/*
* Initiate IO on given buffer.
*/
int
xfs_buf_iorequest(struct xfs_buf *bp)
{
bp->b_flags &= ~(B_INVAL|B_DONE);
bp->b_ioflags &= ~BIO_ERROR;
if (bp->b_flags & B_ASYNC)
BUF_KERNPROC(bp);
if (bp->b_vp == NULL) {
if (bp->b_iocmd == BIO_WRITE) {
bp->b_flags &= ~(B_DELWRI | B_DEFERRED);
bufobj_wref(bp->b_bufobj);
}
bp->b_iooffset = (bp->b_blkno << BBSHIFT);
bstrategy(bp);
} else {
if (bp->b_iocmd == BIO_WRITE) {
/* Mark the buffer clean */
bundirty(bp);
bufobj_wref(bp->b_bufobj);
vfs_busy_pages(bp, 1);
} else if (bp->b_iocmd == BIO_READ) {
vfs_busy_pages(bp, 0);
}
bp->b_iooffset = dbtob(bp->b_blkno);
bstrategy(bp);
}
return 0;
}
void
xfs_buf_free(xfs_buf_t *bp)
{
bp->b_flags |= B_INVAL;
BUF_KERNPROC(bp); /* ugly hack #1 */
if (bp->b_kvasize == 0) {
bp->b_saveaddr = bp->b_kvabase; /* ugly hack #2 */
bp->b_data = bp->b_saveaddr;
bp->b_bcount = 0;
bp->b_bufsize = 0;
}
brelse(bp);
}
/*
* Read data to a buf, including read-ahead if we find this to be beneficial.
* cluster_read replaces bread.
*/
int
cluster_read(struct vnode *vp, u_quad_t filesize, daddr_t lblkno, long size,
struct ucred *cred, long totread, int seqcount, int gbflags,
struct buf **bpp)
{
struct buf *bp, *rbp, *reqbp;
struct bufobj *bo;
daddr_t blkno, origblkno;
int maxra, racluster;
int error, ncontig;
int i;
error = 0;
bo = &vp->v_bufobj;
if (!unmapped_buf_allowed)
gbflags &= ~GB_UNMAPPED;
/*
* Try to limit the amount of read-ahead by a few
* ad-hoc parameters. This needs work!!!
*/
racluster = vp->v_mount->mnt_iosize_max / size;
maxra = seqcount;
maxra = min(read_max, maxra);
maxra = min(nbuf/8, maxra);
if (((u_quad_t)(lblkno + maxra + 1) * size) > filesize)
maxra = (filesize / size) - lblkno;
/*
* get the requested block
*/
*bpp = reqbp = bp = getblk(vp, lblkno, size, 0, 0, gbflags);
if (bp == NULL)
return (EBUSY);
origblkno = lblkno;
/*
* if it is in the cache, then check to see if the reads have been
* sequential. If they have, then try some read-ahead, otherwise
* back-off on prospective read-aheads.
*/
if (bp->b_flags & B_CACHE) {
if (!seqcount) {
return 0;
} else if ((bp->b_flags & B_RAM) == 0) {
return 0;
} else {
bp->b_flags &= ~B_RAM;
BO_RLOCK(bo);
for (i = 1; i < maxra; i++) {
/*
* Stop if the buffer does not exist or it
* is invalid (about to go away?)
*/
rbp = gbincore(&vp->v_bufobj, lblkno+i);
if (rbp == NULL || (rbp->b_flags & B_INVAL))
break;
/*
* Set another read-ahead mark so we know
* to check again. (If we can lock the
* buffer without waiting)
*/
if ((((i % racluster) == (racluster - 1)) ||
(i == (maxra - 1)))
&& (0 == BUF_LOCK(rbp,
LK_EXCLUSIVE | LK_NOWAIT, NULL))) {
rbp->b_flags |= B_RAM;
BUF_UNLOCK(rbp);
}
}
BO_RUNLOCK(bo);
if (i >= maxra) {
return 0;
}
lblkno += i;
}
reqbp = bp = NULL;
/*
* If it isn't in the cache, then get a chunk from
* disk if sequential, otherwise just get the block.
*/
} else {
off_t firstread = bp->b_offset;
int nblks;
long minread;
KASSERT(bp->b_offset != NOOFFSET,
("cluster_read: no buffer offset"));
ncontig = 0;
/*
* Adjust totread if needed
*/
minread = read_min * size;
if (minread > totread)
totread = minread;
/*
* Compute the total number of blocks that we should read
* synchronously.
*/
if (firstread + totread > filesize)
totread = filesize - firstread;
nblks = howmany(totread, size);
if (nblks > racluster)
nblks = racluster;
/*
* Now compute the number of contiguous blocks.
*/
if (nblks > 1) {
error = VOP_BMAP(vp, lblkno, NULL,
&blkno, &ncontig, NULL);
/*
* If this failed to map just do the original block.
*/
if (error || blkno == -1)
ncontig = 0;
}
/*
* If we have contiguous data available do a cluster
* otherwise just read the requested block.
