static void
vnode_pager_generic_getpages_done_async(struct buf *bp)
{
int error;
error = vnode_pager_generic_getpages_done(bp);
bp->b_pager.pg_iodone(bp->b_caller1, bp->b_pages,
bp->b_pager.pg_reqpage, error);
for (int i = 0; i < bp->b_npages; i++)
bp->b_pages[i] = NULL;
bp->b_vp = NULL;
pbrelbo(bp);
relpbuf(bp, &vnode_async_pbuf_freecnt);
}
static void
vnode_pager_generic_getpages_done_async(struct buf *bp)
{
int error;
error = vnode_pager_generic_getpages_done(bp);
/* Run the iodone upon the requested range. */
bp->b_pgiodone(bp->b_caller1, bp->b_pages + bp->b_pgbefore,
bp->b_npages - bp->b_pgbefore - bp->b_pgafter, error);
for (int i = 0; i < bp->b_npages; i++)
bp->b_pages[i] = NULL;
bp->b_vp = NULL;
pbrelbo(bp);
relpbuf(bp, &vnode_async_pbuf_freecnt);
}
int
physio(struct cdev *dev, struct uio *uio, int ioflag)
{
int i;
int error;
caddr_t sa;
u_int iolen;
struct buf *bp;
/* Keep the process UPAGES from being swapped. XXX: why ? */
PHOLD(curproc);
bp = getpbuf(NULL);
sa = bp->b_data;
error = 0;
/* XXX: sanity check */
if(dev->si_iosize_max < PAGE_SIZE) {
printf("WARNING: %s si_iosize_max=%d, using DFLTPHYS.\n",
devtoname(dev), dev->si_iosize_max);
dev->si_iosize_max = DFLTPHYS;
}
for (i = 0; i < uio->uio_iovcnt; i++) {
while (uio->uio_iov[i].iov_len) {
bp->b_flags = 0;
if (uio->uio_rw == UIO_READ) {
bp->b_iocmd = BIO_READ;
curthread->td_ru.ru_inblock++;
} else {
bp->b_iocmd = BIO_WRITE;
curthread->td_ru.ru_oublock++;
}
bp->b_iodone = bdone;
bp->b_data = uio->uio_iov[i].iov_base;
bp->b_bcount = uio->uio_iov[i].iov_len;
bp->b_offset = uio->uio_offset;
bp->b_iooffset = uio->uio_offset;
bp->b_saveaddr = sa;
/* Don't exceed drivers iosize limit */
if (bp->b_bcount > dev->si_iosize_max)
bp->b_bcount = dev->si_iosize_max;
/*
* Make sure the pbuf can map the request
* XXX: The pbuf has kvasize = MAXPHYS so a request
* XXX: larger than MAXPHYS - PAGE_SIZE must be
* XXX: page aligned or it will be fragmented.
*/
iolen = ((vm_offset_t) bp->b_data) & PAGE_MASK;
if ((bp->b_bcount + iolen) > bp->b_kvasize) {
bp->b_bcount = bp->b_kvasize;
if (iolen != 0)
bp->b_bcount -= PAGE_SIZE;
}
bp->b_bufsize = bp->b_bcount;
bp->b_blkno = btodb(bp->b_offset);
if (uio->uio_segflg == UIO_USERSPACE)
if (vmapbuf(bp) < 0) {
error = EFAULT;
goto doerror;
}
dev_strategy(dev, bp);
if (uio->uio_rw == UIO_READ)
bwait(bp, PRIBIO, "physrd");
else
bwait(bp, PRIBIO, "physwr");
if (uio->uio_segflg == UIO_USERSPACE)
vunmapbuf(bp);
iolen = bp->b_bcount - bp->b_resid;
if (iolen == 0 && !(bp->b_ioflags & BIO_ERROR))
goto doerror; /* EOF */
uio->uio_iov[i].iov_len -= iolen;
uio->uio_iov[i].iov_base =
(char *)uio->uio_iov[i].iov_base + iolen;
uio->uio_resid -= iolen;
uio->uio_offset += iolen;
if( bp->b_ioflags & BIO_ERROR) {
error = bp->b_error;
goto doerror;
}
}
}
doerror:
relpbuf(bp, NULL);
PRELE(curproc);
return (error);
}
static int
ffs_rawread_main(struct vnode *vp, struct uio *uio)
{
int error, nerror;
struct buf *bp, *nbp, *tbp;
int iolen;
caddr_t udata;
int resid;
off_t offset;
udata = uio->uio_iov->iov_base;
resid = uio->uio_resid;
offset = uio->uio_offset;
error = 0;
nerror = 0;
bp = NULL;
nbp = NULL;
while (resid > 0) {
if (bp == NULL) { /* Setup first read */
/* XXX: Leave some bufs for swap */
bp = getpbuf_kva(&vp->v_mount->mnt_pbuf_count);
error = ffs_rawread_readahead(vp, udata, offset,
resid, bp);
if (error != 0)
break;
if (resid > bp->b_bufsize) { /* Setup fist readahead */
/* XXX: Leave bufs for swap */
if (rawreadahead != 0)
nbp = trypbuf_kva(&vp->v_mount->mnt_pbuf_count);
else
nbp = NULL;
if (nbp != NULL) {
nerror = ffs_rawread_readahead(
vp,
udata + bp->b_bufsize,
offset + bp->b_bufsize,
resid - bp->b_bufsize,
nbp);
if (nerror) {
relpbuf(nbp, &vp->v_mount->mnt_pbuf_count);
nbp = NULL;
}
}
}
}
biowait(&bp->b_bio1, "rawrd");
vunmapbuf(bp);
iolen = bp->b_bcount - bp->b_resid;
if (iolen == 0 && (bp->b_flags & B_ERROR) == 0) {
nerror = 0; /* Ignore possible beyond EOF error */
break; /* EOF */
}
if ((bp->b_flags & B_ERROR) != 0) {
error = bp->b_error;
break;
}
clearbiocache(&bp->b_bio2);
resid -= iolen;
udata += iolen;
offset += iolen;
if (iolen < bp->b_bufsize) {
/* Incomplete read. Try to read remaining part */
error = ffs_rawread_readahead(
vp, udata, offset,
bp->b_bufsize - iolen, bp);
if (error != 0)
break;
} else if (nbp != NULL) { /* Complete read with readahead */
tbp = bp;
bp = nbp;
nbp = tbp;
clearbiocache(&nbp->b_bio2);
if (resid <= bp->b_bufsize) { /* No more readaheads */
relpbuf(nbp, &vp->v_mount->mnt_pbuf_count);
nbp = NULL;
} else { /* Setup next readahead */
nerror = ffs_rawread_readahead(
vp, udata + bp->b_bufsize,
offset + bp->b_bufsize,
resid - bp->b_bufsize,
nbp);
if (nerror != 0) {
relpbuf(nbp, &vp->v_mount->mnt_pbuf_count);
nbp = NULL;
}
}
} else if (nerror != 0) {/* Deferred Readahead error */
break;
} else if (resid > 0) { /* More to read, no readahead */
error = ffs_rawread_readahead(vp, udata, offset,
resid, bp);
if (error != 0)
break;
}
}
if (bp != NULL)
relpbuf(bp, &vp->v_mount->mnt_pbuf_count);
if (nbp != NULL) { /* Run down readahead buffer */
biowait(&nbp->b_bio1, "rawrd");
vunmapbuf(nbp);
relpbuf(nbp, &vp->v_mount->mnt_pbuf_count);
}
if (error == 0)
error = nerror;
uio->uio_iov->iov_base = udata;
uio->uio_resid = resid;
uio->uio_offset = offset;
return error;
}
int
physio(struct cdev *dev, struct uio *uio, int ioflag)
{
struct buf *bp;
struct cdevsw *csw;
caddr_t sa;
u_int iolen;
int error, i, mapped;
/* Keep the process UPAGES from being swapped. XXX: why ? */
PHOLD(curproc);
bp = getpbuf(NULL);
sa = bp->b_data;
error = 0;
/* XXX: sanity check */
if(dev->si_iosize_max < PAGE_SIZE) {
printf("WARNING: %s si_iosize_max=%d, using DFLTPHYS.\n",
devtoname(dev), dev->si_iosize_max);
dev->si_iosize_max = DFLTPHYS;
}
/*
* If the driver does not want I/O to be split, that means that we
* need to reject any requests that will not fit into one buffer.
