sendfile: Use the pager size to determine the file extent when possible

Previously sendfile would issue a VOP_GETATTR and use the returned size,
i.e., the file size.  When paging in file data, sendfile_swapin() will
use the pager to determine whether it needs to zero-fill, most often
because of a hole in a sparse file.  An attempt to page in beyond the
end of a file is treated this way, and occurs when the requested page is
past the end of the pager.  In other words, both the file size and pager
size were used interchangeably.

With ZFS, updates to the pager and file sizes are not synchronized by
the exclusive vnode lock, at least partially due to its use of
MNTK_SHARED_WRITES.  In particular, the pager size is updated after the
file size, so in the presence of a writer concurrently extending the
file, sendfile could incorrectly instantiate "holes" in the page cache
pages backing the file, which manifests as data corruption when reading
the file back from the page cache.  The on-disk copy is unaffected.

Fix this by consistently using the pager size when available.

Reported by:	dumbbell
Reviewed by:	chs, kib
Tested by:	dumbbell, pho
MFC after:	1 week
Sponsored by:	The FreeBSD Foundation
Differential Revision:	https://reviews.freebsd.org/D28811
This commit is contained in:
Mark Johnston 2021-02-25 10:04:44 -05:00
parent 13ad237a19
commit faa998f6ff

View File

@ -588,28 +588,40 @@ sendfile_getobj(struct thread *td, struct file *fp, vm_object_t *obj_res,
goto out;
}
*bsize = vp->v_mount->mnt_stat.f_iosize;
error = VOP_GETATTR(vp, &va, td->td_ucred);
if (error != 0)
goto out;
*obj_size = va.va_size;
obj = vp->v_object;
if (obj == NULL) {
error = EINVAL;
goto out;
}
/*
* Use the pager size when available to simplify synchronization
* with filesystems, which otherwise must atomically update both
* the vnode pager size and file size.
*/
if (obj->type == OBJT_VNODE) {
VM_OBJECT_RLOCK(obj);
*obj_size = obj->un_pager.vnp.vnp_size;
} else {
error = VOP_GETATTR(vp, &va, td->td_ucred);
if (error != 0)
goto out;
*obj_size = va.va_size;
VM_OBJECT_RLOCK(obj);
}
} else if (fp->f_type == DTYPE_SHM) {
error = 0;
shmfd = fp->f_data;
obj = shmfd->shm_object;
VM_OBJECT_RLOCK(obj);
*obj_size = shmfd->shm_size;
} else {
error = EINVAL;
goto out;
}
VM_OBJECT_WLOCK(obj);
if ((obj->flags & OBJ_DEAD) != 0) {
VM_OBJECT_WUNLOCK(obj);
VM_OBJECT_RUNLOCK(obj);
error = EBADF;
goto out;
}
@ -620,7 +632,7 @@ sendfile_getobj(struct thread *td, struct file *fp, vm_object_t *obj_res,
* immediately destroy it.
*/
vm_object_reference_locked(obj);
VM_OBJECT_WUNLOCK(obj);
VM_OBJECT_RUNLOCK(obj);
*obj_res = obj;
*vp_res = vp;
*shmfd_res = shmfd;
@ -679,7 +691,7 @@ vn_sendfile(struct file *fp, int sockfd, struct uio *hdr_uio,
struct shmfd *shmfd;
struct sendfile_sync *sfs;
struct vattr va;
off_t off, sbytes, rem, obj_size;
off_t off, sbytes, rem, obj_size, nobj_size;
int bsize, error, ext_pgs_idx, hdrlen, max_pgs, softerr;
#ifdef KERN_TLS
int tls_enq_cnt;
@ -852,15 +864,30 @@ vn_sendfile(struct file *fp, int sockfd, struct uio *hdr_uio,
error = vn_lock(vp, LK_SHARED);
if (error != 0)
goto done;
error = VOP_GETATTR(vp, &va, td->td_ucred);
if (error != 0 || off >= va.va_size) {
/*
* Check to see if the file size has changed.
*/
if (obj->type == OBJT_VNODE) {
VM_OBJECT_RLOCK(obj);
nobj_size = obj->un_pager.vnp.vnp_size;
VM_OBJECT_RUNLOCK(obj);
} else {
error = VOP_GETATTR(vp, &va, td->td_ucred);
if (error != 0) {
VOP_UNLOCK(vp);
goto done;
}
nobj_size = va.va_size;
}
if (off >= nobj_size) {
VOP_UNLOCK(vp);
goto done;
}
if (va.va_size != obj_size) {
obj_size = va.va_size;
rem = nbytes ?
omin(nbytes + offset, obj_size) : obj_size;
if (nobj_size != obj_size) {
obj_size = nobj_size;
rem = nbytes ? omin(nbytes + offset, obj_size) :
obj_size;
rem -= off;
}
}