*/
if (ncontig) {
/* Account for our first block. */
ncontig = min(ncontig + 1, nblks);
if (ncontig < nblks)
nblks = ncontig;
bp = cluster_rbuild(vp, filesize, lblkno,
blkno, size, nblks, gbflags, bp);
lblkno += (bp->b_bufsize / size);
} else {
bp->b_flags |= B_RAM;
bp->b_iocmd = BIO_READ;
lblkno += 1;
}
}
/*
* handle the synchronous read so that it is available ASAP.
*/
if (bp) {
if ((bp->b_flags & B_CLUSTER) == 0) {
vfs_busy_pages(bp, 0);
}
bp->b_flags &= ~B_INVAL;
bp->b_ioflags &= ~BIO_ERROR;
if ((bp->b_flags & B_ASYNC) || bp->b_iodone != NULL)
BUF_KERNPROC(bp);
bp->b_iooffset = dbtob(bp->b_blkno);
bstrategy(bp);
#ifdef RACCT
if (racct_enable) {
PROC_LOCK(curproc);
racct_add_buf(curproc, bp, 0);
PROC_UNLOCK(curproc);
}
#endif /* RACCT */
curthread->td_ru.ru_inblock++;
}
/*
* If we have been doing sequential I/O, then do some read-ahead.
*/
while (lblkno < (origblkno + maxra)) {
error = VOP_BMAP(vp, lblkno, NULL, &blkno, &ncontig, NULL);
if (error)
break;
if (blkno == -1)
break;
/*
* We could throttle ncontig here by maxra but we might as
* well read the data if it is contiguous. We're throttled
* by racluster anyway.
*/
if (ncontig) {
ncontig = min(ncontig + 1, racluster);
rbp = cluster_rbuild(vp, filesize, lblkno, blkno,
size, ncontig, gbflags, NULL);
lblkno += (rbp->b_bufsize / size);
if (rbp->b_flags & B_DELWRI) {
bqrelse(rbp);
continue;
}
} else {
rbp = getblk(vp, lblkno, size, 0, 0, gbflags);
lblkno += 1;
if (rbp->b_flags & B_DELWRI) {
bqrelse(rbp);
continue;
}
rbp->b_flags |= B_ASYNC | B_RAM;
rbp->b_iocmd = BIO_READ;
rbp->b_blkno = blkno;
}
if (rbp->b_flags & B_CACHE) {
rbp->b_flags &= ~B_ASYNC;
bqrelse(rbp);
continue;
}
if ((rbp->b_flags & B_CLUSTER) == 0) {
vfs_busy_pages(rbp, 0);
}
rbp->b_flags &= ~B_INVAL;
rbp->b_ioflags &= ~BIO_ERROR;
if ((rbp->b_flags & B_ASYNC) || rbp->b_iodone != NULL)
BUF_KERNPROC(rbp);
rbp->b_iooffset = dbtob(rbp->b_blkno);
bstrategy(rbp);
#ifdef RACCT
if (racct_enable) {
PROC_LOCK(curproc);
racct_add_buf(curproc, rbp, 0);
PROC_UNLOCK(curproc);
}
#endif /* RACCT */
curthread->td_ru.ru_inblock++;
}
if (reqbp) {
/*
* Like bread, always brelse() the buffer when
* returning an error.
*/
error = bufwait(reqbp);
if (error != 0) {
brelse(reqbp);
*bpp = NULL;
}
}
return (error);
}
/*
* If blocks are contiguous on disk, use this to provide clustered
* read ahead. We will read as many blocks as possible sequentially
* and then parcel them up into logical blocks in the buffer hash table.
*/
static struct buf *
cluster_rbuild(struct vnode *vp, u_quad_t filesize, daddr_t lbn,
daddr_t blkno, long size, int run, int gbflags, struct buf *fbp)
{
struct buf *bp, *tbp;
daddr_t bn;
off_t off;
long tinc, tsize;
int i, inc, j, k, toff;
KASSERT(size == vp->v_mount->mnt_stat.f_iosize,
("cluster_rbuild: size %ld != f_iosize %jd\n",
size, (intmax_t)vp->v_mount->mnt_stat.f_iosize));
/*
* avoid a division
*/
while ((u_quad_t) size * (lbn + run) > filesize) {
--run;
}
if (fbp) {
tbp = fbp;
tbp->b_iocmd = BIO_READ;
} else {
tbp = getblk(vp, lbn, size, 0, 0, gbflags);
if (tbp->b_flags & B_CACHE)
return tbp;
tbp->b_flags |= B_ASYNC | B_RAM;
tbp->b_iocmd = BIO_READ;
}
tbp->b_blkno = blkno;
if( (tbp->b_flags & B_MALLOC) ||
((tbp->b_flags & B_VMIO) == 0) || (run <= 1) )
return tbp;
bp = trypbuf(&cluster_pbuf_freecnt);
if (bp == NULL)
return tbp;
/*
* We are synthesizing a buffer out of vm_page_t's, but
* if the block size is not page aligned then the starting
* address may not be either. Inherit the b_data offset
* from the original buffer.
