[linux-block.git] / include / linux / futex.h

#ifndef _LINUX_FUTEX_H
#define _LINUX_FUTEX_H

#include <linux/compiler.h>
#include <linux/types.h>

struct inode;
struct mm_struct;
struct task_struct;
union ktime;

/* Second argument to futex syscall */


#define FUTEX_WAIT		0
#define FUTEX_WAKE		1
#define FUTEX_FD		2
#define FUTEX_REQUEUE		3
#define FUTEX_CMP_REQUEUE	4
#define FUTEX_WAKE_OP		5
#define FUTEX_LOCK_PI		6
#define FUTEX_UNLOCK_PI		7
#define FUTEX_TRYLOCK_PI	8
#define FUTEX_WAIT_BITSET	9
#define FUTEX_WAKE_BITSET	10

#define FUTEX_PRIVATE_FLAG	128
#define FUTEX_CLOCK_REALTIME	256
#define FUTEX_CMD_MASK		~(FUTEX_PRIVATE_FLAG | FUTEX_CLOCK_REALTIME)

#define FUTEX_WAIT_PRIVATE	(FUTEX_WAIT | FUTEX_PRIVATE_FLAG)
#define FUTEX_WAKE_PRIVATE	(FUTEX_WAKE | FUTEX_PRIVATE_FLAG)
#define FUTEX_REQUEUE_PRIVATE	(FUTEX_REQUEUE | FUTEX_PRIVATE_FLAG)
#define FUTEX_CMP_REQUEUE_PRIVATE (FUTEX_CMP_REQUEUE | FUTEX_PRIVATE_FLAG)
#define FUTEX_WAKE_OP_PRIVATE	(FUTEX_WAKE_OP | FUTEX_PRIVATE_FLAG)
#define FUTEX_LOCK_PI_PRIVATE	(FUTEX_LOCK_PI | FUTEX_PRIVATE_FLAG)
#define FUTEX_UNLOCK_PI_PRIVATE	(FUTEX_UNLOCK_PI | FUTEX_PRIVATE_FLAG)
#define FUTEX_TRYLOCK_PI_PRIVATE (FUTEX_TRYLOCK_PI | FUTEX_PRIVATE_FLAG)
#define FUTEX_WAIT_BITSET_PRIVATE	(FUTEX_WAIT_BITS | FUTEX_PRIVATE_FLAG)
#define FUTEX_WAKE_BITSET_PRIVATE	(FUTEX_WAKE_BITS | FUTEX_PRIVATE_FLAG)

/*
 * Support for robust futexes: the kernel cleans up held futexes at
 * thread exit time.
 */

/*
 * Per-lock list entry - embedded in user-space locks, somewhere close
 * to the futex field. (Note: user-space uses a double-linked list to
 * achieve O(1) list add and remove, but the kernel only needs to know
 * about the forward link)
 *
 * NOTE: this structure is part of the syscall ABI, and must not be
 * changed.
 */
struct robust_list {
	struct robust_list __user *next;
};

/*
 * Per-thread list head:
 *
 * NOTE: this structure is part of the syscall ABI, and must only be
 * changed if the change is first communicated with the glibc folks.
 * (When an incompatible change is done, we'll increase the structure
 *  size, which glibc will detect)
 */
struct robust_list_head {
	/*
	 * The head of the list. Points back to itself if empty:
	 */
	struct robust_list list;

	/*
	 * This relative offset is set by user-space, it gives the kernel
	 * the relative position of the futex field to examine. This way
	 * we keep userspace flexible, to freely shape its data-structure,
	 * without hardcoding any particular offset into the kernel:
	 */
	long futex_offset;

	/*
	 * The death of the thread may race with userspace setting
	 * up a lock's links. So to handle this race, userspace first
	 * sets this field to the address of the to-be-taken lock,
	 * then does the lock acquire, and then adds itself to the
	 * list, and then clears this field. Hence the kernel will
	 * always have full knowledge of all locks that the thread
	 * _might_ have taken. We check the owner TID in any case,
	 * so only truly owned locks will be handled.
	 */
	struct robust_list __user *list_op_pending;
};

