#ifndef LIBLUSTRE_H__
#define LIBLUSTRE_H__
+#ifdef __KERNEL__
+#error Kernel files should not #include <liblustre.h>
+#else
+/*
+ * The userspace implementations of linux/spinlock.h vary; we just
+ * include our own for all of them
+ */
+#define __LINUX_SPINLOCK_H
+#endif
+
#include <sys/mman.h>
#ifdef HAVE_STDINT_H
# include <stdint.h>
#ifdef HAVE_SYS_USER_H
# include <sys/user.h>
#endif
-
-#include "ioctl.h"
+#ifdef HAVE_SYS_IOCTL_H
+# include <sys/ioctl.h>
+#endif
+#ifndef _IOWR
+# include "ioctl.h"
+#endif
#include <stdio.h>
#include <sys/ioctl.h>
#include <string.h>
#include <errno.h>
#include <sys/stat.h>
-#include <sys/vfs.h>
+#ifdef HAVE_SYS_VFS_H
+# include <sys/vfs.h>
+#endif
+#include <unistd.h>
+#include <fcntl.h>
+#include <assert.h>
#include <libcfs/list.h>
-#include <portals/p30.h>
+#include <lnet/lnet.h>
#include <libcfs/kp30.h>
/* definitions for liblustre */
#ifdef __CYGWIN__
-#define PAGE_SHIFT 12
-#define PAGE_SIZE (1UL << PAGE_SHIFT)
-#define PAGE_MASK (~(PAGE_SIZE-1))
+#define CFS_PAGE_SHIFT 12
+#define CFS_PAGE_SIZE (1UL << CFS_PAGE_SHIFT)
+#define CFS_PAGE_MASK (~((__u64)CFS_PAGE_SIZE-1))
#define loff_t long long
#define ERESTART 2001
typedef unsigned short umode_t;
#endif
+#ifndef CURRENT_SECONDS
+# define CURRENT_SECONDS time(0)
+#endif
+
+#ifndef ARRAY_SIZE
+#define ARRAY_SIZE(a) ((sizeof (a))/(sizeof ((a)[0])))
+#endif
+
/* This is because lprocfs_status.h gets included here indirectly. It would
* be much better to just avoid lprocfs being included into liblustre entirely
* but that requires more header surgery than I can handle right now.
#ifndef smp_processor_id
#define smp_processor_id() 0
#endif
+#ifndef smp_num_cpus
+#define smp_num_cpus 1
+#endif
/* always adopt 2.5 definitions */
#define KERNEL_VERSION(a,b,c) ((a)*100+(b)*10+c)
-#define LINUX_VERSION_CODE (2*200+5*10+0)
+#define LINUX_VERSION_CODE KERNEL_VERSION(2,5,0)
+
+#ifndef page_private
+#define page_private(page) ((page)->private)
+#define set_page_private(page, v) ((page)->private = (v))
+#endif
+
static inline void inter_module_put(void *a)
{
return;
}
-extern ptl_handle_ni_t tcpnal_ni;
-
void *inter_module_get(char *arg);
/* cheats for now */
#define GFP_HIGHUSER 1
#define GFP_ATOMIC 1
#define GFP_NOFS 1
-#define IS_ERR(a) (((a) && abs((long)(a)) < 500) ? 1 : 0)
+#define IS_ERR(a) ((unsigned long)(a) > (unsigned long)-1000L)
#define PTR_ERR(a) ((long)(a))
#define ERR_PTR(a) ((void*)((long)(a)))
-typedef struct {
- void *cwd;
-}mm_segment_t;
-
typedef int (read_proc_t)(char *page, char **start, off_t off,
int count, int *eof, void *data);
#endif /* __LITTLE_ENDIAN */
/* bits ops */
-static __inline__ int set_bit(int nr,long * addr)
+
+/* a long can be more than 32 bits, so use BITS_PER_LONG
+ * to allow the compiler to adjust the bit shifting accordingly
+ */
+
+/* test if bit nr is set in bitmap addr; returns previous value of bit nr */
+static __inline__ int set_bit(int nr, long * addr)
{
- int mask, retval;
+ long mask;
- addr += nr >> 5;
- mask = 1 << (nr & 0x1f);
- retval = (mask & *addr) != 0;
- *addr |= mask;
- return retval;
+ addr += nr / BITS_PER_LONG;
