PRIO_PGRP needs the tasklist_lock mainly to serialize vs setpgid(2), to protect against any concurrent change_pid(PIDTYPE_PGID) that can move the task from one hlist to another while iterating. However, the remaining can only rely only on RCU: PRIO_PROCESS only does the task lookup and never iterates over tasklist and we already have an rcu-aware stable pointer. PRIO_USER is already racy vs setuid(2) so with creds being rcu protected, we can end up seeing stale data. When removing the tasklist_lock there can be a race with (i) fork but this is benign as the child's nice is inherited and the new task is not observable by the user yet either, hence the return semantics do not differ. And (ii) a race with exit, which is a small window and can cause us to miss a task which was removed from the list and it had the highest nice. Similarly change the buggy do_each_thread/while_each_thread combo in PRIO_USER for the rcu-safe for_each_process_thread flavor, which doesn't make use of next_thread/p->thread_group. [akpm@linux-foundation.org: coding style fixes]
Unified · Split
| @@ -220,7 +220,6 @@ SYSCALL_DEFINE3(setpriority, int, which, int, who, int, niceval) · 219 unmodified lines | |||
| 220 | niceval = MAX_NICE; | 220 | niceval = MAX_NICE; |
| 221 | 221 | ||
| 222 | rcu_read_lock(); | 222 | rcu_read_lock(); |
| 223 | read_lock(&tasklist_lock); | ||
| 224 | switch (which) { | 223 | switch (which) { |
| 225 | case PRIO_PROCESS: | 224 | case PRIO_PROCESS: |
| 226 | if (who) | 225 | if (who) |
| @@ -235,9 +234,11 @@ SYSCALL_DEFINE3(setpriority, int, which, int, who, int, niceval) · 8 unmodified lines | |||
| 235 | pgrp = find_vpid(who); | 234 | pgrp = find_vpid(who); |
| 236 | else | 235 | else |
| 237 | pgrp = task_pgrp(current); | 236 | pgrp = task_pgrp(current); |
| 237 | read_lock(&tasklist_lock); | ||
| 238 | do_each_pid_thread(pgrp, PIDTYPE_PGID, p) { | 238 | do_each_pid_thread(pgrp, PIDTYPE_PGID, p) { |
| 239 | error = set_one_prio(p, niceval, error); | 239 | error = set_one_prio(p, niceval, error); |
| 240 | } while_each_pid_thread(pgrp, PIDTYPE_PGID, p); | 240 | } while_each_pid_thread(pgrp, PIDTYPE_PGID, p); |
| 241 | read_unlock(&tasklist_lock); | ||
| 241 | break; | 242 | break; |
| 242 | case PRIO_USER: | 243 | case PRIO_USER: |
| 243 | uid = make_kuid(cred->user_ns, who); | 244 | uid = make_kuid(cred->user_ns, who); |
| @@ -249,16 +250,15 @@ SYSCALL_DEFINE3(setpriority, int, which, int, who, int, niceval) · 5 unmodified lines | |||
| 249 | if (!user) | 250 | if (!user) |
| 250 | goto out_unlock; /* No processes for this user */ | 251 | goto out_unlock; /* No processes for this user */ |
| 251 | } | 252 | } |
| 252 | do_each_thread(g, p) { | 253 | for_each_process_thread(g, p) { |
| 253 | if (uid_eq(task_uid(p), uid) && task_pid_vnr(p)) | 254 | if (uid_eq(task_uid(p), uid) && task_pid_vnr(p)) |
| 254 | error = set_one_prio(p, niceval, error); | 255 | error = set_one_prio(p, niceval, error); |
| 255 | } while_each_thread(g, p); | 256 | } |
