On Wed, 2011-01-26 at 12:46 -0500, Jeff Garzik wrote: > On 01/26/2011 09:08 AM, James Bottomley wrote: > > I think better might be to move the eh thread functionality into block. > > What libata really wants is one eh thread per phy. It gets this by > [...] > > machinery, everyone should get what they want, and libata could dispense > > with its current corruption of multiple SCSI hosts per physical bus > > attachment. > > That's not the whole picture. libata originally chose one-host-per-port > because that winds up being the best queueing/queue-busy arrangement for > legacy IDE ports. Right, but we count things at the lun, the target and the host. For a two level queue setup, you could use the target. > They behave quite similarly to independent > controllers -- even to the point of having separate irqs -- even though > multiple IDE ports are shoehorned into a single PCI device. So interrupts don't really have much to do with the host, that's all part of driver setup. IDE is really just using the two lines as a simple completion signal ... much like the current multi queue cards do (which also have multiple interrupts per host). James -- To unsubscribe from this list: send the line "unsubscribe linux-ide" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
