tsk->cputime_expires.virt_exp = cputime_zero;
933 tsk->cputime_expires.sched_exp = 0; 938 tsk->cputime_expires.sched_exp = 0;
934 INIT_LIST_HEAD(&tsk->cpu_timers[0]); 939 INIT_LIST_HEAD(&tsk->cpu_timers[0]);
935 INIT_LIST_HEAD(&tsk->cpu_timers[1]); 940 INIT_LIST_HEAD(&tsk->cpu_timers[1]);
936 INIT_LIST_HEAD(&tsk->cpu_timers[2]); 941 INIT_LIST_HEAD(&tsk->cpu_timers[2]);
937 } 942 }
938 943
939 /* 944 /*
940 * This creates a new process as a copy of the old one, 945 * This creates a new process as a copy of the old one,
941 * but does not actually start it yet. 946 * but does not actually start it yet.
942 * 947 *
943 * It copies the registers, and all the appropriate 948 * It copies the registers, and all the appropriate
944 * parts of the process environment (as per the clone 949 * parts of the process environment (as per the clone
945 * flags). The actual kick-off is left to the caller. 950 * flags). The actual kick-off is left to the caller.
946 */ 951 */
947 static struct task_struct *copy_process(unsigned long clone_flags, 952 static struct task_struct *copy_process(unsigned long clone_flags,
948 unsigned long stack_start, 953 unsigned long stack_start,
949 struct pt_regs *regs, 954 struct pt_regs *regs,
950 unsigned long stack_size, 955 unsigned long stack_size,
951 int __user *child_tidptr, 956 int __user *child_tidptr,
952 struct pid *pid, 957 struct pid *pid,
953 int trace) 958 int trace)
954 { 959 {
955 int retval; 960 int retval;
956 struct task_struct *p; 961 struct task_struct *p;
957 int cgroup_callbacks_done = 0; 962 int cgroup_callbacks_done = 0;
958 963
959 if ((clone_flags & (CLONE_NEWNS|CLONE_FS)) == (CLONE_NEWNS|CLONE_FS)) 964 if ((clone_flags & (CLONE_NEWNS|CLONE_FS)) == (CLONE_NEWNS|CLONE_FS))
960 return ERR_PTR(-EINVAL); 965 return ERR_PTR(-EINVAL);
961 966
962 /* 967 /*
963 * Thread groups must share signals as well, and detached threads 968 * Thread groups must share signals as well, and detached threads
964 * can only be started up within the thread group. 969 * can only be started up within the thread group.
965 */ 970 */
966 if ((clone_flags & CLONE_THREAD) && !(clone_flags & CLONE_SIGHAND)) 971 if ((clone_flags & CLONE_THREAD) && !(clone_flags & CLONE_SIGHAND))
967 return ERR_PTR(-EINVAL); 972 return ERR_PTR(-EINVAL);
968 973
969 /* 974 /*
970 * Shared signal handlers imply shared VM. By way of the above, 975 * Shared signal handlers imply shared VM. By way of the above,
971 * thread groups also imply shared VM. Blocking this case allows 976 * thread groups also imply shared VM. Blocking this case allows
972 * for various simplifications in other code. 977 * for various simplifications in other code.
973 */ 978 */
974 if ((clone_flags & CLONE_SIGHAND) && !(clone_flags & CLONE_VM)) 979 if ((clone_flags & CLONE_SIGHAND) && !(clone_flags & CLONE_VM))
975 return ERR_PTR(-EINVAL); 980 return ERR_PTR(-EINVAL);
976 981
977 retval = security_task_create(clone_flags); 982 retval = security_task_create(clone_flags);
978 if (retval) 983 if (retval)
979 goto fork_out; 984 goto fork_out;
980 985
981 retval = -ENOMEM; 986 retval = -ENOMEM;
982 p = dup_task_struct(current); 987 p = dup_task_struct(current);
983 if (!p) 988 if (!p)
984 goto fork_out; 989 goto fork_out;
985 990
986 rt_mutex_init_task(p); 991 rt_mutex_init_task(p);
987 992
988 #ifdef CONFIG_PROVE_LOCKING 993 #ifdef CONFIG_PROVE_LOCKING
989 DEBUG_LOCKS_WARN_ON(!p->hardirqs_enabled); 994 DEBUG_LOCKS_WARN_ON(!p->hardirqs_enabled);
990 DEBUG_LOCKS_WARN_ON(!p->softirqs_enabled); 995 DEBUG_LOCKS_WARN_ON(!p->softirqs_enabled);
991 #endif 996 #endif
992 retval = -EAGAIN; 997 retval = -EAGAIN;
993 if (atomic_read(&p->real_cred->user->processes) >= 998 if (atomic_read(&p->real_cred->user->processes) >=
994 p->signal->rlim[RLIMIT_NPROC].rlim_cur) { 999 p->signal->rlim[RLIMIT_NPROC].rlim_cur) {
995 if (!capable(CAP_SYS_ADMIN) && !capable(CAP_SYS_RESOURCE) && 1000 if (!capable(CAP_SYS_ADMIN) && !capable(CAP_SYS_RESOURCE) &&
996 p->real_cred->user != INIT_USER) 1001 p->real_cred->user != INIT_USER)
997 goto bad_fork_free; 1002 goto bad_fork_free;
998 } 1003 }
999 1004
1000 retval = copy_creds(p, clone_flags); 1005 retval = copy_creds(p, clone_flags);
1001 if (retval < 0) 1006 if (retval < 0)
1002 goto bad_fork_free; 1007 goto bad_fork_free;
1003 1008
1004 /* 1009 /*
1005 * If multiple threads are within copy_process(), then this check 1010 * If multiple threads are within copy_process(), then this check
1006 * triggers too late. This doesn't hurt, the check is only there 1011 * triggers too late. This doesn't hurt, the check is only there
1007 * to stop root fork bombs. 1012 * to stop root fork bombs.
