diff options
Diffstat (limited to 'kernel')
-rw-r--r-- | kernel/audit.c | 160 | ||||
-rw-r--r-- | kernel/audit.h | 10 | ||||
-rw-r--r-- | kernel/auditfilter.c | 289 | ||||
-rw-r--r-- | kernel/auditsc.c | 269 | ||||
-rw-r--r-- | kernel/cpuset.c | 25 | ||||
-rw-r--r-- | kernel/exit.c | 5 | ||||
-rw-r--r-- | kernel/extable.c | 2 | ||||
-rw-r--r-- | kernel/fork.c | 11 | ||||
-rw-r--r-- | kernel/hrtimer.c | 10 | ||||
-rw-r--r-- | kernel/irq/manage.c | 6 | ||||
-rw-r--r-- | kernel/kprobes.c | 3 | ||||
-rw-r--r-- | kernel/module.c | 12 | ||||
-rw-r--r-- | kernel/power/main.c | 2 | ||||
-rw-r--r-- | kernel/power/pm.c | 20 | ||||
-rw-r--r-- | kernel/power/snapshot.c | 9 | ||||
-rw-r--r-- | kernel/profile.c | 2 | ||||
-rw-r--r-- | kernel/ptrace.c | 64 | ||||
-rw-r--r-- | kernel/rcupdate.c | 23 | ||||
-rw-r--r-- | kernel/sched.c | 64 | ||||
-rw-r--r-- | kernel/signal.c | 4 | ||||
-rw-r--r-- | kernel/softirq.c | 4 | ||||
-rw-r--r-- | kernel/softlockup.c | 4 | ||||
-rw-r--r-- | kernel/timer.c | 20 | ||||
-rw-r--r-- | kernel/uid16.c | 59 | ||||
-rw-r--r-- | kernel/workqueue.c | 2 |
25 files changed, 736 insertions, 343 deletions
diff --git a/kernel/audit.c b/kernel/audit.c index c8ccbd09048f..df57b493e1cb 100644 --- a/kernel/audit.c +++ b/kernel/audit.c | |||
@@ -55,6 +55,9 @@ | |||
55 | #include <net/netlink.h> | 55 | #include <net/netlink.h> |
56 | #include <linux/skbuff.h> | 56 | #include <linux/skbuff.h> |
57 | #include <linux/netlink.h> | 57 | #include <linux/netlink.h> |
58 | #include <linux/selinux.h> | ||
59 | |||
60 | #include "audit.h" | ||
58 | 61 | ||
59 | /* No auditing will take place until audit_initialized != 0. | 62 | /* No auditing will take place until audit_initialized != 0. |
60 | * (Initialization happens after skb_init is called.) */ | 63 | * (Initialization happens after skb_init is called.) */ |
@@ -227,49 +230,103 @@ void audit_log_lost(const char *message) | |||
227 | } | 230 | } |
228 | } | 231 | } |
229 | 232 | ||
230 | static int audit_set_rate_limit(int limit, uid_t loginuid) | 233 | static int audit_set_rate_limit(int limit, uid_t loginuid, u32 sid) |
231 | { | 234 | { |
232 | int old = audit_rate_limit; | 235 | int old = audit_rate_limit; |
233 | audit_rate_limit = limit; | 236 | |
234 | audit_log(NULL, GFP_KERNEL, AUDIT_CONFIG_CHANGE, | 237 | if (sid) { |
238 | char *ctx = NULL; | ||
239 | u32 len; | ||
240 | int rc; | ||
241 | if ((rc = selinux_ctxid_to_string(sid, &ctx, &len))) | ||
242 | return rc; | ||
243 | else | ||
244 | audit_log(NULL, GFP_KERNEL, AUDIT_CONFIG_CHANGE, | ||
245 | "audit_rate_limit=%d old=%d by auid=%u subj=%s", | ||
246 | limit, old, loginuid, ctx); | ||
247 | kfree(ctx); | ||
248 | } else | ||
249 | audit_log(NULL, GFP_KERNEL, AUDIT_CONFIG_CHANGE, | ||
235 | "audit_rate_limit=%d old=%d by auid=%u", | 250 | "audit_rate_limit=%d old=%d by auid=%u", |
236 | audit_rate_limit, old, loginuid); | 251 | limit, old, loginuid); |
252 | audit_rate_limit = limit; | ||
237 | return old; | 253 | return old; |
238 | } | 254 | } |
239 | 255 | ||
240 | static int audit_set_backlog_limit(int limit, uid_t loginuid) | 256 | static int audit_set_backlog_limit(int limit, uid_t loginuid, u32 sid) |
241 | { | 257 | { |
242 | int old = audit_backlog_limit; | 258 | int old = audit_backlog_limit; |
243 | audit_backlog_limit = limit; | 259 | |
244 | audit_log(NULL, GFP_KERNEL, AUDIT_CONFIG_CHANGE, | 260 | if (sid) { |
261 | char *ctx = NULL; | ||
262 | u32 len; | ||
263 | int rc; | ||
264 | if ((rc = selinux_ctxid_to_string(sid, &ctx, &len))) | ||
265 | return rc; | ||
266 | else | ||
267 | audit_log(NULL, GFP_KERNEL, AUDIT_CONFIG_CHANGE, | ||
268 | "audit_backlog_limit=%d old=%d by auid=%u subj=%s", | ||
269 | limit, old, loginuid, ctx); | ||
270 | kfree(ctx); | ||
271 | } else | ||
272 | audit_log(NULL, GFP_KERNEL, AUDIT_CONFIG_CHANGE, | ||
245 | "audit_backlog_limit=%d old=%d by auid=%u", | 273 | "audit_backlog_limit=%d old=%d by auid=%u", |
246 | audit_backlog_limit, old, loginuid); | 274 | limit, old, loginuid); |
275 | audit_backlog_limit = limit; | ||
247 | return old; | 276 | return old; |
248 | } | 277 | } |
249 | 278 | ||
250 | static int audit_set_enabled(int state, uid_t loginuid) | 279 | static int audit_set_enabled(int state, uid_t loginuid, u32 sid) |
251 | { | 280 | { |
252 | int old = audit_enabled; | 281 | int old = audit_enabled; |
282 | |||
253 | if (state != 0 && state != 1) | 283 | if (state != 0 && state != 1) |
254 | return -EINVAL; | 284 | return -EINVAL; |
255 | audit_enabled = state; | 285 | |
256 | audit_log(NULL, GFP_KERNEL, AUDIT_CONFIG_CHANGE, | 286 | if (sid) { |
287 | char *ctx = NULL; | ||
288 | u32 len; | ||
289 | int rc; | ||
290 | if ((rc = selinux_ctxid_to_string(sid, &ctx, &len))) | ||
291 | return rc; | ||
292 | else | ||
293 | audit_log(NULL, GFP_KERNEL, AUDIT_CONFIG_CHANGE, | ||
294 | "audit_enabled=%d old=%d by auid=%u subj=%s", | ||
295 | state, old, loginuid, ctx); | ||
296 | kfree(ctx); | ||
297 | } else | ||
298 | audit_log(NULL, GFP_KERNEL, AUDIT_CONFIG_CHANGE, | ||
257 | "audit_enabled=%d old=%d by auid=%u", | 299 | "audit_enabled=%d old=%d by auid=%u", |
258 | audit_enabled, old, loginuid); | 300 | state, old, loginuid); |
301 | audit_enabled = state; | ||
259 | return old; | 302 | return old; |
260 | } | 303 | } |
261 | 304 | ||
262 | static int audit_set_failure(int state, uid_t loginuid) | 305 | static int audit_set_failure(int state, uid_t loginuid, u32 sid) |
263 | { | 306 | { |
264 | int old = audit_failure; | 307 | int old = audit_failure; |
308 | |||
265 | if (state != AUDIT_FAIL_SILENT | 309 | if (state != AUDIT_FAIL_SILENT |
266 | && state != AUDIT_FAIL_PRINTK | 310 | && state != AUDIT_FAIL_PRINTK |
267 | && state != AUDIT_FAIL_PANIC) | 311 | && state != AUDIT_FAIL_PANIC) |
268 | return -EINVAL; | 312 | return -EINVAL; |
269 | audit_failure = state; | 313 | |
270 | audit_log(NULL, GFP_KERNEL, AUDIT_CONFIG_CHANGE, | 314 | if (sid) { |
315 | char *ctx = NULL; | ||
316 | u32 len; | ||
317 | int rc; | ||
318 | if ((rc = selinux_ctxid_to_string(sid, &ctx, &len))) | ||
319 | return rc; | ||
320 | else | ||
321 | audit_log(NULL, GFP_KERNEL, AUDIT_CONFIG_CHANGE, | ||
322 | "audit_failure=%d old=%d by auid=%u subj=%s", | ||
323 | state, old, loginuid, ctx); | ||
324 | kfree(ctx); | ||
325 | } else | ||
326 | audit_log(NULL, GFP_KERNEL, AUDIT_CONFIG_CHANGE, | ||
271 | "audit_failure=%d old=%d by auid=%u", | 327 | "audit_failure=%d old=%d by auid=%u", |
272 | audit_failure, old, loginuid); | 328 | state, old, loginuid); |
329 | audit_failure = state; | ||
273 | return old; | 330 | return old; |
274 | } | 331 | } |
275 | 332 | ||
@@ -387,7 +444,7 @@ static int audit_netlink_ok(kernel_cap_t eff_cap, u16 msg_type) | |||
387 | 444 | ||
388 | static int audit_receive_msg(struct sk_buff *skb, struct nlmsghdr *nlh) | 445 | static int audit_receive_msg(struct sk_buff *skb, struct nlmsghdr *nlh) |
389 | { | 446 | { |
390 | u32 uid, pid, seq; | 447 | u32 uid, pid, seq, sid; |
391 | void *data; | 448 | void *data; |
392 | struct audit_status *status_get, status_set; | 449 | struct audit_status *status_get, status_set; |
393 | int err; | 450 | int err; |
@@ -413,6 +470,7 @@ static int audit_receive_msg(struct sk_buff *skb, struct nlmsghdr *nlh) | |||
413 | pid = NETLINK_CREDS(skb)->pid; | 470 | pid = NETLINK_CREDS(skb)->pid; |
414 | uid = NETLINK_CREDS(skb)->uid; | 471 | uid = NETLINK_CREDS(skb)->uid; |
415 | loginuid = NETLINK_CB(skb).loginuid; | 472 | loginuid = NETLINK_CB(skb).loginuid; |
473 | sid = NETLINK_CB(skb).sid; | ||
416 | seq = nlh->nlmsg_seq; | 474 | seq = nlh->nlmsg_seq; |
417 | data = NLMSG_DATA(nlh); | 475 | data = NLMSG_DATA(nlh); |
418 | 476 | ||
@@ -433,25 +491,43 @@ static int audit_receive_msg(struct sk_buff *skb, struct nlmsghdr *nlh) | |||
433 | return -EINVAL; | 491 | return -EINVAL; |
434 | status_get = (struct audit_status *)data; | 492 | status_get = (struct audit_status *)data; |
435 | if (status_get->mask & AUDIT_STATUS_ENABLED) { | 493 | if (status_get->mask & AUDIT_STATUS_ENABLED) { |
436 | err = audit_set_enabled(status_get->enabled, loginuid); | 494 | err = audit_set_enabled(status_get->enabled, |
495 | loginuid, sid); | ||
437 | if (err < 0) return err; | 496 | if (err < 0) return err; |
438 | } | 497 | } |
439 | if (status_get->mask & AUDIT_STATUS_FAILURE) { | 498 | if (status_get->mask & AUDIT_STATUS_FAILURE) { |
440 | err = audit_set_failure(status_get->failure, loginuid); | 499 | err = audit_set_failure(status_get->failure, |
500 | loginuid, sid); | ||
441 | if (err < 0) return err; | 501 | if (err < 0) return err; |
442 | } | 502 | } |
443 | if (status_get->mask & AUDIT_STATUS_PID) { | 503 | if (status_get->mask & AUDIT_STATUS_PID) { |
444 | int old = audit_pid; | 504 | int old = audit_pid; |
505 | if (sid) { | ||
506 | char *ctx = NULL; | ||
507 | u32 len; | ||
508 | int rc; | ||
509 | if ((rc = selinux_ctxid_to_string( | ||
510 | sid, &ctx, &len))) | ||
511 | return rc; | ||
512 | else | ||
513 | audit_log(NULL, GFP_KERNEL, | ||
514 | AUDIT_CONFIG_CHANGE, | ||
515 | "audit_pid=%d old=%d by auid=%u subj=%s", | ||
516 | status_get->pid, old, | ||
517 | loginuid, ctx); | ||
518 | kfree(ctx); | ||
519 | } else | ||
520 | audit_log(NULL, GFP_KERNEL, AUDIT_CONFIG_CHANGE, | ||
521 | "audit_pid=%d old=%d by auid=%u", | ||
522 | status_get->pid, old, loginuid); | ||
445 | audit_pid = status_get->pid; | 523 | audit_pid = status_get->pid; |
446 | audit_log(NULL, GFP_KERNEL, AUDIT_CONFIG_CHANGE, | ||
447 | "audit_pid=%d old=%d by auid=%u", | ||
448 | audit_pid, old, loginuid); | ||
449 | } | 524 | } |
450 | if (status_get->mask & AUDIT_STATUS_RATE_LIMIT) | 525 | if (status_get->mask & AUDIT_STATUS_RATE_LIMIT) |
451 | audit_set_rate_limit(status_get->rate_limit, loginuid); | 526 | audit_set_rate_limit(status_get->rate_limit, |
527 | loginuid, sid); | ||
452 | if (status_get->mask & AUDIT_STATUS_BACKLOG_LIMIT) | 528 | if (status_get->mask & AUDIT_STATUS_BACKLOG_LIMIT) |
453 | audit_set_backlog_limit(status_get->backlog_limit, | 529 | audit_set_backlog_limit(status_get->backlog_limit, |
454 | loginuid); | 530 | loginuid, sid); |
455 | break; | 531 | break; |
456 | case AUDIT_USER: | 532 | case AUDIT_USER: |
457 | case AUDIT_FIRST_USER_MSG...AUDIT_LAST_USER_MSG: | 533 | case AUDIT_FIRST_USER_MSG...AUDIT_LAST_USER_MSG: |
@@ -465,8 +541,23 @@ static int audit_receive_msg(struct sk_buff *skb, struct nlmsghdr *nlh) | |||
465 | ab = audit_log_start(NULL, GFP_KERNEL, msg_type); | 541 | ab = audit_log_start(NULL, GFP_KERNEL, msg_type); |
466 | if (ab) { | 542 | if (ab) { |
467 | audit_log_format(ab, | 543 | audit_log_format(ab, |
468 | "user pid=%d uid=%u auid=%u msg='%.1024s'", | 544 | "user pid=%d uid=%u auid=%u", |
469 | pid, uid, loginuid, (char *)data); | 545 | pid, uid, loginuid); |
546 | if (sid) { | ||
547 | char *ctx = NULL; | ||
548 | u32 len; | ||
549 | if (selinux_ctxid_to_string( | ||
550 | sid, &ctx, &len)) { | ||
551 | audit_log_format(ab, | ||
552 | " ssid=%u", sid); | ||
553 | /* Maybe call audit_panic? */ | ||
554 | } else | ||
555 | audit_log_format(ab, | ||
556 | " subj=%s", ctx); | ||
557 | kfree(ctx); | ||
558 | } | ||
559 | audit_log_format(ab, " msg='%.1024s'", | ||
560 | (char *)data); | ||
470 | audit_set_pid(ab, pid); | 561 | audit_set_pid(ab, pid); |
471 | audit_log_end(ab); | 562 | audit_log_end(ab); |
472 | } | 563 | } |
@@ -480,7 +571,7 @@ static int audit_receive_msg(struct sk_buff *skb, struct nlmsghdr *nlh) | |||
480 | case AUDIT_LIST: | 571 | case AUDIT_LIST: |
481 | err = audit_receive_filter(nlh->nlmsg_type, NETLINK_CB(skb).pid, | 572 | err = audit_receive_filter(nlh->nlmsg_type, NETLINK_CB(skb).pid, |
482 | uid, seq, data, nlmsg_len(nlh), | 573 | uid, seq, data, nlmsg_len(nlh), |
483 | loginuid); | 574 | loginuid, sid); |
484 | break; | 575 | break; |
485 | case AUDIT_ADD_RULE: | 576 | case AUDIT_ADD_RULE: |
486 | case AUDIT_DEL_RULE: | 577 | case AUDIT_DEL_RULE: |
@@ -490,7 +581,7 @@ static int audit_receive_msg(struct sk_buff *skb, struct nlmsghdr *nlh) | |||
490 | case AUDIT_LIST_RULES: | 581 | case AUDIT_LIST_RULES: |
