[mirror_ubuntu-bionic-kernel.git] / fs / ioprio.c

/*
 * fs/ioprio.c
 *
 * Copyright (C) 2004 Jens Axboe <axboe@kernel.dk>
 *
 * 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>
#include <linux/capability.h>
#include <linux/syscalls.h>
#include <linux/security.h>
#include <linux/pid_namespace.h>

static int set_task_ioprio(struct task_struct *task, int ioprio)
{
	int err;
	struct io_context *ioc;
	const struct cred *cred = current_cred(), *tcred;

	rcu_read_lock();
	tcred = __task_cred(task);
	if (tcred->uid != cred->euid &&
	    tcred->uid != cred->uid && !capable(CAP_SYS_NICE)) {
		rcu_read_unlock();
		return -EPERM;
	}
	rcu_read_unlock();

	err = security_task_setioprio(task, ioprio);
	if (err)
		return err;

	task_lock(task);
	do {
		ioc = task->io_context;
		/* see wmb() in current_io_context() */
		smp_read_barrier_depends();
		if (ioc)
			break;

		ioc = alloc_io_context(GFP_ATOMIC, -1);
		if (!ioc) {
			err = -ENOMEM;
			break;
		}
		task->io_context = ioc;
	} while (1);

	if (!err) {
		ioc->ioprio = ioprio;
		ioc->ioprio_changed = 1;
	}

	task_unlock(task);
	return err;
}

asmlinkage long sys_ioprio_set(int which, int who, int ioprio)
{
	int class = IOPRIO_PRIO_CLASS(ioprio);
	int data = IOPRIO_PRIO_DATA(ioprio);
	struct task_struct *p, *g;
	struct user_struct *user;
	struct pid *pgrp;
	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:
			break;
		case IOPRIO_CLASS_NONE:
			if (data)
				return -EINVAL;
			break;
		default:
			return -EINVAL;
	}

	ret = -ESRCH;
	/*
	 * 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);
	switch (which) {
		case IOPRIO_WHO_PROCESS:
			if (!who)
				p = current;
			else
				p = find_task_by_vpid(who);
			if (p)
				ret = set_task_ioprio(p, ioprio);
			break;
		case IOPRIO_WHO_PGRP:
			if (!who)
				pgrp = task_pgrp(current);
			else
				pgrp = find_vpid(who);
			do_each_pid_thread(pgrp, PIDTYPE_PGID, p) {
				ret = set_task_ioprio(p, ioprio);
				if (ret)
					break;
			} while_each_pid_thread(pgrp, 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 (__task_cred(p)->uid != who)
					continue;
				ret = set_task_ioprio(p, ioprio);
				if (ret)
					goto free_uid;
			} while_each_thread(g, p);
free_uid:
			if (who)
				free_uid(user);
			break;
		default:
			ret = -EINVAL;
	}

	read_unlock(&tasklist_lock);
	return ret;
}

static int get_task_ioprio(struct task_struct *p)
{
	int ret;

	ret = security_task_getioprio(p);
	if (ret)
		goto out;
	ret = IOPRIO_PRIO_VALUE(IOPRIO_CLASS_NONE, IOPRIO_NORM);
	if (p->io_context)
		ret = p->io_context->ioprio;
out:
	return ret;
}

int ioprio_best(unsigned short aprio, unsigned short bprio)
{
	unsigned short aclass = IOPRIO_PRIO_CLASS(aprio);
	unsigned short bclass = IOPRIO_PRIO_CLASS(bprio);

	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;
}

asmlinkage long sys_ioprio_get(int which, int who)
{
	struct task_struct *g, *p;
	struct user_struct *user;
	struct pid *pgrp;
	int ret = -ESRCH;
	int tmpio;

	read_lock(&tasklist_lock);
	switch (which) {
		case IOPRIO_WHO_PROCESS:
			if (!who)
				p = current;
			else
				p = find_task_by_vpid(who);
			if (p)
				ret = get_task_ioprio(p);
			break;
		case IOPRIO_WHO_PGRP:
			if (!who)
				pgrp = task_pgrp(current);
			else
				pgrp = find_vpid(who);
			do_each_pid_thread(pgrp, PIDTYPE_PGID, p) {
				tmpio = get_task_ioprio(p);
				if (tmpio < 0)
					continue;
				if (ret == -ESRCH)
					ret = tmpio;
				else
					ret = ioprio_best(ret, tmpio);
			} while_each_pid_thread(pgrp, 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 (__task_cred(p)->uid != user->uid)
					continue;
				tmpio = get_task_ioprio(p);
				if (tmpio < 0)
					continue;
				if (ret == -ESRCH)
					ret = tmpio;
				else
					ret = ioprio_best(ret, tmpio);
			} while_each_thread(g, p);

