Re: [PATCH] libata: add TRIM support

[Mark Lord <liml@rtr.ca>]

Alan Cox wrote:
>>> For those that pee themselves you don't need a host flag, you need a
>>> qc_issue hook - you need that hook for the purpose of actually making the
>>> command work, so in the short term making it just return AC_ERR_DEV isn't
>>> exactly hard. Not having the flag and fixing those odd few drivers would
>>> be far more productive.
>> We are talking about chips whose command set is essentially fixed at 
>> manufacture time.
> 
> There are only two of those I know of
> - IT821x which already uses qc_issue to filter.
> - pata_st412, which uses qc_issue and isn't in tree anyway and probably
>   never will be.
..

The sata_mv driver will choke the chipset if the TRIM command
involves more than one sector of TRIM data.

Cheers