491 | err = audit_receive_filter(nlh->nlmsg_type, NETLINK_CB(skb).pid, | 582 | err = audit_receive_filter(nlh->nlmsg_type, NETLINK_CB(skb).pid, |
492 | uid, seq, data, nlmsg_len(nlh), | 583 | uid, seq, data, nlmsg_len(nlh), |
493 | loginuid); | 584 | loginuid, sid); |
494 | break; | 585 | break; |
495 | case AUDIT_SIGNAL_INFO: | 586 | case AUDIT_SIGNAL_INFO: |
496 | sig_data.uid = audit_sig_uid; | 587 | sig_data.uid = audit_sig_uid; |
@@ -564,6 +655,11 @@ static int __init audit_init(void) | |||
564 | skb_queue_head_init(&audit_skb_queue); | 655 | skb_queue_head_init(&audit_skb_queue); |
565 | audit_initialized = 1; | 656 | audit_initialized = 1; |
566 | audit_enabled = audit_default; | 657 | audit_enabled = audit_default; |
658 | |||
659 | /* Register the callback with selinux. This callback will be invoked | ||
660 | * when a new policy is loaded. */ | ||
661 | selinux_audit_set_callback(&selinux_audit_rule_update); | ||
662 | |||
567 | audit_log(NULL, GFP_KERNEL, AUDIT_KERNEL, "initialized"); | 663 | audit_log(NULL, GFP_KERNEL, AUDIT_KERNEL, "initialized"); |
568 | return 0; | 664 | return 0; |
569 | } | 665 | } |
diff --git a/kernel/audit.h b/kernel/audit.h index bc5392076e2b..6f733920fd32 100644 --- a/kernel/audit.h +++ b/kernel/audit.h | |||
@@ -54,9 +54,11 @@ enum audit_state { | |||
54 | 54 | ||
55 | /* Rule lists */ | 55 | /* Rule lists */ |
56 | struct audit_field { | 56 | struct audit_field { |
57 | u32 type; | 57 | u32 type; |
58 | u32 val; | 58 | u32 val; |
59 | u32 op; | 59 | u32 op; |
60 | char *se_str; | ||
61 | struct selinux_audit_rule *se_rule; | ||
60 | }; | 62 | }; |
61 | 63 | ||
62 | struct audit_krule { | 64 | struct audit_krule { |
@@ -86,3 +88,5 @@ extern void audit_send_reply(int pid, int seq, int type, | |||
86 | extern void audit_log_lost(const char *message); | 88 | extern void audit_log_lost(const char *message); |
87 | extern void audit_panic(const char *message); | 89 | extern void audit_panic(const char *message); |
88 | extern struct mutex audit_netlink_mutex; | 90 | extern struct mutex audit_netlink_mutex; |
91 | |||
92 | extern int selinux_audit_rule_update(void); | ||
diff --git a/kernel/auditfilter.c b/kernel/auditfilter.c index d3a8539f3a83..7c134906d689 100644 --- a/kernel/auditfilter.c +++ b/kernel/auditfilter.c | |||
@@ -23,6 +23,7 @@ | |||
23 | #include <linux/audit.h> | 23 | #include <linux/audit.h> |
24 | #include <linux/kthread.h> | 24 | #include <linux/kthread.h> |
25 | #include <linux/netlink.h> | 25 | #include <linux/netlink.h> |
26 | #include <linux/selinux.h> | ||
26 | #include "audit.h" | 27 | #include "audit.h" |
27 | 28 | ||
28 | /* There are three lists of rules -- one to search at task creation | 29 | /* There are three lists of rules -- one to search at task creation |
@@ -42,6 +43,13 @@ struct list_head audit_filter_list[AUDIT_NR_FILTERS] = { | |||
42 | 43 | ||
43 | static inline void audit_free_rule(struct audit_entry *e) | 44 | static inline void audit_free_rule(struct audit_entry *e) |
44 | { | 45 | { |
46 | int i; | ||
47 | if (e->rule.fields) | ||
48 | for (i = 0; i < e->rule.field_count; i++) { | ||
49 | struct audit_field *f = &e->rule.fields[i]; | ||
50 | kfree(f->se_str); | ||
51 | selinux_audit_rule_free(f->se_rule); | ||
52 | } | ||
45 | kfree(e->rule.fields); | 53 | kfree(e->rule.fields); |
46 | kfree(e); | 54 | kfree(e); |
47 | } | 55 | } |
@@ -52,9 +60,29 @@ static inline void audit_free_rule_rcu(struct rcu_head *head) | |||
52 | audit_free_rule(e); | 60 | audit_free_rule(e); |
53 | } | 61 | } |
54 | 62 | ||
63 | /* Initialize an audit filterlist entry. */ | ||
64 | static inline struct audit_entry *audit_init_entry(u32 field_count) | ||
65 | { | ||
66 | struct audit_entry *entry; | ||
67 | struct audit_field *fields; | ||
68 | |||
69 | entry = kzalloc(sizeof(*entry), GFP_KERNEL); | ||
70 | if (unlikely(!entry)) | ||
71 | return NULL; | ||
72 | |||
73 | fields = kzalloc(sizeof(*fields) * field_count, GFP_KERNEL); | ||
74 | if (unlikely(!fields)) { | ||
75 | kfree(entry); | ||
76 | return NULL; | ||
77 | } | ||
78 | entry->rule.fields = fields; | ||
79 | |||
80 | return entry; | ||
81 | } | ||
82 | |||
55 | /* Unpack a filter field's string representation from user-space | 83 | /* Unpack a filter field's string representation from user-space |
56 | * buffer. */ | 84 | * buffer. */ |
57 | static __attribute__((unused)) char *audit_unpack_string(void **bufp, size_t *remain, size_t len) | 85 | static char *audit_unpack_string(void **bufp, size_t *remain, size_t len) |
58 | { | 86 | { |
59 | char *str; | 87 | char *str; |
60 | 88 | ||
@@ -84,7 +112,6 @@ static inline struct audit_entry *audit_to_entry_common(struct audit_rule *rule) | |||
84 | { | 112 | { |
85 | unsigned listnr; | 113 | unsigned listnr; |
86 | struct audit_entry *entry; | 114 | struct audit_entry *entry; |
87 | struct audit_field *fields; | ||
88 | int i, err; | 115 | int i, err; |
89 | 116 | ||
90 | err = -EINVAL; | 117 | err = -EINVAL; |
@@ -108,23 +135,14 @@ static inline struct audit_entry *audit_to_entry_common(struct audit_rule *rule) | |||
108 | goto exit_err; | 135 | goto exit_err; |
109 | 136 | ||
110 | err = -ENOMEM; | 137 | err = -ENOMEM; |
111 | entry = kmalloc(sizeof(*entry), GFP_KERNEL); | 138 | entry = audit_init_entry(rule->field_count); |
112 | if (unlikely(!entry)) | 139 | if (!entry) |
113 | goto exit_err; | ||
114 | fields = kmalloc(sizeof(*fields) * rule->field_count, GFP_KERNEL); | ||
115 | if (unlikely(!fields)) { | ||
116 | kfree(entry); | ||
117 | goto exit_err; | 140 | goto exit_err; |
118 | } | ||
119 | |||
120 | memset(&entry->rule, 0, sizeof(struct audit_krule)); | ||
121 | memset(fields, 0, sizeof(struct audit_field)); | ||
122 | 141 | ||
123 | entry->rule.flags = rule->flags & AUDIT_FILTER_PREPEND; | 142 | entry->rule.flags = rule->flags & AUDIT_FILTER_PREPEND; |
124 | entry->rule.listnr = listnr; | 143 | entry->rule.listnr = listnr; |
125 | entry->rule.action = rule->action; | 144 | entry->rule.action = rule->action; |
126 | entry->rule.field_count = rule->field_count; | 145 | entry->rule.field_count = rule->field_count; |
127 | entry->rule.fields = fields; | ||
128 | 146 | ||
129 | for (i = 0; i < AUDIT_BITMASK_SIZE; i++) | 147 | for (i = 0; i < AUDIT_BITMASK_SIZE; i++) |
130 | entry->rule.mask[i] = rule->mask[i]; | 148 | entry->rule.mask[i] = rule->mask[i]; |
@@ -150,15 +168,20 @@ static struct audit_entry *audit_rule_to_entry(struct audit_rule *rule) | |||
150 | for (i = 0; i < rule->field_count; i++) { | 168 | for (i = 0; i < rule->field_count; i++) { |
151 | struct audit_field *f = &entry->rule.fields[i]; | 169 | struct audit_field *f = &entry->rule.fields[i]; |
152 | 170 | ||
153 | if (rule->fields[i] & AUDIT_UNUSED_BITS) { | ||
154 | err = -EINVAL; | ||
155 | goto exit_free; | ||
156 | } | ||
157 | |||
158 | f->op = rule->fields[i] & (AUDIT_NEGATE|AUDIT_OPERATORS); | 171 | f->op = rule->fields[i] & (AUDIT_NEGATE|AUDIT_OPERATORS); |
159 | f->type = rule->fields[i] & ~(AUDIT_NEGATE|AUDIT_OPERATORS); | 172 | f->type = rule->fields[i] & ~(AUDIT_NEGATE|AUDIT_OPERATORS); |
160 | f->val = rule->values[i]; | 173 | f->val = rule->values[i]; |
161 | 174 | ||
175 | if (f->type & AUDIT_UNUSED_BITS || | ||
176 | f->type == AUDIT_SE_USER || | ||
177 | f->type == AUDIT_SE_ROLE || | ||
178 | f->type == AUDIT_SE_TYPE || | ||
179 | f->type == AUDIT_SE_SEN || | ||
180 | f->type == AUDIT_SE_CLR) { | ||
181 | err = -EINVAL; | ||
182 | goto exit_free; | ||
183 | } | ||
184 | |||
162 | entry->rule.vers_ops = (f->op & AUDIT_OPERATORS) ? 2 : 1; | 185 | entry->rule.vers_ops = (f->op & AUDIT_OPERATORS) ? 2 : 1; |
163 | 186 | ||
164 | /* Support for legacy operators where | 187 | /* Support for legacy operators where |
@@ -188,8 +211,9 @@ static struct audit_entry *audit_data_to_entry(struct audit_rule_data *data, | |||
188 | int err = 0; | 211 | int err = 0; |
189 | struct audit_entry *entry; | 212 | struct audit_entry *entry; |
190 | void *bufp; | 213 | void *bufp; |
191 | /* size_t remain = datasz - sizeof(struct audit_rule_data); */ | 214 | size_t remain = datasz - sizeof(struct audit_rule_data); |
192 | int i; | 215 | int i; |
216 | char *str; | ||
193 | 217 | ||
194 | entry = audit_to_entry_common((struct audit_rule *)data); | 218 | entry = audit_to_entry_common((struct audit_rule *)data); |
195 | if (IS_ERR(entry)) | 219 | if (IS_ERR(entry)) |
@@ -207,10 +231,35 @@ static struct audit_entry *audit_data_to_entry(struct audit_rule_data *data, | |||
207 | 231 | ||
208 | f->op = data->fieldflags[i] & AUDIT_OPERATORS; | 232 | f->op = data->fieldflags[i] & AUDIT_OPERATORS; |
209 | f->type = data->fields[i]; | 233 | f->type = data->fields[i]; |
234 | f->val = data->values[i]; | ||
235 | f->se_str = NULL; | ||
236 | f->se_rule = NULL; | ||
210 | switch(f->type) { | 237 | switch(f->type) { |
211 | /* call type-specific conversion routines here */ | 238 | case AUDIT_SE_USER: |
212 | default: | 239 | case AUDIT_SE_ROLE: |
213 | f->val = data->values[i]; | 240 | case AUDIT_SE_TYPE: |
241 | case AUDIT_SE_SEN: | ||
242 | case AUDIT_SE_CLR: | ||
243 | str = audit_unpack_string(&bufp, &remain, f->val); | ||
244 | if (IS_ERR(str)) | ||
245 | goto exit_free; | ||
246 | entry->rule.buflen += f->val; | ||
247 | |||
248 | err = selinux_audit_rule_init(f->type, f->op, str, | ||
249 | &f->se_rule); | ||
250 | /* Keep currently invalid fields around in case they | ||
251 | * become valid after a policy reload. */ | ||
252 | if (err == -EINVAL) { | ||
253 | printk(KERN_WARNING "audit rule for selinux " | ||
254 | "\'%s\' is invalid\n", str); | ||
255 | err = 0; | ||
256 | } | ||
257 | if (err) { | ||
258 | kfree(str); | ||
259 | goto exit_free; | ||
260 | } else | ||
261 | f->se_str = str; | ||
262 | break; | ||
214 | } | 263 | } |
215 | } | 264 | } |
216 | 265 | ||
@@ -286,7 +335,14 @@ static struct audit_rule_data *audit_krule_to_data(struct audit_krule *krule) | |||
286 | data->fields[i] = f->type; | 335 | data->fields[i] = f->type; |
287 | data->fieldflags[i] = f->op; | 336 | data->fieldflags[i] = f->op; |
288 | switch(f->type) { | 337 | switch(f->type) { |
289 | /* call type-specific conversion routines here */ | 338 | case AUDIT_SE_USER: |
339 | case AUDIT_SE_ROLE: | ||
340 | case AUDIT_SE_TYPE: | ||
341 | case AUDIT_SE_SEN: | ||
342 | case AUDIT_SE_CLR: | ||
343 | data->buflen += data->values[i] = | ||
344 | audit_pack_string(&bufp, f->se_str); | ||
345 | break; | ||
290 | default: | 346 | default: |
291 | data->values[i] = f->val; | 347 | data->values[i] = f->val; |
292 | } | 348 | } |
@@ -314,7 +370,14 @@ static int audit_compare_rule(struct audit_krule *a, struct audit_krule *b) | |||
314 | return 1; | 370 | return 1; |
315 | 371 | ||
316 | switch(a->fields[i].type) { | 372 | switch(a->fields[i].type) { |
317 | /* call type-specific comparison routines here */ | 373 | case AUDIT_SE_USER: |
374 | case AUDIT_SE_ROLE: | ||
375 | case AUDIT_SE_TYPE: | ||
376 | case AUDIT_SE_SEN: | ||
377 | case AUDIT_SE_CLR: | ||
378 | if (strcmp(a->fields[i].se_str, b->fields[i].se_str)) | ||
379 | return 1; | ||
380 | break; | ||
318 | default: | 381 | default: |
319 | if (a->fields[i].val != b->fields[i].val) | 382 | if (a->fields[i].val != b->fields[i].val) |
320 | return 1; | 383 | return 1; |
@@ -328,6 +391,81 @@ static int audit_compare_rule(struct audit_krule *a, struct audit_krule *b) | |||
328 | return 0; | 391 | return 0; |
329 | } | 392 | } |
330 | 393 | ||
394 | /* Duplicate selinux field information. The se_rule is opaque, so must be | ||
395 | * re-initialized. */ | ||
396 | static inline int audit_dupe_selinux_field(struct audit_field *df, | ||
397 | struct audit_field *sf) | ||
398 | { | ||
399 | int ret = 0; | ||
400 | char *se_str; | ||
401 | |||
402 | /* our own copy of se_str */ | ||
403 | se_str = kstrdup(sf->se_str, GFP_KERNEL); | ||
404 | if (unlikely(IS_ERR(se_str))) | ||
405 | return -ENOMEM; | ||
406 | df->se_str = se_str; | ||
407 | |||
408 | /* our own (refreshed) copy of se_rule */ | ||
409 | ret = selinux_audit_rule_init(df->type, df->op, df->se_str, | ||
410 | &df->se_rule); | ||
411 | /* Keep currently invalid fields around in case they | ||
412 | * become valid after a policy reload. */ | ||
413 | if (ret == -EINVAL) { | ||
