Adding upstream version 2.33.1.upstream/2.33.1upstream

Signed-off-by: Daniel Baumann <daniel.baumann@progress-linux.org>

7 files changed, 1550 insertions, 0 deletions

diff --git a/schedutils/Makemodule.am b/schedutils/Makemodule.am
new file mode 100644
index 0000000..f32d2b3
--- /dev/null
+++ b/schedutils/Makemodule.am
@@ -0,0 +1,20 @@
+if BUILD_CHRT
+usrbin_exec_PROGRAMS += chrt
+dist_man_MANS += schedutils/chrt.1
+chrt_SOURCES = schedutils/chrt.c
+chrt_LDADD = $(LDADD) libcommon.la
+endif
+
+if BUILD_IONICE
+usrbin_exec_PROGRAMS += ionice
+dist_man_MANS += schedutils/ionice.1
+ionice_SOURCES = schedutils/ionice.c
+ionice_LDADD = $(LDADD) libcommon.la
+endif
+
+if BUILD_TASKSET
+usrbin_exec_PROGRAMS += taskset
+dist_man_MANS += schedutils/taskset.1
+taskset_SOURCES = schedutils/taskset.c
+taskset_LDADD = $(LDADD) libcommon.la
+endif
diff --git a/schedutils/chrt.1 b/schedutils/chrt.1
new file mode 100644
index 0000000..166579e
--- /dev/null
+++ b/schedutils/chrt.1
@@ -0,0 +1,173 @@
+.\" chrt(1) manpage
+.\"
+.\" Copyright (C) 2004 Robert Love
+.\" Copyright (C) 2015 Karel Zak <kzak@redhat.com>
+.\"
+.\" This is free documentation; you can redistribute it and/or
+.\" modify it under the terms of the GNU General Public License,
+.\" version 2, as published by the Free Software Foundation.
+.\"
+.\" The GNU General Public License's references to "object code"
+.\" and "executables" are to be interpreted as the output of any
+.\" document formatting or typesetting system, including
+.\" intermediate and printed output.
+.\"
+.\" This manual is distributed in the hope that it will be useful,
+.\" but WITHOUT ANY WARRANTY; without even the implied warranty of
+.\" MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+.\" GNU General Public License for more details.
+.\"
+.\" You should have received a copy of the GNU General Public License along
+.\" with this program; if not, write to the Free Software Foundation, Inc.,
+.\" 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
+.\"
+.TH CHRT 1 "January 2016" "util-linux" "User Commands"
+.SH NAME
+chrt \- manipulate the real-time attributes of a process
+.SH SYNOPSIS
+.B chrt
+[options]
+.IR priority\ command\  [ argument ...]
+.br
+.B chrt
+[options]
+.B \-p
+.RI [ priority ]\  pid
+.SH DESCRIPTION
+.PP
+.B chrt
+sets or retrieves the real-time scheduling attributes of an existing \fIpid\fR,
+or runs \fIcommand\fR with the given attributes.
+
+.SH POLICIES
+.TP
+\fB\-o\fR, \fB\-\-other\fR
+Set scheduling policy to
+.BR SCHED_OTHER .
+This is the default Linux scheduling policy.
+.TP
+\fB\-f\fR, \fB\-\-fifo\fR
+Set scheduling policy to \fBSCHED_FIFO\fR.
+.TP
+\fB\-r\fR, \fB\-\-rr\fR
+Set scheduling policy to
+.BR SCHED_RR .
+When no policy is defined, the
+.B SCHED_RR
+is used as the default.
+.TP
+\fB\-b\fR, \fB\-\-batch\fR
+Set scheduling policy to
+.B SCHED_BATCH
+(Linux-specific, supported since 2.6.16).  The priority argument has to be set to zero.
+.TP
+\fB\-i\fR, \fB\-\-idle\fR
+Set scheduling policy to
+.B SCHED_IDLE
+(Linux-specific, supported since 2.6.23).  The priority argument has to be set to zero.
+.TP
+.BR -d ,\  --deadline
+Set scheduling policy to
+.B SCHED_DEADLINE
+(Linux-specific, supported since 3.14).  The priority argument has to be set to zero.
+See also \fB\-\-sched\-runtime\fR, \fB\-\-sched\-deadline\fR and
+\fB\-\-sched\-period\fR.  The relation between the options required by the kernel is
+runtime <= deadline <= period.
+.B chrt
+copies \fIperiod\fR to \fIdeadline\fR if \fB\-\-sched\-deadline\fR is not specified and
+\fIdeadline\fR to \fIruntime\fR if \fB\-\-sched\-runtime\fR is not specified.
+It means that at least \fB\-\-sched\-period\fR has to be specified.  See
+.BR sched (7)
+for more details.
+
+.SH "SCHEDULING OPTIONS"
+.TP
+\fB\-T\fR, \fB\-\-sched\-runtime\fR \fInanoseconds\fR
+Specifies runtime parameter for SCHED_DEADLINE policy (Linux-specific).
+.TP
+\fB\-P\fR, \fB\-\-sched\-period\fR \fInanoseconds\fR
+Specifies period parameter for SCHED_DEADLINE policy (Linux-specific).
+.TP
+\fB\-D\fR, \fB\-\-sched\-deadline\fR \fInanoseconds\fR
+Specifies deadline parameter for SCHED_DEADLINE policy (Linux-specific).
+.TP
+\fB\-R\fR, \fB\-\-reset-on-fork\fR
+Add
+.B SCHED_RESET_ON_FORK
+flag to the
+.B SCHED_FIFO
+or
+.B SCHED_RR
+scheduling policy (Linux-specific, supported since 2.6.31).
+
+.SH OPTIONS
+.TP
+.BR -a ,\  --all-tasks
+Set or retrieve the scheduling attributes of all the tasks (threads) for a
+given PID.
+.TP
+.BR -m ,\  --max
+Show minimum and maximum valid priorities, then exit.
+.TP
+.BR -p ,\  --pid
+Operate on an existing PID and do not launch a new task.
+.TP
+.BR -v ,\  --verbose
+Show status information.
+.TP
+.BR -V ,\  --version
+Display version information and exit.
+.TP
+.BR -h ,\  --help
+Display help text and exit.
+.SH USAGE
+.TP
+The default behavior is to run a new command:
+.B chrt
+.I priority
+.IR command\  [ arguments ]
+.TP
+You can also retrieve the real-time attributes of an existing task:
+.B chrt \-p
+.I pid
+.TP
+Or set them:
+.B chrt \-r \-p
+.I priority pid
+.SH PERMISSIONS
+A user must possess
+.B CAP_SYS_NICE
+to change the scheduling attributes of a process.  Any user can retrieve the
+scheduling information.
+
+.SH NOTES
+Only
+.BR SCHED_FIFO ,
+.B SCHED_OTHER
+and
+.B SCHED_RR
+are part of POSIX 1003.1b Process Scheduling.  The other scheduling attributes
+may be ignored on some systems.
+.P
+Linux' default scheduling policy is
+.BR SCHED_OTHER .
+.SH SEE ALSO
+.BR nice (1),
+.BR renice (1),
+.BR taskset (1),
+.BR sched (7)
+.sp
+See
+.BR sched_setscheduler (2)
+for a description of the Linux scheduling scheme.