| 256 | if (!uid_eq(uid, cred->uid)) | 257 | if (!uid_eq(uid, cred->uid)) |
| 257 | free_uid(user); /* For find_user() */ | 258 | free_uid(user); /* For find_user() */ |
| 258 | break; | 259 | break; |
| 259 | } | 260 | } |
| 260 | out_unlock: | 261 | out_unlock: |
| 261 | read_unlock(&tasklist_lock); | ||
| 262 | rcu_read_unlock(); | 262 | rcu_read_unlock(); |
| 263 | out: | 263 | out: |
| 264 | return error; | 264 | return error; |
| @@ -283,7 +283,6 @@ SYSCALL_DEFINE2(getpriority, int, which, int, who) · 18 unmodified lines | |||
| 283 | return -EINVAL; | 283 | return -EINVAL; |
| 284 | 284 | ||
| 285 | rcu_read_lock(); | 285 | rcu_read_lock(); |
| 286 | read_lock(&tasklist_lock); | ||
| 287 | switch (which) { | 286 | switch (which) { |
| 288 | case PRIO_PROCESS: | 287 | case PRIO_PROCESS: |
| 289 | if (who) | 288 | if (who) |
| @@ -301,11 +300,13 @@ SYSCALL_DEFINE2(getpriority, int, which, int, who) · 11 unmodified lines | |||
| 301 | pgrp = find_vpid(who); | 300 | pgrp = find_vpid(who); |
| 302 | else | 301 | else |
| 303 | pgrp = task_pgrp(current); | 302 | pgrp = task_pgrp(current); |
| 303 | read_lock(&tasklist_lock); | ||
| 304 | do_each_pid_thread(pgrp, PIDTYPE_PGID, p) { | 304 | do_each_pid_thread(pgrp, PIDTYPE_PGID, p) { |
| 305 | niceval = nice_to_rlimit(task_nice(p)); | 305 | niceval = nice_to_rlimit(task_nice(p)); |
| 306 | if (niceval > retval) | 306 | if (niceval > retval) |
| 307 | retval = niceval; | 307 | retval = niceval; |
| 308 | } while_each_pid_thread(pgrp, PIDTYPE_PGID, p); | 308 | } while_each_pid_thread(pgrp, PIDTYPE_PGID, p); |
| 309 | read_unlock(&tasklist_lock); | ||
| 309 | break; | 310 | break; |
| 310 | case PRIO_USER: | 311 | case PRIO_USER: |
| 311 | uid = make_kuid(cred->user_ns, who); | 312 | uid = make_kuid(cred->user_ns, who); |
| @@ -317,19 +318,18 @@ SYSCALL_DEFINE2(getpriority, int, which, int, who) · 5 unmodified lines | |||
| 317 | if (!user) | 318 | if (!user) |
| 318 | goto out_unlock; /* No processes for this user */ | 319 | goto out_unlock; /* No processes for this user */ |
| 319 | } | 320 | } |
| 320 | do_each_thread(g, p) { | 321 | for_each_process_thread(g, p) { |
| 321 | if (uid_eq(task_uid(p), uid) && task_pid_vnr(p)) { | 322 | if (uid_eq(task_uid(p), uid) && task_pid_vnr(p)) { |
| 322 | niceval = nice_to_rlimit(task_nice(p)); | 323 | niceval = nice_to_rlimit(task_nice(p)); |
| 323 | if (niceval > retval) | 324 | if (niceval > retval) |
| 324 | retval = niceval; | 325 | retval = niceval; |
| 325 | } | 326 | } |
| 326 | } while_each_thread(g, p); | 327 | } |
| 327 | if (!uid_eq(uid, cred->uid)) | 328 | if (!uid_eq(uid, cred->uid)) |
| 328 | free_uid(user); /* for find_user() */ | 329 | free_uid(user); /* for find_user() */ |
| 329 | break; | 330 | break; |
| 330 | } | 331 | } |
| 331 | out_unlock: | 332 | out_unlock: |
| 332 | read_unlock(&tasklist_lock); | ||
| 333 | rcu_read_unlock(); | 333 | rcu_read_unlock(); |
| 334 | 334 | ||
| 335 | return retval; | 335 | return retval; |