/*
 * Are there any waiters for this robust futex:
 */
#define FUTEX_WAITERS		0x80000000

/*
 * The kernel signals via this bit that a thread holding a futex
 * has exited without unlocking the futex. The kernel also does
 * a FUTEX_WAKE on such futexes, after setting the bit, to wake
 * up any possible waiters:
 */
#define FUTEX_OWNER_DIED	0x40000000

/*
 * The rest of the robust-futex field is for the TID:
 */
#define FUTEX_TID_MASK		0x3fffffff

/*
 * This limit protects against a deliberately circular list.
 * (Not worth introducing an rlimit for it)
 */
#define ROBUST_LIST_LIMIT	2048

/*
 * bitset with all bits set for the FUTEX_xxx_BITSET OPs to request a
 * match of any bit.
 */
#define FUTEX_BITSET_MATCH_ANY	0xffffffff

#ifdef __KERNEL__
long do_futex(u32 __user *uaddr, int op, u32 val, union ktime *timeout,
	      u32 __user *uaddr2, u32 val2, u32 val3);

extern int
handle_futex_death(u32 __user *uaddr, struct task_struct *curr, int pi);

/*
 * Futexes are matched on equal values of this key.
 * The key type depends on whether it's a shared or private mapping.
 * Don't rearrange members without looking at hash_futex().
 *
 * offset is aligned to a multiple of sizeof(u32) (== 4) by definition.
 * We use the two low order bits of offset to tell what is the kind of key :
 *  00 : Private process futex (PTHREAD_PROCESS_PRIVATE)
 *       (no reference on an inode or mm)
 *  01 : Shared futex (PTHREAD_PROCESS_SHARED)
 *	mapped on a file (reference on the underlying inode)
 *  10 : Shared futex (PTHREAD_PROCESS_SHARED)
 *       (but private mapping on an mm, and reference taken on it)
*/

#define FUT_OFF_INODE    1 /* We set bit 0 if key has a reference on inode */
#define FUT_OFF_MMSHARED 2 /* We set bit 1 if key has a reference on mm */

union futex_key {
	struct {
		unsigned long pgoff;
		struct inode *inode;
		int offset;
	} shared;
	struct {
		unsigned long address;
		struct mm_struct *mm;
		int offset;
	} private;
	struct {
		unsigned long word;
		void *ptr;
		int offset;
	} both;
};

#define FUTEX_KEY_INIT (union futex_key) { .both = { .ptr = NULL } }

#ifdef CONFIG_FUTEX
extern void exit_robust_list(struct task_struct *curr);
extern void exit_pi_state_list(struct task_struct *curr);
extern int futex_cmpxchg_enabled;
#else
static inline void exit_robust_list(struct task_struct *curr)
{
}
static inline void exit_pi_state_list(struct task_struct *curr)
{
}
#endif
#endif /* __KERNEL__ */

#define FUTEX_OP_SET		0	/* *(int *)UADDR2 = OPARG; */
#define FUTEX_OP_ADD		1	/* *(int *)UADDR2 += OPARG; */
#define FUTEX_OP_OR		2	/* *(int *)UADDR2 |= OPARG; */
#define FUTEX_OP_ANDN		3	/* *(int *)UADDR2 &= ~OPARG; */
#define FUTEX_OP_XOR		4	/* *(int *)UADDR2 ^= OPARG; */

