1/*-
  2 * SPDX-License-Identifier: BSD-3-Clause
  3 *
  4 * Copyright (c) 1982, 1986, 1989, 1993
  5 *	The Regents of the University of California.  All rights reserved.
  6 * (c) UNIX System Laboratories, Inc.
  7 * All or some portions of this file are derived from material licensed
  8 * to the University of California by American Telephone and Telegraph
  9 * Co. or Unix System Laboratories, Inc. and are reproduced herein with
 10 * the permission of UNIX System Laboratories, Inc.
 11 *
 12 * Redistribution and use in source and binary forms, with or without
 13 * modification, are permitted provided that the following conditions
 14 * are met:
 15 * 1. Redistributions of source code must retain the above copyright
 16 *    notice, this list of conditions and the following disclaimer.
 17 * 2. Redistributions in binary form must reproduce the above copyright
 18 *    notice, this list of conditions and the following disclaimer in the
 19 *    documentation and/or other materials provided with the distribution.
 20 * 3. Neither the name of the University nor the names of its contributors
 21 *    may be used to endorse or promote products derived from this software
 22 *    without specific prior written permission.
 23 *
 24 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
 25 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
 26 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
 27 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
 28 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
 29 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
 30 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
 31 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
 32 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
 33 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
 34 * SUCH DAMAGE.
 35 *
 36 *	@(#)buf.h	8.9 (Berkeley) 3/30/95
 37 */
 38
 39#ifndef _SYS_BUF_H_
 40#define	_SYS_BUF_H_
 41
 42#include <sys/bufobj.h>
 43#include <sys/queue.h>
 44#include <sys/lock.h>
 45#include <sys/lockmgr.h>
 46#include <vm/uma.h>
 47
 48struct bio;
 49struct buf;
 50struct bufobj;
 51struct mount;
 52struct vnode;
 53struct uio;
 54
 55/*
 56 * To avoid including <ufs/ffs/softdep.h> 
 57 */   
 58LIST_HEAD(workhead, worklist);
 59/*
 60 * These are currently used only by the soft dependency code, hence
 61 * are stored once in a global variable. If other subsystems wanted
 62 * to use these hooks, a pointer to a set of bio_ops could be added
 63 * to each buffer.
 64 */
 65extern struct bio_ops {
 66	void	(*io_start)(struct buf *);
 67	void	(*io_complete)(struct buf *);
 68	void	(*io_deallocate)(struct buf *);
 69	int	(*io_countdeps)(struct buf *, int);
 70} bioops;
 71
 72struct vm_object;
 73struct vm_page;
 74
 75typedef uint32_t b_xflags_t;
 76
 77/*
 78 * The buffer header describes an I/O operation in the kernel.
 79 *
 80 * NOTES:
 81 *	b_bufsize, b_bcount.  b_bufsize is the allocation size of the
 82 *	buffer, either DEV_BSIZE or PAGE_SIZE aligned.  b_bcount is the
 83 *	originally requested buffer size and can serve as a bounds check
 84 *	against EOF.  For most, but not all uses, b_bcount == b_bufsize.
 85 *
 86 *	b_dirtyoff, b_dirtyend.  Buffers support piecemeal, unaligned
 87 *	ranges of dirty data that need to be written to backing store.
 88 *	The range is typically clipped at b_bcount ( not b_bufsize ).
 89 *
 90 *	b_resid.  Number of bytes remaining in I/O.  After an I/O operation
 91 *	completes, b_resid is usually 0 indicating 100% success.
