Currently, if you request a reset through ioctl and it fails,
the kernel escalates like it does for scsi_abort_eh_cmnd:
bus device reset ->
target reset ->
bus reset ->
host reset
This is from scsi_error.c scsi_ioctl_reset:
switch (val) {
case SG_SCSI_RESET_NOTHING:
break;
case SG_SCSI_RESET_DEVICE:
rtn = scsi_try_bus_device_reset(scmd);
if (rtn == SUCCESS)
break;
/* FALLTHROUGH */
case SG_SCSI_RESET_TARGET:
rtn = scsi_try_target_reset(scmd);
if (rtn == SUCCESS)
break;
/* FALLTHROUGH */
case SG_SCSI_RESET_BUS:
rtn = scsi_try_bus_reset(scmd);
if (rtn == SUCCESS)
break;
/* FALLTHROUGH */
case SG_SCSI_RESET_HOST:
rtn = scsi_try_host_reset(scmd);
if (rtn == SUCCESS)
break;
default:
/* FALLTHROUGH */
error = -EIO;
break;
}
Those stronger reset levels affect multiple devices,
which may or may not be desired.
sg-reset includes a --no-esc option to ask the kernel
to not escalate resets, based on a February 2013 thread,
but the matching kernel changes were never merged.
Original issue:
http://marc.info/?l=linux-scsi&m=136000904311597&w=2
Suggested fix:
http://marc.info/?l=linux-scsi&m=136095718729372&w=2
Could that fix be revived?
---
Rob Elliott HP Server Storage
--
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