*/
if (dev->si_flags & SI_NOSPLIT &&
(uio->uio_resid > dev->si_iosize_max || uio->uio_resid > MAXPHYS ||
uio->uio_iovcnt > 1)) {
/*
* Tell the user why his I/O was rejected.
*/
if (uio->uio_resid > dev->si_iosize_max)
uprintf("%s: request size=%zd > si_iosize_max=%d; "
"cannot split request\n", devtoname(dev),
uio->uio_resid, dev->si_iosize_max);
if (uio->uio_resid > MAXPHYS)
uprintf("%s: request size=%zd > MAXPHYS=%d; "
"cannot split request\n", devtoname(dev),
uio->uio_resid, MAXPHYS);
if (uio->uio_iovcnt > 1)
uprintf("%s: request vectors=%d > 1; "
"cannot split request\n", devtoname(dev),
uio->uio_iovcnt);
error = EFBIG;
goto doerror;
}
for (i = 0; i < uio->uio_iovcnt; i++) {
while (uio->uio_iov[i].iov_len) {
bp->b_flags = 0;
if (uio->uio_rw == UIO_READ) {
bp->b_iocmd = BIO_READ;
curthread->td_ru.ru_inblock++;
} else {
bp->b_iocmd = BIO_WRITE;
curthread->td_ru.ru_oublock++;
}
bp->b_iodone = bdone;
bp->b_data = uio->uio_iov[i].iov_base;
bp->b_bcount = uio->uio_iov[i].iov_len;
bp->b_offset = uio->uio_offset;
bp->b_iooffset = uio->uio_offset;
bp->b_saveaddr = sa;
/* Don't exceed drivers iosize limit */
if (bp->b_bcount > dev->si_iosize_max)
bp->b_bcount = dev->si_iosize_max;
/*
* Make sure the pbuf can map the request
* XXX: The pbuf has kvasize = MAXPHYS so a request
* XXX: larger than MAXPHYS - PAGE_SIZE must be
* XXX: page aligned or it will be fragmented.
*/
iolen = ((vm_offset_t) bp->b_data) & PAGE_MASK;
if ((bp->b_bcount + iolen) > bp->b_kvasize) {
/*
* This device does not want I/O to be split.
*/
if (dev->si_flags & SI_NOSPLIT) {
uprintf("%s: request ptr %p is not "
"on a page boundary; cannot split "
"request\n", devtoname(dev),
bp->b_data);
error = EFBIG;
goto doerror;
}
bp->b_bcount = bp->b_kvasize;
if (iolen != 0)
bp->b_bcount -= PAGE_SIZE;
}
bp->b_bufsize = bp->b_bcount;
bp->b_blkno = btodb(bp->b_offset);
csw = dev->si_devsw;
if (uio->uio_segflg == UIO_USERSPACE) {
if (dev->si_flags & SI_UNMAPPED)
mapped = 0;
else
mapped = 1;
if (vmapbuf(bp, mapped) < 0) {
error = EFAULT;
goto doerror;
}
}
dev_strategy_csw(dev, csw, bp);
if (uio->uio_rw == UIO_READ)
bwait(bp, PRIBIO, "physrd");
else
bwait(bp, PRIBIO, "physwr");
if (uio->uio_segflg == UIO_USERSPACE)
vunmapbuf(bp);
iolen = bp->b_bcount - bp->b_resid;
if (iolen == 0 && !(bp->b_ioflags & BIO_ERROR))
goto doerror; /* EOF */
uio->uio_iov[i].iov_len -= iolen;
uio->uio_iov[i].iov_base =
(char *)uio->uio_iov[i].iov_base + iolen;
uio->uio_resid -= iolen;
uio->uio_offset += iolen;
if( bp->b_ioflags & BIO_ERROR) {
error = bp->b_error;
goto doerror;
}
}
}
doerror:
relpbuf(bp, NULL);
PRELE(curproc);
return (error);
}
/*
struct vnop_putpages_args {
struct vnode *a_vp;
vm_page_t *a_m;
int a_count;
int a_sync;
int *a_rtvals;
vm_ooffset_t a_offset;
};
*/
static int
fuse_vnop_putpages(struct vop_putpages_args *ap)
{
struct uio uio;
struct iovec iov;
vm_offset_t kva;
struct buf *bp;
int i, error, npages, count;
off_t offset;
int *rtvals;
struct vnode *vp;
struct thread *td;
struct ucred *cred;
vm_page_t *pages;
vm_ooffset_t fsize;
FS_DEBUG2G("heh\n");
vp = ap->a_vp;
KASSERT(vp->v_object, ("objectless vp passed to putpages"));
fsize = vp->v_object->un_pager.vnp.vnp_size;
td = curthread; /* XXX */
cred = curthread->td_ucred; /* XXX */
pages = ap->a_m;
count = ap->a_count;
rtvals = ap->a_rtvals;
npages = btoc(count);
offset = IDX_TO_OFF(pages[0]->pindex);
if (!fsess_opt_mmap(vnode_mount(vp))) {
FS_DEBUG("called on non-cacheable vnode??\n");
}
for (i = 0; i < npages; i++)
rtvals[i] = VM_PAGER_AGAIN;
/*
* When putting pages, do not extend file past EOF.
*/
if (offset + count > fsize) {
count = fsize - offset;
if (count < 0)
count = 0;
}
/*
* We use only the kva address for the buffer, but this is extremely
* convienient and fast.