1008 */ 1013 */
1009 retval = -EAGAIN; 1014 retval = -EAGAIN;
1010 if (nr_threads >= max_threads) 1015 if (nr_threads >= max_threads)
1011 goto bad_fork_cleanup_count; 1016 goto bad_fork_cleanup_count;
1012 1017
1013 if (!try_module_get(task_thread_info(p)->exec_domain->module)) 1018 if (!try_module_get(task_thread_info(p)->exec_domain->module))
1014 goto bad_fork_cleanup_count; 1019 goto bad_fork_cleanup_count;
1015 1020
1016 if (p->binfmt && !try_module_get(p->binfmt->module)) 1021 if (p->binfmt && !try_module_get(p->binfmt->module))
1017 goto bad_fork_cleanup_put_domain; 1022 goto bad_fork_cleanup_put_domain;
1018 1023
1019 p->did_exec = 0; 1024 p->did_exec = 0;
1020 delayacct_tsk_init(p); /* Must remain after dup_task_struct() */ 1025 delayacct_tsk_init(p); /* Must remain after dup_task_struct() */
1021 copy_flags(clone_flags, p); 1026 copy_flags(clone_flags, p);
1022 INIT_LIST_HEAD(&p->children); 1027 INIT_LIST_HEAD(&p->children);
1023 INIT_LIST_HEAD(&p->sibling); 1028 INIT_LIST_HEAD(&p->sibling);
1024 #ifdef CONFIG_PREEMPT_RCU 1029 #ifdef CONFIG_PREEMPT_RCU
1025 p->rcu_read_lock_nesting = 0; 1030 p->rcu_read_lock_nesting = 0;
1026 p->rcu_flipctr_idx = 0; 1031 p->rcu_flipctr_idx = 0;
1027 #endif /* #ifdef CONFIG_PREEMPT_RCU */ 1032 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1028 p->vfork_done = NULL; 1033 p->vfork_done = NULL;
1029 spin_lock_init(&p->alloc_lock); 1034 spin_lock_init(&p->alloc_lock);
1030 1035
1031 clear_tsk_thread_flag(p, TIF_SIGPENDING); 1036 clear_tsk_thread_flag(p, TIF_SIGPENDING);
1032 init_sigpending(&p->pending); 1037 init_sigpending(&p->pending);
1033 1038
1034 p->utime = cputime_zero; 1039 p->utime = cputime_zero;
1035 p->stime = cputime_zero; 1040 p->stime = cputime_zero;
1036 p->gtime = cputime_zero; 1041 p->gtime = cputime_zero;
1037 p->utimescaled = cputime_zero; 1042 p->utimescaled = cputime_zero;
1038 p->stimescaled = cputime_zero; 1043 p->stimescaled = cputime_zero;
1039 p->prev_utime = cputime_zero; 1044 p->prev_utime = cputime_zero;
1040 p->prev_stime = cputime_zero; 1045 p->prev_stime = cputime_zero;
1041 1046
1042 p->default_timer_slack_ns = current->timer_slack_ns; 1047 p->default_timer_slack_ns = current->timer_slack_ns;
1043 1048
1044 #ifdef CONFIG_DETECT_SOFTLOCKUP 1049 #ifdef CONFIG_DETECT_SOFTLOCKUP
1045 p->last_switch_count = 0; 1050 p->last_switch_count = 0;
1046 p->last_switch_timestamp = 0; 1051 p->last_switch_timestamp = 0;
1047 #endif 1052 #endif
1048 1053
1049 task_io_accounting_init(&p->ioac); 1054 task_io_accounting_init(&p->ioac);
1050 acct_clear_integrals(p); 1055 acct_clear_integrals(p);
1051 1056
1052 posix_cpu_timers_init(p); 1057 posix_cpu_timers_init(p);
1053 1058
1054 p->lock_depth = -1; /* -1 = no lock */ 1059 p->lock_depth = -1; /* -1 = no lock */
1055 do_posix_clock_monotonic_gettime(&p->start_time); 1060 do_posix_clock_monotonic_gettime(&p->start_time);
1056 p->real_start_time = p->start_time; 1061 p->real_start_time = p->start_time;
1057 monotonic_to_bootbased(&p->real_start_time); 1062 monotonic_to_bootbased(&p->real_start_time);
1058 p->io_context = NULL; 1063 p->io_context = NULL;
1059 p->audit_context = NULL; 1064 p->audit_context = NULL;
1060 cgroup_fork(p); 1065 cgroup_fork(p);
1061 #ifdef CONFIG_NUMA 1066 #ifdef CONFIG_NUMA
1062 p->mempolicy = mpol_dup(p->mempolicy); 1067 p->mempolicy = mpol_dup(p->mempolicy);
1063 if (IS_ERR(p->mempolicy)) { 1068 if (IS_ERR(p->mempolicy)) {
1064 retval = PTR_ERR(p->mempolicy); 1069 retval = PTR_ERR(p->mempolicy);
1065 p->mempolicy = NULL; 1070 p->mempolicy = NULL;
1066 goto bad_fork_cleanup_cgroup; 1071 goto bad_fork_cleanup_cgroup;
1067 } 1072 }
1068 mpol_fix_fork_child_flag(p); 1073 mpol_fix_fork_child_flag(p);
1069 #endif 1074 #endif
1070 #ifdef CONFIG_TRACE_IRQFLAGS 1075 #ifdef CONFIG_TRACE_IRQFLAGS
1071 p->irq_events = 0; 1076 p->irq_events = 0;
1072 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW 1077 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
1073 p->hardirqs_enabled = 1; 1078 p->hardirqs_enabled = 1;
1074 #else 1079 #else
1075 p->hardirqs_enabled = 0; 1080 p->hardirqs_enabled = 0;
1076 #endif 1081 #endif
1077 p->hardirq_enable_ip = 0; 1082 p->hardirq_enable_ip = 0;
1078 p->hardirq_enable_event = 0; 1083 p->hardirq_enable_event = 0;
1079 p->hardirq_disable_ip = _THIS_IP_; 1084 p->hardirq_disable_ip = _THIS_IP_;
1080 p->hardirq_disable_event = 0; 1085 p->hardirq_disable_event = 0;
1081 p->softirqs_enabled = 1; 1086 p->softirqs_enabled = 1;
1082 p->softirq_enable_ip = _THIS_IP_; 1087 p->softirq_enable_ip = _THIS_IP_;
1083 p->softirq_enable_event = 0; 1088 p->softirq_enable_event = 0;
1084 p->softirq_disable_ip = 0; 1089 p->softirq_disable_ip = 0;
1085 p->softirq_disable_event = 0; 1090 p->softirq_disable_event = 0;
1086 p->hardirq_context = 0; 1091 p->hardirq_context = 0;
1087 p->softirq_context = 0; 1092 p->softirq_context = 0;
1088 #endif 1093 #endif
1089 #ifdef CONFIG_LOCKDEP 1094 #ifdef CONFIG_LOCKDEP
1090 p->lockdep_depth = 0; /* no locks held yet */ 1095 p->lockdep_depth = 0; /* no locks held yet */
1091 p->curr_chain_key = 0; 1096 p->curr_chain_key = 0;
1092 p->lockdep_recursion = 0; 1097 p->lockdep_recursion = 0;
1093 #endif 1098 #endif
1094 1099
1095 #ifdef CONFIG_DEBUG_MUTEXES 1100 #ifdef CONFIG_DEBUG_MUTEXES
1096 p->blocked_on = NULL; /* not blocked yet */ 1101 p->blocked_on = NULL; /* not blocked yet */
1097 #endif 1102 #endif
1098 if (unlikely(current->ptrace)) 1103 if (unlikely(current->ptrace))
1099 ptrace_fork(p, clone_flags); 1104 ptrace_fork(p, clone_flags);
1100 1105
1101 /* Perform scheduler related setup. Assign this task to a CPU. */ 1106 /* Perform scheduler related setup. Assign this task to a CPU. */
1102 sched_fork(p, clone_flags); 1107 sched_fork(p, clone_flags);
1103 1108
1104 if ((retval = audit_alloc(p))) 1109 if ((retval = audit_alloc(p)))
1105 goto bad_fork_cleanup_policy; 1110 goto bad_fork_cleanup_policy;
1106 /* copy all the process information */ 1111 /* copy all the process information */
1107 if ((retval = copy_semundo(clone_flags, p))) 1112 if ((retval = copy_semundo(clone_flags, p)))
1108 goto bad_fork_cleanup_audit; 1113 goto bad_fork_cleanup_audit;
1109 if ((retval = copy_files(clone_flags, p))) 1114 if ((retval = copy_files(clone_flags, p)))
1110 goto bad_fork_cleanup_semundo; 1115 goto bad_fork_cleanup_semundo;
1111 if ((retval = copy_fs(clone_flags, p))) 1116 if ((retval = copy_fs(clone_flags, p)))
