ioprio.c 4.7 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
/*
 * fs/ioprio.c
 *
 * Copyright (C) 2004 Jens Axboe <axboe@suse.de>
 *
 * Helper functions for setting/querying io priorities of processes. The
 * system calls closely mimmick getpriority/setpriority, see the man page for
 * those. The prio argument is a composite of prio class and prio data, where
 * the data argument has meaning within that class. The standard scheduling
 * classes have 8 distinct prio levels, with 0 being the highest prio and 7
 * being the lowest.
 *
 * IOW, setting BE scheduling class with prio 2 is done ala:
 *
 * unsigned int prio = (IOPRIO_CLASS_BE << IOPRIO_CLASS_SHIFT) | 2;
 *
 * ioprio_set(PRIO_PROCESS, pid, prio);
 *
 * See also Documentation/block/ioprio.txt
 *
 */
#include <linux/kernel.h>
#include <linux/ioprio.h>
#include <linux/blkdev.h>
25
#include <linux/capability.h>
26
#include <linux/syscalls.h>
27
#include <linux/security.h>
28 29 30

static int set_task_ioprio(struct task_struct *task, int ioprio)
{
31
	int err;
32 33 34 35 36 37
	struct io_context *ioc;

	if (task->uid != current->euid &&
	    task->uid != current->uid && !capable(CAP_SYS_NICE))
		return -EPERM;

38 39 40 41
	err = security_task_setioprio(task, ioprio);
	if (err)
		return err;

42 43 44 45 46
	task_lock(task);

	task->ioprio = ioprio;

	ioc = task->io_context;
47 48 49
	/* see wmb() in current_io_context() */
	smp_read_barrier_depends();

50 51
	if (ioc)
		ioc->ioprio_changed = 1;
52 53 54 55 56

	task_unlock(task);
	return 0;
}

57
asmlinkage long sys_ioprio_set(int which, int who, int ioprio)
58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75
{
	int class = IOPRIO_PRIO_CLASS(ioprio);
	int data = IOPRIO_PRIO_DATA(ioprio);
	struct task_struct *p, *g;
	struct user_struct *user;
	int ret;

	switch (class) {
		case IOPRIO_CLASS_RT:
			if (!capable(CAP_SYS_ADMIN))
				return -EPERM;
			/* fall through, rt has prio field too */
		case IOPRIO_CLASS_BE:
			if (data >= IOPRIO_BE_NR || data < 0)
				return -EINVAL;

			break;
		case IOPRIO_CLASS_IDLE:
76 77
			if (!capable(CAP_SYS_ADMIN))
				return -EPERM;
78 79 80 81 82 83
			break;
		default:
			return -EINVAL;
	}

	ret = -ESRCH;
84 85 86 87 88 89
	/*
	 * We want IOPRIO_WHO_PGRP/IOPRIO_WHO_USER to be "atomic",
	 * so we can't use rcu_read_lock(). See re-copy of ->ioprio
	 * in copy_process().
	 */
	read_lock(&tasklist_lock);
90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121
	switch (which) {
		case IOPRIO_WHO_PROCESS:
			if (!who)
				p = current;
			else
				p = find_task_by_pid(who);
			if (p)
				ret = set_task_ioprio(p, ioprio);
			break;
		case IOPRIO_WHO_PGRP:
			if (!who)
				who = process_group(current);
			do_each_task_pid(who, PIDTYPE_PGID, p) {
				ret = set_task_ioprio(p, ioprio);
				if (ret)
					break;
			} while_each_task_pid(who, PIDTYPE_PGID, p);
			break;
		case IOPRIO_WHO_USER:
			if (!who)
				user = current->user;
			else
				user = find_user(who);

			if (!user)
				break;

			do_each_thread(g, p) {
				if (p->uid != who)
					continue;
				ret = set_task_ioprio(p, ioprio);
				if (ret)
122
					goto free_uid;
123
			} while_each_thread(g, p);
124
free_uid:
125 126 127 128 129 130 131
			if (who)
				free_uid(user);
			break;
		default:
			ret = -EINVAL;
	}

132
	read_unlock(&tasklist_lock);
133 134 135
	return ret;
}

136 137 138 139 140 141 142 143 144 145 146 147
static int get_task_ioprio(struct task_struct *p)
{
	int ret;

	ret = security_task_getioprio(p);
	if (ret)
		goto out;
	ret = p->ioprio;
out:
	return ret;
}

O
Oleg Nesterov 已提交
148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170
int ioprio_best(unsigned short aprio, unsigned short bprio)
{
	unsigned short aclass = IOPRIO_PRIO_CLASS(aprio);
	unsigned short bclass = IOPRIO_PRIO_CLASS(bprio);

	if (!ioprio_valid(aprio))
		return bprio;
	if (!ioprio_valid(bprio))
		return aprio;

	if (aclass == IOPRIO_CLASS_NONE)
		aclass = IOPRIO_CLASS_BE;
	if (bclass == IOPRIO_CLASS_NONE)
		bclass = IOPRIO_CLASS_BE;

	if (aclass == bclass)
		return min(aprio, bprio);
	if (aclass > bclass)
		return bprio;
	else
		return aprio;
}

171
asmlinkage long sys_ioprio_get(int which, int who)
172 173 174 175
{
	struct task_struct *g, *p;
	struct user_struct *user;
	int ret = -ESRCH;
176
	int tmpio;
177

178
	read_lock(&tasklist_lock);
179 180 181 182 183 184 185
	switch (which) {
		case IOPRIO_WHO_PROCESS:
			if (!who)
				p = current;
			else
				p = find_task_by_pid(who);
			if (p)
186
				ret = get_task_ioprio(p);
187 188 189 190 191
			break;
		case IOPRIO_WHO_PGRP:
			if (!who)
				who = process_group(current);
			do_each_task_pid(who, PIDTYPE_PGID, p) {
192 193 194
				tmpio = get_task_ioprio(p);
				if (tmpio < 0)
					continue;
195
				if (ret == -ESRCH)
196
					ret = tmpio;
197
				else
198
					ret = ioprio_best(ret, tmpio);
199 200 201 202 203 204 205 206 207 208 209 210 211 212
			} while_each_task_pid(who, PIDTYPE_PGID, p);
			break;
		case IOPRIO_WHO_USER:
			if (!who)
				user = current->user;
			else
				user = find_user(who);

			if (!user)
				break;

			do_each_thread(g, p) {
				if (p->uid != user->uid)
					continue;
213 214 215
				tmpio = get_task_ioprio(p);
				if (tmpio < 0)
					continue;
216
				if (ret == -ESRCH)
217
					ret = tmpio;
218
				else
219
					ret = ioprio_best(ret, tmpio);
220 221 222 223 224 225 226 227 228
			} while_each_thread(g, p);

			if (who)
				free_uid(user);
			break;
		default:
			ret = -EINVAL;
	}

229
	read_unlock(&tasklist_lock);
230 231 232
	return ret;
}