Re: [RFT] sata_promise: decode and report error reasons

[Tejun Heo <htejun@gmail.com>]

Hello,

Jeff clarified most things.  Just few more things.

Mikael Pettersson wrote:
> Which SErrror bits are standard?
> It is true that some of the SError bits are diagnostic rather than
> actual error indicators, as Promise's driver only checks a subset
> of them. I'll fix that.

Pretty much most of them.  Just take a look at SError section in
serialATA spec.  libata EH decodes most of error and some of diagnostic
bits in ata_eh_analyze_serror().

> ATA_DMA_CNT_ERR is when a DMA FIS data size differs from total DMA S/G size.
> I think AC_ERR_ATA_BUS is the correct choice for this one too.

AC_ERR_ATA_BUS is a bit special in that if it happens repeatedly it will
trigger transfer speed adjustment pretty quickly.  The error should be
set iff the error is a transmission error on the wire which is likely to
be fixed by slowing down the transfer rate.  Maybe it should have been
named AC_ERR_ATA_TRANSMISSION or something.

As Jeff pointed out, AC_ERR_HSM is usually used when we can receive the
FIS correctly but it doesn't make sense.  This is usually caused by
driver error or seriously brain damaged firmware.  :-)

Thanks.

-- 
tejun