+.SH AUTHORS
+.UR rml@tech9.net
+Robert Love
+.UE
+.br
+.UR kzak@redhat.com
+Karel Zak
+.UE
+.SH AVAILABILITY
+The chrt command is part of the util-linux package and is available from
+https://www.kernel.org/pub/linux/utils/util-linux/.
diff --git a/schedutils/chrt.c b/schedutils/chrt.c
new file mode 100644
index 0000000..b08c78e
--- /dev/null
+++ b/schedutils/chrt.c
@@ -0,0 +1,560 @@
+/*
+ * chrt.c - manipulate a task's real-time attributes
+ *
+ * 27-Apr-2002: initial version - Robert Love <rml@tech9.net>
+ * 04-May-2011: make it thread-aware - Davidlohr Bueso <dave@gnu.org>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License, version 2, as
+ * published by the Free Software Foundation
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License along
+ * with this program; if not, write to the Free Software Foundation, Inc.,
+ * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
+ *
+ * Copyright (C) 2004 Robert Love
+ */
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <sched.h>
+#include <unistd.h>
+#include <getopt.h>
+#include <errno.h>
+#include <sys/time.h>
+#include <sys/resource.h>
+
+#include "c.h"
+#include "nls.h"
+#include "closestream.h"
+#include "strutils.h"
+#include "procutils.h"
+
+/* the SCHED_BATCH is supported since Linux 2.6.16
+ *  -- temporary workaround for people with old glibc headers
+ */
+#if defined (__linux__) && !defined(SCHED_BATCH)
+# define SCHED_BATCH 3
+#endif
+
+/* the SCHED_IDLE is supported since Linux 2.6.23
+ * commit id 0e6aca43e08a62a48d6770e9a159dbec167bf4c6
+ * -- temporary workaround for people with old glibc headers
+ */
+#if defined (__linux__) && !defined(SCHED_IDLE)
+# define SCHED_IDLE 5
+#endif
+
+/* flag by sched_getscheduler() */
+#if defined(__linux__) && !defined(SCHED_RESET_ON_FORK)
+# define SCHED_RESET_ON_FORK 0x40000000
+#endif
+
+/* flag by sched_getattr() */
+#if defined(__linux__) && !defined(SCHED_FLAG_RESET_ON_FORK)
+# define SCHED_FLAG_RESET_ON_FORK 0x01
+#endif
+
+#if defined (__linux__) && !defined(HAVE_SCHED_SETATTR)
+# include <sys/syscall.h>
+#endif
+
+/* usable kernel-headers, but old glibc-headers */
+#if defined (__linux__) && !defined(SYS_sched_setattr) && defined(__NR_sched_setattr)
+# define SYS_sched_setattr __NR_sched_setattr
+#endif
+
+#if defined (__linux__) && !defined(SYS_sched_getattr) && defined(__NR_sched_getattr)
+# define SYS_sched_getattr __NR_sched_getattr
+#endif
+
+#if defined (__linux__) && !defined(HAVE_SCHED_SETATTR) && defined(SYS_sched_setattr)
+# define HAVE_SCHED_SETATTR
+
+struct sched_attr {
+	uint32_t size;
+	uint32_t sched_policy;
+	uint64_t sched_flags;
+
+	/* SCHED_NORMAL, SCHED_BATCH */
+	int32_t sched_nice;
+
+	/* SCHED_FIFO, SCHED_RR */
+	uint32_t sched_priority;
+
+	/* SCHED_DEADLINE (nsec) */
+	uint64_t sched_runtime;
+	uint64_t sched_deadline;
+	uint64_t sched_period;
+};
+
+static int sched_setattr(pid_t pid, const struct sched_attr *attr, unsigned int flags)
+{
+	return syscall(SYS_sched_setattr, pid, attr, flags);
+}
+
+static int sched_getattr(pid_t pid, struct sched_attr *attr, unsigned int size, unsigned int flags)
+{
+	return syscall(SYS_sched_getattr, pid, attr, size, flags);
+}
+#endif
+
+/* the SCHED_DEADLINE is supported since Linux 3.14
+ * commit id aab03e05e8f7e26f51dee792beddcb5cca9215a5
+ * -- sched_setattr() is required for this policy!
+ */
+#if defined (__linux__) && !defined(SCHED_DEADLINE) && defined(HAVE_SCHED_SETATTR)
+# define SCHED_DEADLINE 6
+#endif
+
+/* control struct */
+struct chrt_ctl {
+	pid_t	pid;
+	int	policy;				/* SCHED_* */
+	int	priority;
+
+	uint64_t runtime;			/* --sched-* options */
+	uint64_t deadline;
+	uint64_t period;
+
+	unsigned int all_tasks : 1,		/* all threads of the PID */
+		     reset_on_fork : 1,		/* SCHED_RESET_ON_FORK */
+		     altered : 1,		/* sched_set**() used */
+		     verbose : 1;		/* verbose output */
+};
+
+static void __attribute__((__noreturn__)) usage(void)
+{
+	FILE *out = stdout;
+
+	fputs(_("Show or change the real-time scheduling attributes of a process.\n"), out);
+	fputs(USAGE_SEPARATOR, out);
+	fputs(_("Set policy:\n"
+	" chrt [options] <priority> <command> [<arg>...]\n"
+	" chrt [options] --pid <priority> <pid>\n"), out);
+	fputs(USAGE_SEPARATOR, out);
+	fputs(_("Get policy:\n"
+	" chrt [options] -p <pid>\n"), out);
+
+	fputs(USAGE_SEPARATOR, out);
+	fputs(_("Policy options:\n"), out);
+	fputs(_(" -b, --batch          set policy to SCHED_BATCH\n"), out);
+	fputs(_(" -d, --deadline       set policy to SCHED_DEADLINE\n"), out);
+	fputs(_(" -f, --fifo           set policy to SCHED_FIFO\n"), out);
+	fputs(_(" -i, --idle           set policy to SCHED_IDLE\n"), out);
+	fputs(_(" -o, --other          set policy to SCHED_OTHER\n"), out);
+	fputs(_(" -r, --rr             set policy to SCHED_RR (default)\n"), out);
+
+	fputs(USAGE_SEPARATOR, out);
+	fputs(_("Scheduling options:\n"), out);
+	fputs(_(" -R, --reset-on-fork       set SCHED_RESET_ON_FORK for FIFO or RR\n"), out);
+	fputs(_(" -T, --sched-runtime <ns>  runtime parameter for DEADLINE\n"), out);
+	fputs(_(" -P, --sched-period <ns>   period parameter for DEADLINE\n"), out);
+	fputs(_(" -D, --sched-deadline <ns> deadline parameter for DEADLINE\n"), out);
+
+	fputs(USAGE_SEPARATOR, out);
+	fputs(_("Other options:\n"), out);
