On Wednesday, September 20, 2006 9:43 AM, Mark Haverkamp wrote: > > On Tue, 2006-09-19 at 18:35 -0600, Eric Moore wrote: > > From host_reset, this patch will complete domain validation > > in the same context as the eh threads. I've added passing > > the sleepFlag, which is CAN_SLEEP in the context of eh threads. > > > > Mark pls try this with the patch that James posted earlier that > > prevents re entry to pending dv. > > Something isn't working quite right. Here is the log. The modprobe > gets stuck. > James, this is not going to work. I have to call delayed spi_dv_device from host reset, or your getting this hang, and/or get asyn narrow on all devices. There is one other driver in the kernel tree that does dv from host reset. It is 53c700. It calls spi_schedule_dv_device. So its delaying domain validation outside eh threads. We would be callign that too, if it wasn't for integrated raid support, and we need to quiese IO to hidden disk. Regardless, dv needs to be done outside eh threads, and due to this, domain validation is goign to be nested, and re-enterant. The problem is by the time we do dv on the faulty device, the dv_in_progress flag will off been cleared. So the question is how can spi transport layer better handle issueing domain validation scsi passthrus that timeout, leading to host_reset? Eric Moore - To unsubscribe from this list: send the line "unsubscribe linux-scsi" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