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

	read_unlock(&tasklist_lock);
	return ret;
}
Commit	Line	Data
	1	/*
	2	* fs/ioprio.c
	3	*
	4	* Copyright (C) 2004 Jens Axboe <axboe@kernel.dk>
	5	*
	6	* Helper functions for setting/querying io priorities of processes. The
	7	* system calls closely mimmick getpriority/setpriority, see the man page for
	8	* those. The prio argument is a composite of prio class and prio data, where
	9	* the data argument has meaning within that class. The standard scheduling
	10	* classes have 8 distinct prio levels, with 0 being the highest prio and 7
	11	* being the lowest.
	12	*
	13	* IOW, setting BE scheduling class with prio 2 is done ala:
	14	*
	15	* unsigned int prio = (IOPRIO_CLASS_BE << IOPRIO_CLASS_SHIFT) \| 2;
	16	*
	17	* ioprio_set(PRIO_PROCESS, pid, prio);
	18	*
	19	* See also Documentation/block/ioprio.txt
	20	*
	21	*/
	22	#include <linux/kernel.h>
	23	#include <linux/ioprio.h>
	24	#include <linux/blkdev.h>
	25	#include <linux/capability.h>
	26	#include <linux/syscalls.h>
	27	#include <linux/security.h>
	28	#include <linux/pid_namespace.h>
	29
	30	static int set_task_ioprio(struct task_struct *task, int ioprio)
	31	{
	32	int err;
	33	struct io_context *ioc;
	34	const struct cred cred = current_cred(), tcred;
	35
	36	rcu_read_lock();
	37	tcred = __task_cred(task);
	38	if (tcred->uid != cred->euid &&
	39	tcred->uid != cred->uid && !capable(CAP_SYS_NICE)) {
	40	rcu_read_unlock();
	41	return -EPERM;
	42	}
	43	rcu_read_unlock();
	44
	45	err = security_task_setioprio(task, ioprio);
	46	if (err)
	47	return err;
	48
	49	task_lock(task);
	50	do {
	51	ioc = task->io_context;
	52	/* see wmb() in current_io_context() */
	53	smp_read_barrier_depends();
	54	if (ioc)
	55	break;
	56
	57	ioc = alloc_io_context(GFP_ATOMIC, -1);
	58	if (!ioc) {
	59	err = -ENOMEM;
	60	break;
	61	}
	62	task->io_context = ioc;
	63	} while (1);
	64
	65	if (!err) {
	66	ioc->ioprio = ioprio;
	67	ioc->ioprio_changed = 1;
	68	}
	69
	70	task_unlock(task);
	71	return err;
	72	}
	73
	74	asmlinkage long sys_ioprio_set(int which, int who, int ioprio)
	75	{
	76	int class = IOPRIO_PRIO_CLASS(ioprio);
	77	int data = IOPRIO_PRIO_DATA(ioprio);
	78	struct task_struct p, g;
	79	struct user_struct *user;
	80	struct pid *pgrp;
	81	int ret;
	82
	83	switch (class) {
	84	case IOPRIO_CLASS_RT:
	85	if (!capable(CAP_SYS_ADMIN))
	86	return -EPERM;
	87	/* fall through, rt has prio field too */
	88	case IOPRIO_CLASS_BE:
	89	if (data >= IOPRIO_BE_NR \|\| data < 0)
	90	return -EINVAL;
	91
	92	break;
	93	case IOPRIO_CLASS_IDLE:
	94	break;
	95	case IOPRIO_CLASS_NONE:
	96	if (data)
	97	return -EINVAL;
	98	break;
	99	default:
	100	return -EINVAL;
	101	}
	102
	103	ret = -ESRCH;
	104	/*
	105	* We want IOPRIO_WHO_PGRP/IOPRIO_WHO_USER to be "atomic",
	106	* so we can't use rcu_read_lock(). See re-copy of ->ioprio
	107	* in copy_process().
	108	*/
	109	read_lock(&tasklist_lock);
	110	switch (which) {
	111	case IOPRIO_WHO_PROCESS:
	112	if (!who)
	113	p = current;
	114	else
	115	p = find_task_by_vpid(who);