*/
bp = getpbuf(&fuse_pbuf_freecnt);
kva = (vm_offset_t)bp->b_data;
pmap_qenter(kva, pages, npages);
PCPU_INC(cnt.v_vnodeout);
PCPU_ADD(cnt.v_vnodepgsout, count);
iov.iov_base = (caddr_t)kva;
iov.iov_len = count;
uio.uio_iov = &iov;
uio.uio_iovcnt = 1;
uio.uio_offset = offset;
uio.uio_resid = count;
uio.uio_segflg = UIO_SYSSPACE;
uio.uio_rw = UIO_WRITE;
uio.uio_td = td;
error = fuse_io_dispatch(vp, &uio, IO_DIRECT, cred);
pmap_qremove(kva, npages);
relpbuf(bp, &fuse_pbuf_freecnt);
if (!error) {
int nwritten = round_page(count - uio.uio_resid) / PAGE_SIZE;
for (i = 0; i < nwritten; i++) {
rtvals[i] = VM_PAGER_OK;
VM_OBJECT_WLOCK(pages[i]->object);
vm_page_undirty(pages[i]);
VM_OBJECT_WUNLOCK(pages[i]->object);
}
}
return rtvals[0];
}
/*
struct vnop_getpages_args {
struct vnode *a_vp;
vm_page_t *a_m;
int a_count;
int a_reqpage;
vm_ooffset_t a_offset;
};
*/
static int
fuse_vnop_getpages(struct vop_getpages_args *ap)
{
int i, error, nextoff, size, toff, count, npages;
struct uio uio;
struct iovec iov;
vm_offset_t kva;
struct buf *bp;
struct vnode *vp;
struct thread *td;
struct ucred *cred;
vm_page_t *pages;
FS_DEBUG2G("heh\n");
vp = ap->a_vp;
KASSERT(vp->v_object, ("objectless vp passed to getpages"));
td = curthread; /* XXX */
cred = curthread->td_ucred; /* XXX */
pages = ap->a_m;
count = ap->a_count;
if (!fsess_opt_mmap(vnode_mount(vp))) {
FS_DEBUG("called on non-cacheable vnode??\n");
return (VM_PAGER_ERROR);
}
npages = btoc(count);
/*
* If the requested page is partially valid, just return it and
* allow the pager to zero-out the blanks. Partially valid pages
* can only occur at the file EOF.
*/
VM_OBJECT_WLOCK(vp->v_object);
fuse_vm_page_lock_queues();
if (pages[ap->a_reqpage]->valid != 0) {
for (i = 0; i < npages; ++i) {
if (i != ap->a_reqpage) {
fuse_vm_page_lock(pages[i]);
vm_page_free(pages[i]);
fuse_vm_page_unlock(pages[i]);
}
}
fuse_vm_page_unlock_queues();
VM_OBJECT_WUNLOCK(vp->v_object);
return 0;
}
fuse_vm_page_unlock_queues();
VM_OBJECT_WUNLOCK(vp->v_object);
/*
* We use only the kva address for the buffer, but this is extremely
* convienient and fast.
*/
bp = getpbuf(&fuse_pbuf_freecnt);
kva = (vm_offset_t)bp->b_data;
pmap_qenter(kva, pages, npages);
PCPU_INC(cnt.v_vnodein);
PCPU_ADD(cnt.v_vnodepgsin, npages);
iov.iov_base = (caddr_t)kva;
iov.iov_len = count;
uio.uio_iov = &iov;
uio.uio_iovcnt = 1;
uio.uio_offset = IDX_TO_OFF(pages[0]->pindex);
uio.uio_resid = count;
uio.uio_segflg = UIO_SYSSPACE;
uio.uio_rw = UIO_READ;
uio.uio_td = td;
error = fuse_io_dispatch(vp, &uio, IO_DIRECT, cred);
pmap_qremove(kva, npages);
relpbuf(bp, &fuse_pbuf_freecnt);
if (error && (uio.uio_resid == count)) {
FS_DEBUG("error %d\n", error);
VM_OBJECT_WLOCK(vp->v_object);
fuse_vm_page_lock_queues();
for (i = 0; i < npages; ++i) {
if (i != ap->a_reqpage) {
fuse_vm_page_lock(pages[i]);
vm_page_free(pages[i]);
fuse_vm_page_unlock(pages[i]);
}
}
fuse_vm_page_unlock_queues();
VM_OBJECT_WUNLOCK(vp->v_object);
return VM_PAGER_ERROR;
}
/*
* Calculate the number of bytes read and validate only that number
* of bytes. Note that due to pending writes, size may be 0. This
* does not mean that the remaining data is invalid!
*/
size = count - uio.uio_resid;
VM_OBJECT_WLOCK(vp->v_object);
fuse_vm_page_lock_queues();
for (i = 0, toff = 0; i < npages; i++, toff = nextoff) {
vm_page_t m;
nextoff = toff + PAGE_SIZE;
m = pages[i];
if (nextoff <= size) {
/*
* Read operation filled an entire page
*/
m->valid = VM_PAGE_BITS_ALL;
KASSERT(m->dirty == 0,
("fuse_getpages: page %p is dirty", m));
} else if (size > toff) {
/*
* Read operation filled a partial page.
*/
m->valid = 0;
vm_page_set_valid_range(m, 0, size - toff);
KASSERT(m->dirty == 0,
("fuse_getpages: page %p is dirty", m));
} else {
/*
* Read operation was short. If no error occured
* we may have hit a zero-fill section. We simply
* leave valid set to 0.
*/
;
}
if (i != ap->a_reqpage)
vm_page_readahead_finish(m);
}
fuse_vm_page_unlock_queues();
VM_OBJECT_WUNLOCK(vp->v_object);
return 0;
}
int
physio(dev_t dev, struct uio *uio, int ioflag)
{
int i;
int error;
int spl;
caddr_t sa;
off_t blockno;
u_int iolen;
struct buf *bp;
/* Keep the process UPAGES from being swapped. XXX: why ? */
PHOLD(curproc);
bp = getpbuf(NULL);
sa = bp->b_data;
error = bp->b_error = 0;
/* XXX: sanity check */
if(dev->si_iosize_max < PAGE_SIZE) {
printf("WARNING: %s si_iosize_max=%d, using DFLTPHYS.\n",
devtoname(dev), dev->si_iosize_max);
dev->si_iosize_max = DFLTPHYS;
}
for (i = 0; i < uio->uio_iovcnt; i++) {
while (uio->uio_iov[i].iov_len) {
if (uio->uio_rw == UIO_READ)
bp->b_flags = B_PHYS | B_CALL | B_READ;
else
bp->b_flags = B_PHYS | B_CALL | B_WRITE;
bp->b_dev = dev;
bp->b_iodone = physwakeup;
bp->b_data = uio->uio_iov[i].iov_base;
bp->b_bcount = uio->uio_iov[i].iov_len;
bp->b_offset = uio->uio_offset;
bp->b_saveaddr = sa;
/* Don't exceed drivers iosize limit */
if (bp->b_bcount > dev->si_iosize_max)
bp->b_bcount = dev->si_iosize_max;
/*
* Make sure the pbuf can map the request
* XXX: The pbuf has kvasize = MAXPHYS so a request
* XXX: larger than MAXPHYS - PAGE_SIZE must be
* XXX: page aligned or it will be fragmented.
