bug#51407: parted didn’t check the returned sense data

[Johnson.Yan]

Hi Brian,
Thanks for the reply. 
Seems Kernel called parted to send Read(16) command to the drive, it is more 
likely kernel didn't check the returned sense data and change the drive 
capacity to 0.
I suppose this bug#51407 can be closed. Thanks again.

Regards,
Johnson

-----Original Message-----
From: Brian C. Lane <bcl@redhat.com> 
Sent: Thursday, October 28, 2021 12:03 AM
To: Johnson Yan - A30689 <Johnson.Yan@microchip.com>
Cc: 51407@debbugs.gnu.org
Subject: Re: bug#51407: parted didn’t check the returned sense data

[You don't often get email from bcl@redhat.com. Learn why this is important at 
http://aka.ms/LearnAboutSenderIdentification.]

EXTERNAL EMAIL: Do not click links or open attachments unless you know the 
content is safe

On Wed, Oct 27, 2021 at 09:20:59AM +0000, Johnson.Yan--- via Bug reports for 
the GNU Parted disk partition editor wrote:
> Hi,
> Some further information.
>
>
>   1.  We already tried the latest parted-3.4.tar.xz, issue still exists.
>   2.  For below sense data, the corresponding KCQ=2:04:0E, i.e. SENSE KEY: 2, 
> ADDITIONAL SENSE CODE: 04, ADDITIONAL SENSE CODE QUALIFIER: 0E.
>
> Oct 13 01:30:02 localhost kernel: sd 13:0:3:0: [sde] Sense Key : Not 
> Ready [current] [descriptor]
>
> Oct 13 01:30:02 localhost kernel: sd 13:0:3:0: [sde] Add. Sense: 
> Logical unit not ready, security session in progress
>
> You can find ASC and ASCQ assignments from SPC spec as below.
> [cid:image001.png@01D7CB56.D375FCF0]

Sorry, for some reason I didn't get the first email in this thread.
(https://debbugs.gnu.org/cgi/bugreport.cgi?bug=51407)

The logs you included are from the kernel, not from parted, and parted doesn't 
deal with sense data or changing the drive capacity, that's all in the kernel 
driver for the drive.

If parted it having problems handling this, that might be something we can 
improve. It's possible that whatever errors are returned by the kernel aren't 
being checked, but I don't have any way to tell without either a parted 
traceback, or doing this myself. I don't think I have any drives capable of 
secure erase, so someone with access to one will need to dig into this more 
deeply. eg. running parted via strace to capture any unexpected error codes 
from the kernel.

Brian

--
Brian C. Lane (PST8PDT) - weldr.io - lorax - parted - pykickstart