Re: [PATCH 14/14] libata: use PIO for misc ATAPI commands

[Tejun Heo <htejun@gmail.com>]

Alan Cox wrote:
>> This is about time we make decision on this.  If we're gonna go
> 
> I thought everyone but you had made a decision ;)

Hey, at least Mark was loosely on my side! 8-P

>> DMA-for-everything way, we should lift 16 byte default masking for -mm
>> and -rc's as that only makes debugging difficult and try to blacklist
>> offending devices and controllers individually.
> 
> Would be a useful experiment for -mm not really I think for -rc.

Well, if we change that in #upstream now, -mm will receive it and after
2.6.24 is released, it will end up in -rc1.  We can change it back if
the damage is too grave.

>> My primary concern is about not achieving enough test coverage and
>> isolated cases where it's difficult to tell whether it's the device or
>> the controller.  e.g. Loading recent distro on an ancient and/or strange
>> machine, scratching head while cursing and giving up.  
> 
> Watching the Fedora bugzilla I don't see things that match up to your
> descriptions and fixes. Indeed I've now got verification that the weird
> ALi ATAPI problem a few people have isn't even fixed by your extreme
> changes.

Yeah, ALi seems to be genuine driver problem.  I don't think using PIO
for misc ATAPI commands is extreme considering Windows is doing it.

Another thing is check_atapi_dma in sata_promise.  It mentions losing
interrupt which is exactly what happens if the drive tries to transfer
more data but DMA buffer is short on several controllers.  I wonder
whether this is unnecessary with DMA draining added.

Thanks.

-- 
tejun