*/
iolen = ((vm_offset_t) bp->b_data) & PAGE_MASK;
if ((bp->b_bcount + iolen) > bp->b_kvasize) {
bp->b_bcount = bp->b_kvasize;
if (iolen != 0)
bp->b_bcount -= PAGE_SIZE;
}
bp->b_bufsize = bp->b_bcount;
blockno = bp->b_offset >> DEV_BSHIFT;
if ((daddr_t)blockno != blockno) {
error = EINVAL; /* blockno overflow */
goto doerror;
}
bp->b_blkno = blockno;
if (uio->uio_segflg == UIO_USERSPACE) {
if (!useracc(bp->b_data, bp->b_bufsize,
bp->b_flags & B_READ ?
VM_PROT_WRITE : VM_PROT_READ)) {
error = EFAULT;
goto doerror;
}
vmapbuf(bp);
}
BUF_STRATEGY(bp, 0);
spl = splbio();
while ((bp->b_flags & B_DONE) == 0)
tsleep((caddr_t)bp, PRIBIO, "physstr", 0);
splx(spl);
if (uio->uio_segflg == UIO_USERSPACE)
vunmapbuf(bp);
iolen = bp->b_bcount - bp->b_resid;
if (iolen == 0 && !(bp->b_flags & B_ERROR))
goto doerror; /* EOF */
uio->uio_iov[i].iov_len -= iolen;
uio->uio_iov[i].iov_base += iolen;
uio->uio_resid -= iolen;
uio->uio_offset += iolen;
if( bp->b_flags & B_ERROR) {
error = bp->b_error;
goto doerror;
}
}
}
doerror:
relpbuf(bp, NULL);
PRELE(curproc);
return (error);
}
/*
* Vnode op for VM getpages.
* Wish wish .... get rid from multiple IO routines
*
* nwfs_getpages(struct vnode *a_vp, vm_page_t *a_m, int a_count,
* int a_reqpage, vm_ooffset_t a_offset)
*/
int
nwfs_getpages(struct vop_getpages_args *ap)
{
#ifndef NWFS_RWCACHE
return vnode_pager_generic_getpages(ap->a_vp, ap->a_m, ap->a_count,
ap->a_reqpage, ap->a_seqaccess);
#else
int i, error, npages;
size_t nextoff, toff;
size_t count;
size_t size;
struct uio uio;
struct iovec iov;
vm_offset_t kva;
struct buf *bp;
struct vnode *vp;
struct thread *td = curthread; /* XXX */
struct ucred *cred;
struct nwmount *nmp;
struct nwnode *np;
vm_page_t *pages;
KKASSERT(td->td_proc);
cred = td->td_proc->p_ucred;
vp = ap->a_vp;
np = VTONW(vp);
nmp = VFSTONWFS(vp->v_mount);
pages = ap->a_m;
count = (size_t)ap->a_count;
if (vp->v_object == NULL) {
kprintf("nwfs_getpages: called with non-merged cache vnode??\n");
return VM_PAGER_ERROR;
}
bp = getpbuf_kva(&nwfs_pbuf_freecnt);
npages = btoc(count);
kva = (vm_offset_t) bp->b_data;
pmap_qenter(kva, pages, npages);
iov.iov_base = (caddr_t) kva;
iov.iov_len = count;
uio.uio_iov = &iov;
uio.uio_iovcnt = 1;
uio.uio_offset = IDX_TO_OFF(pages[0]->pindex);
uio.uio_resid = count;
uio.uio_segflg = UIO_SYSSPACE;
uio.uio_rw = UIO_READ;
uio.uio_td = td;
error = ncp_read(NWFSTOCONN(nmp), &np->n_fh, &uio,cred);
pmap_qremove(kva, npages);
relpbuf(bp, &nwfs_pbuf_freecnt);
if (error && (uio.uio_resid == count)) {
kprintf("nwfs_getpages: error %d\n",error);
for (i = 0; i < npages; i++) {
if (ap->a_reqpage != i)
vnode_pager_freepage(pages[i]);
}
return VM_PAGER_ERROR;
}
size = count - uio.uio_resid;
for (i = 0, toff = 0; i < npages; i++, toff = nextoff) {
vm_page_t m;
nextoff = toff + PAGE_SIZE;
m = pages[i];
m->flags &= ~PG_ZERO;
/*
* NOTE: pmap dirty bit should have already been cleared.
* We do not clear it here.
*/
if (nextoff <= size) {
m->valid = VM_PAGE_BITS_ALL;
m->dirty = 0;
} else {
int nvalid = ((size + DEV_BSIZE - 1) - toff) &
~(DEV_BSIZE - 1);
vm_page_set_validclean(m, 0, nvalid);
}
if (i != ap->a_reqpage) {
/*
* Whether or not to leave the page activated is up in
* the air, but we should put the page on a page queue
* somewhere (it already is in the object). Result:
* It appears that emperical results show that
* deactivating pages is best.
*/
/*
* Just in case someone was asking for this page we
* now tell them that it is ok to use.
*/
if (!error) {
if (m->flags & PG_REFERENCED)
vm_page_activate(m);
else
vm_page_deactivate(m);
vm_page_wakeup(m);
} else {
vnode_pager_freepage(m);
}
}
}
return 0;
#endif /* NWFS_RWCACHE */
}
/*
* Vnode op for VM getpages.
* Wish wish .... get rid from multiple IO routines
*
* smbfs_getpages(struct vnode *a_vp, vm_page_t *a_m, int a_count,
* int a_reqpage, vm_ooffset_t a_offset)
*/
int
smbfs_getpages(struct vop_getpages_args *ap)
{
#ifdef SMBFS_RWGENERIC
return vop_stdgetpages(ap);
#else
int i, error, npages;
int doclose;
size_t size, toff, nextoff, count;
struct uio uio;
struct iovec iov;
vm_offset_t kva;
struct buf *bp;
struct vnode *vp;
struct thread *td = curthread; /* XXX */
struct ucred *cred;
struct smbmount *smp;
struct smbnode *np;
struct smb_cred scred;
vm_page_t *pages;
KKASSERT(td->td_proc);
vp = ap->a_vp;
cred = td->td_proc->p_ucred;
np = VTOSMB(vp);
smp = VFSTOSMBFS(vp->v_mount);
pages = ap->a_m;
count = (size_t)ap->a_count;
if (vp->v_object == NULL) {
kprintf("smbfs_getpages: called with non-merged cache vnode??\n");
return VM_PAGER_ERROR;
}
smb_makescred(&scred, td, cred);
bp = getpbuf_kva(&smbfs_pbuf_freecnt);
npages = btoc(count);
kva = (vm_offset_t) bp->b_data;
pmap_qenter(kva, pages, npages);
iov.iov_base = (caddr_t) kva;
iov.iov_len = count;
uio.uio_iov = &iov;
uio.uio_iovcnt = 1;
uio.uio_offset = IDX_TO_OFF(pages[0]->pindex);
uio.uio_resid = count;
uio.uio_segflg = UIO_SYSSPACE;
uio.uio_rw = UIO_READ;
uio.uio_td = td;
/*
* This is kinda nasty. Since smbfs is physically closing the
* fid on close(), we have to reopen it if necessary. There are
* other races here too, such as if another process opens the same
* file while we are blocked in read. XXX
*/
error = 0;
doclose = 0;
if (np->n_opencount == 0) {
error = smbfs_smb_open(np, SMB_AM_OPENREAD, &scred);
if (error == 0)
doclose = 1;
}
if (error == 0)
error = smb_read(smp->sm_share, np->n_fid, &uio, &scred);
if (doclose)
smbfs_smb_close(smp->sm_share, np->n_fid, NULL, &scred);
pmap_qremove(kva, npages);
relpbuf(bp, &smbfs_pbuf_freecnt);
if (error && (uio.uio_resid == count)) {
kprintf("smbfs_getpages: error %d\n",error);
for (i = 0; i < npages; i++) {
if (ap->a_reqpage != i)
vnode_pager_freepage(pages[i]);
}
return VM_PAGER_ERROR;
}
size = count - uio.uio_resid;
for (i = 0, toff = 0; i < npages; i++, toff = nextoff) {
vm_page_t m;
nextoff = toff + PAGE_SIZE;
m = pages[i];
m->flags &= ~PG_ZERO;
/*
* NOTE: pmap dirty bit should have already been cleared.