 92 *
 93 *	All fields are protected by the buffer lock except those marked:
 94 *		V - Protected by owning bufobj lock
 95 *		Q - Protected by the buf queue lock
 96 *		D - Protected by an dependency implementation specific lock
 97 */
 98struct buf {
 99	struct bufobj	*b_bufobj;
100	long		b_bcount;
101	void		*b_caller1;
102	caddr_t		b_data;
103	int		b_error;
104	uint16_t	b_iocmd;	/* BIO_* bio_cmd from bio.h */
105	uint16_t	b_ioflags;	/* BIO_* bio_flags from bio.h */
106	off_t		b_iooffset;
107	long		b_resid;
108	void	(*b_iodone)(struct buf *);
109	void	(*b_ckhashcalc)(struct buf *);
110	uint64_t	b_ckhash;	/* B_CKHASH requested check-hash */
111	daddr_t b_blkno;		/* Underlying physical block number. */
112	off_t	b_offset;		/* Offset into file. */
113	TAILQ_ENTRY(buf) b_bobufs;	/* (V) Buffer's associated vnode. */
114	uint32_t	b_vflags;	/* (V) BV_* flags */
115	uint8_t		b_qindex;	/* (Q) buffer queue index */
116	uint8_t		b_domain;	/* (Q) buf domain this resides in */
117	uint16_t	b_subqueue;	/* (Q) per-cpu q if any */
118	uint32_t	b_flags;	/* B_* flags. */
119	b_xflags_t b_xflags;		/* extra flags */
120	struct lock b_lock;		/* Buffer lock */
121	long	b_bufsize;		/* Allocated buffer size. */
122	int	b_runningbufspace;	/* when I/O is running, pipelining */
123	int	b_kvasize;		/* size of kva for buffer */
124	int	b_dirtyoff;		/* Offset in buffer of dirty region. */
125	int	b_dirtyend;		/* Offset of end of dirty region. */
126	caddr_t	b_kvabase;		/* base kva for buffer */
127	daddr_t b_lblkno;		/* Logical block number. */
128	struct	vnode *b_vp;		/* Device vnode. */
129	struct	ucred *b_rcred;		/* Read credentials reference. */
130	struct	ucred *b_wcred;		/* Write credentials reference. */
131	union {
132		TAILQ_ENTRY(buf) b_freelist; /* (Q) */
133		struct {
134			void	(*b_pgiodone)(void *, struct vm_page **,
135				    int, int);
136			int	b_pgbefore;
137			int	b_pgafter;
138		};
139	};
140	union	cluster_info {
141		TAILQ_HEAD(cluster_list_head, buf) cluster_head;
142		TAILQ_ENTRY(buf) cluster_entry;
143	} b_cluster;
144	int		b_npages;
145	struct	workhead b_dep;		/* (D) List of filesystem dependencies. */
146	void	*b_fsprivate1;
147	void	*b_fsprivate2;
148	void	*b_fsprivate3;
149
150#if defined(FULL_BUF_TRACKING)
151#define BUF_TRACKING_SIZE	32
152#define BUF_TRACKING_ENTRY(x)	((x) & (BUF_TRACKING_SIZE - 1))
153	const char	*b_io_tracking[BUF_TRACKING_SIZE];
154	uint32_t	b_io_tcnt;
155#elif defined(BUF_TRACKING)
156	const char	*b_io_tracking;
157#endif
158	struct	vm_page *b_pages[];
159};
160
161#define b_object	b_bufobj->bo_object
162
163/*
164 * These flags are kept in b_flags.
165 *
166 * Notes:
167 *
168 *	B_ASYNC		VOP calls on bp's are usually async whether or not
169 *			B_ASYNC is set, but some subsystems, such as NFS, like 
170 *			to know what is best for the caller so they can
171 *			optimize the I/O.
172 *
173 *	B_PAGING	Indicates that bp is being used by the paging system or
174 *			some paging system and that the bp is not linked into
175 *			the b_vp's clean/dirty linked lists or ref counts.
176 *			Buffer vp reassignments are illegal in this case.
177 *
178 *	B_CACHE		This may only be set if the buffer is entirely valid.
179 *			The situation where B_DELWRI is set and B_CACHE is
180 *			clear MUST be committed to disk by getblk() so 
181 *			B_DELWRI can also be cleared.  See the comments for
182 *			getblk() in kern/vfs_bio.c.  If B_CACHE is clear,
183 *			the caller is expected to clear BIO_ERROR and B_INVAL,
184 *			set BIO_READ, and initiate an I/O.
185 *
186 *			The 'entire buffer' is defined to be the range from
187 *			0 through b_bcount.
188 *
189 *	B_MALLOC	Request that the buffer be allocated from the malloc
190 *			pool, DEV_BSIZE aligned instead of PAGE_SIZE aligned.