*/
bp->b_flags = B_ASYNC | B_CLUSTER | B_VMIO;
if ((gbflags & GB_UNMAPPED) != 0) {
bp->b_data = unmapped_buf;
} else {
bp->b_data = (char *)((vm_offset_t)bp->b_data |
((vm_offset_t)tbp->b_data & PAGE_MASK));
}
bp->b_iocmd = BIO_READ;
bp->b_iodone = cluster_callback;
bp->b_blkno = blkno;
bp->b_lblkno = lbn;
bp->b_offset = tbp->b_offset;
KASSERT(bp->b_offset != NOOFFSET, ("cluster_rbuild: no buffer offset"));
pbgetvp(vp, bp);
TAILQ_INIT(&bp->b_cluster.cluster_head);
bp->b_bcount = 0;
bp->b_bufsize = 0;
bp->b_npages = 0;
inc = btodb(size);
for (bn = blkno, i = 0; i < run; ++i, bn += inc) {
if (i == 0) {
VM_OBJECT_WLOCK(tbp->b_bufobj->bo_object);
vfs_drain_busy_pages(tbp);
vm_object_pip_add(tbp->b_bufobj->bo_object,
tbp->b_npages);
for (k = 0; k < tbp->b_npages; k++)
vm_page_sbusy(tbp->b_pages[k]);
VM_OBJECT_WUNLOCK(tbp->b_bufobj->bo_object);
} else {
if ((bp->b_npages * PAGE_SIZE) +
round_page(size) > vp->v_mount->mnt_iosize_max) {
break;
}
tbp = getblk(vp, lbn + i, size, 0, 0, GB_LOCK_NOWAIT |
(gbflags & GB_UNMAPPED));
/* Don't wait around for locked bufs. */
if (tbp == NULL)
break;
/*
* Stop scanning if the buffer is fully valid
* (marked B_CACHE), or locked (may be doing a
* background write), or if the buffer is not
* VMIO backed. The clustering code can only deal
* with VMIO-backed buffers. The bo lock is not
* required for the BKGRDINPROG check since it
* can not be set without the buf lock.
*/
if ((tbp->b_vflags & BV_BKGRDINPROG) ||
(tbp->b_flags & B_CACHE) ||
(tbp->b_flags & B_VMIO) == 0) {
bqrelse(tbp);
break;
}
/*
* The buffer must be completely invalid in order to
* take part in the cluster. If it is partially valid
* then we stop.
*/
off = tbp->b_offset;
tsize = size;
VM_OBJECT_WLOCK(tbp->b_bufobj->bo_object);
for (j = 0; tsize > 0; j++) {
toff = off & PAGE_MASK;
tinc = tsize;
if (toff + tinc > PAGE_SIZE)
tinc = PAGE_SIZE - toff;
VM_OBJECT_ASSERT_WLOCKED(tbp->b_pages[j]->object);
if ((tbp->b_pages[j]->valid &
vm_page_bits(toff, tinc)) != 0)
break;
if (vm_page_xbusied(tbp->b_pages[j]))
break;
vm_object_pip_add(tbp->b_bufobj->bo_object, 1);
vm_page_sbusy(tbp->b_pages[j]);
off += tinc;
tsize -= tinc;
}
if (tsize > 0) {
clean_sbusy:
vm_object_pip_add(tbp->b_bufobj->bo_object, -j);
for (k = 0; k < j; k++)
vm_page_sunbusy(tbp->b_pages[k]);
VM_OBJECT_WUNLOCK(tbp->b_bufobj->bo_object);
bqrelse(tbp);
break;
}
VM_OBJECT_WUNLOCK(tbp->b_bufobj->bo_object);
/*
* Set a read-ahead mark as appropriate
*/
if ((fbp && (i == 1)) || (i == (run - 1)))
tbp->b_flags |= B_RAM;
/*
* Set the buffer up for an async read (XXX should
* we do this only if we do not wind up brelse()ing?).
* Set the block number if it isn't set, otherwise
* if it is make sure it matches the block number we
* expect.
*/
tbp->b_flags |= B_ASYNC;
tbp->b_iocmd = BIO_READ;
if (tbp->b_blkno == tbp->b_lblkno) {
tbp->b_blkno = bn;
} else if (tbp->b_blkno != bn) {
VM_OBJECT_WLOCK(tbp->b_bufobj->bo_object);
goto clean_sbusy;
}
}
/*
* XXX fbp from caller may not be B_ASYNC, but we are going
* to biodone() it in cluster_callback() anyway
*/
BUF_KERNPROC(tbp);
TAILQ_INSERT_TAIL(&bp->b_cluster.cluster_head,
tbp, b_cluster.cluster_entry);
VM_OBJECT_WLOCK(tbp->b_bufobj->bo_object);
for (j = 0; j < tbp->b_npages; j += 1) {
vm_page_t m;
m = tbp->b_pages[j];
if ((bp->b_npages == 0) ||
(bp->b_pages[bp->b_npages-1] != m)) {
bp->b_pages[bp->b_npages] = m;
bp->b_npages++;
}
if (m->valid == VM_PAGE_BITS_ALL)
tbp->b_pages[j] = bogus_page;
}
VM_OBJECT_WUNLOCK(tbp->b_bufobj->bo_object);
/*
* Don't inherit tbp->b_bufsize as it may be larger due to
* a non-page-aligned size. Instead just aggregate using
* 'size'.
