2 * Copyright (c) 2000-2003 Tor Egge
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
8 * 1. Redistributions of source code must retain the above copyright
9 * notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 * notice, this list of conditions and the following disclaimer in the
12 * documentation and/or other materials provided with the distribution.
14 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
15 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
16 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
17 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
18 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
19 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
20 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
21 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
22 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
23 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27 #include <sys/cdefs.h>
28 __FBSDID("$FreeBSD$");
30 #include <sys/param.h>
31 #include <sys/systm.h>
32 #include <sys/fcntl.h>
36 #include <sys/limits.h>
37 #include <sys/mount.h>
38 #include <sys/namei.h>
39 #include <sys/vnode.h>
41 #include <sys/filio.h>
42 #include <sys/ttycom.h>
45 #include <sys/rwlock.h>
46 #include <ufs/ufs/extattr.h>
47 #include <ufs/ufs/quota.h>
48 #include <ufs/ufs/inode.h>
49 #include <ufs/ufs/ufsmount.h>
50 #include <ufs/ufs/ufs_extern.h>
51 #include <ufs/ffs/fs.h>
52 #include <ufs/ffs/ffs_extern.h>
55 #include <vm/vm_extern.h>
56 #include <vm/vm_object.h>
57 #include <sys/kernel.h>
58 #include <sys/sysctl.h>
60 static int ffs_rawread_readahead(struct vnode *vp,
67 static int ffs_rawread_main(struct vnode *vp,
70 static int ffs_rawread_sync(struct vnode *vp);
72 int ffs_rawread(struct vnode *vp, struct uio *uio, int *workdone);
74 void ffs_rawread_setup(void);
76 SYSCTL_DECL(_vfs_ffs);
78 static int ffsrawbufcnt = 4;
79 SYSCTL_INT(_vfs_ffs, OID_AUTO, ffsrawbufcnt, CTLFLAG_RD, &ffsrawbufcnt, 0,
80 "Buffers available for raw reads");
82 static int allowrawread = 1;
83 SYSCTL_INT(_vfs_ffs, OID_AUTO, allowrawread, CTLFLAG_RW, &allowrawread, 0,
84 "Flag to enable raw reads");
86 static int rawreadahead = 1;
87 SYSCTL_INT(_vfs_ffs, OID_AUTO, rawreadahead, CTLFLAG_RW, &rawreadahead, 0,
88 "Flag to enable readahead for long raw reads");
92 ffs_rawread_setup(void)
94 ffsrawbufcnt = (nswbuf > 100 ) ? (nswbuf - (nswbuf >> 4)) : nswbuf - 8;
99 ffs_rawread_sync(struct vnode *vp)
107 /* Check for dirty mmap, pending writes and dirty buffers */
111 if (bo->bo_numoutput > 0 ||
112 bo->bo_dirty.bv_cnt > 0 ||
113 ((obj = vp->v_object) != NULL &&
114 (obj->flags & OBJ_MIGHTBEDIRTY) != 0)) {
118 if (vn_start_write(vp, &mp, V_NOWAIT) != 0) {
119 if (VOP_ISLOCKED(vp) != LK_EXCLUSIVE)
124 (void) vn_start_write(vp, &mp, V_WAIT);
125 VOP_LOCK(vp, LK_EXCLUSIVE);
126 } else if (VOP_ISLOCKED(vp) != LK_EXCLUSIVE) {
128 /* Upgrade to exclusive lock, this might block */
129 VOP_LOCK(vp, LK_UPGRADE);
135 /* Check if vnode was reclaimed while unlocked. */
136 if ((vp->v_iflag & VI_DOOMED) != 0) {
139 VOP_LOCK(vp, LK_DOWNGRADE);
140 vn_finished_write(mp);
143 /* Attempt to msync mmap() regions to clean dirty mmap */
144 if ((obj = vp->v_object) != NULL &&
145 (obj->flags & OBJ_MIGHTBEDIRTY) != 0) {
147 VM_OBJECT_WLOCK(obj);
148 vm_object_page_clean(obj, 0, 0, OBJPC_SYNC);
149 VM_OBJECT_WUNLOCK(obj);
153 /* Wait for pending writes to complete */
155 error = bufobj_wwait(&vp->v_bufobj, 0, 0);
157 /* XXX: can't happen with a zero timeout ??? */
160 VOP_LOCK(vp, LK_DOWNGRADE);
161 vn_finished_write(mp);
164 /* Flush dirty buffers */
165 if (bo->bo_dirty.bv_cnt > 0) {
167 if ((error = ffs_syncvnode(vp, MNT_WAIT, 0)) != 0) {
169 VOP_LOCK(vp, LK_DOWNGRADE);
170 vn_finished_write(mp);
174 if (bo->bo_numoutput > 0 || bo->bo_dirty.bv_cnt > 0)
175 panic("ffs_rawread_sync: dirty bufs");
179 VOP_LOCK(vp, LK_DOWNGRADE);
180 vn_finished_write(mp);
190 ffs_rawread_readahead(struct vnode *vp,
208 bsize = vp->v_mount->mnt_stat.f_iosize;
213 iolen = ((vm_offset_t) udata) & PAGE_MASK;
215 if (bp->b_bcount + iolen > bp->b_kvasize) {
216 bp->b_bcount = bp->b_kvasize;
218 bp->b_bcount -= PAGE_SIZE;
