diff mbox

nfsd: make nfsd4_scsi_identify_device retry with a larger buffer

Message ID 20180507130131.7221-1-smayhew@redhat.com (mailing list archive)
State New, archived
Headers show

Commit Message

Scott Mayhew May 7, 2018, 1:01 p.m. UTC
nfsd4_scsi_identify_device() performs a single IDENTIFY command for the
device identification VPD page using a small buffer.  If the reply is
too large to fit in this buffer then the GETDEVICEINFO reply will not
contain any info for the SCSI volume aside from the registration key.
This can happen for example if the device has descriptors using long
SCSI name strings.

When the initial reply from the device indicates a larger buffer is
needed, retry once using the page length from that reply.

Signed-off-by: Scott Mayhew <smayhew@redhat.com>
---
 fs/nfsd/blocklayout.c | 18 +++++++++++++-----
 1 file changed, 13 insertions(+), 5 deletions(-)

Comments

J. Bruce Fields May 7, 2018, 6:45 p.m. UTC | #1
On Mon, May 07, 2018 at 09:01:31AM -0400, Scott Mayhew wrote:
> nfsd4_scsi_identify_device() performs a single IDENTIFY command for the
> device identification VPD page using a small buffer.  If the reply is
> too large to fit in this buffer then the GETDEVICEINFO reply will not
> contain any info for the SCSI volume aside from the registration key.
> This can happen for example if the device has descriptors using long
> SCSI name strings.
> 
> When the initial reply from the device indicates a larger buffer is
> needed, retry once using the page length from that reply.

Looks good, but how did you choose 65532 as the maximum?  Does the scsi
layer give us any maximum?

Also, I don't know why this is bothering me, but: I'd prefer "buflen" to
"bufflen".

--b.

> 
> Signed-off-by: Scott Mayhew <smayhew@redhat.com>
> ---
>  fs/nfsd/blocklayout.c | 18 +++++++++++++-----
>  1 file changed, 13 insertions(+), 5 deletions(-)
> 
> diff --git a/fs/nfsd/blocklayout.c b/fs/nfsd/blocklayout.c
> index 70b8bf7..0cd9249 100644
> --- a/fs/nfsd/blocklayout.c
> +++ b/fs/nfsd/blocklayout.c
> @@ -216,13 +216,14 @@ static int nfsd4_scsi_identify_device(struct block_device *bdev,
>  	struct request_queue *q = bdev->bd_disk->queue;
>  	struct request *rq;
>  	struct scsi_request *req;
> -	size_t bufflen = 252, len, id_len;
> +	size_t bufflen = 252, maxlen = 65532, len, id_len;
>  	u8 *buf, *d, type, assoc;
> -	int error;
> +	int retries = 1, error;
>  
>  	if (WARN_ON_ONCE(!blk_queue_scsi_passthrough(q)))
>  		return -EINVAL;
>  
> +again:
>  	buf = kzalloc(bufflen, GFP_KERNEL);
>  	if (!buf)
>  		return -ENOMEM;
> @@ -255,9 +256,16 @@ static int nfsd4_scsi_identify_device(struct block_device *bdev,
>  
>  	len = (buf[2] << 8) + buf[3] + 4;
>  	if (len > bufflen) {
> -		pr_err("pNFS: INQUIRY 0x83 response invalid (len = %zd)\n",
> -			len);
> -		goto out_put_request;
> +		if (len < maxlen && retries--) {
> +			blk_put_request(rq);
> +			kfree(buf);
> +			bufflen = len;
> +			goto again;
> +		} else {
> +			pr_err("pNFS: INQUIRY 0x83 response invalid (len = %zd)\n",
> +				len);
> +			goto out_put_request;
> +		}
>  	}
>  
>  	d = buf + 4;
> -- 
> 2.9.5
--
To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Scott Mayhew May 7, 2018, 8:17 p.m. UTC | #2
On Mon, 07 May 2018, J. Bruce Fields wrote:

> On Mon, May 07, 2018 at 09:01:31AM -0400, Scott Mayhew wrote:
> > nfsd4_scsi_identify_device() performs a single IDENTIFY command for the
> > device identification VPD page using a small buffer.  If the reply is
> > too large to fit in this buffer then the GETDEVICEINFO reply will not
> > contain any info for the SCSI volume aside from the registration key.
> > This can happen for example if the device has descriptors using long
> > SCSI name strings.
> > 
> > When the initial reply from the device indicates a larger buffer is
> > needed, retry once using the page length from that reply.
> 
> Looks good, but how did you choose 65532 as the maximum?  Does the scsi
> layer give us any maximum?

