在 2009-05-26二的 08:56 +0100，Alan Cox写道： > > > In any case I have yet to see a CS5536 system whose BIOS sets the cable > > > detection bit. If that's the problem I guess we could add a quirk flag > > > to override the (lack of) BIOS setting. > > > > > This may be the cause that udma5 cannot be set. > > > If the system doesn't have working BIOS cable detect as per the > documentation for the chip then just return the "unknown" cable type and > libata will do device side cable detect. After force cable type as 80 wire, udma5 can be set OK now. after some raw test, the legacy and native driver have no obvious difference in performance. Remain issue: this native driver still have suspend/resume issue. > > Alan