414 | printk(KERN_WARNING "audit rule for selinux \'%s\' is " | ||
415 | "invalid\n", df->se_str); | ||
416 | ret = 0; | ||
417 | } | ||
418 | |||
419 | return ret; | ||
420 | } | ||
421 | |||
422 | /* Duplicate an audit rule. This will be a deep copy with the exception | ||
423 | * of the watch - that pointer is carried over. The selinux specific fields | ||
424 | * will be updated in the copy. The point is to be able to replace the old | ||
425 | * rule with the new rule in the filterlist, then free the old rule. */ | ||
426 | static struct audit_entry *audit_dupe_rule(struct audit_krule *old) | ||
427 | { | ||
428 | u32 fcount = old->field_count; | ||
429 | struct audit_entry *entry; | ||
430 | struct audit_krule *new; | ||
431 | int i, err = 0; | ||
432 | |||
433 | entry = audit_init_entry(fcount); | ||
434 | if (unlikely(!entry)) | ||
435 | return ERR_PTR(-ENOMEM); | ||
436 | |||
437 | new = &entry->rule; | ||
438 | new->vers_ops = old->vers_ops; | ||
439 | new->flags = old->flags; | ||
440 | new->listnr = old->listnr; | ||
441 | new->action = old->action; | ||
442 | for (i = 0; i < AUDIT_BITMASK_SIZE; i++) | ||
443 | new->mask[i] = old->mask[i]; | ||
444 | new->buflen = old->buflen; | ||
445 | new->field_count = old->field_count; | ||
446 | memcpy(new->fields, old->fields, sizeof(struct audit_field) * fcount); | ||
447 | |||
448 | /* deep copy this information, updating the se_rule fields, because | ||
449 | * the originals will all be freed when the old rule is freed. */ | ||
450 | for (i = 0; i < fcount; i++) { | ||
451 | switch (new->fields[i].type) { | ||
452 | case AUDIT_SE_USER: | ||
453 | case AUDIT_SE_ROLE: | ||
454 | case AUDIT_SE_TYPE: | ||
455 | case AUDIT_SE_SEN: | ||
456 | case AUDIT_SE_CLR: | ||
457 | err = audit_dupe_selinux_field(&new->fields[i], | ||
458 | &old->fields[i]); | ||
459 | } | ||
460 | if (err) { | ||
461 | audit_free_rule(entry); | ||
462 | return ERR_PTR(err); | ||
463 | } | ||
464 | } | ||
465 | |||
466 | return entry; | ||
467 | } | ||
468 | |||
331 | /* Add rule to given filterlist if not a duplicate. Protected by | 469 | /* Add rule to given filterlist if not a duplicate. Protected by |
332 | * audit_netlink_mutex. */ | 470 | * audit_netlink_mutex. */ |
333 | static inline int audit_add_rule(struct audit_entry *entry, | 471 | static inline int audit_add_rule(struct audit_entry *entry, |
@@ -448,9 +586,10 @@ static int audit_list_rules(void *_dest) | |||
448 | * @data: payload data | 586 | * @data: payload data |
449 | * @datasz: size of payload data | 587 | * @datasz: size of payload data |
450 | * @loginuid: loginuid of sender | 588 | * @loginuid: loginuid of sender |
589 | * @sid: SE Linux Security ID of sender | ||
451 | */ | 590 | */ |
452 | int audit_receive_filter(int type, int pid, int uid, int seq, void *data, | 591 | int audit_receive_filter(int type, int pid, int uid, int seq, void *data, |
453 | size_t datasz, uid_t loginuid) | 592 | size_t datasz, uid_t loginuid, u32 sid) |
454 | { | 593 | { |
455 | struct task_struct *tsk; | 594 | struct task_struct *tsk; |
456 | int *dest; | 595 | int *dest; |
@@ -493,9 +632,23 @@ int audit_receive_filter(int type, int pid, int uid, int seq, void *data, | |||
493 | 632 | ||
494 | err = audit_add_rule(entry, | 633 | err = audit_add_rule(entry, |
495 | &audit_filter_list[entry->rule.listnr]); | 634 | &audit_filter_list[entry->rule.listnr]); |
496 | audit_log(NULL, GFP_KERNEL, AUDIT_CONFIG_CHANGE, | 635 | if (sid) { |
497 | "auid=%u add rule to list=%d res=%d\n", | 636 | char *ctx = NULL; |
498 | loginuid, entry->rule.listnr, !err); | 637 | u32 len; |
638 | if (selinux_ctxid_to_string(sid, &ctx, &len)) { | ||
639 | /* Maybe call audit_panic? */ | ||
640 | audit_log(NULL, GFP_KERNEL, AUDIT_CONFIG_CHANGE, | ||
641 | "auid=%u ssid=%u add rule to list=%d res=%d", | ||
642 | loginuid, sid, entry->rule.listnr, !err); | ||
643 | } else | ||
644 | audit_log(NULL, GFP_KERNEL, AUDIT_CONFIG_CHANGE, | ||
645 | "auid=%u subj=%s add rule to list=%d res=%d", | ||
646 | loginuid, ctx, entry->rule.listnr, !err); | ||
647 | kfree(ctx); | ||
648 | } else | ||
649 | audit_log(NULL, GFP_KERNEL, AUDIT_CONFIG_CHANGE, | ||
650 | "auid=%u add rule to list=%d res=%d", | ||
651 | loginuid, entry->rule.listnr, !err); | ||
499 | 652 | ||
500 | if (err) | 653 | if (err) |
501 | audit_free_rule(entry); | 654 | audit_free_rule(entry); |
@@ -511,9 +664,24 @@ int audit_receive_filter(int type, int pid, int uid, int seq, void *data, | |||
511 | 664 | ||
512 | err = audit_del_rule(entry, | 665 | err = audit_del_rule(entry, |
513 | &audit_filter_list[entry->rule.listnr]); | 666 | &audit_filter_list[entry->rule.listnr]); |
514 | audit_log(NULL, GFP_KERNEL, AUDIT_CONFIG_CHANGE, | 667 | |
515 | "auid=%u remove rule from list=%d res=%d\n", | 668 | if (sid) { |
516 | loginuid, entry->rule.listnr, !err); | 669 | char *ctx = NULL; |
670 | u32 len; | ||
671 | if (selinux_ctxid_to_string(sid, &ctx, &len)) { | ||
672 | /* Maybe call audit_panic? */ | ||
673 | audit_log(NULL, GFP_KERNEL, AUDIT_CONFIG_CHANGE, | ||
674 | "auid=%u ssid=%u remove rule from list=%d res=%d", | ||
675 | loginuid, sid, entry->rule.listnr, !err); | ||
676 | } else | ||
677 | audit_log(NULL, GFP_KERNEL, AUDIT_CONFIG_CHANGE, | ||
678 | "auid=%u subj=%s remove rule from list=%d res=%d", | ||
679 | loginuid, ctx, entry->rule.listnr, !err); | ||
680 | kfree(ctx); | ||
681 | } else | ||
682 | audit_log(NULL, GFP_KERNEL, AUDIT_CONFIG_CHANGE, | ||
683 | "auid=%u remove rule from list=%d res=%d", | ||
684 | loginuid, entry->rule.listnr, !err); | ||
517 | 685 | ||
518 | audit_free_rule(entry); | 686 | audit_free_rule(entry); |
519 | break; | 687 | break; |
@@ -628,3 +796,62 @@ unlock_and_return: | |||
628 | rcu_read_unlock(); | 796 | rcu_read_unlock(); |
629 | return result; | 797 | return result; |
630 | } | 798 | } |
799 | |||
800 | /* Check to see if the rule contains any selinux fields. Returns 1 if there | ||
801 | are selinux fields specified in the rule, 0 otherwise. */ | ||
802 | static inline int audit_rule_has_selinux(struct audit_krule *rule) | ||
803 | { | ||
804 | int i; | ||
805 | |||
806 | for (i = 0; i < rule->field_count; i++) { | ||
807 | struct audit_field *f = &rule->fields[i]; | ||
808 | switch (f->type) { | ||
809 | case AUDIT_SE_USER: | ||
810 | case AUDIT_SE_ROLE: | ||
811 | case AUDIT_SE_TYPE: | ||
812 | case AUDIT_SE_SEN: | ||
813 | case AUDIT_SE_CLR: | ||
814 | return 1; | ||
815 | } | ||
816 | } | ||
817 | |||
818 | return 0; | ||
819 | } | ||
820 | |||
821 | /* This function will re-initialize the se_rule field of all applicable rules. | ||
822 | * It will traverse the filter lists serarching for rules that contain selinux | ||
823 | * specific filter fields. When such a rule is found, it is copied, the | ||
824 | * selinux field is re-initialized, and the old rule is replaced with the | ||
825 | * updated rule. */ | ||
826 | int selinux_audit_rule_update(void) | ||
827 | { | ||
828 | struct audit_entry *entry, *n, *nentry; | ||
829 | int i, err = 0; | ||
830 | |||
831 | /* audit_netlink_mutex synchronizes the writers */ | ||
832 | mutex_lock(&audit_netlink_mutex); | ||
833 | |||
834 | for (i = 0; i < AUDIT_NR_FILTERS; i++) { | ||
835 | list_for_each_entry_safe(entry, n, &audit_filter_list[i], list) { | ||
836 | if (!audit_rule_has_selinux(&entry->rule)) | ||
837 | continue; | ||
838 | |||
839 | nentry = audit_dupe_rule(&entry->rule); | ||
840 | if (unlikely(IS_ERR(nentry))) { | ||
841 | /* save the first error encountered for the | ||
842 | * return value */ | ||
843 | if (!err) | ||
844 | err = PTR_ERR(nentry); | ||
845 | audit_panic("error updating selinux filters"); | ||
846 | list_del_rcu(&entry->list); | ||
847 | } else { | ||
848 | list_replace_rcu(&entry->list, &nentry->list); | ||
849 | } | ||
850 | call_rcu(&entry->rcu, audit_free_rule_rcu); | ||
851 | } | ||
852 | } | ||
853 | |||
854 | mutex_unlock(&audit_netlink_mutex); | ||
855 | |||
856 | return err; | ||
857 | } | ||
diff --git a/kernel/auditsc.c b/kernel/auditsc.c index 7f160df21a23..1c03a4ed1b27 100644 --- a/kernel/auditsc.c +++ b/kernel/auditsc.c | |||
@@ -58,6 +58,7 @@ | |||
58 | #include <linux/security.h> | 58 | #include <linux/security.h> |
59 | #include <linux/list.h> | 59 | #include <linux/list.h> |
60 | #include <linux/tty.h> | 60 | #include <linux/tty.h> |
61 | #include <linux/selinux.h> | ||
61 | 62 | ||
62 | #include "audit.h" | 63 | #include "audit.h" |
63 | 64 | ||
@@ -89,7 +90,7 @@ struct audit_names { | |||
89 | uid_t uid; | 90 | uid_t uid; |
90 | gid_t gid; | 91 | gid_t gid; |
91 | dev_t rdev; | 92 | dev_t rdev; |
92 | char *ctx; | 93 | u32 osid; |
93 | }; | 94 | }; |
94 | 95 | ||
95 | struct audit_aux_data { | 96 | struct audit_aux_data { |
@@ -106,7 +107,7 @@ struct audit_aux_data_ipcctl { | |||
106 | uid_t uid; | 107 | uid_t uid; |
107 | gid_t gid; | 108 | gid_t gid; |
108 | mode_t mode; | 109 | mode_t mode; |
109 | char *ctx; | 110 | u32 osid; |
110 | }; | 111 | }; |
111 | 112 | ||
112 | struct audit_aux_data_socketcall { | 113 | struct audit_aux_data_socketcall { |
@@ -167,7 +168,8 @@ static int audit_filter_rules(struct task_struct *tsk, | |||
167 | struct audit_context *ctx, | 168 | struct audit_context *ctx, |
168 | enum audit_state *state) | 169 | enum audit_state *state) |
169 | { | 170 | { |
170 | int i, j; | 171 | int i, j, need_sid = 1; |
172 | u32 sid; | ||
171 | 173 | ||
172 | for (i = 0; i < rule->field_count; i++) { | 174 | for (i = 0; i < rule->field_count; i++) { |
173 | struct audit_field *f = &rule->fields[i]; | 175 | struct audit_field *f = &rule->fields[i]; |
@@ -257,6 +259,27 @@ static int audit_filter_rules(struct task_struct *tsk, | |||
257 | if (ctx) | 259 | if (ctx) |
258 | result = audit_comparator(ctx->loginuid, f->op, f->val); | 260 | result = audit_comparator(ctx->loginuid, f->op, f->val); |
259 | break; | 261 | break; |
262 | case AUDIT_SE_USER: | ||
263 | case AUDIT_SE_ROLE: | ||
264 | case AUDIT_SE_TYPE: | ||
265 | case AUDIT_SE_SEN: | ||
266 | case AUDIT_SE_CLR: | ||
267 | /* NOTE: this may return negative values indicating | ||
268 | a temporary error. We simply treat this as a | ||
269 | match for now to avoid losing information that | ||
270 | may be wanted. An error message will also be | ||
271 | logged upon error */ | ||
272 | if (f->se_rule) { | ||
273 | if (need_sid) { | ||
274 | selinux_task_ctxid(tsk, &sid); | ||
275 | need_sid = 0; | ||
276 | } | ||
277 | result = selinux_audit_rule_match(sid, f->type, | ||
278 | f->op, | ||
279 | f->se_rule, | ||
280 | ctx); | ||
281 | } | ||
282 | break; | ||
260 | case AUDIT_ARG0: | 283 | case AUDIT_ARG0: |
261 | case AUDIT_ARG1: | 284 | case AUDIT_ARG1: |
262 | case AUDIT_ARG2: | 285 | case AUDIT_ARG2: |
@@ -329,7 +352,6 @@ static enum audit_state audit_filter_syscall(struct task_struct *tsk, | |||
329 | return AUDIT_BUILD_CONTEXT; | 352 | return AUDIT_BUILD_CONTEXT; |
330 | } | 353 | } |
331 | 354 | ||
332 | /* This should be called with task_lock() held. */ | ||
333 | static inline struct audit_context *audit_get_context(struct task_struct *tsk, | 355 | static inline struct audit_context *audit_get_context(struct task_struct *tsk, |
334 | int return_valid, | 356 | int return_valid, |
335 | int return_code) | 357 | int return_code) |
@@ -391,9 +413,6 @@ static inline void audit_free_names(struct audit_context *context) | |||
391 | #endif | 413 | #endif |
392 | 414 | ||
393 | for (i = 0; i < context->name_count; i++) { | 415 | for (i = 0; i < context->name_count; i++) { |
394 | char *p = context->names[i].ctx; | ||
395 | context->names[i].ctx = NULL; | ||
396 | kfree(p); | ||
397 | if (context->names[i].name) | 416 | if (context->names[i].name) |
398 | __putname(context->names[i].name); | 417 | __putname(context->names[i].name); |
399 | } | 418 | } |
@@ -416,11 +435,6 @@ static inline void audit_free_aux(struct audit_context *context) | |||
416 | dput(axi->dentry); | 435 | dput(axi->dentry); |
417 | mntput(axi->mnt); | 436 | mntput(axi->mnt); |
418 | } | 437 | } |
419 | if ( aux->type == AUDIT_IPC ) { | ||
420 | struct audit_aux_data_ipcctl *axi = (void *)aux; | ||
421 | if (axi->ctx) | ||