	116	if (p)
	117	ret = set_task_ioprio(p, ioprio);
	118	break;
	119	case IOPRIO_WHO_PGRP:
	120	if (!who)
	121	pgrp = task_pgrp(current);
	122	else
	123	pgrp = find_vpid(who);
	124	do_each_pid_thread(pgrp, PIDTYPE_PGID, p) {
	125	ret = set_task_ioprio(p, ioprio);
	126	if (ret)
	127	break;
	128	} while_each_pid_thread(pgrp, PIDTYPE_PGID, p);
	129	break;
	130	case IOPRIO_WHO_USER:
	131	if (!who)
	132	user = current_user();
	133	else
	134	user = find_user(who);
	135
	136	if (!user)
	137	break;
	138
	139	do_each_thread(g, p) {
	140	if (__task_cred(p)->uid != who)
	141	continue;
	142	ret = set_task_ioprio(p, ioprio);
	143	if (ret)
	144	goto free_uid;
	145	} while_each_thread(g, p);
	146	free_uid:
	147	if (who)
	148	free_uid(user);
	149	break;
	150	default:
	151	ret = -EINVAL;
	152	}
	153
	154	read_unlock(&tasklist_lock);
	155	return ret;
	156	}
	157
	158	static int get_task_ioprio(struct task_struct *p)
	159	{
	160	int ret;
	161
	162	ret = security_task_getioprio(p);
	163	if (ret)
	164	goto out;
	165	ret = IOPRIO_PRIO_VALUE(IOPRIO_CLASS_NONE, IOPRIO_NORM);
	166	if (p->io_context)
	167	ret = p->io_context->ioprio;
	168	out:
	169	return ret;
	170	}
	171
	172	int ioprio_best(unsigned short aprio, unsigned short bprio)
	173	{
	174	unsigned short aclass = IOPRIO_PRIO_CLASS(aprio);
	175	unsigned short bclass = IOPRIO_PRIO_CLASS(bprio);
	176
	177	if (aclass == IOPRIO_CLASS_NONE)
	178	aclass = IOPRIO_CLASS_BE;
	179	if (bclass == IOPRIO_CLASS_NONE)
	180	bclass = IOPRIO_CLASS_BE;
	181
	182	if (aclass == bclass)
	183	return min(aprio, bprio);
	184	if (aclass > bclass)
	185	return bprio;
	186	else
	187	return aprio;
	188	}
	189
	190	asmlinkage long sys_ioprio_get(int which, int who)
	191	{
	192	struct task_struct g, p;
	193	struct user_struct *user;
	194	struct pid *pgrp;
	195	int ret = -ESRCH;
	196	int tmpio;
	197
	198	read_lock(&tasklist_lock);
	199	switch (which) {
	200	case IOPRIO_WHO_PROCESS:
	201	if (!who)
	202	p = current;
	203	else
	204	p = find_task_by_vpid(who);
	205	if (p)
	206	ret = get_task_ioprio(p);
	207	break;
	208	case IOPRIO_WHO_PGRP:
	209	if (!who)
	210	pgrp = task_pgrp(current);
	211	else
	212	pgrp = find_vpid(who);
	213	do_each_pid_thread(pgrp, PIDTYPE_PGID, p) {
	214	tmpio = get_task_ioprio(p);
	215	if (tmpio < 0)
	216	continue;
	217	if (ret == -ESRCH)
	218	ret = tmpio;
	219	else
	220	ret = ioprio_best(ret, tmpio);
	221	} while_each_pid_thread(pgrp, PIDTYPE_PGID, p);
	222	break;
	223	case IOPRIO_WHO_USER:
	224	if (!who)
	225	user = current_user();
	226	else
	227	user = find_user(who);
	228
	229	if (!user)
	230	break;
	231
	232	do_each_thread(g, p) {
	233	if (__task_cred(p)->uid != user->uid)
	234	continue;
	235	tmpio = get_task_ioprio(p);
	236	if (tmpio < 0)
	237	continue;
	238	if (ret == -ESRCH)
	239	ret = tmpio;
	240	else
	241	ret = ioprio_best(ret, tmpio);
	242	} while_each_thread(g, p);
	243
	244	if (who)
	245	free_uid(user);
	246	break;
	247	default:
	248	ret = -EINVAL;
	249	}
	250
	251	read_unlock(&tasklist_lock);
	252	return ret;
	253	}
	254