+	fputs(_(" -a, --all-tasks      operate on all the tasks (threads) for a given pid\n"), out);
+	fputs(_(" -m, --max            show min and max valid priorities\n"), out);
+	fputs(_(" -p, --pid            operate on existing given pid\n"), out);
+	fputs(_(" -v, --verbose        display status information\n"), out);
+
+	fputs(USAGE_SEPARATOR, out);
+	printf(USAGE_HELP_OPTIONS(22));
+
+	printf(USAGE_MAN_TAIL("chrt(1)"));
+	exit(EXIT_SUCCESS);
+}
+
+static const char *get_policy_name(int policy)
+{
+	switch (policy) {
+	case SCHED_OTHER:
+		return "SCHED_OTHER";
+	case SCHED_FIFO:
+#ifdef SCHED_RESET_ON_FORK
+	case SCHED_FIFO | SCHED_RESET_ON_FORK:
+#endif
+		return "SCHED_FIFO";
+#ifdef SCHED_IDLE
+	case SCHED_IDLE:
+		return "SCHED_IDLE";
+#endif
+	case SCHED_RR:
+#ifdef SCHED_RESET_ON_FORK
+	case SCHED_RR | SCHED_RESET_ON_FORK:
+#endif
+		return "SCHED_RR";
+#ifdef SCHED_BATCH
+	case SCHED_BATCH:
+		return "SCHED_BATCH";
+#endif
+#ifdef SCHED_DEADLINE
+	case SCHED_DEADLINE:
+		return "SCHED_DEADLINE";
+#endif
+	default:
+		break;
+	}
+
+	return _("unknown");
+}
+
+static void show_sched_pid_info(struct chrt_ctl *ctl, pid_t pid)
+{
+	int policy = -1, reset_on_fork = 0, prio = 0;
+#ifdef SCHED_DEADLINE
+	uint64_t deadline = 0, runtime = 0, period = 0;
+#endif
+
+	/* don't display "pid 0" as that is confusing */
+	if (!pid)
+		pid = getpid();
+
+	errno = 0;
+
+	/*
+	 * New way
+	 */
+#ifdef HAVE_SCHED_SETATTR
+	{
+		struct sched_attr sa;
+
+		if (sched_getattr(pid, &sa, sizeof(sa), 0) != 0) {
+			if (errno == ENOSYS)
+				goto fallback;
+			err(EXIT_FAILURE, _("failed to get pid %d's policy"), pid);
+		}
+
+		policy = sa.sched_policy;
+		prio = sa.sched_priority;
+		reset_on_fork = sa.sched_flags & SCHED_FLAG_RESET_ON_FORK;
+		deadline = sa.sched_deadline;
+		runtime = sa.sched_runtime;
+		period = sa.sched_period;
+	}
+
+	/*
+	 * Old way
+	 */
+fallback:
+	if (errno == ENOSYS)
+#endif
+	{
+		struct sched_param sp;
+
+		policy = sched_getscheduler(pid);
+		if (policy == -1)
+			err(EXIT_FAILURE, _("failed to get pid %d's policy"), pid);
+
+		if (sched_getparam(pid, &sp) != 0)
+			err(EXIT_FAILURE, _("failed to get pid %d's attributes"), pid);
+		else
+			prio = sp.sched_priority;
+# ifdef SCHED_RESET_ON_FORK
+		if (policy == (SCHED_FIFO|SCHED_RESET_ON_FORK) || policy == (SCHED_BATCH|SCHED_RESET_ON_FORK))
+			reset_on_fork = 1;
+# endif
+	}
+
+	if (ctl->altered)
+		printf(_("pid %d's new scheduling policy: %s"), pid, get_policy_name(policy));
+	else
+		printf(_("pid %d's current scheduling policy: %s"), pid, get_policy_name(policy));
+
+	if (reset_on_fork)
+		printf("|SCHED_RESET_ON_FORK");
+	putchar('\n');
+
+	if (ctl->altered)
+		printf(_("pid %d's new scheduling priority: %d\n"), pid, prio);
+	else
+		printf(_("pid %d's current scheduling priority: %d\n"), pid, prio);
+
+#ifdef SCHED_DEADLINE
+	if (policy == SCHED_DEADLINE) {
+		if (ctl->altered)
+			printf(_("pid %d's new runtime/deadline/period parameters: %ju/%ju/%ju\n"),
+					pid, runtime, deadline, period);
+		else
+			printf(_("pid %d's current runtime/deadline/period parameters: %ju/%ju/%ju\n"),
+					pid, runtime, deadline, period);
+	}
+#endif
+}
+
+
+static void show_sched_info(struct chrt_ctl *ctl)
+{
+	if (ctl->all_tasks) {
+		pid_t tid;
+		struct proc_tasks *ts = proc_open_tasks(ctl->pid);
+
+		if (!ts)
+			err(EXIT_FAILURE, _("cannot obtain the list of tasks"));
+
+		while (!proc_next_tid(ts, &tid))
+			show_sched_pid_info(ctl, tid);
+
+		proc_close_tasks(ts);
+	} else
+		show_sched_pid_info(ctl, ctl->pid);
+}
+
+static void show_min_max(void)
+{
+	unsigned long i;
+	int policies[] = {
+		SCHED_OTHER,
+		SCHED_FIFO,
+		SCHED_RR,
+#ifdef SCHED_BATCH
+		SCHED_BATCH,
+#endif
+#ifdef SCHED_IDLE
+		SCHED_IDLE,
+#endif
+#ifdef SCHED_DEADLINE
+		SCHED_DEADLINE,
+#endif
+	};
+
+	for (i = 0; i < ARRAY_SIZE(policies); i++) {
+		int plc = policies[i];
+		int max = sched_get_priority_max(plc);
+		int min = sched_get_priority_min(plc);
+
+		if (max >= 0 && min >= 0)
+			printf(_("%s min/max priority\t: %d/%d\n"),
+					get_policy_name(plc), min, max);
+		else
+			printf(_("%s not supported?\n"), get_policy_name(plc));
+	}
+}
+
+static int set_sched_one_by_setscheduler(struct chrt_ctl *ctl, pid_t pid)
+{
+	struct sched_param sp = { .sched_priority = ctl->priority };
+	int policy = ctl->policy;
+
+	errno = 0;
+# ifdef SCHED_RESET_ON_FORK
+	if (ctl->reset_on_fork)
+		policy |= SCHED_RESET_ON_FORK;
+# endif
+	return sched_setscheduler(pid, policy, &sp);
+}
+
+
+#ifndef HAVE_SCHED_SETATTR
+static int set_sched_one(struct chrt_ctl *ctl, pid_t pid)
+{
+	return set_sched_one_by_setscheduler(ctl, pid);
+}
+
+#else /* !HAVE_SCHED_SETATTR */
+static int set_sched_one(struct chrt_ctl *ctl, pid_t pid)
+{
+	struct sched_attr sa = { .size = sizeof(struct sched_attr) };
+
+	/* old API is good enough for non-deadline */
+	if (ctl->policy != SCHED_DEADLINE)
+		return set_sched_one_by_setscheduler(ctl, pid);
+
+	/* no changeed by chrt, follow the current setting */
+	sa.sched_nice = getpriority(PRIO_PROCESS, pid);
+
+	/* use main() to check if the setting makes sense */
+	sa.sched_policy	  = ctl->policy;
+	sa.sched_priority = ctl->priority;
+	sa.sched_runtime  = ctl->runtime;
+	sa.sched_period   = ctl->period;
+	sa.sched_deadline = ctl->deadline;
+