422 | kfree(axi->ctx); | ||
423 | } | ||
424 | 438 | ||
425 | context->aux = aux->next; | 439 | context->aux = aux->next; |
426 | kfree(aux); | 440 | kfree(aux); |
@@ -506,7 +520,7 @@ static inline void audit_free_context(struct audit_context *context) | |||
506 | printk(KERN_ERR "audit: freed %d contexts\n", count); | 520 | printk(KERN_ERR "audit: freed %d contexts\n", count); |
507 | } | 521 | } |
508 | 522 | ||
509 | static void audit_log_task_context(struct audit_buffer *ab, gfp_t gfp_mask) | 523 | static void audit_log_task_context(struct audit_buffer *ab) |
510 | { | 524 | { |
511 | char *ctx = NULL; | 525 | char *ctx = NULL; |
512 | ssize_t len = 0; | 526 | ssize_t len = 0; |
@@ -518,7 +532,7 @@ static void audit_log_task_context(struct audit_buffer *ab, gfp_t gfp_mask) | |||
518 | return; | 532 | return; |
519 | } | 533 | } |
520 | 534 | ||
521 | ctx = kmalloc(len, gfp_mask); | 535 | ctx = kmalloc(len, GFP_KERNEL); |
522 | if (!ctx) | 536 | if (!ctx) |
523 | goto error_path; | 537 | goto error_path; |
524 | 538 | ||
@@ -536,47 +550,46 @@ error_path: | |||
536 | return; | 550 | return; |
537 | } | 551 | } |
538 | 552 | ||
539 | static void audit_log_task_info(struct audit_buffer *ab, gfp_t gfp_mask) | 553 | static void audit_log_task_info(struct audit_buffer *ab, struct task_struct *tsk) |
540 | { | 554 | { |
541 | char name[sizeof(current->comm)]; | 555 | char name[sizeof(tsk->comm)]; |
542 | struct mm_struct *mm = current->mm; | 556 | struct mm_struct *mm = tsk->mm; |
543 | struct vm_area_struct *vma; | 557 | struct vm_area_struct *vma; |
544 | 558 | ||
545 | get_task_comm(name, current); | 559 | /* tsk == current */ |
560 | |||
561 | get_task_comm(name, tsk); | ||
546 | audit_log_format(ab, " comm="); | 562 | audit_log_format(ab, " comm="); |
547 | audit_log_untrustedstring(ab, name); | 563 | audit_log_untrustedstring(ab, name); |
548 | 564 | ||
549 | if (!mm) | 565 | if (mm) { |
550 | return; | 566 | down_read(&mm->mmap_sem); |
551 | 567 | vma = mm->mmap; | |
552 | /* | 568 | while (vma) { |
553 | * this is brittle; all callers that pass GFP_ATOMIC will have | 569 | if ((vma->vm_flags & VM_EXECUTABLE) && |
554 | * NULL current->mm and we won't get here. | 570 | vma->vm_file) { |
555 | */ | 571 | audit_log_d_path(ab, "exe=", |
556 | down_read(&mm->mmap_sem); | 572 | vma->vm_file->f_dentry, |
557 | vma = mm->mmap; | 573 | vma->vm_file->f_vfsmnt); |
558 | while (vma) { | 574 | break; |
559 | if ((vma->vm_flags & VM_EXECUTABLE) && | 575 | } |
560 | vma->vm_file) { | 576 | vma = vma->vm_next; |
561 | audit_log_d_path(ab, "exe=", | ||
562 | vma->vm_file->f_dentry, | ||
563 | vma->vm_file->f_vfsmnt); | ||
564 | break; | ||
565 | } | 577 | } |
566 | vma = vma->vm_next; | 578 | up_read(&mm->mmap_sem); |
567 | } | 579 | } |
568 | up_read(&mm->mmap_sem); | 580 | audit_log_task_context(ab); |
569 | audit_log_task_context(ab, gfp_mask); | ||
570 | } | 581 | } |
571 | 582 | ||
572 | static void audit_log_exit(struct audit_context *context, gfp_t gfp_mask) | 583 | static void audit_log_exit(struct audit_context *context, struct task_struct *tsk) |
573 | { | 584 | { |
574 | int i; | 585 | int i, call_panic = 0; |
575 | struct audit_buffer *ab; | 586 | struct audit_buffer *ab; |
576 | struct audit_aux_data *aux; | 587 | struct audit_aux_data *aux; |
577 | const char *tty; | 588 | const char *tty; |
578 | 589 | ||
579 | ab = audit_log_start(context, gfp_mask, AUDIT_SYSCALL); | 590 | /* tsk == current */ |
591 | |||
592 | ab = audit_log_start(context, GFP_KERNEL, AUDIT_SYSCALL); | ||
580 | if (!ab) | 593 | if (!ab) |
581 | return; /* audit_panic has been called */ | 594 | return; /* audit_panic has been called */ |
582 | audit_log_format(ab, "arch=%x syscall=%d", | 595 | audit_log_format(ab, "arch=%x syscall=%d", |
@@ -587,8 +600,8 @@ static void audit_log_exit(struct audit_context *context, gfp_t gfp_mask) | |||
587 | audit_log_format(ab, " success=%s exit=%ld", | 600 | audit_log_format(ab, " success=%s exit=%ld", |
588 | (context->return_valid==AUDITSC_SUCCESS)?"yes":"no", | 601 | (context->return_valid==AUDITSC_SUCCESS)?"yes":"no", |
589 | context->return_code); | 602 | context->return_code); |
590 | if (current->signal->tty && current->signal->tty->name) | 603 | if (tsk->signal && tsk->signal->tty && tsk->signal->tty->name) |
591 | tty = current->signal->tty->name; | 604 | tty = tsk->signal->tty->name; |
592 | else | 605 | else |
593 | tty = "(none)"; | 606 | tty = "(none)"; |
594 | audit_log_format(ab, | 607 | audit_log_format(ab, |
@@ -607,12 +620,12 @@ static void audit_log_exit(struct audit_context *context, gfp_t gfp_mask) | |||
607 | context->gid, | 620 | context->gid, |
608 | context->euid, context->suid, context->fsuid, | 621 | context->euid, context->suid, context->fsuid, |
609 | context->egid, context->sgid, context->fsgid, tty); | 622 | context->egid, context->sgid, context->fsgid, tty); |
610 | audit_log_task_info(ab, gfp_mask); | 623 | audit_log_task_info(ab, tsk); |
611 | audit_log_end(ab); | 624 | audit_log_end(ab); |
612 | 625 | ||
613 | for (aux = context->aux; aux; aux = aux->next) { | 626 | for (aux = context->aux; aux; aux = aux->next) { |
614 | 627 | ||
615 | ab = audit_log_start(context, gfp_mask, aux->type); | 628 | ab = audit_log_start(context, GFP_KERNEL, aux->type); |
616 | if (!ab) | 629 | if (!ab) |
617 | continue; /* audit_panic has been called */ | 630 | continue; /* audit_panic has been called */ |
618 | 631 | ||
@@ -620,8 +633,39 @@ static void audit_log_exit(struct audit_context *context, gfp_t gfp_mask) | |||
620 | case AUDIT_IPC: { | 633 | case AUDIT_IPC: { |
621 | struct audit_aux_data_ipcctl *axi = (void *)aux; | 634 | struct audit_aux_data_ipcctl *axi = (void *)aux; |
622 | audit_log_format(ab, | 635 | audit_log_format(ab, |
623 | " qbytes=%lx iuid=%u igid=%u mode=%x obj=%s", | 636 | " qbytes=%lx iuid=%u igid=%u mode=%x", |
624 | axi->qbytes, axi->uid, axi->gid, axi->mode, axi->ctx); | 637 | axi->qbytes, axi->uid, axi->gid, axi->mode); |
638 | if (axi->osid != 0) { | ||
639 | char *ctx = NULL; | ||
640 | u32 len; | ||
641 | if (selinux_ctxid_to_string( | ||
642 | axi->osid, &ctx, &len)) { | ||
643 | audit_log_format(ab, " osid=%u", | ||
644 | axi->osid); | ||
645 | call_panic = 1; | ||
646 | } else | ||
647 | audit_log_format(ab, " obj=%s", ctx); | ||
648 | kfree(ctx); | ||
649 | } | ||
650 | break; } | ||
651 | |||
652 | case AUDIT_IPC_SET_PERM: { | ||
653 | struct audit_aux_data_ipcctl *axi = (void *)aux; | ||
654 | audit_log_format(ab, | ||
655 | " new qbytes=%lx new iuid=%u new igid=%u new mode=%x", | ||
656 | axi->qbytes, axi->uid, axi->gid, axi->mode); | ||
657 | if (axi->osid != 0) { | ||
658 | char *ctx = NULL; | ||
659 | u32 len; | ||
660 | if (selinux_ctxid_to_string( | ||
661 | axi->osid, &ctx, &len)) { | ||
662 | audit_log_format(ab, " osid=%u", | ||
663 | axi->osid); | ||
664 | call_panic = 1; | ||
665 | } else | ||
666 | audit_log_format(ab, " obj=%s", ctx); | ||
667 | kfree(ctx); | ||
668 | } | ||
625 | break; } | 669 | break; } |
626 | 670 | ||
627 | case AUDIT_SOCKETCALL: { | 671 | case AUDIT_SOCKETCALL: { |
@@ -649,7 +693,7 @@ static void audit_log_exit(struct audit_context *context, gfp_t gfp_mask) | |||
649 | } | 693 | } |
650 | 694 | ||
651 | if (context->pwd && context->pwdmnt) { | 695 | if (context->pwd && context->pwdmnt) { |
652 | ab = audit_log_start(context, gfp_mask, AUDIT_CWD); | 696 | ab = audit_log_start(context, GFP_KERNEL, AUDIT_CWD); |
653 | if (ab) { | 697 | if (ab) { |
654 | audit_log_d_path(ab, "cwd=", context->pwd, context->pwdmnt); | 698 | audit_log_d_path(ab, "cwd=", context->pwd, context->pwdmnt); |
655 | audit_log_end(ab); | 699 | audit_log_end(ab); |
@@ -659,7 +703,7 @@ static void audit_log_exit(struct audit_context *context, gfp_t gfp_mask) | |||
659 | unsigned long ino = context->names[i].ino; | 703 | unsigned long ino = context->names[i].ino; |
660 | unsigned long pino = context->names[i].pino; | 704 | unsigned long pino = context->names[i].pino; |
661 | 705 | ||
662 | ab = audit_log_start(context, gfp_mask, AUDIT_PATH); | 706 | ab = audit_log_start(context, GFP_KERNEL, AUDIT_PATH); |
663 | if (!ab) | 707 | if (!ab) |
664 | continue; /* audit_panic has been called */ | 708 | continue; /* audit_panic has been called */ |
665 | 709 | ||
@@ -685,32 +729,35 @@ static void audit_log_exit(struct audit_context *context, gfp_t gfp_mask) | |||
685 | context->names[i].gid, | 729 | context->names[i].gid, |
686 | MAJOR(context->names[i].rdev), | 730 | MAJOR(context->names[i].rdev), |
687 | MINOR(context->names[i].rdev)); | 731 | MINOR(context->names[i].rdev)); |
688 | if (context->names[i].ctx) { | 732 | if (context->names[i].osid != 0) { |
689 | audit_log_format(ab, " obj=%s", | 733 | char *ctx = NULL; |
690 | context->names[i].ctx); | 734 | u32 len; |
735 | if (selinux_ctxid_to_string( | ||
736 | context->names[i].osid, &ctx, &len)) { | ||
737 | audit_log_format(ab, " osid=%u", | ||
738 | context->names[i].osid); | ||
739 | call_panic = 2; | ||
740 | } else | ||
741 | audit_log_format(ab, " obj=%s", ctx); | ||
742 | kfree(ctx); | ||
691 | } | 743 | } |
692 | 744 | ||
693 | audit_log_end(ab); | 745 | audit_log_end(ab); |
694 | } | 746 | } |
747 | if (call_panic) | ||
748 | audit_panic("error converting sid to string"); | ||
695 | } | 749 | } |
696 | 750 | ||
697 | /** | 751 | /** |
698 | * audit_free - free a per-task audit context | 752 | * audit_free - free a per-task audit context |
699 | * @tsk: task whose audit context block to free | 753 | * @tsk: task whose audit context block to free |
700 | * | 754 | * |
701 | * Called from copy_process and __put_task_struct. | 755 | * Called from copy_process and do_exit |
702 | */ | 756 | */ |
703 | void audit_free(struct task_struct *tsk) | 757 | void audit_free(struct task_struct *tsk) |
704 | { | 758 | { |
705 | struct audit_context *context; | 759 | struct audit_context *context; |
706 | 760 | ||
707 | /* | ||
708 | * No need to lock the task - when we execute audit_free() | ||
709 | * then the task has no external references anymore, and | ||
710 | * we are tearing it down. (The locking also confuses | ||
711 | * DEBUG_LOCKDEP - this freeing may occur in softirq | ||
712 | * contexts as well, via RCU.) | ||
713 | */ | ||
714 | context = audit_get_context(tsk, 0, 0); | 761 | context = audit_get_context(tsk, 0, 0); |
715 | if (likely(!context)) | 762 | if (likely(!context)) |
716 | return; | 763 | return; |
@@ -719,8 +766,9 @@ void audit_free(struct task_struct *tsk) | |||
719 | * function (e.g., exit_group), then free context block. | 766 | * function (e.g., exit_group), then free context block. |
720 | * We use GFP_ATOMIC here because we might be doing this | 767 | * We use GFP_ATOMIC here because we might be doing this |
721 | * in the context of the idle thread */ | 768 | * in the context of the idle thread */ |
769 | /* that can happen only if we are called from do_exit() */ | ||
722 | if (context->in_syscall && context->auditable) | 770 | if (context->in_syscall && context->auditable) |
723 | audit_log_exit(context, GFP_ATOMIC); | 771 | audit_log_exit(context, tsk); |
724 | 772 | ||
725 | audit_free_context(context); | 773 | audit_free_context(context); |
726 | } | 774 | } |
@@ -743,10 +791,11 @@ void audit_free(struct task_struct *tsk) | |||
743 | * will only be written if another part of the kernel requests that it | 791 | * will only be written if another part of the kernel requests that it |
744 | * be written). | 792 | * be written). |
745 | */ | 793 | */ |
746 | void audit_syscall_entry(struct task_struct *tsk, int arch, int major, | 794 | void audit_syscall_entry(int arch, int major, |
747 | unsigned long a1, unsigned long a2, | 795 | unsigned long a1, unsigned long a2, |
748 | unsigned long a3, unsigned long a4) | 796 | unsigned long a3, unsigned long a4) |
749 | { | 797 | { |
798 | struct task_struct *tsk = current; | ||
750 | struct audit_context *context = tsk->audit_context; | 799 | struct audit_context *context = tsk->audit_context; |
751 | enum audit_state state; | 800 | enum audit_state state; |
752 | 801 | ||
@@ -824,22 +873,18 @@ void audit_syscall_entry(struct task_struct *tsk, int arch, int major, | |||