191 *
192 *	B_CLUSTEROK	This flag is typically set for B_DELWRI buffers
193 *			by filesystems that allow clustering when the buffer
194 *			is fully dirty and indicates that it may be clustered
195 *			with other adjacent dirty buffers.  Note the clustering
196 *			may not be used with the stage 1 data write under NFS
197 *			but may be used for the commit rpc portion.
198 *
199 *	B_INVALONERR	This flag is set on dirty buffers.  It specifies that a
200 *			write error should forcibly invalidate the buffer
201 *			contents.  This flag should be used with caution, as it
202 *			discards data.  It is incompatible with B_ASYNC.
203 *
204 *	B_VMIO		Indicates that the buffer is tied into an VM object.
205 *			The buffer's data is always PAGE_SIZE aligned even
206 *			if b_bufsize and b_bcount are not.  ( b_bufsize is 
207 *			always at least DEV_BSIZE aligned, though ).
208 *
209 *	B_DIRECT	Hint that we should attempt to completely free
210 *			the pages underlying the buffer.  B_DIRECT is
211 *			sticky until the buffer is released and typically
212 *			only has an effect when B_RELBUF is also set.
213 *
214 */
215
216#define	B_AGE		0x00000001	/* Move to age queue when I/O done. */
217#define	B_NEEDCOMMIT	0x00000002	/* Append-write in progress. */
218#define	B_ASYNC		0x00000004	/* Start I/O, do not wait. */
219#define	B_DIRECT	0x00000008	/* direct I/O flag (pls free vmio) */
220#define	B_DEFERRED	0x00000010	/* Skipped over for cleaning */
221#define	B_CACHE		0x00000020	/* Bread found us in the cache. */
222#define	B_VALIDSUSPWRT	0x00000040	/* Valid write during suspension. */
223#define	B_DELWRI	0x00000080	/* Delay I/O until buffer reused. */
224#define	B_CKHASH	0x00000100	/* checksum hash calculated on read */
225#define	B_DONE		0x00000200	/* I/O completed. */
226#define	B_EINTR		0x00000400	/* I/O was interrupted */
227#define	B_NOREUSE	0x00000800	/* Contents not reused once released. */
228#define	B_REUSE		0x00001000	/* Contents reused, second chance. */
229#define	B_INVAL		0x00002000	/* Does not contain valid info. */
230#define	B_BARRIER	0x00004000	/* Write this and all preceding first. */
231#define	B_NOCACHE	0x00008000	/* Do not cache block after use. */
232#define	B_MALLOC	0x00010000	/* malloced b_data */
233#define	B_CLUSTEROK	0x00020000	/* Pagein op, so swap() can count it. */
234#define	B_INVALONERR	0x00040000	/* Invalidate on write error. */
235#define	B_IOSTARTED	0x00080000	/* buf_start() called */
236#define	B_00100000	0x00100000	/* Available flag. */
237#define	B_MAXPHYS	0x00200000	/* nitems(b_pages[]) = atop(MAXPHYS). */
238#define	B_RELBUF	0x00400000	/* Release VMIO buffer. */
239#define	B_FS_FLAG1	0x00800000	/* Available flag for FS use. */
240#define	B_NOCOPY	0x01000000	/* Don't copy-on-write this buf. */
241#define	B_INFREECNT	0x02000000	/* buf is counted in numfreebufs */
242#define	B_PAGING	0x04000000	/* volatile paging I/O -- bypass VMIO */
243#define B_MANAGED	0x08000000	/* Managed by FS. */
244#define B_RAM		0x10000000	/* Read ahead mark (flag) */
245#define B_VMIO		0x20000000	/* VMIO flag */
246#define B_CLUSTER	0x40000000	/* pagein op, so swap() can count it */
247#define B_REMFREE	0x80000000	/* Delayed bremfree */
248
249#define PRINT_BUF_FLAGS "\20\40remfree\37cluster\36vmio\35ram\34managed" \
250	"\33paging\32infreecnt\31nocopy\30b23\27relbuf\26maxphys\25b20" \
251	"\24iostarted\23invalonerr\22clusterok\21malloc\20nocache\17b14" \
252	"\16inval\15reuse\14noreuse\13eintr\12done\11b8\10delwri" \
253	"\7validsuspwrt\6cache\5deferred\4direct\3async\2needcommit\1age"
254
255/*
256 * These flags are kept in b_xflags.