#define FUTEX_OP_OPARG_SHIFT	8	/* Use (1 << OPARG) instead of OPARG.  */

#define FUTEX_OP_CMP_EQ		0	/* if (oldval == CMPARG) wake */
#define FUTEX_OP_CMP_NE		1	/* if (oldval != CMPARG) wake */
#define FUTEX_OP_CMP_LT		2	/* if (oldval < CMPARG) wake */
#define FUTEX_OP_CMP_LE		3	/* if (oldval <= CMPARG) wake */
#define FUTEX_OP_CMP_GT		4	/* if (oldval > CMPARG) wake */
#define FUTEX_OP_CMP_GE		5	/* if (oldval >= CMPARG) wake */

/* FUTEX_WAKE_OP will perform atomically
   int oldval = *(int *)UADDR2;
   *(int *)UADDR2 = oldval OP OPARG;
   if (oldval CMP CMPARG)
     wake UADDR2;  */

#define FUTEX_OP(op, oparg, cmp, cmparg) \
  (((op & 0xf) << 28) | ((cmp & 0xf) << 24)		\
   | ((oparg & 0xfff) << 12) | (cmparg & 0xfff))

#endif
Commit	Line	Data
1da177e4 LT	1	#ifndef _LINUX_FUTEX_H
	2	#define _LINUX_FUTEX_H
	3
286100a6 AD	4	#include <linux/compiler.h>
286100a6 AD	5	#include <linux/types.h>
0771dfef	6
286100a6 AD	7	struct inode;
	8	struct mm_struct;
	9	struct task_struct;
c19384b5 PP	10	union ktime;
c19384b5 PP	11
1da177e4 LT	12	/* Second argument to futex syscall */
	13
	14
4732efbe JJ	15	#define FUTEX_WAIT 0
	16	#define FUTEX_WAKE 1
	17	#define FUTEX_FD 2
	18	#define FUTEX_REQUEUE 3
	19	#define FUTEX_CMP_REQUEUE 4
	20	#define FUTEX_WAKE_OP 5
c87e2837 IM	21	#define FUTEX_LOCK_PI 6
	22	#define FUTEX_UNLOCK_PI 7
	23	#define FUTEX_TRYLOCK_PI 8
cd689985 TG	24	#define FUTEX_WAIT_BITSET 9
cd689985 TG	25	#define FUTEX_WAKE_BITSET 10
1da177e4	26
34f01cc1	27	#define FUTEX_PRIVATE_FLAG 128
1acdac10 TG	28	#define FUTEX_CLOCK_REALTIME 256
1acdac10 TG	29	#define FUTEX_CMD_MASK ~(FUTEX_PRIVATE_FLAG \| FUTEX_CLOCK_REALTIME)
34f01cc1 ED	30
	31	#define FUTEX_WAIT_PRIVATE (FUTEX_WAIT \| FUTEX_PRIVATE_FLAG)
	32	#define FUTEX_WAKE_PRIVATE (FUTEX_WAKE \| FUTEX_PRIVATE_FLAG)
	33	#define FUTEX_REQUEUE_PRIVATE (FUTEX_REQUEUE \| FUTEX_PRIVATE_FLAG)
	34	#define FUTEX_CMP_REQUEUE_PRIVATE (FUTEX_CMP_REQUEUE \| FUTEX_PRIVATE_FLAG)
	35	#define FUTEX_WAKE_OP_PRIVATE (FUTEX_WAKE_OP \| FUTEX_PRIVATE_FLAG)
	36	#define FUTEX_LOCK_PI_PRIVATE (FUTEX_LOCK_PI \| FUTEX_PRIVATE_FLAG)
	37	#define FUTEX_UNLOCK_PI_PRIVATE (FUTEX_UNLOCK_PI \| FUTEX_PRIVATE_FLAG)
	38	#define FUTEX_TRYLOCK_PI_PRIVATE (FUTEX_TRYLOCK_PI \| FUTEX_PRIVATE_FLAG)
cd689985 TG	39	#define FUTEX_WAIT_BITSET_PRIVATE (FUTEX_WAIT_BITS \| FUTEX_PRIVATE_FLAG)
cd689985 TG	40	#define FUTEX_WAKE_BITSET_PRIVATE (FUTEX_WAKE_BITS \| FUTEX_PRIVATE_FLAG)
34f01cc1	41
0771dfef IM	42	/*
	43	* Support for robust futexes: the kernel cleans up held futexes at
	44	* thread exit time.
	45	*/
	46
	47	/*
	48	* Per-lock list entry - embedded in user-space locks, somewhere close
	49	* to the futex field. (Note: user-space uses a double-linked list to
	50	* achieve O(1) list add and remove, but the kernel only needs to know
	51	* about the forward link)
	52	*