824 | * message), then write out the syscall information. In call cases, | 873 | * message), then write out the syscall information. In call cases, |
825 | * free the names stored from getname(). | 874 | * free the names stored from getname(). |
826 | */ | 875 | */ |
827 | void audit_syscall_exit(struct task_struct *tsk, int valid, long return_code) | 876 | void audit_syscall_exit(int valid, long return_code) |
828 | { | 877 | { |
878 | struct task_struct *tsk = current; | ||
829 | struct audit_context *context; | 879 | struct audit_context *context; |
830 | 880 | ||
831 | get_task_struct(tsk); | ||
832 | task_lock(tsk); | ||
833 | context = audit_get_context(tsk, valid, return_code); | 881 | context = audit_get_context(tsk, valid, return_code); |
834 | task_unlock(tsk); | ||
835 | 882 | ||
836 | /* Not having a context here is ok, since the parent may have | ||
837 | * called __put_task_struct. */ | ||
838 | if (likely(!context)) | 883 | if (likely(!context)) |
839 | goto out; | 884 | return; |
840 | 885 | ||
841 | if (context->in_syscall && context->auditable) | 886 | if (context->in_syscall && context->auditable) |
842 | audit_log_exit(context, GFP_KERNEL); | 887 | audit_log_exit(context, tsk); |
843 | 888 | ||
844 | context->in_syscall = 0; | 889 | context->in_syscall = 0; |
845 | context->auditable = 0; | 890 | context->auditable = 0; |
@@ -854,8 +899,6 @@ void audit_syscall_exit(struct task_struct *tsk, int valid, long return_code) | |||
854 | audit_free_aux(context); | 899 | audit_free_aux(context); |
855 | tsk->audit_context = context; | 900 | tsk->audit_context = context; |
856 | } | 901 | } |
857 | out: | ||
858 | put_task_struct(tsk); | ||
859 | } | 902 | } |
860 | 903 | ||
861 | /** | 904 | /** |
@@ -936,40 +979,11 @@ void audit_putname(const char *name) | |||
936 | #endif | 979 | #endif |
937 | } | 980 | } |
938 | 981 | ||
939 | void audit_inode_context(int idx, const struct inode *inode) | 982 | static void audit_inode_context(int idx, const struct inode *inode) |
940 | { | 983 | { |
941 | struct audit_context *context = current->audit_context; | 984 | struct audit_context *context = current->audit_context; |
942 | const char *suffix = security_inode_xattr_getsuffix(); | ||
943 | char *ctx = NULL; | ||
944 | int len = 0; | ||
945 | |||
946 | if (!suffix) | ||
947 | goto ret; | ||
948 | |||
949 | len = security_inode_getsecurity(inode, suffix, NULL, 0, 0); | ||
950 | if (len == -EOPNOTSUPP) | ||
951 | goto ret; | ||
952 | if (len < 0) | ||
953 | goto error_path; | ||
954 | |||
955 | ctx = kmalloc(len, GFP_KERNEL); | ||
956 | if (!ctx) | ||
957 | goto error_path; | ||
958 | 985 | ||
959 | len = security_inode_getsecurity(inode, suffix, ctx, len, 0); | 986 | selinux_get_inode_sid(inode, &context->names[idx].osid); |
960 | if (len < 0) | ||
961 | goto error_path; | ||
962 | |||
963 | kfree(context->names[idx].ctx); | ||
964 | context->names[idx].ctx = ctx; | ||
965 | goto ret; | ||
966 | |||
967 | error_path: | ||
968 | if (ctx) | ||
969 | kfree(ctx); | ||
970 | audit_panic("error in audit_inode_context"); | ||
971 | ret: | ||
972 | return; | ||
973 | } | 987 | } |
974 | 988 | ||
975 | 989 | ||
@@ -1155,40 +1169,37 @@ uid_t audit_get_loginuid(struct audit_context *ctx) | |||
1155 | return ctx ? ctx->loginuid : -1; | 1169 | return ctx ? ctx->loginuid : -1; |
1156 | } | 1170 | } |
1157 | 1171 | ||
1158 | static char *audit_ipc_context(struct kern_ipc_perm *ipcp) | 1172 | /** |
1173 | * audit_ipc_obj - record audit data for ipc object | ||
1174 | * @ipcp: ipc permissions | ||
1175 | * | ||
1176 | * Returns 0 for success or NULL context or < 0 on error. | ||
1177 | */ | ||
1178 | int audit_ipc_obj(struct kern_ipc_perm *ipcp) | ||
1159 | { | 1179 | { |
1180 | struct audit_aux_data_ipcctl *ax; | ||
1160 | struct audit_context *context = current->audit_context; | 1181 | struct audit_context *context = current->audit_context; |
1161 | char *ctx = NULL; | ||
1162 | int len = 0; | ||
1163 | 1182 | ||
1164 | if (likely(!context)) | 1183 | if (likely(!context)) |
1165 | return NULL; | 1184 | return 0; |
1166 | |||
1167 | len = security_ipc_getsecurity(ipcp, NULL, 0); | ||
1168 | if (len == -EOPNOTSUPP) | ||
1169 | goto ret; | ||
1170 | if (len < 0) | ||
1171 | goto error_path; | ||
1172 | |||
1173 | ctx = kmalloc(len, GFP_ATOMIC); | ||
1174 | if (!ctx) | ||
1175 | goto error_path; | ||
1176 | 1185 | ||
1177 | len = security_ipc_getsecurity(ipcp, ctx, len); | 1186 | ax = kmalloc(sizeof(*ax), GFP_ATOMIC); |
1178 | if (len < 0) | 1187 | if (!ax) |
1179 | goto error_path; | 1188 | return -ENOMEM; |
1180 | 1189 | ||
1181 | return ctx; | 1190 | ax->uid = ipcp->uid; |
1191 | ax->gid = ipcp->gid; | ||
1192 | ax->mode = ipcp->mode; | ||
1193 | selinux_get_ipc_sid(ipcp, &ax->osid); | ||
1182 | 1194 | ||
1183 | error_path: | 1195 | ax->d.type = AUDIT_IPC; |
1184 | kfree(ctx); | 1196 | ax->d.next = context->aux; |
1185 | audit_panic("error in audit_ipc_context"); | 1197 | context->aux = (void *)ax; |
1186 | ret: | 1198 | return 0; |
1187 | return NULL; | ||
1188 | } | 1199 | } |
1189 | 1200 | ||
1190 | /** | 1201 | /** |
1191 | * audit_ipc_perms - record audit data for ipc | 1202 | * audit_ipc_set_perm - record audit data for new ipc permissions |
1192 | * @qbytes: msgq bytes | 1203 | * @qbytes: msgq bytes |
1193 | * @uid: msgq user id | 1204 | * @uid: msgq user id |
1194 | * @gid: msgq group id | 1205 | * @gid: msgq group id |
@@ -1196,7 +1207,7 @@ ret: | |||
1196 | * | 1207 | * |
1197 | * Returns 0 for success or NULL context or < 0 on error. | 1208 | * Returns 0 for success or NULL context or < 0 on error. |
1198 | */ | 1209 | */ |
1199 | int audit_ipc_perms(unsigned long qbytes, uid_t uid, gid_t gid, mode_t mode, struct kern_ipc_perm *ipcp) | 1210 | int audit_ipc_set_perm(unsigned long qbytes, uid_t uid, gid_t gid, mode_t mode, struct kern_ipc_perm *ipcp) |
1200 | { | 1211 | { |
1201 | struct audit_aux_data_ipcctl *ax; | 1212 | struct audit_aux_data_ipcctl *ax; |
1202 | struct audit_context *context = current->audit_context; | 1213 | struct audit_context *context = current->audit_context; |
@@ -1212,9 +1223,9 @@ int audit_ipc_perms(unsigned long qbytes, uid_t uid, gid_t gid, mode_t mode, str | |||
1212 | ax->uid = uid; | 1223 | ax->uid = uid; |
1213 | ax->gid = gid; | 1224 | ax->gid = gid; |
1214 | ax->mode = mode; | 1225 | ax->mode = mode; |
1215 | ax->ctx = audit_ipc_context(ipcp); | 1226 | selinux_get_ipc_sid(ipcp, &ax->osid); |
1216 | 1227 | ||
1217 | ax->d.type = AUDIT_IPC; | 1228 | ax->d.type = AUDIT_IPC_SET_PERM; |
1218 | ax->d.next = context->aux; | 1229 | ax->d.next = context->aux; |
1219 | context->aux = (void *)ax; | 1230 | context->aux = (void *)ax; |
1220 | return 0; | 1231 | return 0; |
diff --git a/kernel/cpuset.c b/kernel/cpuset.c index 72248d1b9e3f..ab81fdd4572b 100644 --- a/kernel/cpuset.c +++ b/kernel/cpuset.c | |||
@@ -2231,19 +2231,25 @@ static const struct cpuset *nearest_exclusive_ancestor(const struct cpuset *cs) | |||
2231 | * So only GFP_KERNEL allocations, if all nodes in the cpuset are | 2231 | * So only GFP_KERNEL allocations, if all nodes in the cpuset are |
2232 | * short of memory, might require taking the callback_mutex mutex. | 2232 | * short of memory, might require taking the callback_mutex mutex. |
2233 | * | 2233 | * |
2234 | * The first loop over the zonelist in mm/page_alloc.c:__alloc_pages() | 2234 | * The first call here from mm/page_alloc:get_page_from_freelist() |
2235 | * calls here with __GFP_HARDWALL always set in gfp_mask, enforcing | 2235 | * has __GFP_HARDWALL set in gfp_mask, enforcing hardwall cpusets, so |
2236 | * hardwall cpusets - no allocation on a node outside the cpuset is | 2236 | * no allocation on a node outside the cpuset is allowed (unless in |
2237 | * allowed (unless in interrupt, of course). | 2237 | * interrupt, of course). |
2238 | * | 2238 | * |
2239 | * The second loop doesn't even call here for GFP_ATOMIC requests | 2239 | * The second pass through get_page_from_freelist() doesn't even call |
2240 | * (if the __alloc_pages() local variable 'wait' is set). That check | 2240 | * here for GFP_ATOMIC calls. For those calls, the __alloc_pages() |
2241 | * and the checks below have the combined affect in the second loop of | 2241 | * variable 'wait' is not set, and the bit ALLOC_CPUSET is not set |
2242 | * the __alloc_pages() routine that: | 2242 | * in alloc_flags. That logic and the checks below have the combined |
2243 | * affect that: | ||
2243 | * in_interrupt - any node ok (current task context irrelevant) | 2244 | * in_interrupt - any node ok (current task context irrelevant) |
2244 | * GFP_ATOMIC - any node ok | 2245 | * GFP_ATOMIC - any node ok |
2245 | * GFP_KERNEL - any node in enclosing mem_exclusive cpuset ok | 2246 | * GFP_KERNEL - any node in enclosing mem_exclusive cpuset ok |
2246 | * GFP_USER - only nodes in current tasks mems allowed ok. | 2247 | * GFP_USER - only nodes in current tasks mems allowed ok. |
2248 | * | ||
2249 | * Rule: | ||
2250 | * Don't call cpuset_zone_allowed() if you can't sleep, unless you | ||
2251 | * pass in the __GFP_HARDWALL flag set in gfp_flag, which disables | ||
2252 | * the code that might scan up ancestor cpusets and sleep. | ||
2247 | **/ | 2253 | **/ |
2248 | 2254 | ||
2249 | int __cpuset_zone_allowed(struct zone *z, gfp_t gfp_mask) | 2255 | int __cpuset_zone_allowed(struct zone *z, gfp_t gfp_mask) |
@@ -2255,6 +2261,7 @@ int __cpuset_zone_allowed(struct zone *z, gfp_t gfp_mask) | |||
2255 | if (in_interrupt()) | 2261 | if (in_interrupt()) |
2256 | return 1; | 2262 | return 1; |
2257 | node = z->zone_pgdat->node_id; | 2263 | node = z->zone_pgdat->node_id; |
2264 | might_sleep_if(!(gfp_mask & __GFP_HARDWALL)); | ||
2258 | if (node_isset(node, current->mems_allowed)) | 2265 | if (node_isset(node, current->mems_allowed)) |
2259 | return 1; | 2266 | return 1; |
2260 | if (gfp_mask & __GFP_HARDWALL) /* If hardwall request, stop here */ | 2267 | if (gfp_mask & __GFP_HARDWALL) /* If hardwall request, stop here */ |
diff --git a/kernel/exit.c b/kernel/exit.c index 1a9787ac6173..e95b93282210 100644 --- a/kernel/exit.c +++ b/kernel/exit.c | |||
@@ -35,6 +35,7 @@ | |||
35 | #include <linux/futex.h> | 35 | #include <linux/futex.h> |
36 | #include <linux/compat.h> | 36 | #include <linux/compat.h> |
37 | #include <linux/pipe_fs_i.h> | 37 | #include <linux/pipe_fs_i.h> |
38 | #include <linux/audit.h> /* for audit_free() */ | ||
38 | 39 | ||
39 | #include <asm/uaccess.h> | 40 | #include <asm/uaccess.h> |
40 | #include <asm/unistd.h> | 41 | #include <asm/unistd.h> |
@@ -56,7 +57,7 @@ static void __unhash_process(struct task_struct *p) | |||
56 | detach_pid(p, PIDTYPE_PGID); | 57 | detach_pid(p, PIDTYPE_PGID); |
57 | detach_pid(p, PIDTYPE_SID); | 58 | detach_pid(p, PIDTYPE_SID); |
58 | 59 | ||
59 | list_del_init(&p->tasks); | 60 | list_del_rcu(&p->tasks); |
60 | __get_cpu_var(process_counts)--; | 61 | __get_cpu_var(process_counts)--; |
61 | } | 62 | } |
62 | list_del_rcu(&p->thread_group); | 63 | list_del_rcu(&p->thread_group); |
@@ -910,6 +911,8 @@ fastcall NORET_TYPE void do_exit(long code) | |||
910 | if (unlikely(tsk->compat_robust_list)) | 911 | if (unlikely(tsk->compat_robust_list)) |
911 | compat_exit_robust_list(tsk); | 912 | compat_exit_robust_list(tsk); |
912 | #endif | 913 | #endif |
914 | if (unlikely(tsk->audit_context)) | ||
915 | audit_free(tsk); | ||
913 | exit_mm(tsk); | 916 | exit_mm(tsk); |
914 | 917 | ||
915 | exit_sem(tsk); | 918 | exit_sem(tsk); |
diff --git a/kernel/extable.c b/kernel/extable.c index 7501b531ceed..7fe262855317 100644 --- a/kernel/extable.c +++ b/kernel/extable.c | |||
@@ -40,7 +40,7 @@ const struct exception_table_entry *search_exception_tables(unsigned long addr) | |||
40 | return e; | 40 | return e; |
41 | } | 41 | } |
42 | 42 | ||
43 | static int core_kernel_text(unsigned long addr) | 43 | int core_kernel_text(unsigned long addr) |
44 | { | 44 | { |
45 | if (addr >= (unsigned long)_stext && | 45 | if (addr >= (unsigned long)_stext && |
46 | addr <= (unsigned long)_etext) | 46 | addr <= (unsigned long)_etext) |