+ mask = 1UL << (nr & (BITS_PER_LONG - 1));
+ nr = (mask & *addr) != 0;
+ *addr |= mask;
+ return nr;
}
+/* clear bit nr in bitmap addr; returns previous value of bit nr*/
static __inline__ int clear_bit(int nr, long * addr)
{
- int mask, retval;
+ long mask;
- addr += nr >> 5;
- mask = 1 << (nr & 0x1f);
- retval = (mask & *addr) != 0;
- *addr &= ~mask;
- return retval;
+ addr += nr / BITS_PER_LONG;
+ mask = 1UL << (nr & (BITS_PER_LONG - 1));
+ nr = (mask & *addr) != 0;
+ *addr &= ~mask;
+ return nr;
}
-static __inline__ int test_bit(int nr, long * addr)
+static __inline__ int test_bit(int nr, const long * addr)
{
- int mask;
-
- addr += nr >> 5;
- mask = 1 << (nr & 0x1f);
- return ((mask & *addr) != 0);
+ return ((1UL << (nr & (BITS_PER_LONG - 1))) & ((addr)[nr / BITS_PER_LONG])) != 0;
}
static __inline__ int ext2_set_bit(int nr, void *addr)
int count;
};
-#define MODULE_AUTHOR(name)
-#define MODULE_DESCRIPTION(name)
-#define MODULE_LICENSE(name)
-
-#define module_init(init)
-#define module_exit(exit)
+static inline void MODULE_AUTHOR(char *name)
+{
+ printf("%s\n", name);
+}
+#define MODULE_DESCRIPTION(name) MODULE_AUTHOR(name)
+#define MODULE_LICENSE(name) MODULE_AUTHOR(name)
#define THIS_MODULE NULL
#define __init
#define __exit
-#define __user
/* devices */
extern int osc_init(void);
extern int lov_init(void);
extern int mdc_init(void);
+extern int lmv_init(void);
+extern int mgc_init(void);
extern int echo_client_init(void);
#define EXPORT_SYMBOL(S)
+struct rcu_head { };
+
typedef struct { } spinlock_t;
typedef __u64 kdev_t;
#define SPIN_LOCK_UNLOCKED (spinlock_t) { }
#define LASSERT_SPIN_LOCKED(lock) do {} while(0)
+#define LASSERT_SEM_LOCKED(sem) do {} while(0)
static inline void spin_lock(spinlock_t *l) {return;}
static inline void spin_unlock(spinlock_t *l) {return;}
static inline void spin_lock_irqsave(spinlock_t *a, unsigned long b) {}
static inline void spin_unlock_irqrestore(spinlock_t *a, unsigned long b) {}
-typedef struct { } rwlock_t;
-#define rwlock_init(x) do {} while(0)
-#define RW_LOCK_UNLOCKED (rwlock_t) {}
-#define read_lock(l)
-#define read_unlock(l)
-#define write_lock(l)
-#define write_unlock(l)
+typedef spinlock_t rwlock_t;
+#define RW_LOCK_UNLOCKED SPIN_LOCK_UNLOCKED
+#define read_lock(l) spin_lock(l)
+#define read_unlock(l) spin_unlock(l)
+#define write_lock(l) spin_lock(l)
+#define write_unlock(l) spin_unlock(l)
+
#define min(x,y) ((x)<(y) ? (x) : (y))
#define max(x,y) ((x)>(y) ? (x) : (y))
#ifndef min_t
#define min_t(type,x,y) \
- ({ type __x = (x); type __y = (y); __x < __y ? __x: __y; })
+ ({ type __x = (x); type __y = (y); __x < __y ? __x: __y; })
#endif
#ifndef max_t
#define max_t(type,x,y) \
- ({ type __x = (x); type __y = (y); __x > __y ? __x: __y; })
+ ({ type __x = (x); type __y = (y); __x > __y ? __x: __y; })
#endif
-#define container_of(ptr, type, member) ({ \
- const typeof( ((type *)0)->member ) *__mptr = (ptr); \
- (type *)( (char *)__mptr - offsetof(type,member) );})
+#define simple_strtol strtol
/* registering symbols */
-
+#ifndef ERESTARTSYS
#define ERESTARTSYS ERESTART
+#endif
#define HZ 1
/* random */
/* memory */
-/* FIXME */