* We do not clear it here.
*/
if (nextoff <= size) {
m->valid = VM_PAGE_BITS_ALL;
m->dirty = 0;
} else {
int nvalid = ((size + DEV_BSIZE - 1) - toff) &
~(DEV_BSIZE - 1);
vm_page_set_validclean(m, 0, nvalid);
}
if (i != ap->a_reqpage) {
/*
* Whether or not to leave the page activated is up in
* the air, but we should put the page on a page queue
* somewhere (it already is in the object). Result:
* It appears that emperical results show that
* deactivating pages is best.
*/
/*
* Just in case someone was asking for this page we
* now tell them that it is ok to use.
*/
if (!error) {
if (m->flags & PG_REFERENCED)
vm_page_activate(m);
else
vm_page_deactivate(m);
vm_page_wakeup(m);
} else {
vnode_pager_freepage(m);
}
}
}
return 0;
#endif /* SMBFS_RWGENERIC */
}
/*
* Vnode op for VM putpages.
* possible bug: all IO done in sync mode
* Note that vop_close always invalidate pages before close, so it's
* not necessary to open vnode.
*
* smbfs_putpages(struct vnode *a_vp, vm_page_t *a_m, int a_count, int a_sync,
* int *a_rtvals, vm_ooffset_t a_offset)
*/
int
smbfs_putpages(struct vop_putpages_args *ap)
{
int error;
struct vnode *vp = ap->a_vp;
struct thread *td = curthread; /* XXX */
struct ucred *cred;
#ifdef SMBFS_RWGENERIC
KKASSERT(td->td_proc);
cred = td->td_proc->p_ucred;
VOP_OPEN(vp, FWRITE, cred, NULL);
error = vop_stdputpages(ap);
VOP_CLOSE(vp, FWRITE, cred);
return error;
#else
struct uio uio;
struct iovec iov;
vm_offset_t kva;
struct buf *bp;
int i, npages, count;
int doclose;
int *rtvals;
struct smbmount *smp;
struct smbnode *np;
struct smb_cred scred;
vm_page_t *pages;
KKASSERT(td->td_proc);
cred = td->td_proc->p_ucred;
/* VOP_OPEN(vp, FWRITE, cred, NULL);*/
np = VTOSMB(vp);
smp = VFSTOSMBFS(vp->v_mount);
pages = ap->a_m;
count = ap->a_count;
rtvals = ap->a_rtvals;
npages = btoc(count);
for (i = 0; i < npages; i++) {
rtvals[i] = VM_PAGER_AGAIN;
}
bp = getpbuf_kva(&smbfs_pbuf_freecnt);
kva = (vm_offset_t) bp->b_data;
pmap_qenter(kva, pages, npages);
iov.iov_base = (caddr_t) kva;
iov.iov_len = count;
uio.uio_iov = &iov;
uio.uio_iovcnt = 1;
uio.uio_offset = IDX_TO_OFF(pages[0]->pindex);
uio.uio_resid = count;
uio.uio_segflg = UIO_SYSSPACE;
uio.uio_rw = UIO_WRITE;
uio.uio_td = td;
SMBVDEBUG("ofs=%d,resid=%d\n",(int)uio.uio_offset, uio.uio_resid);
smb_makescred(&scred, td, cred);
/*
* This is kinda nasty. Since smbfs is physically closing the
* fid on close(), we have to reopen it if necessary. There are
* other races here too, such as if another process opens the same
* file while we are blocked in read, or the file is open read-only
* XXX
*/
error = 0;
doclose = 0;
if (np->n_opencount == 0) {
error = smbfs_smb_open(np, SMB_AM_OPENRW, &scred);
if (error == 0)
doclose = 1;
}
if (error == 0)
error = smb_write(smp->sm_share, np->n_fid, &uio, &scred);
if (doclose)
smbfs_smb_close(smp->sm_share, np->n_fid, NULL, &scred);
/* VOP_CLOSE(vp, FWRITE, cred);*/
SMBVDEBUG("paged write done: %d\n", error);
pmap_qremove(kva, npages);
relpbuf(bp, &smbfs_pbuf_freecnt);
if (!error) {
int nwritten = round_page(count - uio.uio_resid) / PAGE_SIZE;
for (i = 0; i < nwritten; i++) {
rtvals[i] = VM_PAGER_OK;
vm_page_undirty(pages[i]);
}
}
return rtvals[0];
#endif /* SMBFS_RWGENERIC */
}
/*
* spec_getpages() - get pages associated with device vnode.
*
* Note that spec_read and spec_write do not use the buffer cache, so we
* must fully implement getpages here.
*/
static int
devfs_spec_getpages(struct vop_getpages_args *ap)
{
vm_offset_t kva;
int error;
int i, pcount, size;
struct buf *bp;
vm_page_t m;
vm_ooffset_t offset;
int toff, nextoff, nread;
struct vnode *vp = ap->a_vp;
int blksiz;
int gotreqpage;
error = 0;
pcount = round_page(ap->a_count) / PAGE_SIZE;
/*
* Calculate the offset of the transfer and do sanity check.
*/
offset = IDX_TO_OFF(ap->a_m[0]->pindex) + ap->a_offset;
/*
* Round up physical size for real devices. We cannot round using
* v_mount's block size data because v_mount has nothing to do with
* the device. i.e. it's usually '/dev'. We need the physical block
* size for the device itself.
*
* We can't use v_rdev->si_mountpoint because it only exists when the
* block device is mounted. However, we can use v_rdev.
*/
if (vn_isdisk(vp, NULL))
blksiz = vp->v_rdev->si_bsize_phys;
else
blksiz = DEV_BSIZE;
size = (ap->a_count + blksiz - 1) & ~(blksiz - 1);
bp = getpbuf_kva(NULL);
kva = (vm_offset_t)bp->b_data;
/*
* Map the pages to be read into the kva.