diff --git a/kernel/fork.c b/kernel/fork.c index 3384eb89cb1c..ac8100e3088a 100644 --- a/kernel/fork.c +++ b/kernel/fork.c | |||
@@ -114,8 +114,6 @@ void __put_task_struct(struct task_struct *tsk) | |||
114 | WARN_ON(atomic_read(&tsk->usage)); | 114 | WARN_ON(atomic_read(&tsk->usage)); |
115 | WARN_ON(tsk == current); | 115 | WARN_ON(tsk == current); |
116 | 116 | ||
117 | if (unlikely(tsk->audit_context)) | ||
118 | audit_free(tsk); | ||
119 | security_task_free(tsk); | 117 | security_task_free(tsk); |
120 | free_uid(tsk->user); | 118 | free_uid(tsk->user); |
121 | put_group_info(tsk->group_info); | 119 | put_group_info(tsk->group_info); |
@@ -124,12 +122,6 @@ void __put_task_struct(struct task_struct *tsk) | |||
124 | free_task(tsk); | 122 | free_task(tsk); |
125 | } | 123 | } |
126 | 124 | ||
127 | void __put_task_struct_cb(struct rcu_head *rhp) | ||
128 | { | ||
129 | struct task_struct *tsk = container_of(rhp, struct task_struct, rcu); | ||
130 | __put_task_struct(tsk); | ||
131 | } | ||
132 | |||
133 | void __init fork_init(unsigned long mempages) | 125 | void __init fork_init(unsigned long mempages) |
134 | { | 126 | { |
135 | #ifndef __HAVE_ARCH_TASK_STRUCT_ALLOCATOR | 127 | #ifndef __HAVE_ARCH_TASK_STRUCT_ALLOCATOR |
@@ -186,6 +178,7 @@ static struct task_struct *dup_task_struct(struct task_struct *orig) | |||
186 | atomic_set(&tsk->usage,2); | 178 | atomic_set(&tsk->usage,2); |
187 | atomic_set(&tsk->fs_excl, 0); | 179 | atomic_set(&tsk->fs_excl, 0); |
188 | tsk->btrace_seq = 0; | 180 | tsk->btrace_seq = 0; |
181 | tsk->splice_pipe = NULL; | ||
189 | return tsk; | 182 | return tsk; |
190 | } | 183 | } |
191 | 184 | ||
@@ -1210,7 +1203,7 @@ static task_t *copy_process(unsigned long clone_flags, | |||
1210 | attach_pid(p, PIDTYPE_PGID, process_group(p)); | 1203 | attach_pid(p, PIDTYPE_PGID, process_group(p)); |
1211 | attach_pid(p, PIDTYPE_SID, p->signal->session); | 1204 | attach_pid(p, PIDTYPE_SID, p->signal->session); |
1212 | 1205 | ||
1213 | list_add_tail(&p->tasks, &init_task.tasks); | 1206 | list_add_tail_rcu(&p->tasks, &init_task.tasks); |
1214 | __get_cpu_var(process_counts)++; | 1207 | __get_cpu_var(process_counts)++; |
1215 | } | 1208 | } |
1216 | attach_pid(p, PIDTYPE_PID, p->pid); | 1209 | attach_pid(p, PIDTYPE_PID, p->pid); |
diff --git a/kernel/hrtimer.c b/kernel/hrtimer.c index d2a7296c8251..01fa2ae98a85 100644 --- a/kernel/hrtimer.c +++ b/kernel/hrtimer.c | |||
@@ -456,6 +456,7 @@ hrtimer_start(struct hrtimer *timer, ktime_t tim, const enum hrtimer_mode mode) | |||
456 | 456 | ||
457 | return ret; | 457 | return ret; |
458 | } | 458 | } |
459 | EXPORT_SYMBOL_GPL(hrtimer_start); | ||
459 | 460 | ||
460 | /** | 461 | /** |
461 | * hrtimer_try_to_cancel - try to deactivate a timer | 462 | * hrtimer_try_to_cancel - try to deactivate a timer |
@@ -484,6 +485,7 @@ int hrtimer_try_to_cancel(struct hrtimer *timer) | |||
484 | return ret; | 485 | return ret; |
485 | 486 | ||
486 | } | 487 | } |
488 | EXPORT_SYMBOL_GPL(hrtimer_try_to_cancel); | ||
487 | 489 | ||
488 | /** | 490 | /** |
489 | * hrtimer_cancel - cancel a timer and wait for the handler to finish. | 491 | * hrtimer_cancel - cancel a timer and wait for the handler to finish. |
@@ -504,6 +506,7 @@ int hrtimer_cancel(struct hrtimer *timer) | |||
504 | cpu_relax(); | 506 | cpu_relax(); |
505 | } | 507 | } |
506 | } | 508 | } |
509 | EXPORT_SYMBOL_GPL(hrtimer_cancel); | ||
507 | 510 | ||
508 | /** | 511 | /** |
509 | * hrtimer_get_remaining - get remaining time for the timer | 512 | * hrtimer_get_remaining - get remaining time for the timer |
@@ -522,6 +525,7 @@ ktime_t hrtimer_get_remaining(const struct hrtimer *timer) | |||
522 | 525 | ||
523 | return rem; | 526 | return rem; |
524 | } | 527 | } |
528 | EXPORT_SYMBOL_GPL(hrtimer_get_remaining); | ||
525 | 529 | ||
526 | #ifdef CONFIG_NO_IDLE_HZ | 530 | #ifdef CONFIG_NO_IDLE_HZ |
527 | /** | 531 | /** |
@@ -580,6 +584,7 @@ void hrtimer_init(struct hrtimer *timer, clockid_t clock_id, | |||
580 | timer->base = &bases[clock_id]; | 584 | timer->base = &bases[clock_id]; |
581 | timer->node.rb_parent = HRTIMER_INACTIVE; | 585 | timer->node.rb_parent = HRTIMER_INACTIVE; |
582 | } | 586 | } |
587 | EXPORT_SYMBOL_GPL(hrtimer_init); | ||
583 | 588 | ||
584 | /** | 589 | /** |
585 | * hrtimer_get_res - get the timer resolution for a clock | 590 | * hrtimer_get_res - get the timer resolution for a clock |
@@ -599,6 +604,7 @@ int hrtimer_get_res(const clockid_t which_clock, struct timespec *tp) | |||
599 | 604 | ||
600 | return 0; | 605 | return 0; |
601 | } | 606 | } |
607 | EXPORT_SYMBOL_GPL(hrtimer_get_res); | ||
602 | 608 | ||
603 | /* | 609 | /* |
604 | * Expire the per base hrtimer-queue: | 610 | * Expire the per base hrtimer-queue: |
@@ -836,7 +842,7 @@ static void migrate_hrtimers(int cpu) | |||
836 | } | 842 | } |
837 | #endif /* CONFIG_HOTPLUG_CPU */ | 843 | #endif /* CONFIG_HOTPLUG_CPU */ |
838 | 844 | ||
839 | static int __devinit hrtimer_cpu_notify(struct notifier_block *self, | 845 | static int hrtimer_cpu_notify(struct notifier_block *self, |
840 | unsigned long action, void *hcpu) | 846 | unsigned long action, void *hcpu) |
841 | { | 847 | { |
842 | long cpu = (long)hcpu; | 848 | long cpu = (long)hcpu; |
@@ -860,7 +866,7 @@ static int __devinit hrtimer_cpu_notify(struct notifier_block *self, | |||
860 | return NOTIFY_OK; | 866 | return NOTIFY_OK; |
861 | } | 867 | } |
862 | 868 | ||
863 | static struct notifier_block __devinitdata hrtimers_nb = { | 869 | static struct notifier_block hrtimers_nb = { |
864 | .notifier_call = hrtimer_cpu_notify, | 870 | .notifier_call = hrtimer_cpu_notify, |
865 | }; | 871 | }; |
866 | 872 | ||
diff --git a/kernel/irq/manage.c b/kernel/irq/manage.c index ac766ad573e8..1279e3499534 100644 --- a/kernel/irq/manage.c +++ b/kernel/irq/manage.c | |||
@@ -246,8 +246,10 @@ int setup_irq(unsigned int irq, struct irqaction * new) | |||
246 | 246 | ||
247 | mismatch: | 247 | mismatch: |
248 | spin_unlock_irqrestore(&desc->lock, flags); | 248 | spin_unlock_irqrestore(&desc->lock, flags); |
249 | printk(KERN_ERR "%s: irq handler mismatch\n", __FUNCTION__); | 249 | if (!(new->flags & SA_PROBEIRQ)) { |
250 | dump_stack(); | 250 | printk(KERN_ERR "%s: irq handler mismatch\n", __FUNCTION__); |
251 | dump_stack(); | ||
252 | } | ||
251 | return -EBUSY; | 253 | return -EBUSY; |
252 | } | 254 | } |
253 | 255 | ||
diff --git a/kernel/kprobes.c b/kernel/kprobes.c index 1156eb0977d0..1fbf466a29aa 100644 --- a/kernel/kprobes.c +++ b/kernel/kprobes.c | |||
@@ -585,6 +585,9 @@ int __kprobes register_kretprobe(struct kretprobe *rp) | |||
585 | int i; | 585 | int i; |
586 | 586 | ||
587 | rp->kp.pre_handler = pre_handler_kretprobe; | 587 | rp->kp.pre_handler = pre_handler_kretprobe; |
588 | rp->kp.post_handler = NULL; | ||
589 | rp->kp.fault_handler = NULL; | ||
590 | rp->kp.break_handler = NULL; | ||
588 | 591 | ||
589 | /* Pre-allocate memory for max kretprobe instances */ | 592 | /* Pre-allocate memory for max kretprobe instances */ |
590 | if (rp->maxactive <= 0) { | 593 | if (rp->maxactive <= 0) { |
diff --git a/kernel/module.c b/kernel/module.c index d24deb0dbbc9..bbe04862e1b0 100644 --- a/kernel/module.c +++ b/kernel/module.c | |||
@@ -705,14 +705,14 @@ EXPORT_SYMBOL(__symbol_put); | |||
705 | 705 | ||
706 | void symbol_put_addr(void *addr) | 706 | void symbol_put_addr(void *addr) |
707 | { | 707 | { |
708 | unsigned long flags; | 708 | struct module *modaddr; |
709 | 709 | ||
710 | spin_lock_irqsave(&modlist_lock, flags); | 710 | if (core_kernel_text((unsigned long)addr)) |
711 | if (!kernel_text_address((unsigned long)addr)) | 711 | return; |
712 | BUG(); | ||
713 | 712 | ||
714 | module_put(module_text_address((unsigned long)addr)); | 713 | if (!(modaddr = module_text_address((unsigned long)addr))) |
715 | spin_unlock_irqrestore(&modlist_lock, flags); | 714 | BUG(); |
715 | module_put(modaddr); | ||
716 | } | 716 | } |
717 | EXPORT_SYMBOL_GPL(symbol_put_addr); | 717 | EXPORT_SYMBOL_GPL(symbol_put_addr); |
718 | 718 | ||
diff --git a/kernel/power/main.c b/kernel/power/main.c index ee371f50ccaa..a6d9ef46009e 100644 --- a/kernel/power/main.c +++ b/kernel/power/main.c | |||
@@ -272,7 +272,7 @@ static ssize_t state_store(struct subsystem * subsys, const char * buf, size_t n | |||
272 | if (*s && !strncmp(buf, *s, len)) | 272 | if (*s && !strncmp(buf, *s, len)) |
273 | break; | 273 | break; |
274 | } | 274 | } |
275 | if (*s) | 275 | if (state < PM_SUSPEND_MAX && *s) |
276 | error = enter_state(state); | 276 | error = enter_state(state); |
277 | else | 277 | else |
278 | error = -EINVAL; | 278 | error = -EINVAL; |
diff --git a/kernel/power/pm.c b/kernel/power/pm.c index 0f6908cce1dd..84063ac8fcfc 100644 --- a/kernel/power/pm.c +++ b/kernel/power/pm.c | |||
@@ -75,25 +75,6 @@ struct pm_dev *pm_register(pm_dev_t type, | |||
75 | return dev; | 75 | return dev; |
76 | } | 76 | } |
77 | 77 | ||
78 | /** | ||
79 | * pm_unregister - unregister a device with power management | ||
80 | * @dev: device to unregister | ||
81 | * | ||
82 | * Remove a device from the power management notification lists. The | ||
83 | * dev passed must be a handle previously returned by pm_register. | ||
84 | */ | ||
85 | |||
86 | void pm_unregister(struct pm_dev *dev) | ||
87 | { | ||
88 | if (dev) { | ||
89 | mutex_lock(&pm_devs_lock); | ||
90 | list_del(&dev->entry); | ||
91 | mutex_unlock(&pm_devs_lock); | ||
92 | |||
93 | kfree(dev); | ||
94 | } | ||
95 | } | ||
96 | |||
97 | static void __pm_unregister(struct pm_dev *dev) | 78 | static void __pm_unregister(struct pm_dev *dev) |
98 | { | 79 | { |
99 | if (dev) { | 80 | if (dev) { |
@@ -258,7 +239,6 @@ int pm_send_all(pm_request_t rqst, void *data) | |||
258 | } | 239 | } |
259 | 240 | ||
260 | EXPORT_SYMBOL(pm_register); | 241 | EXPORT_SYMBOL(pm_register); |
261 | EXPORT_SYMBOL(pm_unregister); | ||
262 | EXPORT_SYMBOL(pm_unregister_all); | 242 | EXPORT_SYMBOL(pm_unregister_all); |
263 | EXPORT_SYMBOL(pm_send_all); | 243 | EXPORT_SYMBOL(pm_send_all); |
264 | EXPORT_SYMBOL(pm_active); | 244 | EXPORT_SYMBOL(pm_active); |
diff --git a/kernel/power/snapshot.c b/kernel/power/snapshot.c index c5863d02c89e..3eeedbb13b78 100644 --- a/kernel/power/snapshot.c +++ b/kernel/power/snapshot.c | |||
@@ -240,14 +240,15 @@ static void copy_data_pages(struct pbe *pblist) | |||
240 | * free_pagedir - free pages allocated with alloc_pagedir() | 240 | * free_pagedir - free pages allocated with alloc_pagedir() |
241 | */ | 241 | */ |
242 | 242 | ||
243 | static void free_pagedir(struct pbe *pblist) | 243 | static void free_pagedir(struct pbe *pblist, int clear_nosave_free) |
244 | { | 244 | { |
245 | struct pbe *pbe; | 245 | struct pbe *pbe; |
246 | 246 | ||
247 | while (pblist) { | 247 | while (pblist) { |
248 | pbe = (pblist + PB_PAGE_SKIP)->next; | 248 | pbe = (pblist + PB_PAGE_SKIP)->next; |
249 | ClearPageNosave(virt_to_page(pblist)); | 249 | ClearPageNosave(virt_to_page(pblist)); |
250 | ClearPageNosaveFree(virt_to_page(pblist)); | 250 | if (clear_nosave_free) |
251 | ClearPageNosaveFree(virt_to_page(pblist)); | ||
251 | free_page((unsigned long)pblist); | 252 | free_page((unsigned long)pblist); |
252 | pblist = pbe; | 253 | pblist = pbe; |
253 | } | 254 | } |
@@ -389,7 +390,7 @@ struct pbe *alloc_pagedir(unsigned int nr_pages, gfp_t gfp_mask, int safe_needed | |||
389 | pbe->next = alloc_image_page(gfp_mask, safe_needed); | 390 | pbe->next = alloc_image_page(gfp_mask, safe_needed); |
390 | } | 391 | } |
391 | if (!pbe) { /* get_zeroed_page() failed */ | 392 | if (!pbe) { /* get_zeroed_page() failed */ |
392 | free_pagedir(pblist); | 393 | free_pagedir(pblist, 1); |
393 | pblist = NULL; | 394 | pblist = NULL; |
394 | } else | 395 | } else |
395 | create_pbe_list(pblist, nr_pages); | 396 | create_pbe_list(pblist, nr_pages); |
@@ -736,7 +737,7 @@ static int create_image(struct snapshot_handle *handle) | |||
736 | pblist = alloc_pagedir(nr_copy_pages, GFP_ATOMIC, 1); | 737 | pblist = alloc_pagedir(nr_copy_pages, GFP_ATOMIC, 1); |
737 | if (pblist) | 738 | if (pblist) |
738 | copy_page_backup_list(pblist, p); | 739 | copy_page_backup_list(pblist, p); |
739 | free_pagedir(p); | 740 | free_pagedir(p, 0); |