+# ifdef SCHED_RESET_ON_FORK
+	if (ctl->reset_on_fork)
+		sa.sched_flags |= SCHED_RESET_ON_FORK;
+# endif
+	errno = 0;
+	return sched_setattr(pid, &sa, 0);
+}
+#endif /* HAVE_SCHED_SETATTR */
+
+static void set_sched(struct chrt_ctl *ctl)
+{
+	if (ctl->all_tasks) {
+		pid_t tid;
+		struct proc_tasks *ts = proc_open_tasks(ctl->pid);
+
+		if (!ts)
+			err(EXIT_FAILURE, _("cannot obtain the list of tasks"));
+
+		while (!proc_next_tid(ts, &tid))
+			if (set_sched_one(ctl, tid) == -1)
+				err(EXIT_FAILURE, _("failed to set tid %d's policy"), tid);
+
+		proc_close_tasks(ts);
+
+	} else if (set_sched_one(ctl, ctl->pid) == -1)
+		err(EXIT_FAILURE, _("failed to set pid %d's policy"), ctl->pid);
+
+	ctl->altered = 1;
+}
+
+int main(int argc, char **argv)
+{
+	struct chrt_ctl _ctl = { .pid = -1, .policy = SCHED_RR }, *ctl = &_ctl;
+	int c;
+
+	static const struct option longopts[] = {
+		{ "all-tasks",  no_argument, NULL, 'a' },
+		{ "batch",	no_argument, NULL, 'b' },
+		{ "deadline",   no_argument, NULL, 'd' },
+		{ "fifo",	no_argument, NULL, 'f' },
+		{ "idle",	no_argument, NULL, 'i' },
+		{ "pid",	no_argument, NULL, 'p' },
+		{ "help",	no_argument, NULL, 'h' },
+		{ "max",        no_argument, NULL, 'm' },
+		{ "other",	no_argument, NULL, 'o' },
+		{ "rr",		no_argument, NULL, 'r' },
+		{ "sched-runtime",  required_argument, NULL, 'T' },
+		{ "sched-period",   required_argument, NULL, 'P' },
+		{ "sched-deadline", required_argument, NULL, 'D' },
+		{ "reset-on-fork",  no_argument,       NULL, 'R' },
+		{ "verbose",	no_argument, NULL, 'v' },
+		{ "version",	no_argument, NULL, 'V' },
+		{ NULL,		no_argument, NULL, 0 }
+	};
+
+	setlocale(LC_ALL, "");
+	bindtextdomain(PACKAGE, LOCALEDIR);
+	textdomain(PACKAGE);
+	atexit(close_stdout);
+
+	while((c = getopt_long(argc, argv, "+abdD:fiphmoP:T:rRvV", longopts, NULL)) != -1)
+	{
+		switch (c) {
+		case 'a':
+			ctl->all_tasks = 1;
+			break;
+		case 'b':
+#ifdef SCHED_BATCH
+			ctl->policy = SCHED_BATCH;
+#endif
+			break;
+
+		case 'd':
+#ifdef SCHED_DEADLINE
+			ctl->policy = SCHED_DEADLINE;
+#endif
+			break;
+		case 'f':
+			ctl->policy = SCHED_FIFO;
+			break;
+		case 'R':
+			ctl->reset_on_fork = 1;
+			break;
+		case 'i':
+#ifdef SCHED_IDLE
+			ctl->policy = SCHED_IDLE;
+#endif
+			break;
+		case 'm':
+			show_min_max();
+			return EXIT_SUCCESS;
+		case 'o':
+			ctl->policy = SCHED_OTHER;
+			break;
+		case 'p':
+			errno = 0;
+			ctl->pid = strtos32_or_err(argv[argc - 1], _("invalid PID argument"));
+			break;
+		case 'r':
+			ctl->policy = SCHED_RR;
+			break;
+		case 'v':
+			ctl->verbose = 1;
+			break;
+		case 'T':
+			ctl->runtime = strtou64_or_err(optarg, _("invalid runtime argument"));
+			break;
+		case 'P':
+			ctl->period = strtou64_or_err(optarg, _("invalid period argument"));
+			break;
+		case 'D':
+			ctl->deadline = strtou64_or_err(optarg, _("invalid deadline argument"));
+			break;
+		case 'V':
+			printf(UTIL_LINUX_VERSION);
+			return EXIT_SUCCESS;
+		case 'h':
+			usage();
+		default:
+			errtryhelp(EXIT_FAILURE);
+		}
+	}
+
+	if (((ctl->pid > -1) && argc - optind < 1) ||
+	    ((ctl->pid == -1) && argc - optind < 2)) {
+		warnx(_("bad usage"));
+		errtryhelp(EXIT_FAILURE);
+}
+
+	if ((ctl->pid > -1) && (ctl->verbose || argc - optind == 1)) {
+		show_sched_info(ctl);
+		if (argc - optind == 1)
+			return EXIT_SUCCESS;
+	}
+
+	errno = 0;
+	ctl->priority = strtos32_or_err(argv[optind], _("invalid priority argument"));
+
+#ifdef SCHED_RESET_ON_FORK
+	if (ctl->reset_on_fork && ctl->policy != SCHED_FIFO && ctl->policy != SCHED_RR)
+		errx(EXIT_FAILURE, _("--reset-on-fork option is supported for "
+				     "SCHED_FIFO and SCHED_RR policies only"));
+#endif
+#ifdef SCHED_DEADLINE
+	if ((ctl->runtime || ctl->deadline || ctl->period) && ctl->policy != SCHED_DEADLINE)
+		errx(EXIT_FAILURE, _("--sched-{runtime,deadline,period} options "
+				     "are supported for SCHED_DEADLINE only"));
+	if (ctl->policy == SCHED_DEADLINE) {
+		/* The basic rule is runtime <= deadline <= period, so we can
+		 * make deadline and runtime optional on command line. Note we
+		 * don't check any values or set any defaults, it's kernel
+		 * responsibility.
+		 */
+		if (ctl->deadline == 0)
+			ctl->deadline = ctl->period;
+		if (ctl->runtime == 0)
+			ctl->runtime = ctl->deadline;
+	}
+#else
+	if (ctl->runtime || ctl->deadline || ctl->period)
+		errx(EXIT_FAILURE, _("SCHED_DEADLINE is unsupported"));
+#endif
+	if (ctl->pid == -1)
+		ctl->pid = 0;
+	if (ctl->priority < sched_get_priority_min(ctl->policy) ||
+	    sched_get_priority_max(ctl->policy) < ctl->priority)
+		errx(EXIT_FAILURE,
+		     _("unsupported priority value for the policy: %d: see --max for valid range"),
+		     ctl->priority);
+	set_sched(ctl);
+
+	if (ctl->verbose)
+		show_sched_info(ctl);
+
+	if (!ctl->pid) {
+		argv += optind + 1;
+		execvp(argv[0], argv);
+		errexec(argv[0]);
+	}
+
+	return EXIT_SUCCESS;
+}
diff --git a/schedutils/ionice.1 b/schedutils/ionice.1
new file mode 100644
index 0000000..1b075d3
--- /dev/null
+++ b/schedutils/ionice.1
@@ -0,0 +1,142 @@
+.TH IONICE 1 "July 2011" "util-linux" "User Commands"
+.SH NAME
+ionice \- set or get process I/O scheduling class and priority
+.SH SYNOPSIS
+.B ionice
+.RB [ \-c
+.IR class ]
+.RB [ \-n
+.IR level ]
+.RB [ \-t ]
+.B \-p
+.IR PID ...