257 *
258 * BX_FSPRIV reserves a set of eight flags that may be used by individual
259 * filesystems for their own purpose. Their specific definitions are
260 * found in the header files for each filesystem that uses them.
261 */
262#define	BX_VNDIRTY	0x00000001	/* On vnode dirty list */
263#define	BX_VNCLEAN	0x00000002	/* On vnode clean list */
264#define	BX_CVTENXIO	0x00000004	/* Convert errors to ENXIO */
265#define	BX_BKGRDWRITE	0x00000010	/* Do writes in background */
266#define	BX_BKGRDMARKER	0x00000020	/* Mark buffer for splay tree */
267#define	BX_ALTDATA	0x00000040	/* Holds extended data */
268#define	BX_FSPRIV	0x00FF0000	/* Filesystem-specific flags mask */
269
270#define	PRINT_BUF_XFLAGS "\20\7altdata\6bkgrdmarker\5bkgrdwrite\3cvtenxio" \
271	"\2clean\1dirty"
272
273#define	NOOFFSET	(-1LL)		/* No buffer offset calculated yet */
274
275/*
276 * These flags are kept in b_vflags.
277 */
278#define	BV_SCANNED	0x00000001	/* VOP_FSYNC funcs mark written bufs */
279#define	BV_BKGRDINPROG	0x00000002	/* Background write in progress */
280#define	BV_BKGRDWAIT	0x00000004	/* Background write waiting */
281#define	BV_BKGRDERR	0x00000008	/* Error from background write */
282
283#define	PRINT_BUF_VFLAGS "\20\4bkgrderr\3bkgrdwait\2bkgrdinprog\1scanned"
284
285#ifdef _KERNEL
286
287#ifndef NSWBUF_MIN
288#define	NSWBUF_MIN	16
289#endif
290
291/*
292 * Buffer locking
293 */
294#include <sys/proc.h>			/* XXX for curthread */
295#include <sys/mutex.h>
296
297/*
298 * Initialize a lock.
299 */
300#define BUF_LOCKINIT(bp, wmesg)						\
301	lockinit(&(bp)->b_lock, PRIBIO + 4, wmesg, 0, LK_NEW)
302/*
303 *
304 * Get a lock sleeping non-interruptably until it becomes available.
305 */
306#define	BUF_LOCK(bp, locktype, interlock)				\
307	_lockmgr_args_rw(&(bp)->b_lock, (locktype), (interlock),	\
308	    LK_WMESG_DEFAULT, LK_PRIO_DEFAULT, LK_TIMO_DEFAULT,		\
309	    LOCK_FILE, LOCK_LINE)
310
311/*
312 * Get a lock sleeping with specified interruptably and timeout.
313 */
314#define	BUF_TIMELOCK(bp, locktype, interlock, wmesg, catch, timo)	\
315	_lockmgr_args_rw(&(bp)->b_lock, (locktype) | LK_TIMELOCK,	\
316	    (interlock), (wmesg), (PRIBIO + 4) | (catch), (timo),	\
317	    LOCK_FILE, LOCK_LINE)
318
319/*
320 * Release a lock. Only the acquiring process may free the lock unless
321 * it has been handed off to biodone.
322 */
323#define	BUF_UNLOCK(bp) do {						\
324	KASSERT(((bp)->b_flags & B_REMFREE) == 0,			\
325	    ("BUF_UNLOCK %p while B_REMFREE is still set.", (bp)));	\
326									\
327	BUF_UNLOCK_RAW((bp));						\
328} while (0)
329#define	BUF_UNLOCK_RAW(bp) do {						\
330	(void)_lockmgr_args(&(bp)->b_lock, LK_RELEASE, NULL,		\
331	    LK_WMESG_DEFAULT, LK_PRIO_DEFAULT, LK_TIMO_DEFAULT,		\
332	    LOCK_FILE, LOCK_LINE);					\
333} while (0)
334
335/*
336 * Check if a buffer lock is recursed.
337 */
338#define	BUF_LOCKRECURSED(bp)						\
339	lockmgr_recursed(&(bp)->b_lock)
340
341/*
342 * Check if a buffer lock is currently held.