-#define num_physpages (16 * 1024)
+/* memory size: used for some client tunables */
+#define num_physpages (256 * 1024) /* 1GB */
static inline int copy_from_user(void *a,void *b, int c)
{
return 0;
}
-static inline long strncpy_from_user(char *dest, const char *src, long n)
-{
- char *s;
- s = strncpy(dest, src, n);
- return strnlen(s, n);
-}
/* slabs */
typedef struct {
free(a);
return 0;
}
-#define kmem_cache_alloc(cache, prio) malloc(cache->size)
-#define kmem_cache_free(cache, obj) free(obj)
-
-#define PAGE_CACHE_SIZE PAGE_SIZE
-#define PAGE_CACHE_SHIFT PAGE_SHIFT
-#define PAGE_CACHE_MASK PAGE_MASK
/* struct page decl moved out from here into portals/include/libcfs/user-prim.h */
#define PAGE_LIST_ENTRY list
#define PAGE_LIST(page) ((page)->list)
-#define page_address(page) ((page)->addr)
#define kmap(page) (page)->addr
#define kunmap(a) do {} while (0)
-static inline struct page *alloc_pages(int mask, unsigned long order)
+static inline cfs_page_t *alloc_pages(int mask, unsigned long order)
{
- struct page *pg = malloc(sizeof(*pg));
+ cfs_page_t *pg = malloc(sizeof(*pg));
if (!pg)
return NULL;
#if 0 //#ifdef MAP_ANONYMOUS
pg->addr = mmap(0, PAGE_SIZE << order, PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, 0, 0);
#else
- pg->addr = malloc(PAGE_SIZE << order);
+ pg->addr = malloc(CFS_PAGE_SIZE << order);
#endif
if (!pg->addr) {
}
return pg;
}
+#define cfs_alloc_pages(mask, order) alloc_pages((mask), (order))
-#define alloc_page(mask) alloc_pages((mask), 0)
+#define alloc_page(mask) alloc_pages((mask), 0)
+#define cfs_alloc_page(mask) alloc_page(mask)
-static inline void __free_pages(struct page *pg, int what)
+static inline void __free_pages(cfs_page_t *pg, int what)
{
#if 0 //#ifdef MAP_ANONYMOUS
munmap(pg->addr, PAGE_SIZE);
#endif
free(pg);
}
+#define __cfs_free_pages(pg, order) __free_pages((pg), (order))
#define __free_page(page) __free_pages((page), 0)
#define free_page(page) __free_page(page)
+#define __cfs_free_page(page) __cfs_free_pages((page), 0)
-static inline struct page* __grab_cache_page(unsigned long index)
+static inline cfs_page_t* __grab_cache_page(unsigned long index)
{
- struct page *pg = alloc_pages(0, 0);
+ cfs_page_t *pg = alloc_pages(0, 0);
if (pg)
pg->index = index;
#define ATTR_ATTR_FLAG 0x0400
#define ATTR_RAW 0x0800 /* file system, not vfs will massage attrs */
#define ATTR_FROM_OPEN 0x1000 /* called from open path, ie O_TRUNC */
-/* ATTR_CTIME_SET has been defined in lustre_idl.h */
+#define ATTR_CTIME_SET 0x2000
+#define ATTR_BLOCKS 0x4000
struct iattr {
unsigned int ia_valid;
unsigned int ia_attr_flags;
};
+#define ll_iattr iattr
+
#define IT_OPEN 0x0001
#define IT_CREAT 0x0002
#define IT_READDIR 0x0004
#define IT_GETXATTR 0x0040
#define IT_EXEC 0x0080
#define IT_PIN 0x0100
-#define IT_CHDIR 0x0200
#define IT_FL_LOCKED 0x0001
#define IT_FL_FOLLOWED 0x0002 /* set by vfs_follow_link */
#define INTENT_MAGIC 0x19620323
+struct lustre_intent_data {
+ int it_disposition;
+ int it_status;
+ __u64 it_lock_handle;
+ void *it_data;
+ int it_lock_mode;
+ int it_int_flags;
+};
struct lookup_intent {
int it_magic;
void (*it_op_release)(struct lookup_intent *);