*/
pmap_qenter(kva, ap->a_m, pcount);
/* Build a minimal buffer header. */
bp->b_cmd = BUF_CMD_READ;
bp->b_bcount = size;
bp->b_resid = 0;
bsetrunningbufspace(bp, size);
bp->b_bio1.bio_offset = offset;
bp->b_bio1.bio_done = devfs_spec_getpages_iodone;
mycpu->gd_cnt.v_vnodein++;
mycpu->gd_cnt.v_vnodepgsin += pcount;
/* Do the input. */
vn_strategy(ap->a_vp, &bp->b_bio1);
crit_enter();
/* We definitely need to be at splbio here. */
while (bp->b_cmd != BUF_CMD_DONE)
tsleep(bp, 0, "spread", 0);
crit_exit();
if (bp->b_flags & B_ERROR) {
if (bp->b_error)
error = bp->b_error;
else
error = EIO;
}
/*
* If EOF is encountered we must zero-extend the result in order
* to ensure that the page does not contain garabge. When no
* error occurs, an early EOF is indicated if b_bcount got truncated.
* b_resid is relative to b_bcount and should be 0, but some devices
* might indicate an EOF with b_resid instead of truncating b_bcount.
*/
nread = bp->b_bcount - bp->b_resid;
if (nread < ap->a_count)
bzero((caddr_t)kva + nread, ap->a_count - nread);
pmap_qremove(kva, pcount);
gotreqpage = 0;
for (i = 0, toff = 0; i < pcount; i++, toff = nextoff) {
nextoff = toff + PAGE_SIZE;
m = ap->a_m[i];
m->flags &= ~PG_ZERO;
/*
* NOTE: vm_page_undirty/clear_dirty etc do not clear the
* pmap modified bit. pmap modified bit should have
* already been cleared.
*/
if (nextoff <= nread) {
m->valid = VM_PAGE_BITS_ALL;
vm_page_undirty(m);
} else if (toff < nread) {
/*
* Since this is a VM request, we have to supply the
* unaligned offset to allow vm_page_set_valid()
* to zero sub-DEV_BSIZE'd portions of the page.
*/
vm_page_set_valid(m, 0, nread - toff);
vm_page_clear_dirty_end_nonincl(m, 0, nread - toff);
} else {
m->valid = 0;
vm_page_undirty(m);
}
if (i != ap->a_reqpage) {
/*
* Just in case someone was asking for this page we
* now tell them that it is ok to use.
*/
if (!error || (m->valid == VM_PAGE_BITS_ALL)) {
if (m->valid) {
if (m->flags & PG_REFERENCED) {
vm_page_activate(m);
} else {
vm_page_deactivate(m);
}
vm_page_wakeup(m);
} else {
vm_page_free(m);
}
} else {
vm_page_free(m);
}
} else if (m->valid) {
gotreqpage = 1;
/*
* Since this is a VM request, we need to make the
* entire page presentable by zeroing invalid sections.
*/
if (m->valid != VM_PAGE_BITS_ALL)
vm_page_zero_invalid(m, FALSE);
}
}
if (!gotreqpage) {
m = ap->a_m[ap->a_reqpage];
devfs_debug(DEVFS_DEBUG_WARNING,
"spec_getpages:(%s) I/O read failure: (error=%d) bp %p vp %p\n",
devtoname(vp->v_rdev), error, bp, bp->b_vp);
devfs_debug(DEVFS_DEBUG_WARNING,
" size: %d, resid: %d, a_count: %d, valid: 0x%x\n",
size, bp->b_resid, ap->a_count, m->valid);
devfs_debug(DEVFS_DEBUG_WARNING,
" nread: %d, reqpage: %d, pindex: %lu, pcount: %d\n",
nread, ap->a_reqpage, (u_long)m->pindex, pcount);
/*
* Free the buffer header back to the swap buffer pool.
*/
relpbuf(bp, NULL);
return VM_PAGER_ERROR;
}
/*
* Free the buffer header back to the swap buffer pool.
*/
relpbuf(bp, NULL);
if (DEVFS_NODE(ap->a_vp))
nanotime(&DEVFS_NODE(ap->a_vp)->mtime);
return VM_PAGER_OK;
}
int
mbrinit(cdev_t dev, struct disk_info *info, struct diskslices **sspp)
{
struct buf *bp;
u_char *cp;
int dospart;
struct dos_partition *dp;
struct dos_partition *dp0;
struct dos_partition dpcopy[NDOSPART];
int error;
int max_ncyls;
int max_nsectors;
int max_ntracks;
u_int64_t mbr_offset;
char partname[2];
u_long secpercyl;
char *sname = "tempname";
struct diskslice *sp;
struct diskslices *ssp;
cdev_t wdev;
mbr_offset = DOSBBSECTOR;
reread_mbr:
/*
* Don't bother if the block size is weird or the
* media size is 0 (probably means no media present).
*/
if (info->d_media_blksize & DEV_BMASK)
return (EIO);
if (info->d_media_size == 0)
return (EIO);
/*
* Read master boot record.
*/
wdev = dev;
bp = getpbuf_mem(NULL);
KKASSERT((int)info->d_media_blksize <= bp->b_bufsize);
bp->b_bio1.bio_offset = (off_t)mbr_offset * info->d_media_blksize;
bp->b_bio1.bio_done = biodone_sync;
bp->b_bio1.bio_flags |= BIO_SYNC;
bp->b_bcount = info->d_media_blksize;
bp->b_cmd = BUF_CMD_READ;
bp->b_flags |= B_FAILONDIS;
dev_dstrategy(wdev, &bp->b_bio1);
if (biowait(&bp->b_bio1, "mbrrd") != 0) {
if ((info->d_dsflags & DSO_MBRQUIET) == 0) {
diskerr(&bp->b_bio1, wdev,
"reading primary partition table: error",
LOG_PRINTF, 0);
kprintf("\n");
}
error = EIO;
goto done;
}
/* Weakly verify it. */
cp = bp->b_data;
sname = dsname(dev, 0, 0, 0, NULL);
if (cp[0x1FE] != 0x55 || cp[0x1FF] != 0xAA) {
if (bootverbose)
kprintf("%s: invalid primary partition table: no magic\n",
sname);
error = EINVAL;
goto done;
}
/* Make a copy of the partition table to avoid alignment problems. */
memcpy(&dpcopy[0], cp + DOSPARTOFF, sizeof(dpcopy));
dp0 = &dpcopy[0];
/*
* Check for "Ontrack Diskmanager" or GPT. If a GPT is found in
* the first dos partition, ignore the rest of the MBR and go
* to GPT processing.
*/
for (dospart = 0, dp = dp0; dospart < NDOSPART; dospart++, dp++) {
if (dospart == 0 && dp->dp_typ == DOSPTYP_PMBR) {
if (bootverbose)
kprintf(
"%s: Found GPT in slice #%d\n", sname, dospart + 1);
error = gptinit(dev, info, sspp);
goto done;
}
if (dp->dp_typ == DOSPTYP_ONTRACK) {
if (bootverbose)
kprintf(
"%s: Found \"Ontrack Disk Manager\" on this disk.\n", sname);
bp->b_flags |= B_INVAL | B_AGE;
brelse(bp);
mbr_offset = 63;
goto reread_mbr;
}
}
if (bcmp(dp0, historical_bogus_partition_table,
sizeof historical_bogus_partition_table) == 0 ||
bcmp(dp0, historical_bogus_partition_table_fixed,
sizeof historical_bogus_partition_table_fixed) == 0) {
#if 0
TRACE(("%s: invalid primary partition table: historical\n",
sname));
#endif /* 0 */
if (bootverbose)
kprintf(
"%s: invalid primary partition table: Dangerously Dedicated (ignored)\n",
sname);
error = EINVAL;
goto done;
}
/* Guess the geometry. */
/*
* TODO:
* Perhaps skip entries with 0 size.