1112 goto bad_fork_cleanup_files; 1117 goto bad_fork_cleanup_files;
1113 if ((retval = copy_sighand(clone_flags, p))) 1118 if ((retval = copy_sighand(clone_flags, p)))
1114 goto bad_fork_cleanup_fs; 1119 goto bad_fork_cleanup_fs;
1115 if ((retval = copy_signal(clone_flags, p))) 1120 if ((retval = copy_signal(clone_flags, p)))
1116 goto bad_fork_cleanup_sighand; 1121 goto bad_fork_cleanup_sighand;
1117 if ((retval = copy_mm(clone_flags, p))) 1122 if ((retval = copy_mm(clone_flags, p)))
1118 goto bad_fork_cleanup_signal; 1123 goto bad_fork_cleanup_signal;
1119 if ((retval = copy_namespaces(clone_flags, p))) 1124 if ((retval = copy_namespaces(clone_flags, p)))
1120 goto bad_fork_cleanup_mm; 1125 goto bad_fork_cleanup_mm;
1121 if ((retval = copy_io(clone_flags, p))) 1126 if ((retval = copy_io(clone_flags, p)))
1122 goto bad_fork_cleanup_namespaces; 1127 goto bad_fork_cleanup_namespaces;
1123 retval = copy_thread(0, clone_flags, stack_start, stack_size, p, regs); 1128 retval = copy_thread(0, clone_flags, stack_start, stack_size, p, regs);
1124 if (retval) 1129 if (retval)
1125 goto bad_fork_cleanup_io; 1130 goto bad_fork_cleanup_io;
1126 1131
1127 if (pid != &init_struct_pid) { 1132 if (pid != &init_struct_pid) {
1128 retval = -ENOMEM; 1133 retval = -ENOMEM;
1129 pid = alloc_pid(p->nsproxy->pid_ns); 1134 pid = alloc_pid(p->nsproxy->pid_ns);
1130 if (!pid) 1135 if (!pid)
1131 goto bad_fork_cleanup_io; 1136 goto bad_fork_cleanup_io;
1132 1137
1133 if (clone_flags & CLONE_NEWPID) { 1138 if (clone_flags & CLONE_NEWPID) {
1134 retval = pid_ns_prepare_proc(p->nsproxy->pid_ns); 1139 retval = pid_ns_prepare_proc(p->nsproxy->pid_ns);
1135 if (retval < 0) 1140 if (retval < 0)
1136 goto bad_fork_free_pid; 1141 goto bad_fork_free_pid;
1137 } 1142 }
1138 } 1143 }
1139 1144
1140 ftrace_graph_init_task(p); 1145 ftrace_graph_init_task(p);
1141 1146
1142 p->pid = pid_nr(pid); 1147 p->pid = pid_nr(pid);
1143 p->tgid = p->pid; 1148 p->tgid = p->pid;
1144 if (clone_flags & CLONE_THREAD) 1149 if (clone_flags & CLONE_THREAD)
1145 p->tgid = current->tgid; 1150 p->tgid = current->tgid;
1146 1151
1147 if (current->nsproxy != p->nsproxy) { 1152 if (current->nsproxy != p->nsproxy) {
1148 retval = ns_cgroup_clone(p, pid); 1153 retval = ns_cgroup_clone(p, pid);
1149 if (retval) 1154 if (retval)
1150 goto bad_fork_free_graph; 1155 goto bad_fork_free_graph;
1151 } 1156 }
1152 1157
1153 p->set_child_tid = (clone_flags & CLONE_CHILD_SETTID) ? child_tidptr : NULL; 1158 p->set_child_tid = (clone_flags & CLONE_CHILD_SETTID) ? child_tidptr : NULL;
1154 /* 1159 /*
1155 * Clear TID on mm_release()? 1160 * Clear TID on mm_release()?
1156 */ 1161 */
1157 p->clear_child_tid = (clone_flags & CLONE_CHILD_CLEARTID) ? child_tidptr: NULL; 1162 p->clear_child_tid = (clone_flags & CLONE_CHILD_CLEARTID) ? child_tidptr: NULL;
1158 #ifdef CONFIG_FUTEX 1163 #ifdef CONFIG_FUTEX
1159 p->robust_list = NULL; 1164 p->robust_list = NULL;
1160 #ifdef CONFIG_COMPAT 1165 #ifdef CONFIG_COMPAT
1161 p->compat_robust_list = NULL; 1166 p->compat_robust_list = NULL;
1162 #endif 1167 #endif
1163 INIT_LIST_HEAD(&p->pi_state_list); 1168 INIT_LIST_HEAD(&p->pi_state_list);
1164 p->pi_state_cache = NULL; 1169 p->pi_state_cache = NULL;
1165 #endif 1170 #endif
1166 /* 1171 /*
1167 * sigaltstack should be cleared when sharing the same VM 1172 * sigaltstack should be cleared when sharing the same VM
1168 */ 1173 */
1169 if ((clone_flags & (CLONE_VM|CLONE_VFORK)) == CLONE_VM) 1174 if ((clone_flags & (CLONE_VM|CLONE_VFORK)) == CLONE_VM)
1170 p->sas_ss_sp = p->sas_ss_size = 0; 1175 p->sas_ss_sp = p->sas_ss_size = 0;
1171 1176
1172 /* 1177 /*
1173 * Syscall tracing should be turned off in the child regardless 1178 * Syscall tracing should be turned off in the child regardless
1174 * of CLONE_PTRACE. 1179 * of CLONE_PTRACE.
1175 */ 1180 */
1176 clear_tsk_thread_flag(p, TIF_SYSCALL_TRACE); 1181 clear_tsk_thread_flag(p, TIF_SYSCALL_TRACE);
1177 #ifdef TIF_SYSCALL_EMU 1182 #ifdef TIF_SYSCALL_EMU
1178 clear_tsk_thread_flag(p, TIF_SYSCALL_EMU); 1183 clear_tsk_thread_flag(p, TIF_SYSCALL_EMU);
1179 #endif 1184 #endif
1180 clear_all_latency_tracing(p); 1185 clear_all_latency_tracing(p);
1181 1186
1182 /* ok, now we should be set up.. */ 1187 /* ok, now we should be set up.. */
1183 p->exit_signal = (clone_flags & CLONE_THREAD) ? -1 : (clone_flags & CSIGNAL); 1188 p->exit_signal = (clone_flags & CLONE_THREAD) ? -1 : (clone_flags & CSIGNAL);
1184 p->pdeath_signal = 0; 1189 p->pdeath_signal = 0;
1185 p->exit_state = 0; 1190 p->exit_state = 0;
1186 1191
1187 /* 1192 /*
1188 * Ok, make it visible to the rest of the system. 1193 * Ok, make it visible to the rest of the system.
1189 * We dont wake it up yet. 1194 * We dont wake it up yet.
1190 */ 1195 */
1191 p->group_leader = p; 1196 p->group_leader = p;
1192 INIT_LIST_HEAD(&p->thread_group); 1197 INIT_LIST_HEAD(&p->thread_group);
1193 1198
1194 /* Now that the task is set up, run cgroup callbacks if 1199 /* Now that the task is set up, run cgroup callbacks if
1195 * necessary. We need to run them before the task is visible 1200 * necessary. We need to run them before the task is visible
1196 * on the tasklist. */ 1201 * on the tasklist. */
1197 cgroup_fork_callbacks(p); 1202 cgroup_fork_callbacks(p);
1198 cgroup_callbacks_done = 1; 1203 cgroup_callbacks_done = 1;
1199 1204
1200 /* Need tasklist lock for parent etc handling! */ 1205 /* Need tasklist lock for parent etc handling! */
1201 write_lock_irq(&tasklist_lock); 1206 write_lock_irq(&tasklist_lock);
1202 1207
1203 /* 1208 /*
1204 * The task hasn't been attached yet, so its cpus_allowed mask will 1209 * The task hasn't been attached yet, so its cpus_allowed mask will
1205 * not be changed, nor will its assigned CPU. 1210 * not be changed, nor will its assigned CPU.
1206 * 1211 *
1207 * The cpus_allowed mask of the parent may have changed after it was 1212 * The cpus_allowed mask of the parent may have changed after it was
1208 * copied first time - so re-copy it here, then check the child's CPU 1213 * copied first time - so re-copy it here, then check the child's CPU
1209 * to ensure it is on a valid CPU (and if not, just force it back to 1214 * to ensure it is on a valid CPU (and if not, just force it back to
1210 * parent's CPU). This avoids alot of nasty races. 1215 * parent's CPU). This avoids alot of nasty races.