So I wasn't quite sure about that.  The allocation length field for an
INQUIRY command is two bytes.  There are various fields that the SPC-3
doc says have to be a multiple of 4, but I couldn't find anything that
says whether that applies to the allocation length.  Since the initial
value 252 is a muliple of 4, I figured it wouldn't hurt anything to have
the maximum be a multiple of 4 too.

OTOH if you look at scsi_get_vpd_buf() in drivers/scsi/scsi.c, the
initial buffer size is SCSI_VPD_PG_LEN, which is 255.

-Scott

> 
> Also, I don't know why this is bothering me, but: I'd prefer "buflen" to
> "bufflen".
> 
> --b.
> 
> > 
> > Signed-off-by: Scott Mayhew <smayhew@redhat.com>
> > ---
> >  fs/nfsd/blocklayout.c | 18 +++++++++++++-----
> >  1 file changed, 13 insertions(+), 5 deletions(-)
> > 
> > diff --git a/fs/nfsd/blocklayout.c b/fs/nfsd/blocklayout.c
> > index 70b8bf7..0cd9249 100644
> > --- a/fs/nfsd/blocklayout.c
> > +++ b/fs/nfsd/blocklayout.c
> > @@ -216,13 +216,14 @@ static int nfsd4_scsi_identify_device(struct block_device *bdev,
> >  	struct request_queue *q = bdev->bd_disk->queue;
> >  	struct request *rq;
> >  	struct scsi_request *req;
> > -	size_t bufflen = 252, len, id_len;
> > +	size_t bufflen = 252, maxlen = 65532, len, id_len;
> >  	u8 *buf, *d, type, assoc;
> > -	int error;
> > +	int retries = 1, error;
> >  
> >  	if (WARN_ON_ONCE(!blk_queue_scsi_passthrough(q)))
> >  		return -EINVAL;
> >  
> > +again:
> >  	buf = kzalloc(bufflen, GFP_KERNEL);
> >  	if (!buf)
> >  		return -ENOMEM;
> > @@ -255,9 +256,16 @@ static int nfsd4_scsi_identify_device(struct block_device *bdev,
> >  
> >  	len = (buf[2] << 8) + buf[3] + 4;
> >  	if (len > bufflen) {
> > -		pr_err("pNFS: INQUIRY 0x83 response invalid (len = %zd)\n",
> > -			len);
> > -		goto out_put_request;
> > +		if (len < maxlen && retries--) {
> > +			blk_put_request(rq);
> > +			kfree(buf);
> > +			bufflen = len;
> > +			goto again;
> > +		} else {
> > +			pr_err("pNFS: INQUIRY 0x83 response invalid (len = %zd)\n",
> > +				len);
> > +			goto out_put_request;
> > +		}
> >  	}
> >  
> >  	d = buf + 4;
> > -- 
> > 2.9.5
--
To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
J. Bruce Fields May 10, 2018, 2:16 p.m. UTC | #3
On Mon, May 07, 2018 at 04:17:40PM -0400, Scott Mayhew wrote:
> On Mon, 07 May 2018, J. Bruce Fields wrote:
> 
> > On Mon, May 07, 2018 at 09:01:31AM -0400, Scott Mayhew wrote:
> > > nfsd4_scsi_identify_device() performs a single IDENTIFY command for the
> > > device identification VPD page using a small buffer.  If the reply is
> > > too large to fit in this buffer then the GETDEVICEINFO reply will not
> > > contain any info for the SCSI volume aside from the registration key.
> > > This can happen for example if the device has descriptors using long
> > > SCSI name strings.
> > > 
> > > When the initial reply from the device indicates a larger buffer is
> > > needed, retry once using the page length from that reply.
> > 
> > Looks good, but how did you choose 65532 as the maximum?  Does the scsi
> > layer give us any maximum?
> 
> So I wasn't quite sure about that.  The allocation length field for an
> INQUIRY command is two bytes.  There are various fields that the SPC-3
> doc says have to be a multiple of 4, but I couldn't find anything that
> says whether that applies to the allocation length.  Since the initial
> value 252 is a muliple of 4, I figured it wouldn't hurt anything to have
> the maximum be a multiple of 4 too.
> 
> OTOH if you look at scsi_get_vpd_buf() in drivers/scsi/scsi.c, the
> initial buffer size is SCSI_VPD_PG_LEN, which is 255.

