[Scsirastools-developers] RE: sgdiag, STOP_UNIT, and ATA-in-USB caddies

[Alex B. <ale...@as...>]

Hi Andy -

On Mon, 22 Nov 2004, Cress, Andrew R wrote:

> You could test it by using sgdiag:
> c - to compose a command
> 4 = length
> E4   (or 99)
> 00
> 00
> 00
> Response Data Length = 2 (this length probably doesn't matter)
> Output Data Length = 0
>
> Then, if it works, you can see how this is coded in sgdiag.c and make
> your own version of the program with a special ATA sleep option.

Thanks for the prompt reply. Unfortunately, no dice. I've no clue as to
whether it's Linux's usb-storage layer or the USB/1394->ATA bridge (it's a
Prolific-branded chipset) that's failing to pass the command on.

I guess I'll just have to live with it for now. Ah well. :-)

> Andy

Cheers,
Alex.

>
> -----Original Message-----
> From: Alex Butcher [mailto:ale...@as...]
> Sent: Saturday, November 20, 2004 9:27 AM
> To: Cress, Andrew R
> Subject: RE: sgdiag, STOP_UNIT, and ATA-in-USB caddies
>
>
> Hi Andy -
>
> On Thu, 18 Nov 2004, Cress, Andrew R wrote:
>
>> You may need to contact Maxtor to get a tool or API function that will
>> accept the spin down for ATA.  If there is a command that the ATA
> device
>> will accept, it could be sent via sg encapsulation, I think.  In fact,
>> the 'c' option would allow such a command to be composed.
>
> According to the source for hdparm and
> <http://my.execpc.com/~geezer/osd/misc/ata-idle.c>, the equivalent
> command
> is SLEEP, which is defined as 0xe6 or 0x99.
>
> hdparm does the following to send a drive to sleep:
>
>         if (set_sleepnow) {
> #ifndef WIN_SLEEPNOW1
> #define WIN_SLEEPNOW1 0xE6
> #endif
> #ifndef WIN_SLEEPNOW2
> #define WIN_SLEEPNOW2 0x99
> #endif
>                 unsigned char args1[4] = {WIN_SLEEPNOW1,0,0,0};
>                 unsigned char args2[4] = {WIN_SLEEPNOW2,0,0,0};
>                 if (get_sleepnow)
>                         printf(" issuing sleep command\n");
>                 if (ioctl(fd, HDIO_DRIVE_CMD, &args1)
>                  && ioctl(fd, HDIO_DRIVE_CMD, &args2))
>                         perror(" HDIO_DRIVE_CMD(sleep) failed");
>         }
>
> With the exception of Seagate, these commands appear to be standard ATA.
> Now, any ideas on how I can construct an equivalent SCSI CDB?
>
>> Andy
>
> Cheers,
> Alex.
>
>>
>> -----Original Message-----
>> From: Alex Butcher [mailto:ale...@as...]
>> Sent: Thursday, November 18, 2004 2:12 PM
>> To: Cress, Andrew R
>> Subject: sgdiag, STOP_UNIT, and ATA-in-USB caddies
>>
>>
>> Hi -
>>
>> I've been looking for a tool to spin down the various ATA discs that I
>> have
>> USB caddies. scsi-spin and scsi-idle use an ioctl that's deprecated
>> under
>> Linux kernel 2.6.9 (the kernel logs a message saying to port to
> SG_IO),
>> but
>> your sgdiag appears to fit the bill. Unfortunately, when I try to send
>> the
>> STOP_UNIT command, I get the following:
>>
>>
> ========================================================================
>> ==
>>                 sgdiag utility v1.12 for SCSI disks
>>               ******************************************
>> Log file /var/log/sgdiag.log is open, debug=0
>>
>> Num Name [bus:ch:id:lun] Type Vendor Device_Model     FW   Serial#
> Size
>>  0 /dev/sg0 [1:0:0:0] Disk  Maxtor 5 3073U6           DA62  [em]
>>
>> c = Compose Command to send  1 = Do bug 1 (sense_len = 18)
>> r = Reset SCSI bus           2 = Do bug 2 (INQ hang)
>> i = Special Ser# Inquiries   3 = Do bug 3 (fmt w short timeout)
>> f = Format SCSI disk         w = Wipe SCSI disk (fmt wo glist)
>> s = Start Unit               d = Send Diagnostic self-test
>> t = sTop Unit
>> Enter Selection ('q' to quit) : t
>>
>> Device Num (0 to 0) : 0
>> Issuing stop_unit for /dev/sg0
>> stop_unit: Check Condition Driver_status=0x08
> (DRIVER_SENSE,SUGGEST_OK)
>> sense error, key=05 asc=20 ascq=00 Illegal Request, Invalid command
>> operation code
>> Info valid=0, Current
>> stop_unit complete, status = -2
>>
> ========================================================================
>> ==
>>
>> I know this is not the intended use of this tool, but I'd appreciate
>> your
>> opinions on what's up, and whether it's possible to workaround it.
>> Normally,
>> I'd use hdparm -Y to spin down ATA drives, but as they show up as
>> pseudo-SCSI
>> devices when used with Linux's usb-storage driver, hdparm refuses to
>> handle
>> them. :-(
>>
>> Best Regards,
>> Alex.
>>
>
>

-- 
Alex Butcher      Brainbench MVP for Internet Security: www.brainbench.com
Bristol, UK                      Need reliable and secure network systems?
PGP/GnuPG ID:0x271fd950                         <http://www.assursys.com/>