1211 */ 1216 */
1212 p->cpus_allowed = current->cpus_allowed; 1217 p->cpus_allowed = current->cpus_allowed;
1213 p->rt.nr_cpus_allowed = current->rt.nr_cpus_allowed; 1218 p->rt.nr_cpus_allowed = current->rt.nr_cpus_allowed;
1214 if (unlikely(!cpu_isset(task_cpu(p), p->cpus_allowed) || 1219 if (unlikely(!cpu_isset(task_cpu(p), p->cpus_allowed) ||
1215 !cpu_online(task_cpu(p)))) 1220 !cpu_online(task_cpu(p))))
1216 set_task_cpu(p, smp_processor_id()); 1221 set_task_cpu(p, smp_processor_id());
1217 1222
1218 /* CLONE_PARENT re-uses the old parent */ 1223 /* CLONE_PARENT re-uses the old parent */
1219 if (clone_flags & (CLONE_PARENT|CLONE_THREAD)) { 1224 if (clone_flags & (CLONE_PARENT|CLONE_THREAD)) {
1220 p->real_parent = current->real_parent; 1225 p->real_parent = current->real_parent;
1221 p->parent_exec_id = current->parent_exec_id; 1226 p->parent_exec_id = current->parent_exec_id;
1222 } else { 1227 } else {
1223 p->real_parent = current; 1228 p->real_parent = current;
1224 p->parent_exec_id = current->self_exec_id; 1229 p->parent_exec_id = current->self_exec_id;
1225 } 1230 }
1226 1231
1227 spin_lock(&current->sighand->siglock); 1232 spin_lock(&current->sighand->siglock);
1228 1233
1229 /* 1234 /*
1230 * Process group and session signals need to be delivered to just the 1235 * Process group and session signals need to be delivered to just the
1231 * parent before the fork or both the parent and the child after the 1236 * parent before the fork or both the parent and the child after the
1232 * fork. Restart if a signal comes in before we add the new process to 1237 * fork. Restart if a signal comes in before we add the new process to
1233 * it's process group. 1238 * it's process group.
1234 * A fatal signal pending means that current will exit, so the new 1239 * A fatal signal pending means that current will exit, so the new
1235 * thread can't slip out of an OOM kill (or normal SIGKILL). 1240 * thread can't slip out of an OOM kill (or normal SIGKILL).
1236 */ 1241 */
1237 recalc_sigpending(); 1242 recalc_sigpending();
1238 if (signal_pending(current)) { 1243 if (signal_pending(current)) {
1239 spin_unlock(&current->sighand->siglock); 1244 spin_unlock(&current->sighand->siglock);
1240 write_unlock_irq(&tasklist_lock); 1245 write_unlock_irq(&tasklist_lock);
1241 retval = -ERESTARTNOINTR; 1246 retval = -ERESTARTNOINTR;
1242 goto bad_fork_free_graph; 1247 goto bad_fork_free_graph;
1243 } 1248 }
1244 1249
1245 if (clone_flags & CLONE_THREAD) { 1250 if (clone_flags & CLONE_THREAD) {
1246 p->group_leader = current->group_leader; 1251 p->group_leader = current->group_leader;
1247 list_add_tail_rcu(&p->thread_group, &p->group_leader->thread_group); 1252 list_add_tail_rcu(&p->thread_group, &p->group_leader->thread_group);
1248 } 1253 }
1249 1254
1250 if (likely(p->pid)) { 1255 if (likely(p->pid)) {
1251 list_add_tail(&p->sibling, &p->real_parent->children); 1256 list_add_tail(&p->sibling, &p->real_parent->children);
1252 tracehook_finish_clone(p, clone_flags, trace); 1257 tracehook_finish_clone(p, clone_flags, trace);
1253 1258
1254 if (thread_group_leader(p)) { 1259 if (thread_group_leader(p)) {
1255 if (clone_flags & CLONE_NEWPID) 1260 if (clone_flags & CLONE_NEWPID)
1256 p->nsproxy->pid_ns->child_reaper = p; 1261 p->nsproxy->pid_ns->child_reaper = p;
1257 1262
1258 p->signal->leader_pid = pid; 1263 p->signal->leader_pid = pid;
1259 tty_kref_put(p->signal->tty); 1264 tty_kref_put(p->signal->tty);
1260 p->signal->tty = tty_kref_get(current->signal->tty); 1265 p->signal->tty = tty_kref_get(current->signal->tty);
1261 set_task_pgrp(p, task_pgrp_nr(current)); 1266 set_task_pgrp(p, task_pgrp_nr(current));
1262 set_task_session(p, task_session_nr(current)); 1267 set_task_session(p, task_session_nr(current));
1263 attach_pid(p, PIDTYPE_PGID, task_pgrp(current)); 1268 attach_pid(p, PIDTYPE_PGID, task_pgrp(current));
1264 attach_pid(p, PIDTYPE_SID, task_session(current)); 1269 attach_pid(p, PIDTYPE_SID, task_session(current));
1265 list_add_tail_rcu(&p->tasks, &init_task.tasks); 1270 list_add_tail_rcu(&p->tasks, &init_task.tasks);
1266 __get_cpu_var(process_counts)++; 1271 __get_cpu_var(process_counts)++;
1267 } 1272 }
1268 attach_pid(p, PIDTYPE_PID, pid); 1273 attach_pid(p, PIDTYPE_PID, pid);
1269 nr_threads++; 1274 nr_threads++;
1270 } 1275 }
1271 1276
1272 total_forks++; 1277 total_forks++;
1273 spin_unlock(&current->sighand->siglock); 1278 spin_unlock(&current->sighand->siglock);
1274 write_unlock_irq(&tasklist_lock); 1279 write_unlock_irq(&tasklist_lock);
1275 proc_fork_connector(p); 1280 proc_fork_connector(p);
1276 cgroup_post_fork(p); 1281 cgroup_post_fork(p);
1277 return p; 1282 return p;
1278 1283
1279 bad_fork_free_graph: 1284 bad_fork_free_graph:
1280 ftrace_graph_exit_task(p); 1285 ftrace_graph_exit_task(p);
1281 bad_fork_free_pid: 1286 bad_fork_free_pid:
1282 if (pid != &init_struct_pid) 1287 if (pid != &init_struct_pid)
1283 free_pid(pid); 1288 free_pid(pid);
1284 bad_fork_cleanup_io: 1289 bad_fork_cleanup_io:
1285 put_io_context(p->io_context); 1290 put_io_context(p->io_context);
1286 bad_fork_cleanup_namespaces: 1291 bad_fork_cleanup_namespaces:
1287 exit_task_namespaces(p); 1292 exit_task_namespaces(p);
1288 bad_fork_cleanup_mm: 1293 bad_fork_cleanup_mm:
1289 if (p->mm) 1294 if (p->mm)
1290 mmput(p->mm); 1295 mmput(p->mm);
1291 bad_fork_cleanup_signal: 1296 bad_fork_cleanup_signal:
1292 cleanup_signal(p); 1297 cleanup_signal(p);
1293 bad_fork_cleanup_sighand: 1298 bad_fork_cleanup_sighand:
1294 __cleanup_sighand(p->sighand); 1299 __cleanup_sighand(p->sighand);
1295 bad_fork_cleanup_fs: 1300 bad_fork_cleanup_fs:
1296 exit_fs(p); /* blocking */ 1301 exit_fs(p); /* blocking */
1297 bad_fork_cleanup_files: 1302 bad_fork_cleanup_files:
1298 exit_files(p); /* blocking */ 1303 exit_files(p); /* blocking */