343 */
344#define	BUF_ISLOCKED(bp)						\
345	lockstatus(&(bp)->b_lock)
346/*
347 * Free a buffer lock.
348 */
349#define BUF_LOCKFREE(bp) 						\
350	lockdestroy(&(bp)->b_lock)
351
352/*
353 * Print informations on a buffer lock.
354 */
355#define BUF_LOCKPRINTINFO(bp) 						\
356	lockmgr_printinfo(&(bp)->b_lock)
357
358/*
359 * Buffer lock assertions.
360 */
361#if defined(INVARIANTS) && defined(INVARIANT_SUPPORT)
362#define	BUF_ASSERT_LOCKED(bp)						\
363	_lockmgr_assert(&(bp)->b_lock, KA_LOCKED, LOCK_FILE, LOCK_LINE)
364#define	BUF_ASSERT_SLOCKED(bp)						\
365	_lockmgr_assert(&(bp)->b_lock, KA_SLOCKED, LOCK_FILE, LOCK_LINE)
366#define	BUF_ASSERT_XLOCKED(bp)						\
367	_lockmgr_assert(&(bp)->b_lock, KA_XLOCKED, LOCK_FILE, LOCK_LINE)
368#define	BUF_ASSERT_UNLOCKED(bp)						\
369	_lockmgr_assert(&(bp)->b_lock, KA_UNLOCKED, LOCK_FILE, LOCK_LINE)
370#else
371#define	BUF_ASSERT_LOCKED(bp)
372#define	BUF_ASSERT_SLOCKED(bp)
373#define	BUF_ASSERT_XLOCKED(bp)
374#define	BUF_ASSERT_UNLOCKED(bp)
375#endif
376
377#ifdef _SYS_PROC_H_	/* Avoid #include <sys/proc.h> pollution */
378/*
379 * When initiating asynchronous I/O, change ownership of the lock to the
380 * kernel. Once done, the lock may legally released by biodone. The
381 * original owning process can no longer acquire it recursively, but must
382 * wait until the I/O is completed and the lock has been freed by biodone.
383 */
384#define	BUF_KERNPROC(bp)						\
385	_lockmgr_disown(&(bp)->b_lock, LOCK_FILE, LOCK_LINE)
386#endif
387
388#endif /* _KERNEL */
389
390struct buf_queue_head {
391	TAILQ_HEAD(buf_queue, buf) queue;
392	daddr_t last_pblkno;
393	struct	buf *insert_point;
394	struct	buf *switch_point;
395};
396
397/*
398 * This structure describes a clustered I/O. 
399 */
400struct cluster_save {
401	long	bs_bcount;		/* Saved b_bcount. */
402	long	bs_bufsize;		/* Saved b_bufsize. */
403	int	bs_nchildren;		/* Number of associated buffers. */
404	struct buf **bs_children;	/* List of associated buffers. */
405};
406
407/*
408 * Vnode clustering tracker
409 */
410struct vn_clusterw {
411	daddr_t	v_cstart;			/* v start block of cluster */
412	daddr_t	v_lasta;			/* v last allocation  */
413	daddr_t	v_lastw;			/* v last write  */
414	int	v_clen;				/* v length of cur. cluster */
415};
416
417#ifdef _KERNEL
418
419static __inline int
420bwrite(struct buf *bp)
421{
422
423	KASSERT(bp->b_bufobj != NULL, ("bwrite: no bufobj bp=%p", bp));
424	KASSERT(bp->b_bufobj->bo_ops != NULL, ("bwrite: no bo_ops bp=%p", bp));
425	KASSERT(bp->b_bufobj->bo_ops->bop_write != NULL,
426	    ("bwrite: no bop_write bp=%p", bp));
427	return (BO_WRITE(bp->b_bufobj, bp));
428}
429
430static __inline void
431bstrategy(struct buf *bp)
432{
433
434	KASSERT(bp->b_bufobj != NULL, ("bstrategy: no bufobj bp=%p", bp));
435	KASSERT(bp->b_bufobj->bo_ops != NULL,
436	    ("bstrategy: no bo_ops bp=%p", bp));
437	KASSERT(bp->b_bufobj->bo_ops->bop_strategy != NULL,
438	    ("bstrategy: no bop_strategy bp=%p", bp));
439	BO_STRATEGY(bp->b_bufobj, bp);
440}
441
442static __inline void
443buf_start(struct buf *bp)
444{
445	KASSERT((bp->b_flags & B_IOSTARTED) == 0,