+.br
+.B ionice
+.RB [ \-c
+.IR class ]
+.RB [ \-n
+.IR level ]
+.RB [ \-t ]
+.B \-P
+.IR PGID ...
+.br
+.B ionice
+.RB [ \-c
+.IR class ]
+.RB [ \-n
+.IR level ]
+.RB [ \-t ]
+.B \-u
+.IR UID ...
+.br
+.B ionice
+.RB [ \-c
+.IR class ]
+.RB [ \-n
+.IR level ]
+.RB [ \-t ]
+.IR "command " [ argument ...]
+.SH DESCRIPTION
+This program sets or gets the I/O scheduling class and priority for a program.
+If no arguments or just \fB\-p\fR is given, \fBionice\fR will query the current
+I/O scheduling class and priority for that process.
+
+When \fIcommand\fR is given,
+.B ionice
+will run this command with the given arguments.
+If no \fIclass\fR is specified, then
+.I command
+will be executed with the "best-effort" scheduling class.  The default
+priority level is 4.
+
+As of this writing, a process can be in one of three scheduling classes:
+.IP "\fBIdle\fP"
+A program running with idle I/O priority will only get disk time when no other
+program has asked for disk I/O for a defined grace period.  The impact of an
+idle I/O process on normal system activity should be zero.  This scheduling
+class does not take a priority argument.  Presently, this scheduling class
+is permitted for an ordinary user (since kernel 2.6.25).
+.IP "\fBBest-effort\fP"
+This is the effective scheduling class for any process that has not asked for
+a specific I/O priority.
+This class takes a priority argument from \fI0-7\fR, with a lower
+number being higher priority.  Programs running at the same best-effort
+priority are served in a round-robin fashion.
+
+Note that before kernel 2.6.26 a process that has not asked for an I/O priority
+formally uses "\fBnone\fP" as scheduling class, but the I/O scheduler will treat
+such processes as if it were in the best-effort class.  The priority within the
+best-effort class will be dynamically derived from the CPU nice level of the
+process: io_priority = (cpu_nice + 20) / 5.
+
+For kernels after 2.6.26 with the CFQ I/O scheduler, a process that has not asked
+for an I/O priority inherits its CPU scheduling class.  The I/O priority is derived
+from the CPU nice level of the process (same as before kernel 2.6.26).
+
+.IP "\fBRealtime\fP"
+The RT scheduling class is given first access to the disk, regardless of
+what else is going on in the system.  Thus the RT class needs to be used with
+some care, as it can starve other processes.  As with the best-effort class,
+8 priority levels are defined denoting how big a time slice a given process
+will receive on each scheduling window.  This scheduling class is not
+permitted for an ordinary (i.e., non-root) user.
+.SH OPTIONS
+.TP
+.BR \-c , " \-\-class " \fIclass\fR
+Specify the name or number of the scheduling class to use; \fI0\fR for none,
+\fI1\fR for realtime, \fI2\fR for best-effort, \fI3\fR for idle.
+.TP
+.BR \-n , " \-\-classdata " \fIlevel\fR
+Specify the scheduling class data.  This only has an effect if the class
+accepts an argument.  For realtime and best-effort, \fI0-7\fR are valid data
+(priority levels), and \fI0\fR represents the highest priority level.
+.TP
+.BR \-p , " \-\-pid " \fIPID\fR...
+Specify the process IDs of running processes for which to get or set the
+scheduling parameters.
+.TP
+.BR \-P , " \-\-pgid " \fIPGID\fR...
+Specify the process group IDs of running processes for which to get or set the
+scheduling parameters.
+.TP
+.BR \-t , " \-\-ignore"
+Ignore failure to set the requested priority.  If \fIcommand\fR was specified,
+run it even in case it was not possible to set the desired scheduling priority,
+which can happen due to insufficient privileges or an old kernel version.
+.TP
+.BR \-h , " \-\-help"
+Display help text and exit.
+.TP
+.BR \-u , " \-\-uid " \fIUID\fR...
+Specify the user IDs of running processes for which to get or set the
+scheduling parameters.
+.TP
+.BR \-V , " \-\-version"
+Display version information and exit.
+.SH EXAMPLES
+.LP
+.TP 7
+# \fBionice\fP -c 3 -p 89
+.TP 7
+Sets process with PID 89 as an idle I/O process.
+.TP 7
+# \fBionice\fP -c 2 -n 0 bash
+.TP 7
+Runs 'bash' as a best-effort program with highest priority.
+.TP 7
+# \fBionice\fP -p 89 91
+.TP 7
+Prints the class and priority of the processes with PID 89 and 91.
+.SH NOTES
+Linux supports I/O scheduling priorities and classes since 2.6.13 with the CFQ
+I/O scheduler.
+.SH AUTHORS
+.nf
+Jens Axboe <jens@axboe.dk>
+Karel Zak <kzak@redhat.com>
+.fi
+.SH SEE ALSO
+.BR ioprio_set (2)
+.SH AVAILABILITY
+The ionice command is part of the util-linux package and is available from
+https://www.kernel.org/pub/linux/utils/util-linux/.