*/
if (tbp->b_bcount != size)
printf("warning: tbp->b_bcount wrong %ld vs %ld\n", tbp->b_bcount, size);
if (tbp->b_bufsize != size)
printf("warning: tbp->b_bufsize wrong %ld vs %ld\n", tbp->b_bufsize, size);
bp->b_bcount += size;
bp->b_bufsize += size;
}
/*
* Fully valid pages in the cluster are already good and do not need
* to be re-read from disk. Replace the page with bogus_page
*/
VM_OBJECT_WLOCK(bp->b_bufobj->bo_object);
for (j = 0; j < bp->b_npages; j++) {
VM_OBJECT_ASSERT_WLOCKED(bp->b_pages[j]->object);
if (bp->b_pages[j]->valid == VM_PAGE_BITS_ALL)
bp->b_pages[j] = bogus_page;
}
VM_OBJECT_WUNLOCK(bp->b_bufobj->bo_object);
if (bp->b_bufsize > bp->b_kvasize)
panic("cluster_rbuild: b_bufsize(%ld) > b_kvasize(%d)\n",
bp->b_bufsize, bp->b_kvasize);
if (buf_mapped(bp)) {
pmap_qenter(trunc_page((vm_offset_t) bp->b_data),
(vm_page_t *)bp->b_pages, bp->b_npages);
}
return (bp);
}
/*
* This is now called from local media FS's to operate against their
* own vnodes if they fail to implement VOP_GETPAGES.
*/
int
vnode_pager_generic_getpages(struct vnode *vp, vm_page_t *m, int count,
int *a_rbehind, int *a_rahead, vop_getpages_iodone_t iodone, void *arg)
{
vm_object_t object;
struct bufobj *bo;
struct buf *bp;
off_t foff;
#ifdef INVARIANTS
off_t blkno0;
#endif
int bsize, pagesperblock, *freecnt;
int error, before, after, rbehind, rahead, poff, i;
int bytecount, secmask;
KASSERT(vp->v_type != VCHR && vp->v_type != VBLK,
("%s does not support devices", __func__));
if (vp->v_iflag & VI_DOOMED)
return (VM_PAGER_BAD);
object = vp->v_object;
foff = IDX_TO_OFF(m[0]->pindex);
bsize = vp->v_mount->mnt_stat.f_iosize;
pagesperblock = bsize / PAGE_SIZE;
KASSERT(foff < object->un_pager.vnp.vnp_size,
("%s: page %p offset beyond vp %p size", __func__, m[0], vp));
KASSERT(count <= sizeof(bp->b_pages),
("%s: requested %d pages", __func__, count));
/*
* The last page has valid blocks. Invalid part can only
* exist at the end of file, and the page is made fully valid
* by zeroing in vm_pager_get_pages().
*/
if (m[count - 1]->valid != 0 && --count == 0) {
if (iodone != NULL)
iodone(arg, m, 1, 0);
return (VM_PAGER_OK);
}
/*
* Synchronous and asynchronous paging operations use different
* free pbuf counters. This is done to avoid asynchronous requests
* to consume all pbufs.
* Allocate the pbuf at the very beginning of the function, so that
* if we are low on certain kind of pbufs don't even proceed to BMAP,
* but sleep.
*/
freecnt = iodone != NULL ?
&vnode_async_pbuf_freecnt : &vnode_pbuf_freecnt;
bp = getpbuf(freecnt);
/*
* Get the underlying device blocks for the file with VOP_BMAP().
* If the file system doesn't support VOP_BMAP, use old way of
* getting pages via VOP_READ.
*/
error = VOP_BMAP(vp, foff / bsize, &bo, &bp->b_blkno, &after, &before);
if (error == EOPNOTSUPP) {
relpbuf(bp, freecnt);
VM_OBJECT_WLOCK(object);
for (i = 0; i < count; i++) {
PCPU_INC(cnt.v_vnodein);
PCPU_INC(cnt.v_vnodepgsin);
error = vnode_pager_input_old(object, m[i]);
if (error)
break;
}
VM_OBJECT_WUNLOCK(object);
return (error);
} else if (error != 0) {
relpbuf(bp, freecnt);
return (VM_PAGER_ERROR);
}
/*
* If the file system supports BMAP, but blocksize is smaller
* than a page size, then use special small filesystem code.
*/
if (pagesperblock == 0) {
relpbuf(bp, freecnt);
for (i = 0; i < count; i++) {
PCPU_INC(cnt.v_vnodein);
PCPU_INC(cnt.v_vnodepgsin);
error = vnode_pager_input_smlfs(object, m[i]);
if (error)
break;
}
return (error);
}
/*
* A sparse file can be encountered only for a single page request,
* which may not be preceded by call to vm_pager_haspage().
