2 * SPDX-License-Identifier: BSD-2-Clause
4 * Copyright (c) 2000-2003 Tor Egge
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
10 * 1. Redistributions of source code must retain the above copyright
11 * notice, this list of conditions and the following disclaimer.
12 * 2. Redistributions in binary form must reproduce the above copyright
13 * notice, this list of conditions and the following disclaimer in the
14 * documentation and/or other materials provided with the distribution.
16 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
17 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
18 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
19 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
20 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
21 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
22 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
23 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
24 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
25 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
29 #include <sys/cdefs.h>
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,
66 static int ffs_rawread_main(struct vnode *vp,
69 static int ffs_rawread_sync(struct vnode *vp);
71 int ffs_rawread(struct vnode *vp, struct uio *uio, int *workdone);
73 SYSCTL_DECL(_vfs_ffs);
75 static uma_zone_t ffsraw_pbuf_zone;
77 static int allowrawread = 1;
78 SYSCTL_INT(_vfs_ffs, OID_AUTO, allowrawread, CTLFLAG_RW, &allowrawread, 0,
79 "Flag to enable raw reads");
81 static int rawreadahead = 1;
82 SYSCTL_INT(_vfs_ffs, OID_AUTO, rawreadahead, CTLFLAG_RW, &rawreadahead, 0,
83 "Flag to enable readahead for long raw reads");
86 ffs_rawread_setup(void *arg __unused)
89 ffsraw_pbuf_zone = pbuf_zsecond_create("ffsrawpbuf",
90 (nswbuf > 100 ) ? (nswbuf - (nswbuf >> 4)) : nswbuf - 8);
92 SYSINIT(ffs_raw, SI_SUB_VM_CONF, SI_ORDER_ANY, ffs_rawread_setup, NULL);
95 ffs_rawread_sync(struct vnode *vp)
103 /* Check for dirty mmap, pending writes and dirty buffers */
107 if (bo->bo_numoutput > 0 ||
108 bo->bo_dirty.bv_cnt > 0 ||
109 ((obj = vp->v_object) != NULL &&
110 vm_object_mightbedirty(obj))) {
114 if (vn_start_write(vp, &mp, V_NOWAIT) != 0) {
115 if (VOP_ISLOCKED(vp) != LK_EXCLUSIVE)
120 (void) vn_start_write(vp, &mp, V_WAIT);
121 VOP_LOCK(vp, LK_EXCLUSIVE);
122 } else if (VOP_ISLOCKED(vp) != LK_EXCLUSIVE) {
124 /* Upgrade to exclusive lock, this might block */
125 VOP_LOCK(vp, LK_UPGRADE);
131 /* Check if vnode was reclaimed while unlocked. */
132 if (VN_IS_DOOMED(vp)) {
135 VOP_LOCK(vp, LK_DOWNGRADE);
136 vn_finished_write(mp);
139 /* Attempt to msync mmap() regions to clean dirty mmap */
140 if ((obj = vp->v_object) != NULL &&
141 vm_object_mightbedirty(obj)) {
143 VM_OBJECT_WLOCK(obj);
144 vm_object_page_clean(obj, 0, 0, OBJPC_SYNC);
145 VM_OBJECT_WUNLOCK(obj);
149 /* Wait for pending writes to complete */
151 error = bufobj_wwait(&vp->v_bufobj, 0, 0);
153 /* XXX: can't happen with a zero timeout ??? */
156 VOP_LOCK(vp, LK_DOWNGRADE);
157 vn_finished_write(mp);
160 /* Flush dirty buffers */
161 if (bo->bo_dirty.bv_cnt > 0) {
163 if ((error = ffs_syncvnode(vp, MNT_WAIT, 0)) != 0) {
165 VOP_LOCK(vp, LK_DOWNGRADE);
166 vn_finished_write(mp);
170 if (bo->bo_numoutput > 0 || bo->bo_dirty.bv_cnt > 0)
171 panic("ffs_rawread_sync: dirty bufs");
175 VOP_LOCK(vp, LK_DOWNGRADE);
176 vn_finished_write(mp);
185 ffs_rawread_readahead(struct vnode *vp,
202 bsize = vp->v_mount->mnt_stat.f_iosize;
207 iolen = ((vm_offset_t) udata) & PAGE_MASK;
209 if (bp->b_bcount + iolen > bp->b_kvasize) {
210 bp->b_bcount = bp->b_kvasize;
212 bp->b_bcount -= PAGE_SIZE;
214 bp->b_flags = 0; /* XXX necessary ? */
215 bp->b_iocmd = BIO_READ;
216 bp->b_iodone = bdone;
217 blockno = offset / bsize;
218 blockoff = (offset % bsize) / DEV_BSIZE;
219 if ((daddr_t) blockno != blockno) {