* Perhaps only look at entries of type DOSPTYP_386BSD or
* DOSPTYP_DFLYBSD
*/
max_ncyls = 0;
max_nsectors = 0;
max_ntracks = 0;
for (dospart = 0, dp = dp0; dospart < NDOSPART; dospart++, dp++) {
int ncyls;
int nsectors;
int ntracks;
ncyls = DPCYL(dp->dp_ecyl, dp->dp_esect) + 1;
if (max_ncyls < ncyls)
max_ncyls = ncyls;
nsectors = DPSECT(dp->dp_esect);
if (max_nsectors < nsectors)
max_nsectors = nsectors;
ntracks = dp->dp_ehd + 1;
if (max_ntracks < ntracks)
max_ntracks = ntracks;
}
/*
* Check that we have guessed the geometry right by checking the
* partition entries.
*/
/*
* TODO:
* As above.
* Check for overlaps.
* Check against d_secperunit if the latter is reliable.
*/
error = 0;
for (dospart = 0, dp = dp0; dospart < NDOSPART; dospart++, dp++) {
if (dp->dp_scyl == 0 && dp->dp_shd == 0 && dp->dp_ssect == 0
&& dp->dp_start == 0 && dp->dp_size == 0)
continue;
//sname = dsname(dev, dkunit(dev), BASE_SLICE + dospart,
// WHOLE_SLICE_PART, partname);
/*
* Temporarily ignore errors from this check. We could
* simplify things by accepting the table eariler if we
* always ignore errors here. Perhaps we should always
* accept the table if the magic is right but not let
* bad entries affect the geometry.
*/
check_part(sname, dp, mbr_offset, max_nsectors, max_ntracks,
mbr_offset);
}
if (error != 0)
goto done;
/*
* Accept the DOS partition table.
*
* Adjust the disk information structure with updated CHS
* conversion parameters, but only use values extracted from
* the primary partition table.
*
* NOTE! Regardless of our having to deal with this old cruft,
* we do not screw around with the info->d_media* parameters.
*/
secpercyl = (u_long)max_nsectors * max_ntracks;
if (secpercyl != 0 && mbr_offset == DOSBBSECTOR) {
info->d_secpertrack = max_nsectors;
info->d_nheads = max_ntracks;
info->d_secpercyl = secpercyl;
info->d_ncylinders = info->d_media_blocks / secpercyl;
}
/*
* We are passed a pointer to a suitably initialized minimal
* slices "struct" with no dangling pointers in it. Replace it
* by a maximal one. This usually oversizes the "struct", but
* enlarging it while searching for logical drives would be
* inconvenient.
*/
kfree(*sspp, M_DEVBUF);
ssp = dsmakeslicestruct(MAX_SLICES, info);
*sspp = ssp;
/* Initialize normal slices. */
sp = &ssp->dss_slices[BASE_SLICE];
for (dospart = 0, dp = dp0; dospart < NDOSPART; dospart++, dp++, sp++) {
sname = dsname(dev, dkunit(dev), BASE_SLICE + dospart,
WHOLE_SLICE_PART, partname);
(void)mbr_setslice(sname, info, sp, dp, mbr_offset);
}
ssp->dss_nslices = BASE_SLICE + NDOSPART;
/* Handle extended partitions. */
sp -= NDOSPART;
for (dospart = 0; dospart < NDOSPART; dospart++, sp++) {
if (sp->ds_type == DOSPTYP_EXTENDED ||
sp->ds_type == DOSPTYP_EXTENDEDX) {
mbr_extended(wdev, info, ssp,
sp->ds_offset, sp->ds_size, sp->ds_offset,
max_nsectors, max_ntracks, mbr_offset, 1);
}
}
/*
* mbr_extended() abuses ssp->dss_nslices for the number of slices
* that would be found if there were no limit on the number of slices
* in *ssp. Cut it back now.
*/
if (ssp->dss_nslices > MAX_SLICES)
ssp->dss_nslices = MAX_SLICES;
done:
bp->b_flags |= B_INVAL | B_AGE;
relpbuf(bp, NULL);
if (error == EINVAL)
error = 0;
return (error);
}
static
void
mbr_extended(cdev_t dev, struct disk_info *info, struct diskslices *ssp,
u_int64_t ext_offset, u_int64_t ext_size, u_int64_t base_ext_offset,
int nsectors, int ntracks, u_int64_t mbr_offset, int level)
{
struct buf *bp;
u_char *cp;
int dospart;
struct dos_partition *dp;
struct dos_partition dpcopy[NDOSPART];
u_int64_t ext_offsets[NDOSPART];
u_int64_t ext_sizes[NDOSPART];
char partname[2];
int slice;
char *sname;
struct diskslice *sp;
if (level >= 16) {
kprintf(
"%s: excessive recursion in search for slices; aborting search\n",
devtoname(dev));
return;
}
/* Read extended boot record. */
bp = getpbuf_mem(NULL);
KKASSERT((int)info->d_media_blksize <= bp->b_bufsize);
bp->b_bio1.bio_offset = (off_t)ext_offset * info->d_media_blksize;
bp->b_bio1.bio_done = biodone_sync;
bp->b_bio1.bio_flags |= BIO_SYNC;
bp->b_bcount = info->d_media_blksize;
bp->b_cmd = BUF_CMD_READ;
bp->b_flags |= B_FAILONDIS;
dev_dstrategy(dev, &bp->b_bio1);
if (biowait(&bp->b_bio1, "mbrrd") != 0) {
diskerr(&bp->b_bio1, dev,
"reading extended partition table: error",
LOG_PRINTF, 0);
kprintf("\n");
goto done;
}
/* Weakly verify it. */
cp = bp->b_data;
if (cp[0x1FE] != 0x55 || cp[0x1FF] != 0xAA) {
sname = dsname(dev, dkunit(dev), WHOLE_DISK_SLICE, WHOLE_SLICE_PART,
partname);
if (bootverbose)
kprintf("%s: invalid extended partition table: no magic\n",
sname);
goto done;
}
/* Make a copy of the partition table to avoid alignment problems. */
memcpy(&dpcopy[0], cp + DOSPARTOFF, sizeof(dpcopy));
slice = ssp->dss_nslices;
for (dospart = 0, dp = &dpcopy[0]; dospart < NDOSPART;
dospart++, dp++) {
ext_sizes[dospart] = 0;
if (dp->dp_scyl == 0 && dp->dp_shd == 0 && dp->dp_ssect == 0
&& dp->dp_start == 0 && dp->dp_size == 0)
continue;
if (dp->dp_typ == DOSPTYP_EXTENDED ||
