Re: Request: hdparm module with -X <transfer mode>

[Christian Franke]

Nando wrote:
> It's great to have hdparm output added. Another fixup that would be 
> needed would be setting the ATAPI cable mode to be 80pin, rather than 
> the slower 40pin. Users who add hard drives in place of PATA optical 
> drives are liked affected. Eg:
>
> http://forum.notebookreview.com/showthread.php?p=5519419  (HP 2510P / 
> optical drive caddy)
> http://www.gottabemobile.com/forum/forum_posts.asp?TID=8260&PID=55271 
> <http://www.gottabemobile.com/forum/forum_posts.asp?TID=8260&PID=55271> (Dell 
> XT / not sure why)
>
> Comments/suggestions?

AFAICS this is not possible:

Cable detection can be done either by host sensing the PDIAG:CBLID 
signal, by relying on sensing info from the device, or a combination of 
both. This is rather complex such that false results may be not 
unlikely. The device reports the state of PDIAG:CBLID in IDENTIFY word 
93. The original Linux hdparm prints this info as:
"HW reset results: CBLID- above/below Vih".
(I will add this to GRUB hdparm).

There is no standard ATA command to change what the device reports in 
word 93. The host OS driver may or may not use this info from the device 
to detect cable type.

Reference: T13/1532D Vol 2 Rev 4b (ATA-7 Parallel Transport) section 
8.2.11 and annex D.

Christian