	53	* NOTE: this structure is part of the syscall ABI, and must not be
	54	* changed.
	55	*/
	56	struct robust_list {
	57	struct robust_list __user *next;
	58	};
	59
	60	/*
	61	* Per-thread list head:
	62	*
	63	* NOTE: this structure is part of the syscall ABI, and must only be
	64	* changed if the change is first communicated with the glibc folks.
	65	* (When an incompatible change is done, we'll increase the structure
	66	* size, which glibc will detect)
	67	*/
	68	struct robust_list_head {
	69	/*
	70	* The head of the list. Points back to itself if empty:
	71	*/
	72	struct robust_list list;
	73
	74	/*
	75	* This relative offset is set by user-space, it gives the kernel
	76	* the relative position of the futex field to examine. This way
	77	* we keep userspace flexible, to freely shape its data-structure,
	78	* without hardcoding any particular offset into the kernel:
	79	*/
	80	long futex_offset;
	81
	82	/*
	83	* The death of the thread may race with userspace setting
	84	* up a lock's links. So to handle this race, userspace first
	85	* sets this field to the address of the to-be-taken lock,
	86	* then does the lock acquire, and then adds itself to the
	87	* list, and then clears this field. Hence the kernel will
	88	* always have full knowledge of all locks that the thread
	89	* _might_ have taken. We check the owner TID in any case,
	90	* so only truly owned locks will be handled.
	91	*/
	92	struct robust_list __user *list_op_pending;
	93	};
	94
	95	/*
	96	* Are there any waiters for this robust futex:
	97	*/
	98	#define FUTEX_WAITERS 0x80000000
	99
	100	/*
	101	* The kernel signals via this bit that a thread holding a futex
	102	* has exited without unlocking the futex. The kernel also does
	103	* a FUTEX_WAKE on such futexes, after setting the bit, to wake
	104	* up any possible waiters:
	105	*/
106	#define FUTEX_OWNER_DIED 0x40000000
107
0771dfef IM	108	/*
	109	* The rest of the robust-futex field is for the TID:
	110	*/
bd197234	111	#define FUTEX_TID_MASK 0x3fffffff
0771dfef IM	112
0771dfef IM	113	/*
76b81e2b IM	114	* This limit protects against a deliberately circular list.
76b81e2b IM	115	* (Not worth introducing an rlimit for it)
0771dfef	116	*/
76b81e2b	117	#define ROBUST_LIST_LIMIT 2048
0771dfef	118
cd689985 TG	119	/*
	120	* bitset with all bits set for the FUTEX_xxx_BITSET OPs to request a
	121	* match of any bit.
	122	*/
	123	#define FUTEX_BITSET_MATCH_ANY 0xffffffff
	124
58f64d83	125	#ifdef __KERNEL__
c19384b5	126	long do_futex(u32 __user uaddr, int op, u32 val, union ktime timeout,
e2970f2f	127	u32 __user *uaddr2, u32 val2, u32 val3);
1da177e4	128
e3f2ddea IM	129	extern int
e3f2ddea IM	130	handle_futex_death(u32 __user uaddr, struct task_struct curr, int pi);
0771dfef	131
9adef58b RR	132	/*
	133	* Futexes are matched on equal values of this key.