*/
if (bp->b_blkno == -1) {
KASSERT(count == 1,
("%s: array[%d] request to a sparse file %p", __func__,
count, vp));
relpbuf(bp, freecnt);
pmap_zero_page(m[0]);
KASSERT(m[0]->dirty == 0, ("%s: page %p is dirty",
__func__, m[0]));
VM_OBJECT_WLOCK(object);
m[0]->valid = VM_PAGE_BITS_ALL;
VM_OBJECT_WUNLOCK(object);
return (VM_PAGER_OK);
}
#ifdef INVARIANTS
blkno0 = bp->b_blkno;
#endif
bp->b_blkno += (foff % bsize) / DEV_BSIZE;
/* Recalculate blocks available after/before to pages. */
poff = (foff % bsize) / PAGE_SIZE;
before *= pagesperblock;
before += poff;
after *= pagesperblock;
after += pagesperblock - (poff + 1);
if (m[0]->pindex + after >= object->size)
after = object->size - 1 - m[0]->pindex;
KASSERT(count <= after + 1, ("%s: %d pages asked, can do only %d",
__func__, count, after + 1));
after -= count - 1;
/* Trim requested rbehind/rahead to possible values. */
rbehind = a_rbehind ? *a_rbehind : 0;
rahead = a_rahead ? *a_rahead : 0;
rbehind = min(rbehind, before);
rbehind = min(rbehind, m[0]->pindex);
rahead = min(rahead, after);
rahead = min(rahead, object->size - m[count - 1]->pindex);
/*
* Check that total amount of pages fit into buf. Trim rbehind and
* rahead evenly if not.
*/
if (rbehind + rahead + count > nitems(bp->b_pages)) {
int trim, sum;
trim = rbehind + rahead + count - nitems(bp->b_pages) + 1;
sum = rbehind + rahead;
if (rbehind == before) {
/* Roundup rbehind trim to block size. */
rbehind -= roundup(trim * rbehind / sum, pagesperblock);
if (rbehind < 0)
rbehind = 0;
} else
rbehind -= trim * rbehind / sum;
rahead -= trim * rahead / sum;
}
KASSERT(rbehind + rahead + count <= nitems(bp->b_pages),
("%s: behind %d ahead %d count %d", __func__,
rbehind, rahead, count));
/*
* Fill in the bp->b_pages[] array with requested and optional
* read behind or read ahead pages. Read behind pages are looked
* up in a backward direction, down to a first cached page. Same
* for read ahead pages, but there is no need to shift the array
* in case of encountering a cached page.
*/
i = bp->b_npages = 0;
if (rbehind) {
vm_pindex_t startpindex, tpindex;
vm_page_t p;
VM_OBJECT_WLOCK(object);
startpindex = m[0]->pindex - rbehind;
if ((p = TAILQ_PREV(m[0], pglist, listq)) != NULL &&
p->pindex >= startpindex)
startpindex = p->pindex + 1;
/* tpindex is unsigned; beware of numeric underflow. */
for (tpindex = m[0]->pindex - 1;
tpindex >= startpindex && tpindex < m[0]->pindex;
tpindex--, i++) {
p = vm_page_alloc(object, tpindex, VM_ALLOC_NORMAL);
if (p == NULL) {
/* Shift the array. */
for (int j = 0; j < i; j++)
bp->b_pages[j] = bp->b_pages[j +
tpindex + 1 - startpindex];
break;
}
bp->b_pages[tpindex - startpindex] = p;
}
bp->b_pgbefore = i;
bp->b_npages += i;
bp->b_blkno -= IDX_TO_OFF(i) / DEV_BSIZE;
} else
bp->b_pgbefore = 0;
/* Requested pages. */
for (int j = 0; j < count; j++, i++)
bp->b_pages[i] = m[j];
bp->b_npages += count;
if (rahead) {
vm_pindex_t endpindex, tpindex;
vm_page_t p;
if (!VM_OBJECT_WOWNED(object))
VM_OBJECT_WLOCK(object);
endpindex = m[count - 1]->pindex + rahead + 1;
if ((p = TAILQ_NEXT(m[count - 1], listq)) != NULL &&
p->pindex < endpindex)
endpindex = p->pindex;
if (endpindex > object->size)
endpindex = object->size;
for (tpindex = m[count - 1]->pindex + 1;
tpindex < endpindex; i++, tpindex++) {
p = vm_page_alloc(object, tpindex, VM_ALLOC_NORMAL);
if (p == NULL)
break;
bp->b_pages[i] = p;
}
bp->b_pgafter = i - bp->b_npages;
bp->b_npages = i;
} else
bp->b_pgafter = 0;
if (VM_OBJECT_WOWNED(object))
VM_OBJECT_WUNLOCK(object);
/* Report back actual behind/ahead read. */
if (a_rbehind)
*a_rbehind = bp->b_pgbefore;
if (a_rahead)
*a_rahead = bp->b_pgafter;
#ifdef INVARIANTS
KASSERT(bp->b_npages <= nitems(bp->b_pages),
("%s: buf %p overflowed", __func__, bp));
for (int j = 1; j < bp->b_npages; j++)
KASSERT(bp->b_pages[j]->pindex - 1 ==
bp->b_pages[j - 1]->pindex,
("%s: pages array not consecutive, bp %p", __func__, bp));
#endif
/*
* Recalculate first offset and bytecount with regards to read behind.
* Truncate bytecount to vnode real size and round up physical size
* for real devices.