diff --git a/schedutils/ionice.c b/schedutils/ionice.c
new file mode 100644
index 0000000..b4c6a82
--- /dev/null
+++ b/schedutils/ionice.c
@@ -0,0 +1,267 @@
+/*
+ * ionice: set or get process io scheduling class and priority
+ *
+ * Copyright (C) 2005 Jens Axboe <jens@axboe.dk>
+ *
+ * Released under the terms of the GNU General Public License version 2
+ *
+ */
+#include <stdio.h>
+#include <stdlib.h>
+#include <errno.h>
+#include <getopt.h>
+#include <unistd.h>
+#include <sys/syscall.h>
+#include <ctype.h>
+
+#include "nls.h"
+#include "strutils.h"
+#include "c.h"
+#include "closestream.h"
+
+static int tolerant;
+
+static inline int ioprio_set(int which, int who, int ioprio)
+{
+	return syscall(SYS_ioprio_set, which, who, ioprio);
+}
+
+static inline int ioprio_get(int which, int who)
+{
+	return syscall(SYS_ioprio_get, which, who);
+}
+
+enum {
+	IOPRIO_CLASS_NONE,
+	IOPRIO_CLASS_RT,
+	IOPRIO_CLASS_BE,
+	IOPRIO_CLASS_IDLE,
+};
+
+enum {
+	IOPRIO_WHO_PROCESS = 1,
+	IOPRIO_WHO_PGRP,
+	IOPRIO_WHO_USER,
+};
+
+#define IOPRIO_CLASS_SHIFT	(13)
+#define IOPRIO_PRIO_MASK	((1UL << IOPRIO_CLASS_SHIFT) - 1)
+
+#define IOPRIO_PRIO_CLASS(mask)	((mask) >> IOPRIO_CLASS_SHIFT)
+#define IOPRIO_PRIO_DATA(mask)	((mask) & IOPRIO_PRIO_MASK)
+#define IOPRIO_PRIO_VALUE(class, data)	(((class) << IOPRIO_CLASS_SHIFT) | data)
+
+static const char *to_prio[] = {
+	[IOPRIO_CLASS_NONE] = "none",
+	[IOPRIO_CLASS_RT]   = "realtime",
+	[IOPRIO_CLASS_BE]   = "best-effort",
+	[IOPRIO_CLASS_IDLE] = "idle"
+};
+
+static int parse_ioclass(const char *str)
+{
+	size_t i;
+
+	for (i = 0; i < ARRAY_SIZE(to_prio); i++)
+		if (!strcasecmp(str, to_prio[i]))
+			return i;
+	return -1;
+}
+
+static void ioprio_print(int pid, int who)
+{
+	int ioprio = ioprio_get(who, pid);
+
+	if (ioprio == -1)
+		err(EXIT_FAILURE, _("ioprio_get failed"));
+	else {
+		int ioclass = IOPRIO_PRIO_CLASS(ioprio);
+		const char *name = _("unknown");
+
+		if (ioclass >= 0 && (size_t) ioclass < ARRAY_SIZE(to_prio))
+			name = to_prio[ioclass];
+
+		if (ioclass != IOPRIO_CLASS_IDLE)
+			printf(_("%s: prio %lu\n"), name,
+			       IOPRIO_PRIO_DATA(ioprio));
+		else
+			printf("%s\n", name);
+	}
+}
+
+static void ioprio_setid(int which, int ioclass, int data, int who)
+{
+	int rc = ioprio_set(who, which,
+			    IOPRIO_PRIO_VALUE(ioclass, data));
+
+	if (rc == -1 && !tolerant)
+		err(EXIT_FAILURE, _("ioprio_set failed"));
+}
+
+static void __attribute__((__noreturn__)) usage(void)
+{
+	FILE *out = stdout;
+	fputs(USAGE_HEADER, out);
+	fprintf(out,  _(" %1$s [options] -p <pid>...\n"
+			" %1$s [options] -P <pgid>...\n"
+			" %1$s [options] -u <uid>...\n"
+			" %1$s [options] <command>\n"), program_invocation_short_name);
+
+	fputs(USAGE_SEPARATOR, out);
+	fputs(_("Show or change the I/O-scheduling class and priority of a process.\n"), out);
+
+	fputs(USAGE_OPTIONS, out);
+	fputs(_(" -c, --class <class>    name or number of scheduling class,\n"
+		"                          0: none, 1: realtime, 2: best-effort, 3: idle\n"), out);
+	fputs(_(" -n, --classdata <num>  priority (0..7) in the specified scheduling class,\n"
+		"                          only for the realtime and best-effort classes\n"), out);
+	fputs(_(" -p, --pid <pid>...     act on these already running processes\n"), out);
+	fputs(_(" -P, --pgid <pgrp>...   act on already running processes in these groups\n"), out);
+	fputs(_(" -t, --ignore           ignore failures\n"), out);
+	fputs(_(" -u, --uid <uid>...     act on already running processes owned by these users\n"), out);
+
+	fputs(USAGE_SEPARATOR, out);
+	printf(USAGE_HELP_OPTIONS(24));
+
+	printf(USAGE_MAN_TAIL("ionice(1)"));
+
+	exit(EXIT_SUCCESS);
+}
+
+int main(int argc, char **argv)
+{
+	int data = 4, set = 0, ioclass = IOPRIO_CLASS_BE, c;
+	int which = 0, who = 0;
+	const char *invalid_msg = NULL;
+
+	static const struct option longopts[] = {
+		{ "classdata", required_argument, NULL, 'n' },
+		{ "class",     required_argument, NULL, 'c' },
+		{ "help",      no_argument,       NULL, 'h' },
+		{ "ignore",    no_argument,       NULL, 't' },
+		{ "pid",       required_argument, NULL, 'p' },
+		{ "pgid",      required_argument, NULL, 'P' },
+		{ "uid",       required_argument, NULL, 'u' },
+		{ "version",   no_argument,       NULL, 'V' },
+		{ NULL, 0, NULL, 0 }
+	};
+
+	setlocale(LC_ALL, "");
+	bindtextdomain(PACKAGE, LOCALEDIR);
+	textdomain(PACKAGE);
+	atexit(close_stdout);
+
+	while ((c = getopt_long(argc, argv, "+n:c:p:P:u:tVh", longopts, NULL)) != EOF)
+		switch (c) {
+		case 'n':
+			data = strtos32_or_err(optarg, _("invalid class data argument"));
+			set |= 1;
+			break;
+		case 'c':
+			if (isdigit(*optarg))
+				ioclass = strtos32_or_err(optarg,
+						_("invalid class argument"));
+			else {
+				ioclass = parse_ioclass(optarg);
+				if (ioclass < 0)
+					errx(EXIT_FAILURE,
+						_("unknown scheduling class: '%s'"),
+						optarg);
+			}
+			set |= 2;
+			break;
+		case 'p':
+			if (who)
+				errx(EXIT_FAILURE,
+				     _("can handle only one of pid, pgid or uid at once"));
+			invalid_msg = _("invalid PID argument");
+			which = strtos32_or_err(optarg, invalid_msg);
+			who = IOPRIO_WHO_PROCESS;
+			break;
+		case 'P':
+			if (who)
+				errx(EXIT_FAILURE,
+				     _("can handle only one of pid, pgid or uid at once"));
+			invalid_msg = _("invalid PGID argument");
+			which = strtos32_or_err(optarg, invalid_msg);
+			who = IOPRIO_WHO_PGRP;
+			break;
+		case 'u':
+			if (who)
+				errx(EXIT_FAILURE,
+				     _("can handle only one of pid, pgid or uid at once"));
+			invalid_msg = _("invalid UID argument");
+			which = strtos32_or_err(optarg, invalid_msg);
+			who = IOPRIO_WHO_USER;
+			break;
+		case 't':
+			tolerant = 1;
+			break;
+		case 'V':
+			printf(UTIL_LINUX_VERSION);
+			return EXIT_SUCCESS;
+		case 'h':
+			usage();
+		default:
+			errtryhelp(EXIT_FAILURE);
+		}
+
+	switch (ioclass) {
+		case IOPRIO_CLASS_NONE:
+			if ((set & 1) && !tolerant)
+				warnx(_("ignoring given class data for none class"));
+			data = 0;
+			break;
+		case IOPRIO_CLASS_RT:
+		case IOPRIO_CLASS_BE:
+			break;
+		case IOPRIO_CLASS_IDLE:
+			if ((set & 1) && !tolerant)
+				warnx(_("ignoring given class data for idle class"));
+			data = 7;
+			break;
+		default:
+			if (!tolerant)
+				warnx(_("unknown prio class %d"), ioclass);
+			break;
+	}
+
+	if (!set && !which && optind == argc)
+		/*
+		 * ionice without options, print the current ioprio
+		 */
+		ioprio_print(0, IOPRIO_WHO_PROCESS);
+	else if (!set && who) {
+		/*
+		 * ionice -p|-P|-u ID [ID ...]
+		 */
+		ioprio_print(which, who);
+
+		for(; argv[optind]; ++optind) {
+			which = strtos32_or_err(argv[optind], invalid_msg);
+			ioprio_print(which, who);
+		}
+	} else if (set && who) {
+		/*
+		 * ionice -c CLASS -p|-P|-u ID [ID ...]
