Fix for access beyond end of device error

This commit fixes a sign extension bug affecting l2arc devices.  Extremely
large offsets may be passed down to the low level block device driver on
reads, generating errors similar to

    attempt to access beyond end of device
    sdbi1: rw=14, want=36028797014862705, limit=125026959

The unwanted sign extension occurrs because the function arc_read_nolock()
stores the offset as a daddr_t, a 32-bit signed int type in the Linux kernel.
This offset is then passed to zio_read_phys() as a uint64_t argument, causing
sign extension for values of 0x80000000 or greater.  To avoid this, we store
the offset in a uint64_t.

This change also changes a few daddr_t struct members to uint64_t in the libspl
headers to avoid similar bugs cropping up in the future.  We also add an ASSERT
to __vdev_disk_physio() to check for invalid offsets.

Closes #66
Signed-off-by: Brian Behlendorf <[email protected]>

5 files changed, 13 insertions, 11 deletions

diff --git a/lib/libspl/include/sys/dkio.h b/lib/libspl/include/sys/dkio.h
index 32f786565..6b430b81f 100644
--- a/lib/libspl/include/sys/dkio.h
+++ b/lib/libspl/include/sys/dkio.h
@@ -258,13 +258,13 @@ struct defect_header {
  */
 #ifdef _SYSCALL32
 struct part_info32 {
-	daddr32_t	p_start;
+	uint32_t	p_start;
 	int		p_length;
 };
 #endif /* _SYSCALL32 */
 
 struct part_info {
-	daddr_t		p_start;
+	uint64_t	p_start;
 	int		p_length;
 };
 
diff --git a/lib/libspl/include/sys/dklabel.h b/lib/libspl/include/sys/dklabel.h
index 77d5da10e..8772e4a53 100644
--- a/lib/libspl/include/sys/dklabel.h
+++ b/lib/libspl/include/sys/dklabel.h
@@ -81,8 +81,8 @@ extern "C" {
  * Returned in struct dk_allmap by ioctl DKIOC[SG]APART (dkio(7I))
  */
 struct dk_map {
-	daddr_t	dkl_cylno;		/* starting cylinder */
-	daddr_t	dkl_nblk;		/* number of blocks;  if == 0, */
+	uint64_t	dkl_cylno;	/* starting cylinder */
+	uint64_t	dkl_nblk;	/* number of blocks;  if == 0, */
 					/* partition is undefined */
 };
 
diff --git a/lib/libspl/include/sys/vtoc.h b/lib/libspl/include/sys/vtoc.h
index 004b49097..22a652b74 100644
--- a/lib/libspl/include/sys/vtoc.h
+++ b/lib/libspl/include/sys/vtoc.h
@@ -97,10 +97,10 @@ extern "C" {
 #define	VT_EOVERFLOW	(-7)		/* VTOC op. data struct limited */
 
 struct partition	{
-	ushort_t p_tag;			/* ID tag of partition */
-	ushort_t p_flag;		/* permission flags */
-	daddr_t	p_start;		/* start sector no of partition */
-	long	p_size;			/* # of blocks in partition */
+	ushort_t 	p_tag;		/* ID tag of partition */
+	ushort_t 	p_flag;		/* permission flags */
+	uint64_t	p_start;	/* start sector no of partition */
+	long		p_size;		/* # of blocks in partition */
 };
 
 struct vtoc {
@@ -121,7 +121,7 @@ struct extpartition {
 	ushort_t p_flag;		/* permission flags */
 	ushort_t p_pad[2];
 	diskaddr_t p_start;		/* start sector no of partition */
-	diskaddr_t p_size;			/* # of blocks in partition */
+	diskaddr_t p_size;		/* # of blocks in partition */
 };
 
 
@@ -156,7 +156,7 @@ struct extvtoc {
 	for (i = 0; i < V_NUMPAR; i++) {				\
 		v.v_part[i].p_tag = extv.v_part[i].p_tag;		\
 		v.v_part[i].p_flag = extv.v_part[i].p_flag;		\
-		v.v_part[i].p_start = (daddr_t)extv.v_part[i].p_start;	\
+		v.v_part[i].p_start = (uint64_t)extv.v_part[i].p_start;	\
 		v.v_part[i].p_size = (long)extv.v_part[i].p_size;	\
 		v.timestamp[i] = (time_t)extv.timestamp[i];		\
 	}								\
diff --git a/module/zfs/arc.c b/module/zfs/arc.c
index f8dfee2cc..32d99bf39 100644
--- a/module/zfs/arc.c
+++ b/module/zfs/arc.c
@@ -2777,7 +2777,7 @@ top:
 		uint64_t size = BP_GET_LSIZE(bp);
 		arc_callback_t	*acb;
 		vdev_t *vd = NULL;
-		daddr_t addr = -1;
+		uint64_t addr = -1;
 		boolean_t devw = B_FALSE;
 
 		if (hdr == NULL) {
diff --git a/module/zfs/vdev_disk.c b/module/zfs/vdev_disk.c
index 86a089d12..3e59bd226 100644
--- a/module/zfs/vdev_disk.c
+++ b/module/zfs/vdev_disk.c
@@ -360,6 +360,8 @@ __vdev_disk_physio(struct block_device *bdev, zio_t *zio, caddr_t kbuf_ptr,
 	int bio_size, bio_count = 16;
 	int i = 0, error = 0, block_size;
 
+	ASSERT3U(kbuf_offset + kbuf_size, <=, bdev->bd_inode->i_size);
+
 retry:
 	dr = vdev_disk_dio_alloc(bio_count);
 	if (dr == NULL)