fspacectl(2): Changes on rmsr.r_offset's minimum value returned

rmsr.r_offset now is set to rqsr.r_offset plus the number of bytes
zeroed before hitting the end-of-file. After this change rmsr.r_offset
no longer contains the EOF when the requested operation range is
completely beyond the end-of-file. Instead in such case rmsr.r_offset is
equal to rqsr.r_offset.  Callers can obtain the number of bytes zeroed
by subtracting rqsr.r_offset from rmsr.r_offset.

Sponsored by:	The FreeBSD Foundation
Reviewed by:	kib
Differential Revision:	https://reviews.freebsd.org/D31677
This commit is contained in:
Ka Ho Ng 2021-08-26 00:02:27 +08:00
parent dc6ab77d66
commit 9e202d036d
5 changed files with 11 additions and 20 deletions

View File

@ -27,7 +27,7 @@
.\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
.\" SUCH DAMAGE.
.\"
.Dd August 18, 2021
.Dd August 25, 2021
.Dt FSPACECTL 2
.Os
.Sh NAME
@ -73,10 +73,9 @@ operation range,
.Fa "rmsr->r_len"
is updated to be the value 0, and
.Fa "rmsr->r_offset"
is updated to be the smallest of
.Fa "rqsr->r_offset" +
.Fa "rqsr->r_len" ;
and the end-of-file offset.
is updated to be
.Fa "rqsr->r_offset"
plus the number of bytes zeroed before the end-of-file.
The file descriptor's file offset is not used or modified by the system call.
Both
.Fa rqsr

View File

@ -27,7 +27,7 @@
.\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
.\" SUCH DAMAGE.
.\"
.Dd August 11, 2021
.Dd August 25, 2021
.Dt VOP_DEALLOCATE 9
.Os
.Sh NAME
@ -74,14 +74,11 @@ and
are updated to reflect the portion of the range that
still needs to be zeroed/deallocated on return.
Partial result is considered a successful operation.
For a successful completion without an unprocessed portion of the range,
For a non-partial successful completion,
.Fa *len
is updated to be the value 0, and
.Fa *offset
is updated to be the smallest of
.Fa *offset +
.Fa *len
passed to the call and the end-of-file offset.
is incremented by the number of bytes zeroed before the end-of-file.
.Sh LOCKS
The vnode should be locked on entry and will still be locked on exit.
.Sh RETURN VALUES

View File

@ -27,7 +27,7 @@
.\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
.\" SUCH DAMAGE.
.\"
.Dd August 11, 2021
.Dd August 25, 2021
.Dt VN_DEALLOCATE 9
.Os
.Sh NAME
@ -99,10 +99,7 @@ For a successful completion,
.Fa *length
is updated to be the value 0, and
.Fa *offset
is updated to be the smallest of
.Fa *offset +
.Fa *length
passed to the call and the end-of-file offset.
is incremented by the number of bytes zeroed before the end-of-file.
.Sh RETURN VALUES
Upon successful completion, the value 0 is returned; otherwise the
appropriate error is returned.

View File

@ -1917,7 +1917,6 @@ shm_deallocate(struct shmfd *shmfd, off_t *offset, off_t *length, int flags)
/* Handle the case when offset is on or beyond shm size. */
if ((off_t)len <= 0) {
*offset = shmfd->shm_size;
*length = 0;
return (0);
}

View File

@ -1156,6 +1156,7 @@ vop_stddeallocate(struct vop_deallocate_args *ap)
/*
* No more data region to be filled
*/
offset += len;
len = 0;
error = 0;
break;
@ -1185,10 +1186,8 @@ vop_stddeallocate(struct vop_deallocate_args *ap)
break;
}
/* Handle the case when offset is beyond EOF */
if (len < 0) {
offset += len;
if (len < 0)
len = 0;
}
out:
*ap->a_offset = offset;
*ap->a_len = len;