dp->dp_typ == DOSPTYP_EXTENDEDX) {
static char buf[32];
sname = dsname(dev, dkunit(dev), WHOLE_DISK_SLICE,
WHOLE_SLICE_PART, partname);
ksnprintf(buf, sizeof(buf), "%s", sname);
if (strlen(buf) < sizeof buf - 11)
strcat(buf, "<extended>");
check_part(buf, dp, base_ext_offset, nsectors,
ntracks, mbr_offset);
ext_offsets[dospart] = base_ext_offset + dp->dp_start;
ext_sizes[dospart] = dp->dp_size;
} else {
sname = dsname(dev, dkunit(dev), slice, WHOLE_SLICE_PART,
partname);
check_part(sname, dp, ext_offset, nsectors, ntracks,
mbr_offset);
if (slice >= MAX_SLICES) {
kprintf("%s: too many slices\n", sname);
slice++;
continue;
}
sp = &ssp->dss_slices[slice];
if (mbr_setslice(sname, info, sp, dp, ext_offset) != 0)
continue;
slice++;
}
}
ssp->dss_nslices = slice;
/* If we found any more slices, recursively find all the subslices. */
for (dospart = 0; dospart < NDOSPART; dospart++) {
if (ext_sizes[dospart] != 0) {
mbr_extended(dev, info, ssp, ext_offsets[dospart],
ext_sizes[dospart], base_ext_offset,
nsectors, ntracks, mbr_offset, ++level);
}
}
done:
bp->b_flags |= B_INVAL | B_AGE;
relpbuf(bp, NULL);
}
/*
* small block filesystem vnode pager input
*/
static int
vnode_pager_input_smlfs(vm_object_t object, vm_page_t m)
{
struct vnode *vp;
struct bufobj *bo;
struct buf *bp;
struct sf_buf *sf;
daddr_t fileaddr;
vm_offset_t bsize;
vm_page_bits_t bits;
int error, i;
error = 0;
vp = object->handle;
if (vp->v_iflag & VI_DOOMED)
return VM_PAGER_BAD;
bsize = vp->v_mount->mnt_stat.f_iosize;
VOP_BMAP(vp, 0, &bo, 0, NULL, NULL);
sf = sf_buf_alloc(m, 0);
for (i = 0; i < PAGE_SIZE / bsize; i++) {
vm_ooffset_t address;
bits = vm_page_bits(i * bsize, bsize);
if (m->valid & bits)
continue;
address = IDX_TO_OFF(m->pindex) + i * bsize;
if (address >= object->un_pager.vnp.vnp_size) {
fileaddr = -1;
} else {
error = vnode_pager_addr(vp, address, &fileaddr, NULL);
if (error)
break;
}
if (fileaddr != -1) {
bp = getpbuf(&vnode_pbuf_freecnt);
/* build a minimal buffer header */
bp->b_iocmd = BIO_READ;
bp->b_iodone = bdone;
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_data = (caddr_t)sf_buf_kva(sf) + i * bsize;
bp->b_blkno = fileaddr;
pbgetbo(bo, bp);
bp->b_vp = vp;
bp->b_bcount = bsize;
bp->b_bufsize = bsize;
bp->b_runningbufspace = bp->b_bufsize;
atomic_add_long(&runningbufspace, bp->b_runningbufspace);
/* do the input */
bp->b_iooffset = dbtob(bp->b_blkno);
bstrategy(bp);
bwait(bp, PVM, "vnsrd");
if ((bp->b_ioflags & BIO_ERROR) != 0)
error = EIO;
/*
* free the buffer header back to the swap buffer pool
*/
bp->b_vp = NULL;
pbrelbo(bp);
relpbuf(bp, &vnode_pbuf_freecnt);
if (error)
break;
} else
bzero((caddr_t)sf_buf_kva(sf) + i * bsize, bsize);
KASSERT((m->dirty & bits) == 0,
("vnode_pager_input_smlfs: page %p is dirty", m));
VM_OBJECT_WLOCK(object);
m->valid |= bits;
VM_OBJECT_WUNLOCK(object);
}
sf_buf_free(sf);
if (error) {
return VM_PAGER_ERROR;
}
return VM_PAGER_OK;
}
/*
* Vnode op for VM putpages.
* possible bug: all IO done in sync mode
* Note that vop_close always invalidate pages before close, so it's
* not necessary to open vnode.
*
* nwfs_putpages(struct vnode *a_vp, vm_page_t *a_m, int a_count,
* int a_sync, int *a_rtvals, vm_ooffset_t a_offset)
*/
int
nwfs_putpages(struct vop_putpages_args *ap)
{
int error;
struct thread *td = curthread; /* XXX */
struct vnode *vp = ap->a_vp;
struct ucred *cred;
#ifndef NWFS_RWCACHE
KKASSERT(td->td_proc);
cred = td->td_proc->p_ucred; /* XXX */
VOP_OPEN(vp, FWRITE, cred, NULL);
error = vnode_pager_generic_putpages(ap->a_vp, ap->a_m, ap->a_count,
ap->a_sync, ap->a_rtvals);
VOP_CLOSE(vp, FWRITE, cred);
return error;
#else
struct uio uio;
struct iovec iov;
vm_offset_t kva;
struct buf *bp;
int i, npages, count;
int *rtvals;
struct nwmount *nmp;
struct nwnode *np;
vm_page_t *pages;
KKASSERT(td->td_proc);
cred = td->td_proc->p_ucred; /* XXX */
/* VOP_OPEN(vp, FWRITE, cred, NULL);*/
np = VTONW(vp);
nmp = VFSTONWFS(vp->v_mount);
pages = ap->a_m;
count = ap->a_count;
rtvals = ap->a_rtvals;
npages = btoc(count);
for (i = 0; i < npages; i++) {
rtvals[i] = VM_PAGER_AGAIN;
}
bp = getpbuf_kva(&nwfs_pbuf_freecnt);
kva = (vm_offset_t) bp->b_data;
pmap_qenter(kva, pages, npages);
iov.iov_base = (caddr_t) kva;
iov.iov_len = count;
uio.uio_iov = &iov;
uio.uio_iovcnt = 1;
uio.uio_offset = IDX_TO_OFF(pages[0]->pindex);
uio.uio_resid = count;
uio.uio_segflg = UIO_SYSSPACE;
uio.uio_rw = UIO_WRITE;
uio.uio_td = td;
NCPVNDEBUG("ofs=%d,resid=%d\n",(int)uio.uio_offset, uio.uio_resid);
error = ncp_write(NWFSTOCONN(nmp), &np->n_fh, &uio, cred);
/* VOP_CLOSE(vp, FWRITE, cred);*/
NCPVNDEBUG("paged write done: %d\n", error);
pmap_qremove(kva, npages);
relpbuf(bp, &nwfs_pbuf_freecnt);
if (!error) {
int nwritten = round_page(count - uio.uio_resid) / PAGE_SIZE;
for (i = 0; i < nwritten; i++) {
rtvals[i] = VM_PAGER_OK;
vm_page_undirty(pages[i]);
}
}
return rtvals[0];
#endif /* NWFS_RWCACHE */
}
/*
* 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);
}
}
/*
* 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);
}