1299 bad_fork_cleanup_semundo: 1304 bad_fork_cleanup_semundo:
1300 exit_sem(p); 1305 exit_sem(p);
1301 bad_fork_cleanup_audit: 1306 bad_fork_cleanup_audit:
1302 audit_free(p); 1307 audit_free(p);
1303 bad_fork_cleanup_policy: 1308 bad_fork_cleanup_policy:
1304 #ifdef CONFIG_NUMA 1309 #ifdef CONFIG_NUMA
1305 mpol_put(p->mempolicy); 1310 mpol_put(p->mempolicy);
1306 bad_fork_cleanup_cgroup: 1311 bad_fork_cleanup_cgroup:
1307 #endif 1312 #endif
1308 cgroup_exit(p, cgroup_callbacks_done); 1313 cgroup_exit(p, cgroup_callbacks_done);
1309 delayacct_tsk_free(p); 1314 delayacct_tsk_free(p);
1310 if (p->binfmt) 1315 if (p->binfmt)
1311 module_put(p->binfmt->module); 1316 module_put(p->binfmt->module);
1312 bad_fork_cleanup_put_domain: 1317 bad_fork_cleanup_put_domain:
1313 module_put(task_thread_info(p)->exec_domain->module); 1318 module_put(task_thread_info(p)->exec_domain->module);
1314 bad_fork_cleanup_count: 1319 bad_fork_cleanup_count:
1315 atomic_dec(&p->cred->user->processes); 1320 atomic_dec(&p->cred->user->processes);
1316 put_cred(p->real_cred); 1321 put_cred(p->real_cred);
1317 put_cred(p->cred); 1322 put_cred(p->cred);
1318 bad_fork_free: 1323 bad_fork_free:
1319 free_task(p); 1324 free_task(p);
1320 fork_out: 1325 fork_out:
1321 return ERR_PTR(retval); 1326 return ERR_PTR(retval);
1322 } 1327 }
1323 1328
1324 noinline struct pt_regs * __cpuinit __attribute__((weak)) idle_regs(struct pt_regs *regs) 1329 noinline struct pt_regs * __cpuinit __attribute__((weak)) idle_regs(struct pt_regs *regs)
1325 { 1330 {
1326 memset(regs, 0, sizeof(struct pt_regs)); 1331 memset(regs, 0, sizeof(struct pt_regs));
1327 return regs; 1332 return regs;
1328 } 1333 }
1329 1334
1330 struct task_struct * __cpuinit fork_idle(int cpu) 1335 struct task_struct * __cpuinit fork_idle(int cpu)
1331 { 1336 {
1332 struct task_struct *task; 1337 struct task_struct *task;
1333 struct pt_regs regs; 1338 struct pt_regs regs;
1334 1339
1335 task = copy_process(CLONE_VM, 0, idle_regs(&regs), 0, NULL, 1340 task = copy_process(CLONE_VM, 0, idle_regs(&regs), 0, NULL,
1336 &init_struct_pid, 0); 1341 &init_struct_pid, 0);
1337 if (!IS_ERR(task)) 1342 if (!IS_ERR(task))
1338 init_idle(task, cpu); 1343 init_idle(task, cpu);
1339 1344
1340 return task; 1345 return task;
1341 } 1346 }
1342 1347
1343 /* 1348 /*
1344 * Ok, this is the main fork-routine. 1349 * Ok, this is the main fork-routine.
1345 * 1350 *
1346 * It copies the process, and if successful kick-starts 1351 * It copies the process, and if successful kick-starts
1347 * it and waits for it to finish using the VM if required. 1352 * it and waits for it to finish using the VM if required.
1348 */ 1353 */
1349 long do_fork(unsigned long clone_flags, 1354 long do_fork(unsigned long clone_flags,
1350 unsigned long stack_start, 1355 unsigned long stack_start,
1351 struct pt_regs *regs, 1356 struct pt_regs *regs,
1352 unsigned long stack_size, 1357 unsigned long stack_size,
1353 int __user *parent_tidptr, 1358 int __user *parent_tidptr,
1354 int __user *child_tidptr) 1359 int __user *child_tidptr)
1355 { 1360 {
1356 struct task_struct *p; 1361 struct task_struct *p;
1357 int trace = 0; 1362 int trace = 0;
1358 long nr; 1363 long nr;
1359 1364
1360 /* 1365 /*
1361 * Do some preliminary argument and permissions checking before we 1366 * Do some preliminary argument and permissions checking before we
1362 * actually start allocating stuff 1367 * actually start allocating stuff
1363 */ 1368 */
1364 if (clone_flags & CLONE_NEWUSER) { 1369 if (clone_flags & CLONE_NEWUSER) {
1365 if (clone_flags & CLONE_THREAD) 1370 if (clone_flags & CLONE_THREAD)
1366 return -EINVAL; 1371 return -EINVAL;
1367 /* hopefully this check will go away when userns support is 1372 /* hopefully this check will go away when userns support is
1368 * complete 1373 * complete
1369 */ 1374 */
1370 if (!capable(CAP_SYS_ADMIN) || !capable(CAP_SETUID) || 1375 if (!capable(CAP_SYS_ADMIN) || !capable(CAP_SETUID) ||
1371 !capable(CAP_SETGID)) 1376 !capable(CAP_SETGID))
1372 return -EPERM; 1377 return -EPERM;
1373 } 1378 }
1374 1379
1375 /* 1380 /*
1376 * We hope to recycle these flags after 2.6.26 1381 * We hope to recycle these flags after 2.6.26
1377 */ 1382 */
1378 if (unlikely(clone_flags & CLONE_STOPPED)) { 1383 if (unlikely(clone_flags & CLONE_STOPPED)) {
1379 static int __read_mostly count = 100; 1384 static int __read_mostly count = 100;
1380 1385
1381 if (count > 0 && printk_ratelimit()) { 1386 if (count > 0 && printk_ratelimit()) {
1382 char comm[TASK_COMM_LEN]; 1387 char comm[TASK_COMM_LEN];
1383 1388
1384 count--; 1389 count--;
1385 printk(KERN_INFO "fork(): process `%s' used deprecated " 1390 printk(KERN_INFO "fork(): process `%s' used deprecated "
1386 "clone flags 0x%lx\n", 1391 "clone flags 0x%lx\n",
1387 get_task_comm(comm, current), 1392 get_task_comm(comm, current),
1388 clone_flags & CLONE_STOPPED); 1393 clone_flags & CLONE_STOPPED);
1389 } 1394 }
1390 } 1395 }
1391 1396
1392 /* 1397 /*
1393 * When called from kernel_thread, don't do user tracing stuff. 1398 * When called from kernel_thread, don't do user tracing stuff.
1394 */ 1399 */
1395 if (likely(user_mode(regs))) 1400 if (likely(user_mode(regs)))
1396 trace = tracehook_prepare_clone(clone_flags); 1401 trace = tracehook_prepare_clone(clone_flags);
1397 1402
1398 p = copy_process(clone_flags, stack_start, regs, stack_size, 1403 p = copy_process(clone_flags, stack_start, regs, stack_size,
1399 child_tidptr, NULL, trace); 1404 child_tidptr, NULL, trace);
1400 /* 1405 /*
1401 * Do this prior waking up the new thread - the thread pointer 1406 * Do this prior waking up the new thread - the thread pointer
1402 * might get invalid after that point, if the thread exits quickly. 1407 * might get invalid after that point, if the thread exits quickly.