220 return EINVAL; /* blockno overflow */
223 bp->b_lblkno = bp->b_blkno = blockno;
225 error = ufs_bmaparray(vp, bp->b_lblkno, &blkno, NULL, &bforwards, NULL);
229 /* Fill holes with NULs to preserve semantics */
231 if (bp->b_bcount + blockoff * DEV_BSIZE > bsize)
232 bp->b_bcount = bsize - blockoff * DEV_BSIZE;
234 if (vmapbuf(bp, udata, bp->b_bcount, 1) < 0)
238 bzero(bp->b_data, bp->b_bufsize);
240 /* Mark operation completed (similar to bufdone()) */
243 bp->b_flags |= B_DONE;
246 bp->b_blkno = blkno + blockoff;
247 bp->b_offset = bp->b_iooffset = (blkno + blockoff) * DEV_BSIZE;
249 if (bp->b_bcount + blockoff * DEV_BSIZE > bsize * (1 + bforwards))
250 bp->b_bcount = bsize * (1 + bforwards) - blockoff * DEV_BSIZE;
252 if (vmapbuf(bp, udata, bp->b_bcount, 1) < 0)
255 BO_STRATEGY(&dp->v_bufobj, bp);
260 ffs_rawread_main(struct vnode *vp,
264 struct buf *bp, *nbp, *tbp;
271 td = uio->uio_td ? uio->uio_td : curthread;
272 udata = uio->uio_iov->iov_base;
273 resid = uio->uio_resid;
274 offset = uio->uio_offset;
277 * keep the process from being swapped
289 if (bp == NULL) { /* Setup first read */
290 bp = uma_zalloc(ffsraw_pbuf_zone, M_WAITOK);
292 error = ffs_rawread_readahead(vp, udata, offset,
297 if (resid > bp->b_bufsize) { /* Setup fist readahead */
298 if (rawreadahead != 0)
299 nbp = uma_zalloc(ffsraw_pbuf_zone,
306 nerror = ffs_rawread_readahead(vp,
317 uma_zfree(ffsraw_pbuf_zone,
325 bwait(bp, PRIBIO, "rawrd");
328 iolen = bp->b_bcount - bp->b_resid;
329 if (iolen == 0 && (bp->b_ioflags & BIO_ERROR) == 0) {
330 nerror = 0; /* Ignore possible beyond EOF error */
334 if ((bp->b_ioflags & BIO_ERROR) != 0) {
341 if (iolen < bp->b_bufsize) {
342 /* Incomplete read. Try to read remaining part */
343 error = ffs_rawread_readahead(vp,
346 bp->b_bufsize - iolen,
351 } else if (nbp != NULL) { /* Complete read with readahead */
357 if (resid <= bp->b_bufsize) { /* No more readaheads */
359 uma_zfree(ffsraw_pbuf_zone, nbp);
361 } else { /* Setup next readahead */
362 nerror = ffs_rawread_readahead(vp,
373 uma_zfree(ffsraw_pbuf_zone, nbp);
377 } else if (nerror != 0) {/* Deferred Readahead error */
379 } else if (resid > 0) { /* More to read, no readahead */
380 error = ffs_rawread_readahead(vp, udata, offset,
389 uma_zfree(ffsraw_pbuf_zone, bp);
391 if (nbp != NULL) { /* Run down readahead buffer */
392 bwait(nbp, PRIBIO, "rawrd");
395 uma_zfree(ffsraw_pbuf_zone, nbp);
401 uio->uio_iov->iov_base = udata;
402 uio->uio_resid = resid;
403 uio->uio_offset = offset;
408 ffs_rawread(struct vnode *vp,
412 if (allowrawread != 0 &&
413 uio->uio_iovcnt == 1 &&
414 uio->uio_segflg == UIO_USERSPACE &&
415 uio->uio_resid == uio->uio_iov->iov_len &&
416 (((uio->uio_td != NULL) ? uio->uio_td : curthread)->td_pflags &
417 TDP_DEADLKTREAT) == 0) {
418 int secsize; /* Media sector size */
419 off_t filebytes; /* Bytes left of file */
420 int blockbytes; /* Bytes left of file in full blocks */
421 int partialbytes; /* Bytes in last partial block */
422 int skipbytes; /* Bytes not to read in ffs_rawread */
427 /* Only handle sector aligned reads */
429 secsize = ITODEVVP(ip)->v_bufobj.bo_bsize;
430 if ((uio->uio_offset & (secsize - 1)) == 0 &&
431 (uio->uio_resid & (secsize - 1)) == 0) {
433 /* Sync dirty pages and buffers if needed */
434 error = ffs_rawread_sync(vp);
438 /* Check for end of file */
439 if (ip->i_size > uio->uio_offset) {
440 filebytes = ip->i_size - uio->uio_offset;
442 /* No special eof handling needed ? */
443 if (uio->uio_resid <= filebytes) {
445 return ffs_rawread_main(vp, uio);
448 partialbytes = ((unsigned int) ip->i_size) %
450 blockbytes = (int) filebytes - partialbytes;
451 if (blockbytes > 0) {
452 skipbytes = uio->uio_resid -
454 uio->uio_resid = blockbytes;
455 error = ffs_rawread_main(vp, uio);
456 uio->uio_resid += skipbytes;
459 /* Read remaining part using buffer */