Re: [PATCH] libata-core More robust parsing for multi_count(v5)

[Mark Lord <liml@rtr.ca>]

Tejun Heo wrote:
> Jeff Garzik wrote:
>> Yeah, that has long been a design decision -- avoid all the mechanics of
>> PIO-Multi, simply because we did not have much of a need for it (and
>> still don't).
>>
>> Albert Lee worked on PIO-Multi a while ago, which was mainly to get
>> ATA_{12,16} working, IIRC.
>>
>> But overall, I don't see much of a need to introduce it into the main
>> stack?
> 
> Well, there still are awful lot of PIO only devices out in the wild,
> so improving PIO support seems like a good idea?
..


Tons of CompactFlash cards are used PIO-only with Linux,
and systems could benefit from the reduced interrupt overhead
that multiple gives.  Especially since those same systems tend
to have rather modest embedded CPUs.

We *already* do multiple-sector I/O in libata.
But it currently works partially by accident,
and probably gets shut off much of the time
whenever the EH code is invoked and resets anything.

Cheers