This message is from the T13 list server.

too bad no one listened to me 2 years ago when I suggested we add a SATA/PATA bit to 
ID Device.  Glad you managed to find a way to do it...sure would be easier to test a 
bit though wouldn't it.

:-)

gary laatsch

> 
> From: Jeff Garzik <[EMAIL PROTECTED]>
> Date: 2004/08/30 Mon PM 02:13:01 EDT
> To: [EMAIL PROTECTED]
> Subject: [t13] [RFC] detecting PATA device behind SATA bridge
> 
> This message is from the T13 list server.
> 
> 
> 
> So, I have stumbled into a situation where I must determine if a PATA or 
> SATA device is connected to a SATA controller.  The only way a PATA 
> device would be connected to a SATA controller is through a PATA->SATA 
> bridge.
> 
> There does not appear to be a "device is SATA" feature bit in IDENTIFY 
> DEVICE, so I have been forced to come up with hueristic:
> 
>       controller is SATA, and
>       cable is SATA, and
>       (ata version < 5 or word-93 is non-zero)
> 
> Does this look like a good way to detect a PATA device attached to a 
> SATA controller?  Can anyone shoot holes in my theory?
> 
> 
> The reason I need this is that I have found two bridge/SATA-controller 
> combinations (not naming names) that will cause failures if the lba48 
> command set is used to transfer more than 256 sectors per read/write... 
> even when the underlying PATA devices _do_ support lba48 large transfers.
> 
> My OS driver must be able to detect a "transparent" PATA bridge in order 
> to limit transfers to 256 sectors.
> 
> 
> Any insight into detecting a PATA device (versus a SATA device) is welcome.
> 
>       Jeff
> 
> 
> 
> 

Reply via email to