Re: SIL_QUIRK_MOD15WRITE workaround problem on 2.6.14

[Jeff Garzik <jgarzik@pobox.com>]

Tejun Heo wrote:
> Jeff Garzik wrote:
> 
>> Tejun Heo wrote:
>>
>>> Ethan confirmed that it's 1095:3114.  Arghhh....  Maybe we should 
>>> keep m15w quirk for 3114's for the time being?  Better be slow than 
>>> hang. Whatever bug the m15w quirk was hiding.
>>
>>
>>
>> A generic 'slow_down_io' module option is reasonable.
>>
>> It is not appropriate to apply mod15write quirk to hardware that isn't 
>> affected by the chip bug.
>>
>> A better solution is to write a 311x-specific interrupt handler.
>>
> 
> Hello, Jeff.  Hello, Carlos.
> 
> I bought a sii3114 controller yesterday and took out my ST3120026AS for 
> testing.  The drive times out during a WRITE_EXT, and locks up.
> 
> * The ST3120026AS works perfectly on a VIA controller.
> * The sii3114 controller works perfectly with Maxtor 6B080M0 drives.
> 
> I don't know.  It acts and smells like m15w problem.  What are the odds 
> of having the same symptom on the same combination?

A lock-up is very generic, and could be anything.  Tons of problems are 
hidden by slowing things down, so I just feel that presuming this is 
_the_ mod15write problem may lead us down the wrong path.

Key example:  On error, we need to do a channel-reset, and possibly a 
FIFO-reset.  See the FreeBSD code.  sata_sil doesn't do this at all, and 
probably should.  The reset may cure lockups.

	Jeff