+		 */
+		ioprio_setid(which, ioclass, data, who);
+
+		for(; argv[optind]; ++optind) {
+			which = strtos32_or_err(argv[optind], invalid_msg);
+			ioprio_setid(which, ioclass, data, who);
+		}
+	} else if (argv[optind]) {
+		/*
+		 * ionice [-c CLASS] COMMAND
+		 */
+		ioprio_setid(0, ioclass, data, IOPRIO_WHO_PROCESS);
+		execvp(argv[optind], &argv[optind]);
+		errexec(argv[optind]);
+	} else {
+		warnx(_("bad usage"));
+		errtryhelp(EXIT_FAILURE);
+	}
+
+	return EXIT_SUCCESS;
+}
diff --git a/schedutils/taskset.1 b/schedutils/taskset.1
new file mode 100644
index 0000000..6a32978
--- /dev/null
+++ b/schedutils/taskset.1
@@ -0,0 +1,136 @@
+.\" taskset(1) manpage
+.\"
+.\" Copyright (C) 2004 Robert Love
+.\"
+.\" This is free documentation; you can redistribute it and/or
+.\" modify it under the terms of the GNU General Public License,
+.\" version 2, as published by the Free Software Foundation.
+.\"
+.\" The GNU General Public License's references to "object code"
+.\" and "executables" are to be interpreted as the output of any
+.\" document formatting or typesetting system, including
+.\" intermediate and printed output.
+.\"
+.\" This manual is distributed in the hope that it will be useful,
+.\" but WITHOUT ANY WARRANTY; without even the implied warranty of
+.\" MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+.\" GNU General Public License for more details.
+.\"
+.\" You should have received a copy of the GNU General Public License along
+.\" with this program; if not, write to the Free Software Foundation, Inc.,
+.\" 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
+.\"
+.TH TASKSET 1 "August 2014" "util-linux" "User Commands"
+.SH NAME
+taskset \- set or retrieve a process's CPU affinity
+.SH SYNOPSIS
+.B taskset
+[options]
+.IR mask\ command\  [ argument ...]
+.br
+.B taskset
+[options]
+.B \-p
+.RI [ mask ]\  pid
+.SH DESCRIPTION
+.PP
+.B taskset
+is used to set or retrieve the CPU affinity of a running process given its
+\fIpid\fR, or to launch a new \fIcommand\fR with a given CPU affinity.
+CPU affinity is a
+scheduler property that "bonds" a process to a given set of CPUs on the system.
+The Linux scheduler will honor the given CPU affinity and the process will not
+run on any other CPUs.  Note that the Linux scheduler also supports natural
+CPU affinity: the scheduler attempts to keep processes on the same CPU as long
+as practical for performance reasons.  Therefore, forcing a specific CPU
+affinity is useful only in certain applications.
+.sp
+The CPU affinity is represented as a bitmask, with the lowest order bit
+corresponding to the first logical CPU and the highest order bit corresponding
+to the last logical CPU.  Not all CPUs may exist on a given system but a mask
+may specify more CPUs than are present.  A retrieved mask will reflect only the
+bits that correspond to CPUs physically on the system.  If an invalid mask is
+given (i.e., one that corresponds to no valid CPUs on the current system) an
+error is returned.  The masks may be specified in hexadecimal (with or without
+a leading "0x"), or as a CPU list with the
+.B \-\-cpu\-list
+option.  For example,
+.RS 4
+.TP 12
+.B 0x00000001
+is processor #0,
+.TP
+.B 0x00000003
+is processors #0 and #1,
+.TP
+.B 0xFFFFFFFF
+is processors #0 through #31,
+.TP
+.B 32
+is processors #1, #4, and #5,
+.TP
+.B \-\-cpu\-list\ 0-2,6
+is processors #0, #1, #2, and #6.
+.RE
+.PP
+When
+.B taskset
+returns, it is guaranteed that the given program has been scheduled to a legal
+CPU.
+.SH OPTIONS
+.TP
+.BR \-a ,\  \-\-all\-tasks
+Set or retrieve the CPU affinity of all the tasks (threads) for a given PID.
+.TP
+.BR \-c ,\  \-\-cpu\-list
+Interpret \fImask\fR as numerical list of processors instead of a bitmask.
+Numbers are separated by commas and may include ranges.  For example:
+.BR 0,5,8-11 .
+.TP
+.BR \-p ,\  \-\-pid
+Operate on an existing PID and do not launch a new task.
+.TP
+.BR \-V ,\  \-\-version
+Display version information and exit.
+.TP
+.BR \-h ,\  \-\-help
+Display help text and exit.
+.SH USAGE
+.TP
+The default behavior is to run a new command with a given affinity mask:
+.B taskset
+.I mask
+.IR command\  [ arguments ]
+.TP
+You can also retrieve the CPU affinity of an existing task:
+.B taskset \-p
+.I pid
+.TP
+Or set it:
+.B taskset \-p
+.I mask pid
+.SH PERMISSIONS
+A user can change the CPU affinity of a process belonging to the same user.
+A user must possess
+.B CAP_SYS_NICE
+to change the CPU affinity of a process belonging to another user.
+A user can retrieve the affinity mask of any process.
+.SH SEE ALSO
+.BR chrt (1),
+.BR nice (1),
+.BR renice (1),
+.BR sched_getaffinity (2),
+.BR sched_setaffinity (2)
+.sp
+See
+.BR sched (7)
+for a description of the Linux scheduling scheme.
+.SH AUTHOR
+Written by Robert M. Love.
+.SH COPYRIGHT
+Copyright \(co 2004 Robert M. Love.
+This is free software; see the source for copying conditions.  There is NO
+warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
+.SH AVAILABILITY
+The taskset command is part of the util-linux package and is available from
+https://www.kernel.org/pub/linux/utils/util-linux/.