*/
foff = IDX_TO_OFF(bp->b_pages[0]->pindex);
bytecount = bp->b_npages << PAGE_SHIFT;
if ((foff + bytecount) > object->un_pager.vnp.vnp_size)
bytecount = object->un_pager.vnp.vnp_size - foff;
secmask = bo->bo_bsize - 1;
KASSERT(secmask < PAGE_SIZE && secmask > 0,
("%s: sector size %d too large", __func__, secmask + 1));
bytecount = (bytecount + secmask) & ~secmask;
/*
* And map the pages to be read into the kva, if the filesystem
* requires mapped buffers.
*/
if ((vp->v_mount->mnt_kern_flag & MNTK_UNMAPPED_BUFS) != 0 &&
unmapped_buf_allowed) {
bp->b_data = unmapped_buf;
bp->b_offset = 0;
} else {
bp->b_data = bp->b_kvabase;
pmap_qenter((vm_offset_t)bp->b_data, bp->b_pages, bp->b_npages);
}
/* Build a minimal buffer header. */
bp->b_iocmd = BIO_READ;
KASSERT(bp->b_rcred == NOCRED, ("leaking read ucred"));
KASSERT(bp->b_wcred == NOCRED, ("leaking write ucred"));
bp->b_rcred = crhold(curthread->td_ucred);
bp->b_wcred = crhold(curthread->td_ucred);
pbgetbo(bo, bp);
bp->b_vp = vp;
bp->b_bcount = bp->b_bufsize = bp->b_runningbufspace = bytecount;
bp->b_iooffset = dbtob(bp->b_blkno);
KASSERT(IDX_TO_OFF(m[0]->pindex - bp->b_pages[0]->pindex) ==
(blkno0 - bp->b_blkno) * DEV_BSIZE +
IDX_TO_OFF(m[0]->pindex) % bsize,
("wrong offsets bsize %d m[0] %ju b_pages[0] %ju "
"blkno0 %ju b_blkno %ju", bsize,
(uintmax_t)m[0]->pindex, (uintmax_t)bp->b_pages[0]->pindex,
(uintmax_t)blkno0, (uintmax_t)bp->b_blkno));
atomic_add_long(&runningbufspace, bp->b_runningbufspace);
PCPU_INC(cnt.v_vnodein);
PCPU_ADD(cnt.v_vnodepgsin, bp->b_npages);
if (iodone != NULL) { /* async */
bp->b_pgiodone = iodone;
bp->b_caller1 = arg;
bp->b_iodone = vnode_pager_generic_getpages_done_async;
bp->b_flags |= B_ASYNC;
BUF_KERNPROC(bp);
bstrategy(bp);
return (VM_PAGER_OK);
} else {
bp->b_iodone = bdone;
bstrategy(bp);
bwait(bp, PVM, "vnread");
error = vnode_pager_generic_getpages_done(bp);
for (i = 0; i < bp->b_npages; i++)
bp->b_pages[i] = NULL;
bp->b_vp = NULL;
pbrelbo(bp);
relpbuf(bp, &vnode_pbuf_freecnt);
return (error != 0 ? VM_PAGER_ERROR : VM_PAGER_OK);
}
}
/*
* Convert a vnode strategy call into a device strategy call. Vnode strategy
* calls are not limited to device DMA limits so we have to deal with the
* case.
*
* spec_strategy(struct vnode *a_vp, struct bio *a_bio)
*/
static int
devfs_spec_strategy(struct vop_strategy_args *ap)
{
struct bio *bio = ap->a_bio;
struct buf *bp = bio->bio_buf;
struct buf *nbp;
struct vnode *vp;
struct mount *mp;
int chunksize;
int maxiosize;
if (bp->b_cmd != BUF_CMD_READ && LIST_FIRST(&bp->b_dep) != NULL)
buf_start(bp);
/*
* Collect statistics on synchronous and asynchronous read
* and write counts for disks that have associated filesystems.
*/
vp = ap->a_vp;
KKASSERT(vp->v_rdev != NULL); /* XXX */
if (vn_isdisk(vp, NULL) && (mp = vp->v_rdev->si_mountpoint) != NULL) {
if (bp->b_cmd == BUF_CMD_READ) {
if (bp->b_flags & BIO_SYNC)
mp->mnt_stat.f_syncreads++;
else
mp->mnt_stat.f_asyncreads++;
} else {
if (bp->b_flags & BIO_SYNC)
mp->mnt_stat.f_syncwrites++;
else
mp->mnt_stat.f_asyncwrites++;
}
}
/*
* Device iosize limitations only apply to read and write. Shortcut
* the I/O if it fits.
*/
if ((maxiosize = vp->v_rdev->si_iosize_max) == 0) {
devfs_debug(DEVFS_DEBUG_DEBUG,
"%s: si_iosize_max not set!\n",
dev_dname(vp->v_rdev));
maxiosize = MAXPHYS;
}
#if SPEC_CHAIN_DEBUG & 2
maxiosize = 4096;
#endif
if (bp->b_bcount <= maxiosize ||
(bp->b_cmd != BUF_CMD_READ && bp->b_cmd != BUF_CMD_WRITE)) {
dev_dstrategy_chain(vp->v_rdev, bio);
return (0);
}
/*
* Clone the buffer and set up an I/O chain to chunk up the I/O.
