Eric Youngdale wrote: > Yes, I agree - it should be per physical device - we could take > SD_TIMEOUT as a default, and allow the low-level controllers to modify this > as required. This could be handled by adding a field to the Scsi_Device > structure. > > I am not sure what you would want to do if the upper level driver (i.e. > sg) wanted an even larger timeout. My guess is that you take the maximum > value of what the upper level driver wants and the value in SDpnt. We could add a field to the Scsi_Device structure (defaulted to 0) and set it with this extension to the "plan 9"-ish "add-single-device" syntax: % echo "scsi timeout=<to_in_secs> <h> <b> <t> <l>" > /proc/scsi/scsi plus add a corresponding mid level ioctl (and one to get the timeout value). The mid level could pass through the max of this number and whatever the upper level driver requested with each command request. Both st (mainly by command type) and sg (by user input) allow timeouts to be manipulated at the moment. Doug Gilbert - To unsubscribe from this list: send the line "unsubscribe linux-scsi" in the body of a message to [EMAIL PROTECTED]