int it_op;
int it_flags;
int it_create_mode;
- union {
- void *fs_data; /* FS-specific intent data */
- } d;
+ union {
+ struct lustre_intent_data lustre;
+ } d;
};
-struct lustre_intent_data {
- int it_disposition;
- int it_status;
- __u64 it_lock_handle;
- void *it_data;
- int it_lock_mode;
- int it_int_flags;
- void *it_key;
- int it_key_size;
-};
-
-#define LUSTRE_IT(it) ((struct lustre_intent_data *)((it)->d.fs_data))
-
static inline void intent_init(struct lookup_intent *it, int op, int flags)
{
memset(it, 0, sizeof(*it));
it->it_flags = flags;
}
-
struct dentry {
int d_count;
};
/* use the macro's argument to avoid unused warnings */
#define down(a) do { (void)a; } while (0)
+#define mutex_down(a) down(a)
#define up(a) do { (void)a; } while (0)
+#define mutex_up(a) up(a)
#define down_read(a) do { (void)a; } while (0)
#define up_read(a) do { (void)a; } while (0)
#define down_write(a) do { (void)a; } while (0)
{
sema_init(sem, 1);
}
+static inline void init_MUTEX_LOCKED (struct semaphore *sem)
+{
+ sema_init(sem, 0);
+}
+#define init_mutex(s) init_MUTEX(s)
typedef struct {
struct list_head sleepers;
int signal;
};
-struct fs_struct {
- int umask;
-};
-
struct task_struct {
- struct fs_struct *fs;
int state;
struct signal pending;
char comm[32];
+ int uid;
+ int gid;
int pid;
- uid_t uid;
- gid_t gid;
int fsuid;
int fsgid;
int max_groups;
int ngroups;
gid_t *groups;
__u32 cap_effective;
-
- struct fs_struct __fs;
};
+typedef struct task_struct cfs_task_t;
+#define cfs_current() current
+#define cfs_curproc_pid() (current->pid)
+#define cfs_curproc_comm() (current->comm)
+
extern struct task_struct *current;
int in_group_p(gid_t gid);
static inline int capable(int cap)
_ret = tv.tv_sec; \
_ret; \
})
+#define get_jiffies_64() (__u64)jiffies
#define time_after(a, b) ((long)(b) - (long)(a) < 0)
#define time_before(a, b) time_after(b,a)
-
-static inline unsigned long get_seconds(void)
-{
- struct timeval tv;
-
- gettimeofday(&tv, NULL);
- return (tv.tv_sec + tv.tv_usec / 1000000);
-}
+#define time_after_eq(a,b) ((long)(a) - (long)(b) >= 0)
struct timer_list {
struct list_head tl_list;
static inline int timer_pending(struct timer_list *l)
{
- if (l->expires > jiffies)
+ if (time_after(l->expires, jiffies))
return 1;
else
return 0;
#define atomic_read(a) ((a)->counter)
#define atomic_set(a,b) do {(a)->counter = b; } while (0)
#define atomic_dec_and_test(a) ((--((a)->counter)) == 0)
+#define atomic_dec_and_lock(a,b) ((--((a)->counter)) == 0)
#define atomic_inc(a) (((a)->counter)++)
#define atomic_dec(a) do { (a)->counter--; } while (0)
#define atomic_add(b,a) do {(a)->counter += b;} while (0)
#define unlikely(exp) (exp)
#endif
+#define might_sleep()
+#define might_sleep_if(c)
+#define smp_mb()
+
+static inline
+int test_and_set_bit(int nr, unsigned long *addr)
+{
+ int oldbit;
+
+ while (nr >= sizeof(long)) {
+ nr -= sizeof(long);
+ addr++;
+ }
+
+ oldbit = (*addr) & (1 << nr);
+ *addr |= (1 << nr);
+ return oldbit;
+}
+
+static inline
+int test_and_clear_bit(int nr, unsigned long *addr)
+{
+ int oldbit;
+
+ while (nr >= sizeof(long)) {
+ nr -= sizeof(long);