OK.  Well, till we have some input from a scsi expert, would you mind
just adding a brief summary of the above as a code comment?

--b.

> 
> -Scott
> 
> > 
> > Also, I don't know why this is bothering me, but: I'd prefer "buflen" to
> > "bufflen".
> > 
> > --b.
> > 
> > > 
> > > Signed-off-by: Scott Mayhew <smayhew@redhat.com>
> > > ---
> > >  fs/nfsd/blocklayout.c | 18 +++++++++++++-----
> > >  1 file changed, 13 insertions(+), 5 deletions(-)
> > > 
> > > diff --git a/fs/nfsd/blocklayout.c b/fs/nfsd/blocklayout.c
> > > index 70b8bf7..0cd9249 100644
> > > --- a/fs/nfsd/blocklayout.c
> > > +++ b/fs/nfsd/blocklayout.c
> > > @@ -216,13 +216,14 @@ static int nfsd4_scsi_identify_device(struct block_device *bdev,
> > >  	struct request_queue *q = bdev->bd_disk->queue;
> > >  	struct request *rq;
> > >  	struct scsi_request *req;
> > > -	size_t bufflen = 252, len, id_len;
> > > +	size_t bufflen = 252, maxlen = 65532, len, id_len;
> > >  	u8 *buf, *d, type, assoc;
> > > -	int error;
> > > +	int retries = 1, error;
> > >  
> > >  	if (WARN_ON_ONCE(!blk_queue_scsi_passthrough(q)))
> > >  		return -EINVAL;
> > >  
> > > +again:
> > >  	buf = kzalloc(bufflen, GFP_KERNEL);
> > >  	if (!buf)
> > >  		return -ENOMEM;
> > > @@ -255,9 +256,16 @@ static int nfsd4_scsi_identify_device(struct block_device *bdev,
> > >  
> > >  	len = (buf[2] << 8) + buf[3] + 4;
> > >  	if (len > bufflen) {
> > > -		pr_err("pNFS: INQUIRY 0x83 response invalid (len = %zd)\n",
> > > -			len);
> > > -		goto out_put_request;
> > > +		if (len < maxlen && retries--) {
> > > +			blk_put_request(rq);
> > > +			kfree(buf);
> > > +			bufflen = len;
> > > +			goto again;
> > > +		} else {
> > > +			pr_err("pNFS: INQUIRY 0x83 response invalid (len = %zd)\n",
> > > +				len);
> > > +			goto out_put_request;
> > > +		}
> > >  	}
> > >  
> > >  	d = buf + 4;
> > > -- 
> > > 2.9.5
--
To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
diff mbox

Patch

diff --git a/fs/nfsd/blocklayout.c b/fs/nfsd/blocklayout.c
index 70b8bf7..0cd9249 100644
--- a/fs/nfsd/blocklayout.c
+++ b/fs/nfsd/blocklayout.c
@@ -216,13 +216,14 @@  static int nfsd4_scsi_identify_device(struct block_device *bdev,
 	struct request_queue *q = bdev->bd_disk->queue;
 	struct request *rq;
 	struct scsi_request *req;
-	size_t bufflen = 252, len, id_len;
+	size_t bufflen = 252, maxlen = 65532, len, id_len;
 	u8 *buf, *d, type, assoc;
-	int error;
+	int retries = 1, error;
 
 	if (WARN_ON_ONCE(!blk_queue_scsi_passthrough(q)))
 		return -EINVAL;
 
+again:
 	buf = kzalloc(bufflen, GFP_KERNEL);
 	if (!buf)
 		return -ENOMEM;
@@ -255,9 +256,16 @@  static int nfsd4_scsi_identify_device(struct block_device *bdev,
 
 	len = (buf[2] << 8) + buf[3] + 4;
 	if (len > bufflen) {
-		pr_err("pNFS: INQUIRY 0x83 response invalid (len = %zd)\n",
-			len);
-		goto out_put_request;
+		if (len < maxlen && retries--) {
+			blk_put_request(rq);
+			kfree(buf);
+			bufflen = len;
+			goto again;
+		} else {
+			pr_err("pNFS: INQUIRY 0x83 response invalid (len = %zd)\n",
+				len);
+			goto out_put_request;
+		}
 	}
 
 	d = buf + 4;