446	    ("recursed buf_start %p", bp));
447	bp->b_flags |= B_IOSTARTED;
448	if (bioops.io_start)
449		(*bioops.io_start)(bp);
450}
451
452static __inline void
453buf_complete(struct buf *bp)
454{
455	if ((bp->b_flags & B_IOSTARTED) != 0) {
456		bp->b_flags &= ~B_IOSTARTED;
457		if (bioops.io_complete)
458			(*bioops.io_complete)(bp);
459	}
460}
461
462static __inline void
463buf_deallocate(struct buf *bp)
464{
465	if (bioops.io_deallocate)
466		(*bioops.io_deallocate)(bp);
467}
468
469static __inline int
470buf_countdeps(struct buf *bp, int i)
471{
472	if (bioops.io_countdeps)
473		return ((*bioops.io_countdeps)(bp, i));
474	else
475		return (0);
476}
477
478static __inline void
479buf_track(struct buf *bp __unused, const char *location __unused)
480{
481
482#if defined(FULL_BUF_TRACKING)
483	bp->b_io_tracking[BUF_TRACKING_ENTRY(bp->b_io_tcnt++)] = location;
484#elif defined(BUF_TRACKING)
485	bp->b_io_tracking = location;
486#endif
487}
488
489#endif /* _KERNEL */
490
491/*
492 * Zero out the buffer's data area.
493 */
494#define	clrbuf(bp) {							\
495	bzero((bp)->b_data, (u_int)(bp)->b_bcount);			\
496	(bp)->b_resid = 0;						\
497}
498
499/*
500 * Flags for getblk's last parameter.
501 */
502#define	GB_LOCK_NOWAIT	0x0001		/* Fail if we block on a buf lock. */
503#define	GB_NOCREAT	0x0002		/* Don't create a buf if not found. */
504#define	GB_NOWAIT_BD	0x0004		/* Do not wait for bufdaemon. */
505#define	GB_UNMAPPED	0x0008		/* Do not mmap buffer pages. */
506#define	GB_KVAALLOC	0x0010		/* But allocate KVA. */
507#define	GB_CKHASH	0x0020		/* If reading, calc checksum hash */
508#define	GB_NOSPARSE	0x0040		/* Do not instantiate holes */
509#define	GB_CVTENXIO	0x0080		/* Convert errors to ENXIO */
510#define	GB_NOWITNESS	0x0100		/* Do not record for WITNESS */
511
512#ifdef _KERNEL
513extern int	nbuf;			/* The number of buffer headers */
514extern u_long	maxswzone;		/* Max KVA for swap structures */
515extern u_long	maxbcache;		/* Max KVA for buffer cache */
516extern int	maxbcachebuf;		/* Max buffer cache block size */
517extern long	runningbufspace;
518extern long	hibufspace;
519extern int	dirtybufthresh;
520extern int	bdwriteskip;
521extern int	dirtybufferflushes;
522extern int	altbufferflushes;
523extern int	nswbuf;			/* Number of swap I/O buffer headers. */
524extern caddr_t __read_mostly unmapped_buf; /* Data address for unmapped
525					      buffers. */
526
527static inline int
528buf_mapped(struct buf *bp)
529{
530
531	return (bp->b_data != unmapped_buf);
532}
533
534void	runningbufwakeup(struct buf *);
535void	waitrunningbufspace(void);
536caddr_t	kern_vfs_bio_buffer_alloc(caddr_t v, long physmem_est);
537void	bufinit(void);
538void	bufshutdown(int);
539void	bdata2bio(struct buf *bp, struct bio *bip);
540void	bwillwrite(void);
541int	buf_dirty_count_severe(void);
542void	bremfree(struct buf *);
543void	bremfreef(struct buf *);	/* XXX Force bremfree, only for nfs. */
544#define bread(vp, blkno, size, cred, bpp) \
545	    breadn_flags(vp, blkno, blkno, size, NULL, NULL, 0, cred, 0, \
546		NULL, bpp)
547#define bread_gb(vp, blkno, size, cred, gbflags, bpp) \