+ addr++;
+ }
+
+ oldbit = (*addr) & (1 << nr);
+ *addr &= ~(1 << nr);
+ return oldbit;
+}
+
/* FIXME sys/capability will finally included linux/fs.h thus
* cause numerous trouble on x86-64. as temporary solution for
- * build broken at cary, we copy definition we need from capability.h
+ * build broken at Cray, we copy definition we need from capability.h
* FIXME
*/
struct _cap_struct;
cap_t cap_get_proc(void);
int cap_get_flag(cap_t, cap_value_t, cap_flag_t, cap_flag_value_t *);
-
-
/* log related */
static inline int llog_init_commit_master(void) { return 0; }
static inline int llog_cleanup_commit_master(int force) { return 0; }
-static inline void portals_run_lbug_upcall(char *file, const char *fn,
+static inline void libcfs_run_lbug_upcall(char *file, const char *fn,
const int l){}
-#define LBUG() \
- do { \
- printf("!!!LBUG at %s:%d\n", __FILE__, __LINE__); \
- sleep(1000000); \
- } while (0)
-
-
-
/* completion */
struct completion {
unsigned int done;
- wait_queue_head_t wait;
+ cfs_waitq_t wait;
};
#define COMPLETION_INITIALIZER(work) \
struct liblustre_wait_callback {
struct list_head llwc_list;
+ const char *llwc_name;
int (*llwc_fn)(void *arg);
void *llwc_arg;
};
-void *liblustre_register_wait_callback(int (*fn)(void *arg), void *arg);
+void *liblustre_register_wait_callback(const char *name,
+ int (*fn)(void *arg), void *arg);
void liblustre_deregister_wait_callback(void *notifier);
int liblustre_wait_event(int timeout);
-#include <linux/obd_support.h>
-#include <linux/lustre_idl.h>
-#include <linux/lustre_lib.h>
-#include <linux/lustre_import.h>
-#include <linux/lustre_export.h>
-#include <linux/lustre_net.h>
+void *liblustre_register_idle_callback(const char *name,
+ int (*fn)(void *arg), void *arg);
+void liblustre_deregister_idle_callback(void *notifier);
+void liblustre_wait_idle(void);
+
+/* flock related */
+struct nfs_lock_info {
+ __u32 state;
+ __u32 flags;
+ void *host;
+};
+
+typedef struct file_lock {
+ struct file_lock *fl_next; /* singly linked list for this inode */
+ struct list_head fl_link; /* doubly linked list of all locks */
+ struct list_head fl_block; /* circular list of blocked processes */
+ void *fl_owner;
+ unsigned int fl_pid;
+ cfs_waitq_t fl_wait;
+ struct file *fl_file;
+ unsigned char fl_flags;
+ unsigned char fl_type;
+ loff_t fl_start;
+ loff_t fl_end;
+
+ void (*fl_notify)(struct file_lock *); /* unblock callback */
+ void (*fl_insert)(struct file_lock *); /* lock insertion callback */
+ void (*fl_remove)(struct file_lock *); /* lock removal callback */
+
+ void *fl_fasync; /* for lease break notifications */
+ unsigned long fl_break_time; /* for nonblocking lease breaks */
+
+ union {
+ struct nfs_lock_info nfs_fl;
+ } fl_u;
+} cfs_flock_t;
+
+#define cfs_flock_type(fl) ((fl)->fl_type)
+#define cfs_flock_set_type(fl, type) do { (fl)->fl_type = (type); } while(0)
+#define cfs_flock_pid(fl) ((fl)->fl_pid)
+#define cfs_flock_set_pid(fl, pid) do { (fl)->fl_pid = (pid); } while(0)
+#define cfs_flock_start(fl) ((fl)->fl_start)
+#define cfs_flock_set_start(fl, start) do { (fl)->fl_start = (start); } while(0)
