[fio.git] / os / os-linux.h

#ifndef FIO_OS_LINUX_H
#define FIO_OS_LINUX_H

#include <sys/ioctl.h>
#include <sys/uio.h>
#include <sys/syscall.h>
#include <unistd.h>
#include <fcntl.h>
#include <linux/unistd.h>
#include <linux/raw.h>
#include <linux/major.h>

#include "indirect.h"

#define FIO_HAVE_LIBAIO
#define FIO_HAVE_POSIXAIO
#define FIO_HAVE_FADVISE
#define FIO_HAVE_CPU_AFFINITY
#define FIO_HAVE_DISK_UTIL
#define FIO_HAVE_SGIO
#define FIO_HAVE_IOPRIO
#define FIO_HAVE_SPLICE
#define FIO_HAVE_IOSCHED_SWITCH
#define FIO_HAVE_ODIRECT
#define FIO_HAVE_HUGETLB
#define FIO_HAVE_RAWBIND
#define FIO_HAVE_BLKTRACE
#define FIO_HAVE_STRSEP

#define OS_MAP_ANON		(MAP_ANONYMOUS)

#ifndef CLOCK_MONOTONIC
#define CLOCK_MONOTONIC 1
#endif

#ifdef FIO_HAVE_CPU_AFFINITY
typedef cpu_set_t os_cpu_mask_t;
#else
typedef int os_cpu_mask_t;
#endif
typedef struct drand48_data os_random_state_t;

/*
 * we want fadvise64 really, but it's so tangled... later
 */
#ifdef FIO_HAVE_FADVISE
#define fadvise(fd, off, len, advice)	\
	posix_fadvise((fd), (off_t)(off), (len), (advice))
#endif

/*
 * If you are on an ancient glibc (2.3.2), then define GLIBC_2_3_2 if you want
 * the affinity helpers to work.
 */
#ifdef FIO_HAVE_CPU_AFFINITY
#ifndef GLIBC_2_3_2
#define fio_setaffinity(td)		\
	sched_setaffinity((td)->pid, sizeof((td)->o.cpumask), &(td)->o.cpumask)
#define fio_getaffinity(pid, ptr)	\
	sched_getaffinity((pid), sizeof(cpu_set_t), (ptr))
#else
#define fio_setaffinity(td)		\
	sched_setaffinity((td)->pid, &(td)->o.cpumask)
#define fio_getaffinity(pid, ptr)	\
	sched_getaffinity((pid), (ptr))
#endif
#endif

static inline int ioprio_set(int which, int who, int ioprio)
{
	return syscall(__NR_ioprio_set, which, who, ioprio);
}

/*
 * Just check for SPLICE_F_MOVE, if that isn't there, assume the others
 * aren't either.
 */
#ifndef SPLICE_F_MOVE
#define SPLICE_F_MOVE	(0x01)	/* move pages instead of copying */
#define SPLICE_F_NONBLOCK (0x02) /* don't block on the pipe splicing (but */
				 /* we may still block on the fd we splice */
				 /* from/to, of course */
#define SPLICE_F_MORE	(0x04)	/* expect more data */
#define SPLICE_F_GIFT   (0x08)  /* pages passed in are a gift */

static inline int splice(int fdin, loff_t *off_in, int fdout, loff_t *off_out,
			 size_t len, unsigned int flags)
{
	return syscall(__NR_sys_splice, fdin, off_in, fdout, off_out, len, flags);
}

static inline int tee(int fdin, int fdout, size_t len, unsigned int flags)
{
	return syscall(__NR_sys_tee, fdin, fdout, len, flags);
}

static inline int vmsplice(int fd, const struct iovec *iov,
			   unsigned long nr_segs, unsigned int flags)
{
	return syscall(__NR_sys_vmsplice, fd, iov, nr_segs, flags);
}
#endif

#define SPLICE_DEF_SIZE	(64*1024)

#ifdef FIO_HAVE_SYSLET

struct syslet_uatom;
struct async_head_user;

/*
 * syslet stuff
 */
static inline struct syslet_uatom *
async_exec(struct syslet_uatom *atom, struct async_head_user *ahu)
{
	return (void *) syscall(__NR_async_exec, atom, ahu);
}

static inline long
async_wait(unsigned long min_wait_events, unsigned long user_ring_idx,
	   struct async_head_user *ahu)
{
	return syscall(__NR_async_wait, min_wait_events,
			user_ring_idx, ahu);
}

static inline long async_thread(void *event, struct async_head_user *ahu)
{
	return syscall(__NR_async_thread, event, ahu);
}

static inline long umem_add(unsigned long *uptr, unsigned long inc)
{
	return syscall(__NR_umem_add, uptr, inc);
}
#endif /* FIO_HAVE_SYSLET */

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_BITS		16
#define IOPRIO_CLASS_SHIFT	13