*/
nbp = kmalloc(sizeof(*bp), M_DEVBUF, M_INTWAIT|M_ZERO);
initbufbio(nbp);
buf_dep_init(nbp);
BUF_LOCK(nbp, LK_EXCLUSIVE);
BUF_KERNPROC(nbp);
nbp->b_vp = vp;
nbp->b_flags = B_PAGING | (bp->b_flags & B_BNOCLIP);
nbp->b_data = bp->b_data;
nbp->b_bio1.bio_done = devfs_spec_strategy_done;
nbp->b_bio1.bio_offset = bio->bio_offset;
nbp->b_bio1.bio_caller_info1.ptr = bio;
/*
* Start the first transfer
*/
if (vn_isdisk(vp, NULL))
chunksize = vp->v_rdev->si_bsize_phys;
else
chunksize = DEV_BSIZE;
chunksize = maxiosize / chunksize * chunksize;
#if SPEC_CHAIN_DEBUG & 1
devfs_debug(DEVFS_DEBUG_DEBUG,
"spec_strategy chained I/O chunksize=%d\n",
chunksize);
#endif
nbp->b_cmd = bp->b_cmd;
nbp->b_bcount = chunksize;
nbp->b_bufsize = chunksize; /* used to detect a short I/O */
nbp->b_bio1.bio_caller_info2.index = chunksize;
#if SPEC_CHAIN_DEBUG & 1
devfs_debug(DEVFS_DEBUG_DEBUG,
"spec_strategy: chain %p offset %d/%d bcount %d\n",
bp, 0, bp->b_bcount, nbp->b_bcount);
#endif
dev_dstrategy(vp->v_rdev, &nbp->b_bio1);
if (DEVFS_NODE(vp)) {
nanotime(&DEVFS_NODE(vp)->atime);
nanotime(&DEVFS_NODE(vp)->mtime);
}
return (0);
}
/*
* This is now called from local media FS's to operate against their
* own vnodes if they fail to implement VOP_GETPAGES.
*/
int
vnode_pager_generic_getpages(struct vnode *vp, vm_page_t *m, int bytecount,
int reqpage, vop_getpages_iodone_t iodone, void *arg)
{
vm_object_t object;
struct bufobj *bo;
struct buf *bp;
daddr_t firstaddr, reqblock;
off_t foff, pib;
int pbefore, pafter, i, size, bsize, first, last, *freecnt;
int count, error, before, after, secmask;
KASSERT(vp->v_type != VCHR && vp->v_type != VBLK,
("vnode_pager_generic_getpages does not support devices"));
if (vp->v_iflag & VI_DOOMED)
return (VM_PAGER_BAD);
object = vp->v_object;
count = bytecount / PAGE_SIZE;
bsize = vp->v_mount->mnt_stat.f_iosize;
/*
* Synchronous and asynchronous paging operations use different
* free pbuf counters. This is done to avoid asynchronous requests
* to consume all pbufs.
* Allocate the pbuf at the very beginning of the function, so that
* if we are low on certain kind of pbufs don't even proceed to BMAP,
* but sleep.
*/
freecnt = iodone != NULL ?
&vnode_async_pbuf_freecnt : &vnode_pbuf_freecnt;
bp = getpbuf(freecnt);
/*
* Get the underlying device blocks for the file with VOP_BMAP().
* If the file system doesn't support VOP_BMAP, use old way of
* getting pages via VOP_READ.
*/
error = VOP_BMAP(vp, IDX_TO_OFF(m[reqpage]->pindex) / bsize, &bo,
&reqblock, &after, &before);
if (error == EOPNOTSUPP) {
relpbuf(bp, freecnt);
VM_OBJECT_WLOCK(object);
for (i = 0; i < count; i++)
if (i != reqpage) {
vm_page_lock(m[i]);
vm_page_free(m[i]);
vm_page_unlock(m[i]);
}
PCPU_INC(cnt.v_vnodein);
PCPU_INC(cnt.v_vnodepgsin);
error = vnode_pager_input_old(object, m[reqpage]);
VM_OBJECT_WUNLOCK(object);
return (error);
} else if (error != 0) {
relpbuf(bp, freecnt);
vm_pager_free_nonreq(object, m, reqpage, count, FALSE);
return (VM_PAGER_ERROR);
/*
* If the blocksize is smaller than a page size, then use
* special small filesystem code.
*/
} else if ((PAGE_SIZE / bsize) > 1) {
relpbuf(bp, freecnt);
vm_pager_free_nonreq(object, m, reqpage, count, FALSE);
PCPU_INC(cnt.v_vnodein);
PCPU_INC(cnt.v_vnodepgsin);
return (vnode_pager_input_smlfs(object, m[reqpage]));
}
/*
* Since the caller has busied the requested page, that page's valid
* field will not be changed by other threads.
*/
vm_page_assert_xbusied(m[reqpage]);
/*
* If we have a completely valid page available to us, we can
* clean up and return. Otherwise we have to re-read the
* media.
