Re: Debug scsi message in linux 4.19.31

Re: Debug scsi message in linux 4.19.31

[Randy Dunlap]

[adding linux-scsi and mkp]

On 3/24/19 6:07 AM, Hussam Al-Tayeb wrote:
> Hi. I'm seeing a debug message in linux stable kernel 4.19.31 with this
> patch
> scsi-sd-optimal-i-o-size-should-be-a-multiple-of-physical-block-
> size.patch
> 
> I am getting:
> Mar 23 17:40:10 hades kernel: sd 0:0:0:0: [sda] Optimal transfer size 0
> bytes < PAGE_SIZE (4096 bytes)
> Mar 23 17:40:10 hades kernel: sd 2:0:0:0: [sdb] Optimal transfer size 0
> bytes < PAGE_SIZE (4096 bytes)
> Mar 23 17:46:43 hades kernel: sd 4:0:0:0: [sdc] Optimal transfer size 0
> bytes < PAGE_SIZE (4096 bytes)
> 
> sda and sdb are physically attached SATA mechanical disks (Western
> Digital and Seagate, respectively).
> sdc is a kingston usb flash drive. All three disks use ext4 luks
> encrypted partitions.
> Is the fact that it is reporting 0 as optimal transfer size something I
> should worry about? Note that the the three storage devices are from
> three different manufacturers.
> This particular machine is a Lenovo Intel(R) Core(TM) i5-6400 CPU @
> 2.70GHz with 16GB Kingston DDR4 ram.
> 
> I tried moving the root disk to another machine (a DELL Core I7 7700
> with slightly faster ram from Samsung) and I saw the same message.
> 
> This seems like a patch that fits better in a hardware vendor or
> operating system distributor tree than the upstream kernel. Shouldn't
> stable tree kernels only contain direct fixes and regression fixes?
> Does it really belong in an LTS kernel?
> 
> Please CC me in your reply as I am not subscribed to the list.
> Thank you.
> 


-- 
~Randy

Re: Debug scsi message in linux 4.19.31

[Martin K. Petersen]

Randy,

>> Mar 23 17:40:10 hades kernel: sd 0:0:0:0: [sda] Optimal transfer size 0
>> bytes < PAGE_SIZE (4096 bytes)

We're just trying to weed out the devices that report garbage in this
field. But of course none of the devices I tested with put 0 in there.

>> Does it really belong in an LTS kernel?

The patch does fix problems for devices that report a nonsensical
value. But it is a regression that it now prints a warning for devices
that don't report an optimal I/O size at all.

Sorry about that. Will fix tomorrow!

-- 
Martin K. Petersen	Oracle Linux Engineering

[PATCH] scsi: sd: Quiesce warning if device does not report optimal I/O size

[Martin K. Petersen]

Commit a83da8a4509d ("scsi: sd: Optimal I/O size should be a multiple
of physical block size") split one conditional into several separate
statements in an effort to provide more accurate warning messages when
a device reports a nonsensical value. However, this reorganization
accidentally dropped the precondition of the reported value being
larger than zero. This lead to a warning getting emitted on devices
that do not report an optimal I/O size at all.

Remain silent if a device does not report an optimal I/O size.

Fixes: a83da8a4509d ("scsi: sd: Optimal I/O size should be a multiple of physical block size")
Reported-by: Hussam Al-Tayeb <ht990332@gmx.com>
Cc: Randy Dunlap <rdunlap@infradead.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
---
 drivers/scsi/sd.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/drivers/scsi/sd.c b/drivers/scsi/sd.c
index 9d8e15d03d2b..87d542de2db7 100644
--- a/drivers/scsi/sd.c
+++ b/drivers/scsi/sd.c
@@ -3074,6 +3074,9 @@ static bool sd_validate_opt_xfer_size(struct scsi_disk *sdkp,
 	unsigned int opt_xfer_bytes =
 		logical_to_bytes(sdp, sdkp->opt_xfer_blocks);
 
+	if (sdkp->opt_xfer_blocks == 0)
+		return false;
+
 	if (sdkp->opt_xfer_blocks > dev_max) {
 		sd_first_printk(KERN_WARNING, sdkp,
 				"Optimal transfer size %u logical blocks " \
-- 
2.21.0

Re: [PATCH] scsi: sd: Quiesce warning if device does not report optimal I/O size

[Bart Van Assche]

On Wed, 2019-03-27 at 12:11 -0400, Martin K. Petersen wrote:
> Commit a83da8a4509d ("scsi: sd: Optimal I/O size should be a multiple
> of physical block size") split one conditional into several separate
> statements in an effort to provide more accurate warning messages when
> a device reports a nonsensical value. However, this reorganization
> accidentally dropped the precondition of the reported value being
> larger than zero. This lead to a warning getting emitted on devices
> that do not report an optimal I/O size at all.
> 
> Remain silent if a device does not report an optimal I/O size.
> 
> Fixes: a83da8a4509d ("scsi: sd: Optimal I/O size should be a multiple of physical block size")
> Reported-by: Hussam Al-Tayeb <ht990332@gmx.com>
> Cc: Randy Dunlap <rdunlap@infradead.org>
> Cc: <stable@vger.kernel.org>
> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

Reviewed-by: Bart Van Assche <bvanassche@acm.org>

Re: [PATCH] scsi: sd: Quiesce warning if device does not report optimal I/O size

[Hussam Al-Tayeb]

On Wed, 2019-03-27 at 09:46 -0700, Bart Van Assche wrote:
> On Wed, 2019-03-27 at 12:11 -0400, Martin K. Petersen wrote:
> > Commit a83da8a4509d ("scsi: sd: Optimal I/O size should be a
> > multiple
> > of physical block size") split one conditional into several
> > separate
> > statements in an effort to provide more accurate warning messages
> > when
> > a device reports a nonsensical value. However, this reorganization
> > accidentally dropped the precondition of the reported value being
> > larger than zero. This lead to a warning getting emitted on devices
> > that do not report an optimal I/O size at all.
> >
> > Remain silent if a device does not report an optimal I/O size.
> >
> > Fixes: a83da8a4509d ("scsi: sd: Optimal I/O size should be a
> > multiple of physical block size")
> > Reported-by: Hussam Al-Tayeb <ht990332@gmx.com>
> > Cc: Randy Dunlap <rdunlap@infradead.org>
> > Cc: <stable@vger.kernel.org>
> > Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
>
> Reviewed-by: Bart Van Assche <bvanassche@acm.org>
>
>

I tested two SATA disks and three usb flash drives on two machines. The
warnings are gone.
Thank you very much for the patch.