diff --git a/schedutils/taskset.c b/schedutils/taskset.c
new file mode 100644
index 0000000..fe79cd6
--- /dev/null
+++ b/schedutils/taskset.c
@@ -0,0 +1,252 @@
+/*
+ * taskset.c - set or retrieve a task's CPU affinity
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License, version 2, as
+ * published by the Free Software Foundation
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License along
+ * with this program; if not, write to the Free Software Foundation, Inc.,
+ * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
+ *
+ * Copyright (C) 2004 Robert Love
+ * Copyright (C) 2010 Karel Zak <kzak@redhat.com>
+ */
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <unistd.h>
+#include <getopt.h>
+#include <errno.h>
+#include <sched.h>
+#include <stddef.h>
+#include <string.h>
+
+#include "cpuset.h"
+#include "nls.h"
+#include "strutils.h"
+#include "xalloc.h"
+#include "procutils.h"
+#include "c.h"
+#include "closestream.h"
+
+struct taskset {
+	pid_t		pid;		/* task PID */
+	cpu_set_t	*set;		/* task CPU mask */
+	size_t		setsize;
+	char		*buf;		/* buffer for conversion from mask to string */
+	size_t		buflen;
+	unsigned int	use_list:1,	/* use list rather than masks */
+			get_only:1;	/* print the mask, but not modify */
+};
+
+static void __attribute__((__noreturn__)) usage(void)
+{
+	FILE *out = stdout;
+	fprintf(out,
+		_("Usage: %s [options] [mask | cpu-list] [pid|cmd [args...]]\n\n"),
+		program_invocation_short_name);
+
+	fputs(USAGE_SEPARATOR, out);
+	fputs(_("Show or change the CPU affinity of a process.\n"), out);
+	fputs(USAGE_SEPARATOR, out);
+
+	fprintf(out, _(
+		"Options:\n"
+		" -a, --all-tasks         operate on all the tasks (threads) for a given pid\n"
+		" -p, --pid               operate on existing given pid\n"
+		" -c, --cpu-list          display and specify cpus in list format\n"
+		));
+	printf(USAGE_HELP_OPTIONS(25));
+
+	fputs(USAGE_SEPARATOR, out);
+	fprintf(out, _(
+		"The default behavior is to run a new command:\n"
+		"    %1$s 03 sshd -b 1024\n"
+		"You can retrieve the mask of an existing task:\n"
+		"    %1$s -p 700\n"
+		"Or set it:\n"
+		"    %1$s -p 03 700\n"
+		"List format uses a comma-separated list instead of a mask:\n"
+		"    %1$s -pc 0,3,7-11 700\n"
+		"Ranges in list format can take a stride argument:\n"
+		"    e.g. 0-31:2 is equivalent to mask 0x55555555\n"),
+		program_invocation_short_name);
+
+	printf(USAGE_MAN_TAIL("taskset(1)"));
+	exit(EXIT_SUCCESS);
+}
+
+static void print_affinity(struct taskset *ts, int isnew)
+{
+	char *str, *msg;
+
+	if (ts->use_list) {
+		str = cpulist_create(ts->buf, ts->buflen, ts->set, ts->setsize);
+		msg = isnew ? _("pid %d's new affinity list: %s\n") :
+			      _("pid %d's current affinity list: %s\n");
+	} else {
+		str = cpumask_create(ts->buf, ts->buflen, ts->set, ts->setsize);
+		msg = isnew ? _("pid %d's new affinity mask: %s\n") :
+			      _("pid %d's current affinity mask: %s\n");
+	}
+
+	if (!str)
+		errx(EXIT_FAILURE, _("internal error: conversion from cpuset to string failed"));
+
+	printf(msg, ts->pid ? ts->pid : getpid(), str);
+}
+
+static void __attribute__((__noreturn__)) err_affinity(pid_t pid, int set)
+{
+	char *msg;
+
+	msg = set ? _("failed to set pid %d's affinity") :
+		    _("failed to get pid %d's affinity");
+
+	err(EXIT_FAILURE, msg, pid ? pid : getpid());
+}
+
+static void do_taskset(struct taskset *ts, size_t setsize, cpu_set_t *set)
+{
+	/* read the current mask */
+	if (ts->pid) {
+		if (sched_getaffinity(ts->pid, ts->setsize, ts->set) < 0)
+			err_affinity(ts->pid, 1);
+		print_affinity(ts, FALSE);
+	}
+
+	if (ts->get_only)
+		return;
+
+	/* set new mask */
+	if (sched_setaffinity(ts->pid, setsize, set) < 0)
+		err_affinity(ts->pid, 1);
+
+	/* re-read the current mask */
+	if (ts->pid) {
+		if (sched_getaffinity(ts->pid, ts->setsize, ts->set) < 0)
+			err_affinity(ts->pid, 0);
+		print_affinity(ts, TRUE);
+	}
+}
+
+int main(int argc, char **argv)
+{
+	cpu_set_t *new_set;
+	pid_t pid = 0;
+	int c, all_tasks = 0;
+	int ncpus;
+	size_t new_setsize, nbits;
+	struct taskset ts;
+
+	static const struct option longopts[] = {
+		{ "all-tasks",	0, NULL, 'a' },
+		{ "pid",	0, NULL, 'p' },
+		{ "cpu-list",	0, NULL, 'c' },
+		{ "help",	0, NULL, 'h' },
+		{ "version",	0, NULL, 'V' },
+		{ NULL,		0, NULL,  0  }
+	};
+
+	setlocale(LC_ALL, "");
+	bindtextdomain(PACKAGE, LOCALEDIR);
+	textdomain(PACKAGE);
+	atexit(close_stdout);
+
+	memset(&ts, 0, sizeof(ts));
+
+	while ((c = getopt_long(argc, argv, "+apchV", longopts, NULL)) != -1) {
+		switch (c) {
+		case 'a':
+			all_tasks = 1;
+			break;
+		case 'p':
+			pid = strtos32_or_err(argv[argc - 1],
+					    _("invalid PID argument"));
+			break;
+		case 'c':
+			ts.use_list = 1;
+			break;
+		case 'V':
+			printf(UTIL_LINUX_VERSION);
+			return EXIT_SUCCESS;
+		case 'h':
+			usage();
+			break;
+		default:
+			errtryhelp(EXIT_FAILURE);
+		}
+	}
+
+	if ((!pid && argc - optind < 2)
+	    || (pid && (argc - optind < 1 || argc - optind > 2))) {
+		warnx(_("bad usage"));
+		errtryhelp(EXIT_FAILURE);
+	}
+
+	ncpus = get_max_number_of_cpus();
+	if (ncpus <= 0)
+		errx(EXIT_FAILURE, _("cannot determine NR_CPUS; aborting"));
+
+	/*
+	 * the ts->set is always used for the sched_getaffinity call
+	 * On the sched_getaffinity the kernel demands a user mask of
+	 * at least the size of its own cpumask_t.
+	 */
+	ts.set = cpuset_alloc(ncpus, &ts.setsize, &nbits);
+	if (!ts.set)
+		err(EXIT_FAILURE, _("cpuset_alloc failed"));
+
+	/* buffer for conversion from mask to string */
+	ts.buflen = 7 * nbits;
+	ts.buf = xmalloc(ts.buflen);
+
+	/*
+	 * new_set is always used for the sched_setaffinity call
+	 * On the sched_setaffinity the kernel will zero-fill its
+	 * cpumask_t if the user's mask is shorter.
+	 */
+	new_set = cpuset_alloc(ncpus, &new_setsize, NULL);
+	if (!new_set)
+		err(EXIT_FAILURE, _("cpuset_alloc failed"));
+
+	if (argc - optind == 1)
+		ts.get_only = 1;
+
+	else if (ts.use_list) {
+		if (cpulist_parse(argv[optind], new_set, new_setsize, 0))
+			errx(EXIT_FAILURE, _("failed to parse CPU list: %s"),
+			     argv[optind]);
+	} else if (cpumask_parse(argv[optind], new_set, new_setsize)) {
+		errx(EXIT_FAILURE, _("failed to parse CPU mask: %s"),
+		     argv[optind]);
+	}
+
+	if (all_tasks && pid) {
+		struct proc_tasks *tasks = proc_open_tasks(pid);
+		while (!proc_next_tid(tasks, &ts.pid))
+			do_taskset(&ts, new_setsize, new_set);
+		proc_close_tasks(tasks);
+	} else {
+		ts.pid = pid;
+		do_taskset(&ts, new_setsize, new_set);
+	}
+
+	free(ts.buf);
+	cpuset_free(ts.set);
+	cpuset_free(new_set);
+
+	if (!pid) {
+		argv += optind + 1;
+		execvp(argv[0], argv);
+		errexec(argv[0]);
+	}
+
+	return EXIT_SUCCESS;
+}