On Mon, Jun 05, Brian King wrote: > Brian King wrote: > > Douglas Gilbert wrote: > >>> I have a SCSI disk, which I want to spin down when the system is not in > >>> use. I do this by using sdparm, scsi-spin or sg-utils. These tools all > >>> spin down the SCSI drive by using an IOCTL. > >>> > >>> Problem is that the kernel doesn't spin the drive back up. When a > >>> process requests data from the disk (a simple ls), the kernel responds > >>> with an I/O error. After some of these errors, reiserfs marks the drive > >>> read-only. > > > > Setting sdev->allow_restart in struct scsi_device will cause sense > > key/code/qual of 02/04/02 (not ready, initialization command required) > > to wake up the scsi error handler and will force scsi core to issue > > a start unit command to the disk. I added this a while back to handle > > ipr raid arrays which need a start unit command each time the adapter > > gets reset. It would be easy enough for sd to use this, either > > all the time, or only when directed to, or we could add a sysfs attribute > > to the disk to enable/disable this behavior. > > Please try the attached patch. It adds a sysfs device attribute "allow_restart". > If you set it to 1, then scsi core will issue a start unit when it sees 02/04/02 > sense data. The attribute is in the same place as the FUA and cache_type > attributes (eg. /sys/block/sdc/device/scsi_disk:1:1:9:0/allow_restart). Is it safe to have this attribute be 1 per default, and let the user disable it for problematic drives? The IO errors bother me alot... - : 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