740 | if (!pblist) | 741 | if (!pblist) |
741 | error = -ENOMEM; | 742 | error = -ENOMEM; |
742 | } | 743 | } |
diff --git a/kernel/profile.c b/kernel/profile.c index 5a730fdb1a2c..68afe121e507 100644 --- a/kernel/profile.c +++ b/kernel/profile.c | |||
@@ -299,7 +299,7 @@ out: | |||
299 | } | 299 | } |
300 | 300 | ||
301 | #ifdef CONFIG_HOTPLUG_CPU | 301 | #ifdef CONFIG_HOTPLUG_CPU |
302 | static int __devinit profile_cpu_callback(struct notifier_block *info, | 302 | static int profile_cpu_callback(struct notifier_block *info, |
303 | unsigned long action, void *__cpu) | 303 | unsigned long action, void *__cpu) |
304 | { | 304 | { |
305 | int node, cpu = (unsigned long)__cpu; | 305 | int node, cpu = (unsigned long)__cpu; |
diff --git a/kernel/ptrace.c b/kernel/ptrace.c index 0eeb7e66722c..921c22ad16e4 100644 --- a/kernel/ptrace.c +++ b/kernel/ptrace.c | |||
@@ -56,10 +56,6 @@ void ptrace_untrace(task_t *child) | |||
56 | signal_wake_up(child, 1); | 56 | signal_wake_up(child, 1); |
57 | } | 57 | } |
58 | } | 58 | } |
59 | if (child->signal->flags & SIGNAL_GROUP_EXIT) { | ||
60 | sigaddset(&child->pending.signal, SIGKILL); | ||
61 | signal_wake_up(child, 1); | ||
62 | } | ||
63 | spin_unlock(&child->sighand->siglock); | 59 | spin_unlock(&child->sighand->siglock); |
64 | } | 60 | } |
65 | 61 | ||
@@ -81,7 +77,8 @@ void __ptrace_unlink(task_t *child) | |||
81 | add_parent(child); | 77 | add_parent(child); |
82 | } | 78 | } |
83 | 79 | ||
84 | ptrace_untrace(child); | 80 | if (child->state == TASK_TRACED) |
81 | ptrace_untrace(child); | ||
85 | } | 82 | } |
86 | 83 | ||
87 | /* | 84 | /* |
@@ -151,12 +148,34 @@ int ptrace_may_attach(struct task_struct *task) | |||
151 | int ptrace_attach(struct task_struct *task) | 148 | int ptrace_attach(struct task_struct *task) |
152 | { | 149 | { |
153 | int retval; | 150 | int retval; |
154 | task_lock(task); | 151 | |
155 | retval = -EPERM; | 152 | retval = -EPERM; |
156 | if (task->pid <= 1) | 153 | if (task->pid <= 1) |
157 | goto bad; | 154 | goto out; |
158 | if (task->tgid == current->tgid) | 155 | if (task->tgid == current->tgid) |
159 | goto bad; | 156 | goto out; |
157 | |||
158 | repeat: | ||
159 | /* | ||
160 | * Nasty, nasty. | ||
161 | * | ||
162 | * We want to hold both the task-lock and the | ||
163 | * tasklist_lock for writing at the same time. | ||
164 | * But that's against the rules (tasklist_lock | ||
165 | * is taken for reading by interrupts on other | ||
166 | * cpu's that may have task_lock). | ||
167 | */ | ||
168 | task_lock(task); | ||
169 | local_irq_disable(); | ||
170 | if (!write_trylock(&tasklist_lock)) { | ||
171 | local_irq_enable(); | ||
172 | task_unlock(task); | ||
173 | do { | ||
174 | cpu_relax(); | ||
175 | } while (!write_can_lock(&tasklist_lock)); | ||
176 | goto repeat; | ||
177 | } | ||
178 | |||
160 | /* the same process cannot be attached many times */ | 179 | /* the same process cannot be attached many times */ |
161 | if (task->ptrace & PT_PTRACED) | 180 | if (task->ptrace & PT_PTRACED) |
162 | goto bad; | 181 | goto bad; |
@@ -169,17 +188,15 @@ int ptrace_attach(struct task_struct *task) | |||
169 | ? PT_ATTACHED : 0); | 188 | ? PT_ATTACHED : 0); |
170 | if (capable(CAP_SYS_PTRACE)) | 189 | if (capable(CAP_SYS_PTRACE)) |
171 | task->ptrace |= PT_PTRACE_CAP; | 190 | task->ptrace |= PT_PTRACE_CAP; |
172 | task_unlock(task); | ||
173 | 191 | ||
174 | write_lock_irq(&tasklist_lock); | ||
175 | __ptrace_link(task, current); | 192 | __ptrace_link(task, current); |
176 | write_unlock_irq(&tasklist_lock); | ||
177 | 193 | ||
178 | force_sig_specific(SIGSTOP, task); | 194 | force_sig_specific(SIGSTOP, task); |
179 | return 0; | ||
180 | 195 | ||
181 | bad: | 196 | bad: |
197 | write_unlock_irq(&tasklist_lock); | ||
182 | task_unlock(task); | 198 | task_unlock(task); |
199 | out: | ||
183 | return retval; | 200 | return retval; |
184 | } | 201 | } |
185 | 202 | ||
@@ -420,21 +437,22 @@ int ptrace_request(struct task_struct *child, long request, | |||
420 | */ | 437 | */ |
421 | int ptrace_traceme(void) | 438 | int ptrace_traceme(void) |
422 | { | 439 | { |
423 | int ret; | 440 | int ret = -EPERM; |
424 | 441 | ||
425 | /* | 442 | /* |
426 | * Are we already being traced? | 443 | * Are we already being traced? |
427 | */ | 444 | */ |
428 | if (current->ptrace & PT_PTRACED) | 445 | task_lock(current); |
429 | return -EPERM; | 446 | if (!(current->ptrace & PT_PTRACED)) { |
430 | ret = security_ptrace(current->parent, current); | 447 | ret = security_ptrace(current->parent, current); |
431 | if (ret) | 448 | /* |
432 | return -EPERM; | 449 | * Set the ptrace bit in the process ptrace flags. |
433 | /* | 450 | */ |
434 | * Set the ptrace bit in the process ptrace flags. | 451 | if (!ret) |
435 | */ | 452 | current->ptrace |= PT_PTRACED; |
436 | current->ptrace |= PT_PTRACED; | 453 | } |
437 | return 0; | 454 | task_unlock(current); |
455 | return ret; | ||
438 | } | 456 | } |
439 | 457 | ||
440 | /** | 458 | /** |
diff --git a/kernel/rcupdate.c b/kernel/rcupdate.c index 13458bbaa1be..2058f88c7bbb 100644 --- a/kernel/rcupdate.c +++ b/kernel/rcupdate.c | |||
@@ -479,12 +479,31 @@ static int __rcu_pending(struct rcu_ctrlblk *rcp, struct rcu_data *rdp) | |||
479 | return 0; | 479 | return 0; |
480 | } | 480 | } |
481 | 481 | ||
482 | /* | ||
483 | * Check to see if there is any immediate RCU-related work to be done | ||
484 | * by the current CPU, returning 1 if so. This function is part of the | ||
485 | * RCU implementation; it is -not- an exported member of the RCU API. | ||
486 | */ | ||
482 | int rcu_pending(int cpu) | 487 | int rcu_pending(int cpu) |
483 | { | 488 | { |
484 | return __rcu_pending(&rcu_ctrlblk, &per_cpu(rcu_data, cpu)) || | 489 | return __rcu_pending(&rcu_ctrlblk, &per_cpu(rcu_data, cpu)) || |
485 | __rcu_pending(&rcu_bh_ctrlblk, &per_cpu(rcu_bh_data, cpu)); | 490 | __rcu_pending(&rcu_bh_ctrlblk, &per_cpu(rcu_bh_data, cpu)); |
486 | } | 491 | } |
487 | 492 | ||
493 | /* | ||
494 | * Check to see if any future RCU-related work will need to be done | ||
495 | * by the current CPU, even if none need be done immediately, returning | ||
496 | * 1 if so. This function is part of the RCU implementation; it is -not- | ||
497 | * an exported member of the RCU API. | ||
498 | */ | ||
499 | int rcu_needs_cpu(int cpu) | ||
500 | { | ||
501 | struct rcu_data *rdp = &per_cpu(rcu_data, cpu); | ||
502 | struct rcu_data *rdp_bh = &per_cpu(rcu_bh_data, cpu); | ||
503 | |||
504 | return (!!rdp->curlist || !!rdp_bh->curlist || rcu_pending(cpu)); | ||
505 | } | ||
506 | |||
488 | void rcu_check_callbacks(int cpu, int user) | 507 | void rcu_check_callbacks(int cpu, int user) |
489 | { | 508 | { |
490 | if (user || | 509 | if (user || |
@@ -520,7 +539,7 @@ static void __devinit rcu_online_cpu(int cpu) | |||
520 | tasklet_init(&per_cpu(rcu_tasklet, cpu), rcu_process_callbacks, 0UL); | 539 | tasklet_init(&per_cpu(rcu_tasklet, cpu), rcu_process_callbacks, 0UL); |
521 | } | 540 | } |
522 | 541 | ||
523 | static int __devinit rcu_cpu_notify(struct notifier_block *self, | 542 | static int rcu_cpu_notify(struct notifier_block *self, |
524 | unsigned long action, void *hcpu) | 543 | unsigned long action, void *hcpu) |
525 | { | 544 | { |
526 | long cpu = (long)hcpu; | 545 | long cpu = (long)hcpu; |
@@ -537,7 +556,7 @@ static int __devinit rcu_cpu_notify(struct notifier_block *self, | |||
537 | return NOTIFY_OK; | 556 | return NOTIFY_OK; |
538 | } | 557 | } |
539 | 558 | ||
540 | static struct notifier_block __devinitdata rcu_nb = { | 559 | static struct notifier_block rcu_nb = { |
541 | .notifier_call = rcu_cpu_notify, | 560 | .notifier_call = rcu_cpu_notify, |
542 | }; | 561 | }; |
543 | 562 | ||
diff --git a/kernel/sched.c b/kernel/sched.c index 365f0b90b4de..c13f1bd2df7d 100644 --- a/kernel/sched.c +++ b/kernel/sched.c | |||
@@ -665,55 +665,13 @@ static int effective_prio(task_t *p) | |||
665 | } | 665 | } |
666 | 666 | ||
667 | /* | 667 | /* |
668 | * We place interactive tasks back into the active array, if possible. | ||
669 | * | ||
670 | * To guarantee that this does not starve expired tasks we ignore the | ||
671 | * interactivity of a task if the first expired task had to wait more | ||
672 | * than a 'reasonable' amount of time. This deadline timeout is | ||
673 | * load-dependent, as the frequency of array switched decreases with | ||
674 | * increasing number of running tasks. We also ignore the interactivity | ||
675 | * if a better static_prio task has expired, and switch periodically | ||
676 | * regardless, to ensure that highly interactive tasks do not starve | ||
677 | * the less fortunate for unreasonably long periods. | ||
678 | */ | ||
679 | static inline int expired_starving(runqueue_t *rq) | ||
680 | { | ||
681 | int limit; | ||
682 | |||
683 | /* | ||
684 | * Arrays were recently switched, all is well | ||
685 | */ | ||
686 | if (!rq->expired_timestamp) | ||
687 | return 0; | ||
688 | |||
689 | limit = STARVATION_LIMIT * rq->nr_running; | ||
690 | |||
691 | /* | ||
692 | * It's time to switch arrays | ||
693 | */ | ||
694 | if (jiffies - rq->expired_timestamp >= limit) | ||
695 | return 1; | ||
696 | |||
697 | /* | ||
698 | * There's a better selection in the expired array | ||
699 | */ | ||
700 | if (rq->curr->static_prio > rq->best_expired_prio) | ||
701 | return 1; | ||
702 | |||
703 | /* | ||
704 | * All is well | ||
705 | */ | ||
706 | return 0; | ||
707 | } | ||
708 | |||
709 | /* | ||
710 | * __activate_task - move a task to the runqueue. | 668 | * __activate_task - move a task to the runqueue. |
711 | */ | 669 | */ |
712 | static void __activate_task(task_t *p, runqueue_t *rq) | 670 | static void __activate_task(task_t *p, runqueue_t *rq) |
713 | { | 671 | { |
714 | prio_array_t *target = rq->active; | 672 | prio_array_t *target = rq->active; |
715 | 673 | ||
716 | if (unlikely(batch_task(p) || (expired_starving(rq) && !rt_task(p)))) | 674 | if (batch_task(p)) |
717 | target = rq->expired; | 675 | target = rq->expired; |
718 | enqueue_task(p, target); | 676 | enqueue_task(p, target); |
719 | rq->nr_running++; | 677 | rq->nr_running++; |
@@ -2532,6 +2490,22 @@ unsigned long long current_sched_time(const task_t *tsk) | |||
2532 | } | 2490 | } |
2533 | 2491 | ||
2534 | /* | 2492 | /* |
2493 | * We place interactive tasks back into the active array, if possible. | ||
2494 | * | ||
2495 | * To guarantee that this does not starve expired tasks we ignore the | ||
2496 | * interactivity of a task if the first expired task had to wait more | ||
2497 | * than a 'reasonable' amount of time. This deadline timeout is | ||
2498 | * load-dependent, as the frequency of array switched decreases with | ||
2499 | * increasing number of running tasks. We also ignore the interactivity | ||
2500 | * if a better static_prio task has expired: | ||
2501 | */ | ||
2502 | #define EXPIRED_STARVING(rq) \ | ||
2503 | ((STARVATION_LIMIT && ((rq)->expired_timestamp && \ | ||
2504 | (jiffies - (rq)->expired_timestamp >= \ | ||
2505 | STARVATION_LIMIT * ((rq)->nr_running) + 1))) || \ | ||
2506 | ((rq)->curr->static_prio > (rq)->best_expired_prio)) | ||
2507 | |||
2508 | /* | ||
2535 | * Account user cpu time to a process. | 2509 | * Account user cpu time to a process. |
2536 | * @p: the process that the cpu time gets accounted to | 2510 | * @p: the process that the cpu time gets accounted to |
2537 | * @hardirq_offset: the offset to subtract from hardirq_count() | 2511 | * @hardirq_offset: the offset to subtract from hardirq_count() |
@@ -2666,7 +2640,7 @@ void scheduler_tick(void) | |||
2666 | 2640 | ||
2667 | if (!rq->expired_timestamp) | 2641 | if (!rq->expired_timestamp) |
2668 | rq->expired_timestamp = jiffies; | 2642 | rq->expired_timestamp = jiffies; |
2669 | if (!TASK_INTERACTIVE(p) || expired_starving(rq)) { | 2643 | if (!TASK_INTERACTIVE(p) || EXPIRED_STARVING(rq)) { |
2670 | enqueue_task(p, rq->expired); | 2644 | enqueue_task(p, rq->expired); |
2671 | if (p->static_prio < rq->best_expired_prio) | 2645 | if (p->static_prio < rq->best_expired_prio) |
2672 | rq->best_expired_prio = p->static_prio; | 2646 | rq->best_expired_prio = p->static_prio; |