548	    breadn_flags(vp, blkno, blkno, size, NULL, NULL, 0, cred, \
549		gbflags, NULL, bpp)
550#define breadn(vp, blkno, size, rablkno, rabsize, cnt, cred, bpp) \
551	    breadn_flags(vp, blkno, blkno, size, rablkno, rabsize, cnt, cred, \
552		0, NULL, bpp)
553int	breadn_flags(struct vnode *, daddr_t, daddr_t, int, daddr_t *, int *, 
554	    int, struct ucred *, int, void (*)(struct buf *), struct buf **);
555void	bdwrite(struct buf *);
556void	bawrite(struct buf *);
557void	babarrierwrite(struct buf *);
558int	bbarrierwrite(struct buf *);
559void	bdirty(struct buf *);
560void	bundirty(struct buf *);
561void	bufstrategy(struct bufobj *, struct buf *);
562void	brelse(struct buf *);
563void	bqrelse(struct buf *);
564int	vfs_bio_awrite(struct buf *);
565void	vfs_busy_pages_acquire(struct buf *bp);
566void	vfs_busy_pages_release(struct buf *bp);
567struct buf *incore(struct bufobj *, daddr_t);
568bool	inmem(struct vnode *, daddr_t);
569struct buf *gbincore(struct bufobj *, daddr_t);
570struct buf *gbincore_unlocked(struct bufobj *, daddr_t);
571struct buf *getblk(struct vnode *, daddr_t, int, int, int, int);
572int	getblkx(struct vnode *vp, daddr_t blkno, daddr_t dblkno, int size,
573	    int slpflag, int slptimeo, int flags, struct buf **bpp);
574struct buf *geteblk(int, int);
575int	bufwait(struct buf *);
576int	bufwrite(struct buf *);
577void	bufdone(struct buf *);
578void	bd_speedup(void);
579
580extern uma_zone_t pbuf_zone;
581uma_zone_t pbuf_zsecond_create(const char *name, int max);
582
583struct vn_clusterw;
584
585void	cluster_init_vn(struct vn_clusterw *vnc);
586int	cluster_read(struct vnode *, u_quad_t, daddr_t, long,
587	    struct ucred *, long, int, int, struct buf **);
588int	cluster_wbuild(struct vnode *, long, daddr_t, int, int);
589void	cluster_write(struct vnode *, struct vn_clusterw *, struct buf *,
590	    u_quad_t, int, int);
591void	vfs_bio_brelse(struct buf *bp, int ioflags);
592void	vfs_bio_bzero_buf(struct buf *bp, int base, int size);
593void	vfs_bio_clrbuf(struct buf *);
594void	vfs_bio_set_flags(struct buf *bp, int ioflags);
595void	vfs_bio_set_valid(struct buf *, int base, int size);
596void	vfs_busy_pages(struct buf *, int clear_modify);
597void	vfs_unbusy_pages(struct buf *);
598int	vmapbuf(struct buf *, void *, size_t, int);
599void	vunmapbuf(struct buf *);
600void	brelvp(struct buf *);
601void	bgetvp(struct vnode *, struct buf *);
602void	pbgetbo(struct bufobj *bo, struct buf *bp);
603void	pbgetvp(struct vnode *, struct buf *);
604void	pbrelbo(struct buf *);
605void	pbrelvp(struct buf *);
606int	allocbuf(struct buf *bp, int size);
607void	reassignbuf(struct buf *);
608void	bwait(struct buf *, u_char, const char *);
609void	bdone(struct buf *);
610
611typedef daddr_t (vbg_get_lblkno_t)(struct vnode *, vm_ooffset_t);
612typedef int (vbg_get_blksize_t)(struct vnode *, daddr_t, long *);
613int	vfs_bio_getpages(struct vnode *vp, struct vm_page **ma, int count,
614	    int *rbehind, int *rahead, vbg_get_lblkno_t get_lblkno,
615	    vbg_get_blksize_t get_blksize);
616
617#endif /* _KERNEL */
618
619#endif /* !_SYS_BUF_H_ */