Re: [2.6 patch] select ATA_SFF

[Sergei Shtylyov <sshtylyov@ru.mvista.com>]

Mark Lord wrote:

>>>>>>>   Jeff, Tejun, what "sff" in the file name actually means? Isn't 
>>>>>>> it strange that the drivers lacking DMA support or not really 
>>>>>>> compliant with SFF-8038i have to link with this file?

>>>>>> Maybe it should be libata-tf and libata-bmdma, but sff (sans bmdma)
>>>>>> and bmdma is acceptable, hopefully, right?

>>>>>   What's sff sans bmdma?

>>>> Supposed to be TF interface.  IIRC, the SFF term was first from Alan 
>>>> although it's entirely possible that I misunderstood it and used it 
>>>> in the wrong way.  Alan, can you please clear up the confusion?

>>> The SFF/Intel spec is for PCI IDE (BMDMA or otherwise), so it covers and
>>> defines all the common bits of the IDE interface on PCI (and in defining
>>> the legacy interface conveniently documents the extended ST-412 
>>> interface
>>> used by ATA and "pre-ATA" IDE/EIDE controllers).

>>    If you mean SFF-8038i (which can indeed be named "SFF/Intel"), it
>> documents *only* BMDMA.  If you mean something else, please be more 
>> precise.

> ..

> The "Intel PCI IDE Controller Specification Revision 1.0 3/4/94" doesn't 

    This is not an SFF spec.

> mention
> bmdma at all, but does document the taskfile register addresses.
> It defers to ATA-1 for actual taskfile descriptions/functionality, though.

    Yes, it only describes deviation from "historical" IDE, i.e. the missing 
drive address register (port 0x3[7F]7).

> There's nothing particularly bad about the current naming we use, though.

    There wouldn't have been anything bad if that file wasn't covering both 
taskfile and  BMDMA stuff.  This way, it looks misleading (at least for me).

> Cheers

WBR, Sergei