@@ -4814,7 +4788,7 @@ static int migration_call(struct notifier_block *nfb, unsigned long action, | |||
4814 | /* Register at highest priority so that task migration (migrate_all_tasks) | 4788 | /* Register at highest priority so that task migration (migrate_all_tasks) |
4815 | * happens before everything else. | 4789 | * happens before everything else. |
4816 | */ | 4790 | */ |
4817 | static struct notifier_block __devinitdata migration_notifier = { | 4791 | static struct notifier_block migration_notifier = { |
4818 | .notifier_call = migration_call, | 4792 | .notifier_call = migration_call, |
4819 | .priority = 10 | 4793 | .priority = 10 |
4820 | }; | 4794 | }; |
diff --git a/kernel/signal.c b/kernel/signal.c index b14f895027c3..e5f8aea78ffe 100644 --- a/kernel/signal.c +++ b/kernel/signal.c | |||
@@ -1754,9 +1754,9 @@ relock: | |||
1754 | /* Let the debugger run. */ | 1754 | /* Let the debugger run. */ |
1755 | ptrace_stop(signr, signr, info); | 1755 | ptrace_stop(signr, signr, info); |
1756 | 1756 | ||
1757 | /* We're back. Did the debugger cancel the sig or group_exit? */ | 1757 | /* We're back. Did the debugger cancel the sig? */ |
1758 | signr = current->exit_code; | 1758 | signr = current->exit_code; |
1759 | if (signr == 0 || current->signal->flags & SIGNAL_GROUP_EXIT) | 1759 | if (signr == 0) |
1760 | continue; | 1760 | continue; |
1761 | 1761 | ||
1762 | current->exit_code = 0; | 1762 | current->exit_code = 0; |
diff --git a/kernel/softirq.c b/kernel/softirq.c index ec8fed42a86f..336f92d64e2e 100644 --- a/kernel/softirq.c +++ b/kernel/softirq.c | |||
@@ -446,7 +446,7 @@ static void takeover_tasklets(unsigned int cpu) | |||
446 | } | 446 | } |
447 | #endif /* CONFIG_HOTPLUG_CPU */ | 447 | #endif /* CONFIG_HOTPLUG_CPU */ |
448 | 448 | ||
449 | static int __devinit cpu_callback(struct notifier_block *nfb, | 449 | static int cpu_callback(struct notifier_block *nfb, |
450 | unsigned long action, | 450 | unsigned long action, |
451 | void *hcpu) | 451 | void *hcpu) |
452 | { | 452 | { |
@@ -484,7 +484,7 @@ static int __devinit cpu_callback(struct notifier_block *nfb, | |||
484 | return NOTIFY_OK; | 484 | return NOTIFY_OK; |
485 | } | 485 | } |
486 | 486 | ||
487 | static struct notifier_block __devinitdata cpu_nfb = { | 487 | static struct notifier_block cpu_nfb = { |
488 | .notifier_call = cpu_callback | 488 | .notifier_call = cpu_callback |
489 | }; | 489 | }; |
490 | 490 | ||
diff --git a/kernel/softlockup.c b/kernel/softlockup.c index ced91e1ff564..14c7faf02909 100644 --- a/kernel/softlockup.c +++ b/kernel/softlockup.c | |||
@@ -104,7 +104,7 @@ static int watchdog(void * __bind_cpu) | |||
104 | /* | 104 | /* |
105 | * Create/destroy watchdog threads as CPUs come and go: | 105 | * Create/destroy watchdog threads as CPUs come and go: |
106 | */ | 106 | */ |
107 | static int __devinit | 107 | static int |
108 | cpu_callback(struct notifier_block *nfb, unsigned long action, void *hcpu) | 108 | cpu_callback(struct notifier_block *nfb, unsigned long action, void *hcpu) |
109 | { | 109 | { |
110 | int hotcpu = (unsigned long)hcpu; | 110 | int hotcpu = (unsigned long)hcpu; |
@@ -140,7 +140,7 @@ cpu_callback(struct notifier_block *nfb, unsigned long action, void *hcpu) | |||
140 | return NOTIFY_OK; | 140 | return NOTIFY_OK; |
141 | } | 141 | } |
142 | 142 | ||
143 | static struct notifier_block __devinitdata cpu_nfb = { | 143 | static struct notifier_block cpu_nfb = { |
144 | .notifier_call = cpu_callback | 144 | .notifier_call = cpu_callback |
145 | }; | 145 | }; |
146 | 146 | ||
diff --git a/kernel/timer.c b/kernel/timer.c index 883773788836..9e49deed468c 100644 --- a/kernel/timer.c +++ b/kernel/timer.c | |||
@@ -541,6 +541,22 @@ found: | |||
541 | } | 541 | } |
542 | spin_unlock(&base->lock); | 542 | spin_unlock(&base->lock); |
543 | 543 | ||
544 | /* | ||
545 | * It can happen that other CPUs service timer IRQs and increment | ||
546 | * jiffies, but we have not yet got a local timer tick to process | ||
547 | * the timer wheels. In that case, the expiry time can be before | ||
548 | * jiffies, but since the high-resolution timer here is relative to | ||
549 | * jiffies, the default expression when high-resolution timers are | ||
550 | * not active, | ||
551 | * | ||
552 | * time_before(MAX_JIFFY_OFFSET + jiffies, expires) | ||
553 | * | ||
554 | * would falsely evaluate to true. If that is the case, just | ||
555 | * return jiffies so that we can immediately fire the local timer | ||
556 | */ | ||
557 | if (time_before(expires, jiffies)) | ||
558 | return jiffies; | ||
559 | |||
544 | if (time_before(hr_expires, expires)) | 560 | if (time_before(hr_expires, expires)) |
545 | return hr_expires; | 561 | return hr_expires; |
546 | 562 | ||
@@ -1314,7 +1330,7 @@ static void __devinit migrate_timers(int cpu) | |||
1314 | } | 1330 | } |
1315 | #endif /* CONFIG_HOTPLUG_CPU */ | 1331 | #endif /* CONFIG_HOTPLUG_CPU */ |
1316 | 1332 | ||
1317 | static int __devinit timer_cpu_notify(struct notifier_block *self, | 1333 | static int timer_cpu_notify(struct notifier_block *self, |
1318 | unsigned long action, void *hcpu) | 1334 | unsigned long action, void *hcpu) |
1319 | { | 1335 | { |
1320 | long cpu = (long)hcpu; | 1336 | long cpu = (long)hcpu; |
@@ -1334,7 +1350,7 @@ static int __devinit timer_cpu_notify(struct notifier_block *self, | |||
1334 | return NOTIFY_OK; | 1350 | return NOTIFY_OK; |
1335 | } | 1351 | } |
1336 | 1352 | ||
1337 | static struct notifier_block __devinitdata timers_nb = { | 1353 | static struct notifier_block timers_nb = { |
1338 | .notifier_call = timer_cpu_notify, | 1354 | .notifier_call = timer_cpu_notify, |
1339 | }; | 1355 | }; |
1340 | 1356 | ||
diff --git a/kernel/uid16.c b/kernel/uid16.c index aa25605027c8..187e2a423878 100644 --- a/kernel/uid16.c +++ b/kernel/uid16.c | |||
@@ -20,43 +20,67 @@ | |||
20 | 20 | ||
21 | asmlinkage long sys_chown16(const char __user * filename, old_uid_t user, old_gid_t group) | 21 | asmlinkage long sys_chown16(const char __user * filename, old_uid_t user, old_gid_t group) |
22 | { | 22 | { |
23 | return sys_chown(filename, low2highuid(user), low2highgid(group)); | 23 | long ret = sys_chown(filename, low2highuid(user), low2highgid(group)); |
24 | /* avoid REGPARM breakage on x86: */ | ||
25 | prevent_tail_call(ret); | ||
26 | return ret; | ||
24 | } | 27 | } |
25 | 28 | ||
26 | asmlinkage long sys_lchown16(const char __user * filename, old_uid_t user, old_gid_t group) | 29 | asmlinkage long sys_lchown16(const char __user * filename, old_uid_t user, old_gid_t group) |
27 | { | 30 | { |
28 | return sys_lchown(filename, low2highuid(user), low2highgid(group)); | 31 | long ret = sys_lchown(filename, low2highuid(user), low2highgid(group)); |
32 | /* avoid REGPARM breakage on x86: */ | ||
33 | prevent_tail_call(ret); | ||
34 | return ret; | ||
29 | } | 35 | } |
30 | 36 | ||
31 | asmlinkage long sys_fchown16(unsigned int fd, old_uid_t user, old_gid_t group) | 37 | asmlinkage long sys_fchown16(unsigned int fd, old_uid_t user, old_gid_t group) |
32 | { | 38 | { |
33 | return sys_fchown(fd, low2highuid(user), low2highgid(group)); | 39 | long ret = sys_fchown(fd, low2highuid(user), low2highgid(group)); |
40 | /* avoid REGPARM breakage on x86: */ | ||
41 | prevent_tail_call(ret); | ||
42 | return ret; | ||
34 | } | 43 | } |
35 | 44 | ||
36 | asmlinkage long sys_setregid16(old_gid_t rgid, old_gid_t egid) | 45 | asmlinkage long sys_setregid16(old_gid_t rgid, old_gid_t egid) |
37 | { | 46 | { |
38 | return sys_setregid(low2highgid(rgid), low2highgid(egid)); | 47 | long ret = sys_setregid(low2highgid(rgid), low2highgid(egid)); |
48 | /* avoid REGPARM breakage on x86: */ | ||
49 | prevent_tail_call(ret); | ||
50 | return ret; | ||
39 | } | 51 | } |
40 | 52 | ||
41 | asmlinkage long sys_setgid16(old_gid_t gid) | 53 | asmlinkage long sys_setgid16(old_gid_t gid) |
42 | { | 54 | { |
43 | return sys_setgid(low2highgid(gid)); | 55 | long ret = sys_setgid(low2highgid(gid)); |
56 | /* avoid REGPARM breakage on x86: */ | ||
57 | prevent_tail_call(ret); | ||
58 | return ret; | ||
44 | } | 59 | } |
45 | 60 | ||
46 | asmlinkage long sys_setreuid16(old_uid_t ruid, old_uid_t euid) | 61 | asmlinkage long sys_setreuid16(old_uid_t ruid, old_uid_t euid) |
47 | { | 62 | { |
48 | return sys_setreuid(low2highuid(ruid), low2highuid(euid)); | 63 | long ret = sys_setreuid(low2highuid(ruid), low2highuid(euid)); |
64 | /* avoid REGPARM breakage on x86: */ | ||
65 | prevent_tail_call(ret); | ||
66 | return ret; | ||
49 | } | 67 | } |
50 | 68 | ||
51 | asmlinkage long sys_setuid16(old_uid_t uid) | 69 | asmlinkage long sys_setuid16(old_uid_t uid) |
52 | { | 70 | { |
53 | return sys_setuid(low2highuid(uid)); | 71 | long ret = sys_setuid(low2highuid(uid)); |
72 | /* avoid REGPARM breakage on x86: */ | ||
73 | prevent_tail_call(ret); | ||
74 | return ret; | ||
54 | } | 75 | } |
55 | 76 | ||
56 | asmlinkage long sys_setresuid16(old_uid_t ruid, old_uid_t euid, old_uid_t suid) | 77 | asmlinkage long sys_setresuid16(old_uid_t ruid, old_uid_t euid, old_uid_t suid) |
57 | { | 78 | { |
58 | return sys_setresuid(low2highuid(ruid), low2highuid(euid), | 79 | long ret = sys_setresuid(low2highuid(ruid), low2highuid(euid), |
59 | low2highuid(suid)); | 80 | low2highuid(suid)); |
81 | /* avoid REGPARM breakage on x86: */ | ||
82 | prevent_tail_call(ret); | ||
83 | return ret; | ||
60 | } | 84 | } |
61 | 85 | ||
62 | asmlinkage long sys_getresuid16(old_uid_t __user *ruid, old_uid_t __user *euid, old_uid_t __user *suid) | 86 | asmlinkage long sys_getresuid16(old_uid_t __user *ruid, old_uid_t __user *euid, old_uid_t __user *suid) |
@@ -72,8 +96,11 @@ asmlinkage long sys_getresuid16(old_uid_t __user *ruid, old_uid_t __user *euid, | |||
72 | 96 | ||
73 | asmlinkage long sys_setresgid16(old_gid_t rgid, old_gid_t egid, old_gid_t sgid) | 97 | asmlinkage long sys_setresgid16(old_gid_t rgid, old_gid_t egid, old_gid_t sgid) |
74 | { | 98 | { |
75 | return sys_setresgid(low2highgid(rgid), low2highgid(egid), | 99 | long ret = sys_setresgid(low2highgid(rgid), low2highgid(egid), |
76 | low2highgid(sgid)); | 100 | low2highgid(sgid)); |
101 | /* avoid REGPARM breakage on x86: */ | ||
102 | prevent_tail_call(ret); | ||
103 | return ret; | ||
77 | } | 104 | } |
78 | 105 | ||
79 | asmlinkage long sys_getresgid16(old_gid_t __user *rgid, old_gid_t __user *egid, old_gid_t __user *sgid) | 106 | asmlinkage long sys_getresgid16(old_gid_t __user *rgid, old_gid_t __user *egid, old_gid_t __user *sgid) |
@@ -89,12 +116,18 @@ asmlinkage long sys_getresgid16(old_gid_t __user *rgid, old_gid_t __user *egid, | |||
89 | 116 | ||
90 | asmlinkage long sys_setfsuid16(old_uid_t uid) | 117 | asmlinkage long sys_setfsuid16(old_uid_t uid) |
91 | { | 118 | { |
92 | return sys_setfsuid(low2highuid(uid)); | 119 | long ret = sys_setfsuid(low2highuid(uid)); |
120 | /* avoid REGPARM breakage on x86: */ | ||
121 | prevent_tail_call(ret); | ||
122 | return ret; | ||
93 | } | 123 | } |
94 | 124 | ||
95 | asmlinkage long sys_setfsgid16(old_gid_t gid) | 125 | asmlinkage long sys_setfsgid16(old_gid_t gid) |
96 | { | 126 | { |
97 | return sys_setfsgid(low2highgid(gid)); | 127 | long ret = sys_setfsgid(low2highgid(gid)); |
128 | /* avoid REGPARM breakage on x86: */ | ||
129 | prevent_tail_call(ret); | ||
130 | return ret; | ||
98 | } | 131 | } |
99 | 132 | ||
100 | static int groups16_to_user(old_gid_t __user *grouplist, | 133 | static int groups16_to_user(old_gid_t __user *grouplist, |
diff --git a/kernel/workqueue.c b/kernel/workqueue.c index e9e464a90376..880fb415a8f6 100644 --- a/kernel/workqueue.c +++ b/kernel/workqueue.c | |||
@@ -547,7 +547,7 @@ static void take_over_work(struct workqueue_struct *wq, unsigned int cpu) | |||
547 | } | 547 | } |
548 | 548 | ||
549 | /* We're holding the cpucontrol mutex here */ | 549 | /* We're holding the cpucontrol mutex here */ |
550 | static int __devinit workqueue_cpu_callback(struct notifier_block *nfb, | 550 | static int workqueue_cpu_callback(struct notifier_block *nfb, |
551 | unsigned long action, | 551 | unsigned long action, |
552 | void *hcpu) | 552 | void *hcpu) |
553 | { | 553 | { |