1403 */ 1408 */
1404 if (!IS_ERR(p)) { 1409 if (!IS_ERR(p)) {
1405 struct completion vfork; 1410 struct completion vfork;
1406 1411
1407 trace_sched_process_fork(current, p); 1412 trace_sched_process_fork(current, p);
1408 1413
1409 nr = task_pid_vnr(p); 1414 nr = task_pid_vnr(p);
1410 1415
1411 if (clone_flags & CLONE_PARENT_SETTID) 1416 if (clone_flags & CLONE_PARENT_SETTID)
1412 put_user(nr, parent_tidptr); 1417 put_user(nr, parent_tidptr);
1413 1418
1414 if (clone_flags & CLONE_VFORK) { 1419 if (clone_flags & CLONE_VFORK) {
1415 p->vfork_done = &vfork; 1420 p->vfork_done = &vfork;
1416 init_completion(&vfork); 1421 init_completion(&vfork);
1417 } 1422 }
1418 1423
1419 audit_finish_fork(p); 1424 audit_finish_fork(p);
1420 tracehook_report_clone(trace, regs, clone_flags, nr, p); 1425 tracehook_report_clone(trace, regs, clone_flags, nr, p);
1421 1426
1422 /* 1427 /*
1423 * We set PF_STARTING at creation in case tracing wants to 1428 * We set PF_STARTING at creation in case tracing wants to
1424 * use this to distinguish a fully live task from one that 1429 * use this to distinguish a fully live task from one that
1425 * hasn't gotten to tracehook_report_clone() yet. Now we 1430 * hasn't gotten to tracehook_report_clone() yet. Now we
1426 * clear it and set the child going. 1431 * clear it and set the child going.
1427 */ 1432 */
1428 p->flags &= ~PF_STARTING; 1433 p->flags &= ~PF_STARTING;
1429 1434
1430 if (unlikely(clone_flags & CLONE_STOPPED)) { 1435 if (unlikely(clone_flags & CLONE_STOPPED)) {
1431 /* 1436 /*
1432 * We'll start up with an immediate SIGSTOP. 1437 * We'll start up with an immediate SIGSTOP.
1433 */ 1438 */
1434 sigaddset(&p->pending.signal, SIGSTOP); 1439 sigaddset(&p->pending.signal, SIGSTOP);
1435 set_tsk_thread_flag(p, TIF_SIGPENDING); 1440 set_tsk_thread_flag(p, TIF_SIGPENDING);
1436 __set_task_state(p, TASK_STOPPED); 1441 __set_task_state(p, TASK_STOPPED);
1437 } else { 1442 } else {
1438 wake_up_new_task(p, clone_flags); 1443 wake_up_new_task(p, clone_flags);
1439 } 1444 }
1440 1445
1441 tracehook_report_clone_complete(trace, regs, 1446 tracehook_report_clone_complete(trace, regs,
1442 clone_flags, nr, p); 1447 clone_flags, nr, p);
1443 1448
1444 if (clone_flags & CLONE_VFORK) { 1449 if (clone_flags & CLONE_VFORK) {
1445 freezer_do_not_count(); 1450 freezer_do_not_count();
1446 wait_for_completion(&vfork); 1451 wait_for_completion(&vfork);
1447 freezer_count(); 1452 freezer_count();
1448 tracehook_report_vfork_done(p, nr); 1453 tracehook_report_vfork_done(p, nr);
1449 } 1454 }
1450 } else { 1455 } else {
1451 nr = PTR_ERR(p); 1456 nr = PTR_ERR(p);
1452 } 1457 }
1453 return nr; 1458 return nr;
1454 } 1459 }
1455 1460
1456 #ifndef ARCH_MIN_MMSTRUCT_ALIGN 1461 #ifndef ARCH_MIN_MMSTRUCT_ALIGN
1457 #define ARCH_MIN_MMSTRUCT_ALIGN 0 1462 #define ARCH_MIN_MMSTRUCT_ALIGN 0
1458 #endif 1463 #endif
1459 1464
1460 static void sighand_ctor(void *data) 1465 static void sighand_ctor(void *data)
1461 { 1466 {
1462 struct sighand_struct *sighand = data; 1467 struct sighand_struct *sighand = data;
1463 1468
1464 spin_lock_init(&sighand->siglock); 1469 spin_lock_init(&sighand->siglock);
1465 init_waitqueue_head(&sighand->signalfd_wqh); 1470 init_waitqueue_head(&sighand->signalfd_wqh);
1466 } 1471 }
1467 1472
1468 void __init proc_caches_init(void) 1473 void __init proc_caches_init(void)
1469 { 1474 {
1470 sighand_cachep = kmem_cache_create("sighand_cache", 1475 sighand_cachep = kmem_cache_create("sighand_cache",
1471 sizeof(struct sighand_struct), 0, 1476 sizeof(struct sighand_struct), 0,
1472 SLAB_HWCACHE_ALIGN|SLAB_PANIC|SLAB_DESTROY_BY_RCU, 1477 SLAB_HWCACHE_ALIGN|SLAB_PANIC|SLAB_DESTROY_BY_RCU,
1473 sighand_ctor); 1478 sighand_ctor);
1474 signal_cachep = kmem_cache_create("signal_cache", 1479 signal_cachep = kmem_cache_create("signal_cache",
1475 sizeof(struct signal_struct), 0, 1480 sizeof(struct signal_struct), 0,
1476 SLAB_HWCACHE_ALIGN|SLAB_PANIC, NULL); 1481 SLAB_HWCACHE_ALIGN|SLAB_PANIC, NULL);
1477 files_cachep = kmem_cache_create("files_cache", 1482 files_cachep = kmem_cache_create("files_cache",
1478 sizeof(struct files_struct), 0, 1483 sizeof(struct files_struct), 0,
1479 SLAB_HWCACHE_ALIGN|SLAB_PANIC, NULL); 1484 SLAB_HWCACHE_ALIGN|SLAB_PANIC, NULL);
1480 fs_cachep = kmem_cache_create("fs_cache", 1485 fs_cachep = kmem_cache_create("fs_cache",
1481 sizeof(struct fs_struct), 0, 1486 sizeof(struct fs_struct), 0,
1482 SLAB_HWCACHE_ALIGN|SLAB_PANIC, NULL); 1487 SLAB_HWCACHE_ALIGN|SLAB_PANIC, NULL);
1483 mm_cachep = kmem_cache_create("mm_struct", 1488 mm_cachep = kmem_cache_create("mm_struct",
1484 sizeof(struct mm_struct), ARCH_MIN_MMSTRUCT_ALIGN, 1489 sizeof(struct mm_struct), ARCH_MIN_MMSTRUCT_ALIGN,
1485 SLAB_HWCACHE_ALIGN|SLAB_PANIC, NULL); 1490 SLAB_HWCACHE_ALIGN|SLAB_PANIC, NULL);
1486 mmap_init(); 1491 mmap_init();
1487 } 1492 }
1488 1493
1489 /* 1494 /*
1490 * Check constraints on flags passed to the unshare system call and 1495 * Check constraints on flags passed to the unshare system call and
1491 * force unsharing of additional process context as appropriate. 1496 * force unsharing of additional process context as appropriate.
1492 */ 1497 */
1493 static void check_unshare_flags(unsigned long *flags_ptr) 1498 static void check_unshare_flags(unsigned long *flags_ptr)
1494 { 1499 {
1495 /* 1500 /*
1496 * If unsharing a thread from a thread group, must also 1501 * If unsharing a thread from a thread group, must also
1497 * unshare vm. 1502 * unshare vm.
1498 */ 1503 */
1499 if (*flags_ptr & CLONE_THREAD) 1504 if (*flags_ptr & CLONE_THREAD)
1500 *flags_ptr |= CLONE_VM; 1505 *flags_ptr |= CLONE_VM;
1501 1506
1502 /* 1507 /*
1503 * If unsharing vm, must also unshare signal handlers. 1508 * If unsharing vm, must also unshare signal handlers.
