Possible Bug In ATA/SCSI Fixed Sense

Possible Bug In ATA/SCSI Fixed Sense

[Charles Machalow]

Hi all,

As of Linux commit 14970f204b1993af7459d5bd34aaff38dfee6670, In
drivers/ata/libata-scsi.c in ata_gen_passthru_sense() line 1087,
desc[0] is set to tf->feature. This makes sense to me however 4 lines
below the same item in the array is set to 0.

One of these lines must be unintended as no matter what the field will
always be 0 though for a couple lines it is set to tf-feature.

I think that line 1091 is at fault because if I send down a nop
command, feature (error in this case) should be set to 4 in the
returned current task file.

Thanks!

- Charlie Scott Machalow

Re: Possible Bug In ATA/SCSI Fixed Sense

[Tejun Heo]

Hello,

On Fri, Oct 28, 2016 at 11:34:22AM -0700, Charles Machalow wrote:
> As of Linux commit 14970f204b1993af7459d5bd34aaff38dfee6670, In
> drivers/ata/libata-scsi.c in ata_gen_passthru_sense() line 1087,
> desc[0] is set to tf->feature. This makes sense to me however 4 lines
> below the same item in the array is set to 0.
> 
> One of these lines must be unintended as no matter what the field will
> always be 0 though for a couple lines it is set to tf-feature.
> 
> I think that line 1091 is at fault because if I send down a nop
> command, feature (error in this case) should be set to 4 in the
> returned current task file.

This looks to be from 11093cb1ef56 ("libata-scsi: generate correct ATA
pass-through sense").  Hannes?

Thanks.

-- 
tejun

Re: Possible Bug In ATA/SCSI Fixed Sense

[Hannes Reinecke]

On 10/31/2016 04:58 PM, Tejun Heo wrote:
> Hello,
>
> On Fri, Oct 28, 2016 at 11:34:22AM -0700, Charles Machalow wrote:
>> As of Linux commit 14970f204b1993af7459d5bd34aaff38dfee6670, In
>> drivers/ata/libata-scsi.c in ata_gen_passthru_sense() line 1087,
>> desc[0] is set to tf->feature. This makes sense to me however 4 lines
>> below the same item in the array is set to 0.
>>
>> One of these lines must be unintended as no matter what the field will
>> always be 0 though for a couple lines it is set to tf-feature.
>>
>> I think that line 1091 is at fault because if I send down a nop
>> command, feature (error in this case) should be set to 4 in the
>> returned current task file.
>
> This looks to be from 11093cb1ef56 ("libata-scsi: generate correct ATA
> pass-through sense").  Hannes?
>
Indeed, that's an error.
I wanted to set byte 7 to zero, indicating there is no additional sense 
data to be had.
Patch has been sent.

Cheers,

Hannes
-- 
Dr. Hannes Reinecke		      zSeries & Storage
hare@suse.de			      +49 911 74053 688
SUSE LINUX Products GmbH, Maxfeldstr. 5, 90409 Nürnberg
GF: J. Hawn, J. Guild, F. Imendörffer, HRB 16746 (AG Nürnberg)