#ifndef BLKGETSIZE64
#define BLKGETSIZE64	_IOR(0x12,114,size_t)
#endif

#ifndef BLKFLSBUF
#define BLKFLSBUF	_IO(0x12,97)
#endif

static inline int blockdev_invalidate_cache(int fd)
{
	return ioctl(fd, BLKFLSBUF);
}

static inline int blockdev_size(int fd, unsigned long long *bytes)
{
	if (!ioctl(fd, BLKGETSIZE64, bytes))
		return 0;

	return errno;
}

static inline unsigned long long os_phys_mem(void)
{
	long pagesize, pages;

	pagesize = sysconf(_SC_PAGESIZE);
	pages = sysconf(_SC_PHYS_PAGES);
	if (pages == -1 || pagesize == -1)
		return 0;

	return (unsigned long long) pages * (unsigned long long) pagesize;
}

static inline void os_random_seed(unsigned long seed, os_random_state_t *rs)
{
	srand48_r(seed, rs);
}

static inline long os_random_long(os_random_state_t *rs)
{
	long val;

	lrand48_r(rs, &val);
	return val;
}

static inline int fio_lookup_raw(dev_t dev, int *majdev, int *mindev)
{
	struct raw_config_request rq;
	int fd;

	if (major(dev) != RAW_MAJOR)
		return 1;

	/*
	 * we should be able to find /dev/rawctl or /dev/raw/rawctl
	 */
	fd = open("/dev/rawctl", O_RDONLY);
	if (fd < 0) {
		fd = open("/dev/raw/rawctl", O_RDONLY);
		if (fd < 0)
			return 1;
	}

	rq.raw_minor = minor(dev);
	if (ioctl(fd, RAW_GETBIND, &rq) < 0) {
		close(fd);
		return 1;
	}