220 bp->b_flags = 0; /* XXX necessary ? */
221 bp->b_iocmd = BIO_READ;
222 bp->b_iodone = bdone;
225 blockno = offset / bsize;
226 blockoff = (offset % bsize) / DEV_BSIZE;
227 if ((daddr_t) blockno != blockno) {
228 return EINVAL; /* blockno overflow */
231 bp->b_lblkno = bp->b_blkno = blockno;
233 error = ufs_bmaparray(vp, bp->b_lblkno, &blkno, NULL, &bforwards, NULL);
238 /* Fill holes with NULs to preserve semantics */
240 if (bp->b_bcount + blockoff * DEV_BSIZE > bsize)
241 bp->b_bcount = bsize - blockoff * DEV_BSIZE;
242 bp->b_bufsize = bp->b_bcount;
244 if (vmapbuf(bp, 1) < 0)
248 bzero(bp->b_data, bp->b_bufsize);
250 /* Mark operation completed (similar to bufdone()) */
253 bp->b_flags |= B_DONE;
256 bp->b_blkno = blkno + blockoff;
257 bp->b_offset = bp->b_iooffset = (blkno + blockoff) * DEV_BSIZE;
259 if (bp->b_bcount + blockoff * DEV_BSIZE > bsize * (1 + bforwards))
260 bp->b_bcount = bsize * (1 + bforwards) - blockoff * DEV_BSIZE;
261 bp->b_bufsize = bp->b_bcount;
263 if (vmapbuf(bp, 1) < 0)
266 BO_STRATEGY(&dp->v_bufobj, bp);
272 ffs_rawread_main(struct vnode *vp,
276 struct buf *bp, *nbp, *tbp;
277 caddr_t sa, nsa, tsa;
285 td = uio->uio_td ? uio->uio_td : curthread;
286 udata = uio->uio_iov->iov_base;
287 resid = uio->uio_resid;
288 offset = uio->uio_offset;
291 * keep the process from being swapped
305 if (bp == NULL) { /* Setup first read */
306 /* XXX: Leave some bufs for swap */
307 bp = getpbuf(&ffsrawbufcnt);
310 error = ffs_rawread_readahead(vp, udata, offset,
315 if (resid > bp->b_bufsize) { /* Setup fist readahead */
316 /* XXX: Leave bufs for swap */
317 if (rawreadahead != 0)
318 nbp = trypbuf(&ffsrawbufcnt);
325 nerror = ffs_rawread_readahead(vp,
337 relpbuf(nbp, &ffsrawbufcnt);
345 bwait(bp, PRIBIO, "rawrd");
350 iolen = bp->b_bcount - bp->b_resid;
351 if (iolen == 0 && (bp->b_ioflags & BIO_ERROR) == 0) {
352 nerror = 0; /* Ignore possible beyond EOF error */
356 if ((bp->b_ioflags & BIO_ERROR) != 0) {
363 if (iolen < bp->b_bufsize) {
364 /* Incomplete read. Try to read remaining part */
365 error = ffs_rawread_readahead(vp,
368 bp->b_bufsize - iolen,
374 } else if (nbp != NULL) { /* Complete read with readahead */
384 if (resid <= bp->b_bufsize) { /* No more readaheads */
386 relpbuf(nbp, &ffsrawbufcnt);
388 } else { /* Setup next readahead */
389 nerror = ffs_rawread_readahead(vp,
401 relpbuf(nbp, &ffsrawbufcnt);
405 } else if (nerror != 0) {/* Deferred Readahead error */
407 } else if (resid > 0) { /* More to read, no readahead */
408 error = ffs_rawread_readahead(vp, udata, offset,
417 relpbuf(bp, &ffsrawbufcnt);
419 if (nbp != NULL) { /* Run down readahead buffer */
421 bwait(nbp, PRIBIO, "rawrd");
425 relpbuf(nbp, &ffsrawbufcnt);
431 uio->uio_iov->iov_base = udata;
432 uio->uio_resid = resid;
433 uio->uio_offset = offset;
439 ffs_rawread(struct vnode *vp,
443 if (allowrawread != 0 &&
444 uio->uio_iovcnt == 1 &&
445 uio->uio_segflg == UIO_USERSPACE &&
446 uio->uio_resid == uio->uio_iov->iov_len &&
447 (((uio->uio_td != NULL) ? uio->uio_td : curthread)->td_pflags &
448 TDP_DEADLKTREAT) == 0) {
449 int secsize; /* Media sector size */
450 off_t filebytes; /* Bytes left of file */
451 int blockbytes; /* Bytes left of file in full blocks */
452 int partialbytes; /* Bytes in last partial block */
453 int skipbytes; /* Bytes not to read in ffs_rawread */
458 /* Only handle sector aligned reads */
460 secsize = ip->i_devvp->v_bufobj.bo_bsize;
461 if ((uio->uio_offset & (secsize - 1)) == 0 &&
462 (uio->uio_resid & (secsize - 1)) == 0) {
464 /* Sync dirty pages and buffers if needed */
465 error = ffs_rawread_sync(vp);
469 /* Check for end of file */
470 if (ip->i_size > uio->uio_offset) {
471 filebytes = ip->i_size - uio->uio_offset;
473 /* No special eof handling needed ? */
474 if (uio->uio_resid <= filebytes) {
476 return ffs_rawread_main(vp, uio);
479 partialbytes = ((unsigned int) ip->i_size) %
481 blockbytes = (int) filebytes - partialbytes;
482 if (blockbytes > 0) {
483 skipbytes = uio->uio_resid -
485 uio->uio_resid = blockbytes;
486 error = ffs_rawread_main(vp, uio);
487 uio->uio_resid += skipbytes;
490 /* Read remaining part using buffer */