	134	* The key type depends on whether it's a shared or private mapping.
	135	* Don't rearrange members without looking at hash_futex().
	136	*
	137	* offset is aligned to a multiple of sizeof(u32) (== 4) by definition.
34f01cc1 ED	138	* We use the two low order bits of offset to tell what is the kind of key :
	139	* 00 : Private process futex (PTHREAD_PROCESS_PRIVATE)
	140	* (no reference on an inode or mm)
	141	* 01 : Shared futex (PTHREAD_PROCESS_SHARED)
	142	* mapped on a file (reference on the underlying inode)
	143	* 10 : Shared futex (PTHREAD_PROCESS_SHARED)
	144	* (but private mapping on an mm, and reference taken on it)
	145	*/
	146
	147	#define FUT_OFF_INODE 1 /* We set bit 0 if key has a reference on inode */
	148	#define FUT_OFF_MMSHARED 2 /* We set bit 1 if key has a reference on mm */
	149
9adef58b RR	150	union futex_key {
	151	struct {
	152	unsigned long pgoff;
	153	struct inode *inode;
	154	int offset;
	155	} shared;
	156	struct {
	157	unsigned long address;
	158	struct mm_struct *mm;
	159	int offset;
	160	} private;
	161	struct {
	162	unsigned long word;
	163	void *ptr;
	164	int offset;
	165	} both;
	166	};
9adef58b	167
38d47c1b PZ	168	#define FUTEX_KEY_INIT (union futex_key) { .both = { .ptr = NULL } }
38d47c1b PZ	169
0771dfef IM	170	#ifdef CONFIG_FUTEX
0771dfef IM	171	extern void exit_robust_list(struct task_struct *curr);
c87e2837	172	extern void exit_pi_state_list(struct task_struct *curr);
a0c1e907	173	extern int futex_cmpxchg_enabled;
0771dfef IM	174	#else
	175	static inline void exit_robust_list(struct task_struct *curr)
	176	{
	177	}
c87e2837 IM	178	static inline void exit_pi_state_list(struct task_struct *curr)
	179	{
	180	}
0771dfef	181	#endif
58f64d83	182	#endif /* __KERNEL__ */
0771dfef	183
4732efbe JJ	184	#define FUTEX_OP_SET 0 /* (int )UADDR2 = OPARG; */
	185	#define FUTEX_OP_ADD 1 /* (int )UADDR2 += OPARG; */
	186	#define FUTEX_OP_OR 2 /* (int )UADDR2 \|= OPARG; */
	187	#define FUTEX_OP_ANDN 3 /* (int )UADDR2 &= ~OPARG; */
	188	#define FUTEX_OP_XOR 4 /* (int )UADDR2 ^= OPARG; */
	189
	190	#define FUTEX_OP_OPARG_SHIFT 8 /* Use (1 << OPARG) instead of OPARG. */
	191
	192	#define FUTEX_OP_CMP_EQ 0 /* if (oldval == CMPARG) wake */
	193	#define FUTEX_OP_CMP_NE 1 /* if (oldval != CMPARG) wake */
	194	#define FUTEX_OP_CMP_LT 2 /* if (oldval < CMPARG) wake */
	195	#define FUTEX_OP_CMP_LE 3 /* if (oldval <= CMPARG) wake */
	196	#define FUTEX_OP_CMP_GT 4 /* if (oldval > CMPARG) wake */
	197	#define FUTEX_OP_CMP_GE 5 /* if (oldval >= CMPARG) wake */
	198
	199	/* FUTEX_WAKE_OP will perform atomically
	200	int oldval = (int )UADDR2;
	201	(int )UADDR2 = oldval OP OPARG;
	202	if (oldval CMP CMPARG)
	203	wake UADDR2; */
	204
	205	#define FUTEX_OP(op, oparg, cmp, cmparg) \
	206	(((op & 0xf) << 28) \| ((cmp & 0xf) << 24) \
	207	\| ((oparg & 0xfff) << 12) \| (cmparg & 0xfff))
	208
1da177e4	209	#endif