*/
if (m[reqpage]->valid == VM_PAGE_BITS_ALL) {
relpbuf(bp, freecnt);
vm_pager_free_nonreq(object, m, reqpage, count, FALSE);
return (VM_PAGER_OK);
} else if (reqblock == -1) {
relpbuf(bp, freecnt);
pmap_zero_page(m[reqpage]);
KASSERT(m[reqpage]->dirty == 0,
("vnode_pager_generic_getpages: page %p is dirty", m));
VM_OBJECT_WLOCK(object);
m[reqpage]->valid = VM_PAGE_BITS_ALL;
vm_pager_free_nonreq(object, m, reqpage, count, TRUE);
VM_OBJECT_WUNLOCK(object);
return (VM_PAGER_OK);
} else if (m[reqpage]->valid != 0) {
VM_OBJECT_WLOCK(object);
m[reqpage]->valid = 0;
VM_OBJECT_WUNLOCK(object);
}
pib = IDX_TO_OFF(m[reqpage]->pindex) % bsize;
pbefore = ((daddr_t)before * bsize + pib) / PAGE_SIZE;
pafter = ((daddr_t)(after + 1) * bsize - pib) / PAGE_SIZE - 1;
first = reqpage < pbefore ? 0 : reqpage - pbefore;
last = reqpage + pafter >= count ? count - 1 : reqpage + pafter;
if (first > 0 || last + 1 < count) {
VM_OBJECT_WLOCK(object);
for (i = 0; i < first; i++) {
vm_page_lock(m[i]);
vm_page_free(m[i]);
vm_page_unlock(m[i]);
}
for (i = last + 1; i < count; i++) {
vm_page_lock(m[i]);
vm_page_free(m[i]);
vm_page_unlock(m[i]);
}
VM_OBJECT_WUNLOCK(object);
}
/*
* here on direct device I/O
*/
firstaddr = reqblock;
firstaddr += pib / DEV_BSIZE;
firstaddr -= IDX_TO_OFF(reqpage - first) / DEV_BSIZE;
/*
* The first and last page have been calculated now, move
* input pages to be zero based, and adjust the count.
*/
m += first;
reqpage -= first;
count = last - first + 1;
/*
* calculate the file virtual address for the transfer
*/
foff = IDX_TO_OFF(m[0]->pindex);
/*
* calculate the size of the transfer
*/
size = count * PAGE_SIZE;
KASSERT(count > 0, ("zero count"));
if ((foff + size) > object->un_pager.vnp.vnp_size)
size = object->un_pager.vnp.vnp_size - foff;
KASSERT(size > 0, ("zero size"));
/*
* round up physical size for real devices.
*/
secmask = bo->bo_bsize - 1;
KASSERT(secmask < PAGE_SIZE && secmask > 0,
("vnode_pager_generic_getpages: sector size %d too large",
secmask + 1));
size = (size + secmask) & ~secmask;
/*
* and map the pages to be read into the kva, if the filesystem
* requires mapped buffers.
*/
if ((vp->v_mount->mnt_kern_flag & MNTK_UNMAPPED_BUFS) != 0 &&
unmapped_buf_allowed) {
bp->b_data = unmapped_buf;
bp->b_offset = 0;
} else {
bp->b_data = bp->b_kvabase;
pmap_qenter((vm_offset_t)bp->b_data, m, count);
}
/* build a minimal buffer header */
bp->b_iocmd = BIO_READ;
KASSERT(bp->b_rcred == NOCRED, ("leaking read ucred"));
KASSERT(bp->b_wcred == NOCRED, ("leaking write ucred"));
bp->b_rcred = crhold(curthread->td_ucred);
bp->b_wcred = crhold(curthread->td_ucred);
bp->b_blkno = firstaddr;
pbgetbo(bo, bp);
bp->b_vp = vp;
bp->b_bcount = size;
bp->b_bufsize = size;
bp->b_runningbufspace = bp->b_bufsize;
for (i = 0; i < count; i++)
bp->b_pages[i] = m[i];
bp->b_npages = count;
bp->b_pager.pg_reqpage = reqpage;
atomic_add_long(&runningbufspace, bp->b_runningbufspace);
PCPU_INC(cnt.v_vnodein);
PCPU_ADD(cnt.v_vnodepgsin, count);
/* do the input */
bp->b_iooffset = dbtob(bp->b_blkno);
if (iodone != NULL) { /* async */
bp->b_pager.pg_iodone = iodone;
bp->b_caller1 = arg;
bp->b_iodone = vnode_pager_generic_getpages_done_async;
bp->b_flags |= B_ASYNC;
BUF_KERNPROC(bp);
bstrategy(bp);
/* Good bye! */
} else {
bp->b_iodone = bdone;
bstrategy(bp);
bwait(bp, PVM, "vnread");
error = vnode_pager_generic_getpages_done(bp);
for (i = 0; i < bp->b_npages; i++)
bp->b_pages[i] = NULL;
bp->b_vp = NULL;
pbrelbo(bp);
relpbuf(bp, &vnode_pbuf_freecnt);
}
return (error != 0 ? VM_PAGER_ERROR : VM_PAGER_OK);
}