+#define cfs_flock_end(fl) ((fl)->fl_end)
+#define cfs_flock_set_end(fl, end) do { (fl)->fl_end = (end); } while(0)
+
+#ifndef OFFSET_MAX
+#define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
+#define OFFSET_MAX INT_LIMIT(loff_t)
+#endif
+
+/* XXX: defined in kernel */
+#define FL_POSIX 1
+#define FL_SLEEP 128
+
+/* quota */
+#define QUOTA_OK 0
+#define NO_QUOTA 1
+
+/* ACL */
+struct posix_acl_entry {
+ short e_tag;
+ unsigned short e_perm;
+ unsigned int e_id;
+};
+
+struct posix_acl {
+ atomic_t a_refcount;
+ unsigned int a_count;
+ struct posix_acl_entry a_entries[0];
+};
+
+typedef struct {
+ __u16 e_tag;
+ __u16 e_perm;
+ __u32 e_id;
+} xattr_acl_entry;
+
+typedef struct {
+ __u32 a_version;
+ xattr_acl_entry a_entries[0];
+} xattr_acl_header;
+
+static inline size_t xattr_acl_size(int count)
+{
+ return sizeof(xattr_acl_header) + count * sizeof(xattr_acl_entry);
+}
+
+static inline
+struct posix_acl * posix_acl_from_xattr(const void *value, size_t size)
+{
+ return NULL;
+}
+
+static inline
+int posix_acl_valid(const struct posix_acl *acl)
+{
+ return 0;
+}
+
+static inline
+void posix_acl_release(struct posix_acl *acl)
+{
+}
+
+#ifndef ENOTSUPP
+#define ENOTSUPP ENOTSUP
+#endif
+
+typedef int mm_segment_t;
+enum {
+ KERNEL_DS,
+ USER_DS
+};
+static inline mm_segment_t get_fs(void)
+{
+ return USER_DS;
+}
+
+static inline void set_fs(mm_segment_t seg)
+{
+}
+
+#include <obd_support.h>
+#include <lustre/lustre_idl.h>
+#include <lustre_lib.h>
+#include <lustre_import.h>
+#include <lustre_export.h>
+#include <lustre_net.h>
+
+/* Fast hashing routine for a long.
+ (C) 2002 William Lee Irwin III, IBM */
+
+/*
+ * Knuth recommends primes in approximately golden ratio to the maximum
+ * integer representable by a machine word for multiplicative hashing.
+ * Chuck Lever verified the effectiveness of this technique:
+ * http://www.citi.umich.edu/techreports/reports/citi-tr-00-1.pdf
+ *
+ * These primes are chosen to be bit-sparse, that is operations on
+ * them can use shifts and additions instead of multiplications for
+ * machines where multiplications are slow.
+ */
+#if BITS_PER_LONG == 32
+/* 2^31 + 2^29 - 2^25 + 2^22 - 2^19 - 2^16 + 1 */
+#define GOLDEN_RATIO_PRIME 0x9e370001UL
+#elif BITS_PER_LONG == 64
+/* 2^63 + 2^61 - 2^57 + 2^54 - 2^51 - 2^18 + 1 */
+#define GOLDEN_RATIO_PRIME 0x9e37fffffffc0001UL
+#else
+#error Define GOLDEN_RATIO_PRIME for your wordsize.
+#endif
+
+static inline unsigned long hash_long(unsigned long val, unsigned int bits)
+{
+ unsigned long hash = val;
+
+#if BITS_PER_LONG == 64
+ /* Sigh, gcc can't optimise this alone like it does for 32 bits. */
+ unsigned long n = hash;
+ n <<= 18;
+ hash -= n;
+ n <<= 33;
+ hash -= n;
+ n <<= 3;
+ hash += n;
+ n <<= 3;
+ hash -= n;
+ n <<= 4;
+ hash += n;
+ n <<= 2;
+ hash += n;
+#else
+ /* On some cpus multiply is faster, on others gcc will do shifts */
+ hash *= GOLDEN_RATIO_PRIME;
+#endif
+
+ /* High bits are more random, so use them. */
+ return hash >> (BITS_PER_LONG - bits);
+}
+
+static inline unsigned long hash_ptr(void *ptr, unsigned int bits)
+{
+ return hash_long((unsigned long)ptr, bits);
+}
#endif