1504 */ 1509 */
1505 if (*flags_ptr & CLONE_VM) 1510 if (*flags_ptr & CLONE_VM)
1506 *flags_ptr |= CLONE_SIGHAND; 1511 *flags_ptr |= CLONE_SIGHAND;
1507 1512
1508 /* 1513 /*
1509 * If unsharing signal handlers and the task was created 1514 * If unsharing signal handlers and the task was created
1510 * using CLONE_THREAD, then must unshare the thread 1515 * using CLONE_THREAD, then must unshare the thread
1511 */ 1516 */
1512 if ((*flags_ptr & CLONE_SIGHAND) && 1517 if ((*flags_ptr & CLONE_SIGHAND) &&
1513 (atomic_read(&current->signal->count) > 1)) 1518 (atomic_read(&current->signal->count) > 1))
1514 *flags_ptr |= CLONE_THREAD; 1519 *flags_ptr |= CLONE_THREAD;
1515 1520
1516 /* 1521 /*
1517 * If unsharing namespace, must also unshare filesystem information. 1522 * If unsharing namespace, must also unshare filesystem information.
1518 */ 1523 */
1519 if (*flags_ptr & CLONE_NEWNS) 1524 if (*flags_ptr & CLONE_NEWNS)
1520 *flags_ptr |= CLONE_FS; 1525 *flags_ptr |= CLONE_FS;
1521 } 1526 }
1522 1527
1523 /* 1528 /*
1524 * Unsharing of tasks created with CLONE_THREAD is not supported yet 1529 * Unsharing of tasks created with CLONE_THREAD is not supported yet
1525 */ 1530 */
1526 static int unshare_thread(unsigned long unshare_flags) 1531 static int unshare_thread(unsigned long unshare_flags)
1527 { 1532 {
1528 if (unshare_flags & CLONE_THREAD) 1533 if (unshare_flags & CLONE_THREAD)
1529 return -EINVAL; 1534 return -EINVAL;
1530 1535
1531 return 0; 1536 return 0;
1532 } 1537 }
1533 1538
1534 /* 1539 /*
1535 * Unshare the filesystem structure if it is being shared 1540 * Unshare the filesystem structure if it is being shared
1536 */ 1541 */
1537 static int unshare_fs(unsigned long unshare_flags, struct fs_struct **new_fsp) 1542 static int unshare_fs(unsigned long unshare_flags, struct fs_struct **new_fsp)
1538 { 1543 {
1539 struct fs_struct *fs = current->fs; 1544 struct fs_struct *fs = current->fs;
1540 1545
1541 if ((unshare_flags & CLONE_FS) && 1546 if ((unshare_flags & CLONE_FS) &&
1542 (fs && atomic_read(&fs->count) > 1)) { 1547 (fs && atomic_read(&fs->count) > 1)) {
1543 *new_fsp = __copy_fs_struct(current->fs); 1548 *new_fsp = __copy_fs_struct(current->fs);
1544 if (!*new_fsp) 1549 if (!*new_fsp)
1545 return -ENOMEM; 1550 return -ENOMEM;
1546 } 1551 }
1547 1552
1548 return 0; 1553 return 0;
1549 } 1554 }
1550 1555
1551 /* 1556 /*
1552 * Unsharing of sighand is not supported yet 1557 * Unsharing of sighand is not supported yet
1553 */ 1558 */
1554 static int unshare_sighand(unsigned long unshare_flags, struct sighand_struct **new_sighp) 1559 static int unshare_sighand(unsigned long unshare_flags, struct sighand_struct **new_sighp)
1555 { 1560 {
1556 struct sighand_struct *sigh = current->sighand; 1561 struct sighand_struct *sigh = current->sighand;
1557 1562
1558 if ((unshare_flags & CLONE_SIGHAND) && atomic_read(&sigh->count) > 1) 1563 if ((unshare_flags & CLONE_SIGHAND) && atomic_read(&sigh->count) > 1)
1559 return -EINVAL; 1564 return -EINVAL;
1560 else 1565 else
1561 return 0; 1566 return 0;
1562 } 1567 }
1563 1568
1564 /* 1569 /*
1565 * Unshare vm if it is being shared 1570 * Unshare vm if it is being shared
1566 */ 1571 */
1567 static int unshare_vm(unsigned long unshare_flags, struct mm_struct **new_mmp) 1572 static int unshare_vm(unsigned long unshare_flags, struct mm_struct **new_mmp)
1568 { 1573 {
1569 struct mm_struct *mm = current->mm; 1574 struct mm_struct *mm = current->mm;
1570 1575
1571 if ((unshare_flags & CLONE_VM) && 1576 if ((unshare_flags & CLONE_VM) &&
1572 (mm && atomic_read(&mm->mm_users) > 1)) { 1577 (mm && atomic_read(&mm->mm_users) > 1)) {
1573 return -EINVAL; 1578 return -EINVAL;
1574 } 1579 }
1575 1580
1576 return 0; 1581 return 0;
1577 } 1582 }
1578 1583
1579 /* 1584 /*
1580 * Unshare file descriptor table if it is being shared 1585 * Unshare file descriptor table if it is being shared
1581 */ 1586 */
1582 static int unshare_fd(unsigned long unshare_flags, struct files_struct **new_fdp) 1587 static int unshare_fd(unsigned long unshare_flags, struct files_struct **new_fdp)
1583 { 1588 {
1584 struct files_struct *fd = current->files; 1589 struct files_struct *fd = current->files;
1585 int error = 0; 1590 int error = 0;
1586 1591
1587 if ((unshare_flags & CLONE_FILES) && 1592 if ((unshare_flags & CLONE_FILES) &&
1588 (fd && atomic_read(&fd->count) > 1)) { 1593 (fd && atomic_read(&fd->count) > 1)) {
1589 *new_fdp = dup_fd(fd, &error); 1594 *new_fdp = dup_fd(fd, &error);
1590 if (!*new_fdp) 1595 if (!*new_fdp)
1591 return error; 1596 return error;
1592 } 1597 }
1593 1598
1594 return 0; 1599 return 0;
1595 } 1600 }
1596 1601
1597 /* 1602 /*
1598 * unshare allows a process to 'unshare' part of the process 1603 * unshare allows a process to 'unshare' part of the process
1599 * context which was originally shared using clone. copy_* 1604 * context which was originally shared using clone. copy_*
1600 * functions used by do_fork() cannot be used here directly 1605 * functions used by do_fork() cannot be used here directly
1601 * because they modify an inactive task_struct that is being 1606 * because they modify an inactive task_struct that is being
1602 * constructed. Here we are modifying the current, active, 1607 * constructed. Here we are modifying the current, active,
1603 * task_struct. 1608 * task_struct.
1604 */ 1609 */
1605 SYSCALL_DEFINE1(unshare, unsigned long, unshare_flags) 1610 SYSCALL_DEFINE1(unshare, unsigned long, unshare_flags)
1606 { 1611 {
1607 int err = 0; 1612 int err = 0;
1608 struct fs_struct *fs, *new_fs = NULL; 1613 struct fs_struct *fs, *new_fs = NULL;
1609 struct sighand_struct *new_sigh = NULL; 1614 struct sighand_struct *new_sigh = NULL;
1610 struct mm_struct *mm, *new_mm = NULL, *active_mm = NULL; 1615 struct mm_struct *mm, *new_mm = NULL, *active_mm = NULL;
1611 struct files_struct *fd, *new_fd = NULL; 1616 struct files_struct *fd, *new_fd = NULL;
1612 struct nsproxy *new_nsproxy = NULL; 1617 struct nsproxy *new_nsproxy = NULL;
1613 int do_sysvsem = 0; 1618 int do_sysvsem = 0;
1614 1619
1615 check_unshare_flags(&unshare_flags); 1620 check_unshare_flags(&unshare_flags);
1616 1621
1617 /* Return -EINVAL for all unsupported flags */ 1622 /* Return -EINVAL for all unsupported flags */
1618 err = -EINVAL; 1623 err = -EINVAL;
1619 if (unshare_flags & ~(CLONE_THREAD|CLONE_FS|CLONE_NEWNS|CLONE_SIGHAND| 1624 if (unshare_flags & ~(CLONE_THREAD|CLONE_FS|CLONE_NEWNS|CLONE_SIGHAND|
1620 CLONE_VM|CLONE_FILES|CLONE_SYSVSEM| 1625 CLONE_VM|CLONE_FILES|CLONE_SYSVSEM|
1621 CLONE_NEWUTS|CLONE_NEWIPC|CLONE_NEWNET)) 1626 CLONE_NEWUTS|CLONE_NEWIPC|CLONE_NEWNET))
1622 goto bad_unshare_out; 1627 goto bad_unshare_out;
1623 1628
1624 /* 1629 /*
1625 * CLONE_NEWIPC must also detach from the undolist: after switching 1630 * CLONE_NEWIPC must also detach from the undolist: after switching
1626 * to a new ipc namespace, the semaphore arrays from the old 1631 * to a new ipc namespace, the semaphore arrays from the old
1627 * namespace are unreachable. 1632 * namespace are unreachable.