	close(fd);
	*majdev = rq.block_major;
	*mindev = rq.block_minor;
	return 0;
}

#define FIO_O_NOATIME	O_NOATIME

#endif
Commit	Line	Data
	1	#ifndef FIO_OS_LINUX_H
	2	#define FIO_OS_LINUX_H
	3
	4	#include <sys/ioctl.h>
	5	#include <sys/uio.h>
	6	#include <sys/syscall.h>
	7	#include <unistd.h>
	8	#include <fcntl.h>
	9	#include <linux/unistd.h>
	10	#include <linux/raw.h>
	11	#include <linux/major.h>
	12
	13	#include "indirect.h"
	14
	15	#define FIO_HAVE_LIBAIO
	16	#define FIO_HAVE_POSIXAIO
	17	#define FIO_HAVE_FADVISE
	18	#define FIO_HAVE_CPU_AFFINITY
	19	#define FIO_HAVE_DISK_UTIL
	20	#define FIO_HAVE_SGIO
	21	#define FIO_HAVE_IOPRIO
	22	#define FIO_HAVE_SPLICE
	23	#define FIO_HAVE_IOSCHED_SWITCH
	24	#define FIO_HAVE_ODIRECT
	25	#define FIO_HAVE_HUGETLB
	26	#define FIO_HAVE_RAWBIND
	27	#define FIO_HAVE_BLKTRACE
	28	#define FIO_HAVE_STRSEP
	29
	30	#define OS_MAP_ANON (MAP_ANONYMOUS)
	31
	32	#ifndef CLOCK_MONOTONIC
	33	#define CLOCK_MONOTONIC 1
	34	#endif
	35
	36	#ifdef FIO_HAVE_CPU_AFFINITY
	37	typedef cpu_set_t os_cpu_mask_t;
	38	#else
	39	typedef int os_cpu_mask_t;
	40	#endif
	41	typedef struct drand48_data os_random_state_t;
	42
	43	/*
	44	* we want fadvise64 really, but it's so tangled... later
	45	*/
	46	#ifdef FIO_HAVE_FADVISE
	47	#define fadvise(fd, off, len, advice) \
	48	posix_fadvise((fd), (off_t)(off), (len), (advice))
	49	#endif
	50
	51	/*
	52	* If you are on an ancient glibc (2.3.2), then define GLIBC_2_3_2 if you want
	53	* the affinity helpers to work.
	54	*/
	55	#ifdef FIO_HAVE_CPU_AFFINITY
	56	#ifndef GLIBC_2_3_2
	57	#define fio_setaffinity(td) \
	58	sched_setaffinity((td)->pid, sizeof((td)->o.cpumask), &(td)->o.cpumask)
	59	#define fio_getaffinity(pid, ptr) \
	60	sched_getaffinity((pid), sizeof(cpu_set_t), (ptr))
	61	#else
	62	#define fio_setaffinity(td) \
	63	sched_setaffinity((td)->pid, &(td)->o.cpumask)
	64	#define fio_getaffinity(pid, ptr) \
	65	sched_getaffinity((pid), (ptr))
	66	#endif
	67	#endif
	68
	69	static inline int ioprio_set(int which, int who, int ioprio)
	70	{
	71	return syscall(__NR_ioprio_set, which, who, ioprio);
	72	}
	73
	74	/*
	75	* Just check for SPLICE_F_MOVE, if that isn't there, assume the others
	76	* aren't either.
	77	*/
	78	#ifndef SPLICE_F_MOVE
	79	#define SPLICE_F_MOVE (0x01) /* move pages instead of copying */
	80	#define SPLICE_F_NONBLOCK (0x02) /* don't block on the pipe splicing (but */
	81	/* we may still block on the fd we splice */
	82	/* from/to, of course */
	83	#define SPLICE_F_MORE (0x04) /* expect more data */
	84	#define SPLICE_F_GIFT (0x08) /* pages passed in are a gift */
	85
	86	static inline int splice(int fdin, loff_t off_in, int fdout, loff_t off_out,
	87	size_t len, unsigned int flags)
	88	{
	89	return syscall(__NR_sys_splice, fdin, off_in, fdout, off_out, len, flags);
	90	}
	91
	92	static inline int tee(int fdin, int fdout, size_t len, unsigned int flags)
	93	{
	94	return syscall(__NR_sys_tee, fdin, fdout, len, flags);
	95	}
	96
	97	static inline int vmsplice(int fd, const struct iovec *iov,
	98	unsigned long nr_segs, unsigned int flags)
	99	{
	100	return syscall(__NR_sys_vmsplice, fd, iov, nr_segs, flags);
	101	}
	102	#endif
	103
	104	#define SPLICE_DEF_SIZE (64*1024)
	105
	106	#ifdef FIO_HAVE_SYSLET
	107
	108	struct syslet_uatom;
	109	struct async_head_user;
	110
	111	/*
	112	* syslet stuff
	113	*/
	114	static inline struct syslet_uatom *
	115	async_exec(struct syslet_uatom atom, struct async_head_user ahu)
	116	{
	117	return (void *) syscall(__NR_async_exec, atom, ahu);
	118	}
	119
	120	static inline long
	121	async_wait(unsigned long min_wait_events, unsigned long user_ring_idx,
	122	struct async_head_user *ahu)
	123	{
	124	return syscall(__NR_async_wait, min_wait_events,
	125	user_ring_idx, ahu);
	126	}
	127
	128	static inline long async_thread(void event, struct async_head_user ahu)
	129	{
	130	return syscall(__NR_async_thread, event, ahu);
	131	}
	132
	133	static inline long umem_add(unsigned long *uptr, unsigned long inc)
	134	{
	135	return syscall(__NR_umem_add, uptr, inc);
	136	}
	137	#endif /* FIO_HAVE_SYSLET */
	138
	139	enum {
	140	IOPRIO_CLASS_NONE,
	141	IOPRIO_CLASS_RT,
	142	IOPRIO_CLASS_BE,
	143	IOPRIO_CLASS_IDLE,
	144	};
	145
	146	enum {
	147	IOPRIO_WHO_PROCESS = 1,
	148	IOPRIO_WHO_PGRP,
	149	IOPRIO_WHO_USER,
	150	};
	151
	152	#define IOPRIO_BITS 16
	153	#define IOPRIO_CLASS_SHIFT 13
	154
	155	#ifndef BLKGETSIZE64
	156	#define BLKGETSIZE64 _IOR(0x12,114,size_t)
	157	#endif
	158
	159	#ifndef BLKFLSBUF
	160	#define BLKFLSBUF _IO(0x12,97)
	161	#endif
	162
	163	static inline int blockdev_invalidate_cache(int fd)
	164	{
	165	return ioctl(fd, BLKFLSBUF);
	166	}
	167
	168	static inline int blockdev_size(int fd, unsigned long long *bytes)
	169	{
	170	if (!ioctl(fd, BLKGETSIZE64, bytes))
	171	return 0;
	172
	173	return errno;
	174	}
	175
	176	static inline unsigned long long os_phys_mem(void)
	177	{
	178	long pagesize, pages;
	179
	180	pagesize = sysconf(_SC_PAGESIZE);
	181	pages = sysconf(_SC_PHYS_PAGES);
	182	if (pages == -1 \|\| pagesize == -1)
	183	return 0;
	184
	185	return (unsigned long long) pages * (unsigned long long) pagesize;
	186	}
	187
	188	static inline void os_random_seed(unsigned long seed, os_random_state_t *rs)
	189	{
	190	srand48_r(seed, rs);
	191	}
	192
	193	static inline long os_random_long(os_random_state_t *rs)
	194	{
	195	long val;
	196
	197	lrand48_r(rs, &val);
	198	return val;
	199	}
	200
	201	static inline int fio_lookup_raw(dev_t dev, int majdev, int mindev)
	202	{
	203	struct raw_config_request rq;
	204	int fd;
	205
	206	if (major(dev) != RAW_MAJOR)
	207	return 1;
	208
	209	/*
	210	* we should be able to find /dev/rawctl or /dev/raw/rawctl
	211	*/
	212	fd = open("/dev/rawctl", O_RDONLY);
	213	if (fd < 0) {
	214	fd = open("/dev/raw/rawctl", O_RDONLY);
	215	if (fd < 0)
	216	return 1;
	217	}
	218
	219	rq.raw_minor = minor(dev);
	220	if (ioctl(fd, RAW_GETBIND, &rq) < 0) {
	221	close(fd);
	222	return 1;
	223	}
	224
	225	close(fd);
	226	*majdev = rq.block_major;
	227	*mindev = rq.block_minor;
	228	return 0;
	229	}
	230
	231	#define FIO_O_NOATIME O_NOATIME
	232
	233	#endif