1628 */ 1633 */
1629 if (unshare_flags & (CLONE_NEWIPC|CLONE_SYSVSEM)) 1634 if (unshare_flags & (CLONE_NEWIPC|CLONE_SYSVSEM))
1630 do_sysvsem = 1; 1635 do_sysvsem = 1;
1631 if ((err = unshare_thread(unshare_flags))) 1636 if ((err = unshare_thread(unshare_flags)))
1632 goto bad_unshare_out; 1637 goto bad_unshare_out;
1633 if ((err = unshare_fs(unshare_flags, &new_fs))) 1638 if ((err = unshare_fs(unshare_flags, &new_fs)))
1634 goto bad_unshare_cleanup_thread; 1639 goto bad_unshare_cleanup_thread;
1635 if ((err = unshare_sighand(unshare_flags, &new_sigh))) 1640 if ((err = unshare_sighand(unshare_flags, &new_sigh)))
1636 goto bad_unshare_cleanup_fs; 1641 goto bad_unshare_cleanup_fs;
1637 if ((err = unshare_vm(unshare_flags, &new_mm))) 1642 if ((err = unshare_vm(unshare_flags, &new_mm)))
1638 goto bad_unshare_cleanup_sigh; 1643 goto bad_unshare_cleanup_sigh;
1639 if ((err = unshare_fd(unshare_flags, &new_fd))) 1644 if ((err = unshare_fd(unshare_flags, &new_fd)))
1640 goto bad_unshare_cleanup_vm; 1645 goto bad_unshare_cleanup_vm;
1641 if ((err = unshare_nsproxy_namespaces(unshare_flags, &new_nsproxy, 1646 if ((err = unshare_nsproxy_namespaces(unshare_flags, &new_nsproxy,
1642 new_fs))) 1647 new_fs)))
1643 goto bad_unshare_cleanup_fd; 1648 goto bad_unshare_cleanup_fd;
1644 1649
1645 if (new_fs || new_mm || new_fd || do_sysvsem || new_nsproxy) { 1650 if (new_fs || new_mm || new_fd || do_sysvsem || new_nsproxy) {
1646 if (do_sysvsem) { 1651 if (do_sysvsem) {
1647 /* 1652 /*
1648 * CLONE_SYSVSEM is equivalent to sys_exit(). 1653 * CLONE_SYSVSEM is equivalent to sys_exit().
1649 */ 1654 */
1650 exit_sem(current); 1655 exit_sem(current);
1651 } 1656 }
1652 1657
1653 if (new_nsproxy) { 1658 if (new_nsproxy) {
1654 switch_task_namespaces(current, new_nsproxy); 1659 switch_task_namespaces(current, new_nsproxy);
1655 new_nsproxy = NULL; 1660 new_nsproxy = NULL;
1656 } 1661 }
1657 1662
1658 task_lock(current); 1663 task_lock(current);
1659 1664
1660 if (new_fs) { 1665 if (new_fs) {
1661 fs = current->fs; 1666 fs = current->fs;
1662 current->fs = new_fs; 1667 current->fs = new_fs;
1663 new_fs = fs; 1668 new_fs = fs;
1664 } 1669 }
1665 1670
1666 if (new_mm) { 1671 if (new_mm) {
1667 mm = current->mm; 1672 mm = current->mm;
1668 active_mm = current->active_mm; 1673 active_mm = current->active_mm;
1669 current->mm = new_mm; 1674 current->mm = new_mm;
1670 current->active_mm = new_mm; 1675 current->active_mm = new_mm;
1671 activate_mm(active_mm, new_mm); 1676 activate_mm(active_mm, new_mm);
1672 new_mm = mm; 1677 new_mm = mm;
1673 } 1678 }
1674 1679
1675 if (new_fd) { 1680 if (new_fd) {
1676 fd = current->files; 1681 fd = current->files;
1677 current->files = new_fd; 1682 current->files = new_fd;
1678 new_fd = fd; 1683 new_fd = fd;
1679 } 1684 }
1680 1685
1681 task_unlock(current); 1686 task_unlock(current);
1682 } 1687 }
1683 1688
1684 if (new_nsproxy) 1689 if (new_nsproxy)
1685 put_nsproxy(new_nsproxy); 1690 put_nsproxy(new_nsproxy);
1686 1691
1687 bad_unshare_cleanup_fd: 1692 bad_unshare_cleanup_fd:
1688 if (new_fd) 1693 if (new_fd)
1689 put_files_struct(new_fd); 1694 put_files_struct(new_fd);
1690 1695
1691 bad_unshare_cleanup_vm: 1696 bad_unshare_cleanup_vm:
1692 if (new_mm) 1697 if (new_mm)
1693 mmput(new_mm); 1698 mmput(new_mm);
1694 1699
1695 bad_unshare_cleanup_sigh: 1700 bad_unshare_cleanup_sigh:
1696 if (new_sigh) 1701 if (new_sigh)
1697 if (atomic_dec_and_test(&new_sigh->count)) 1702 if (atomic_dec_and_test(&new_sigh->count))
1698 kmem_cache_free(sighand_cachep, new_sigh); 1703 kmem_cache_free(sighand_cachep, new_sigh);
1699 1704
1700 bad_unshare_cleanup_fs: 1705 bad_unshare_cleanup_fs:
1701 if (new_fs) 1706 if (new_fs)
1702 put_fs_struct(new_fs); 1707 put_fs_struct(new_fs);
1703 1708
1704 bad_unshare_cleanup_thread: 1709 bad_unshare_cleanup_thread:
1705 bad_unshare_out: 1710 bad_unshare_out:
1706 return err; 1711 return err;
1707 } 1712 }
1708 1713
1709 /* 1714 /*
1710 * Helper to unshare the files of the current task. 1715 * Helper to unshare the files of the current task.
1711 * We don't want to expose copy_files internals to 1716 * We don't want to expose copy_files internals to
1712 * the exec layer of the kernel. 1717 * the exec layer of the kernel.
1713 */ 1718 */
1714 1719
1715 int unshare_files(struct files_struct **displaced) 1720 int unshare_files(struct files_struct **displaced)
1716 { 1721 {
1717 struct task_struct *task = current; 1722 struct task_struct *task = current;
1718 struct files_struct *copy = NULL; 1723 struct files_struct *copy = NULL;
1719 int error; 1724 int error;
1720 1725
1721 error = unshare_fd(CLONE_FILES, &copy); 1726 error = unshare_fd(CLONE_FILES, &copy);
1722 if (error || !copy) { 1727 if (error || !copy) {
1723 *displaced = NULL; 1728 *displaced = NULL;
1724 return error; 1729 return error;
1725 } 1730 }
1726 *displaced = task->files; 1731 *displaced = task->files;
1727 task_lock(task); 1732 task_lock(task);
1728 task->files = copy; 1733 task->files = copy;
1729 task_unlock(task); 1734 task_unlock(task);