When timer run, if BLK_EH_RESET_TIMER returned by LLD, the retries of block request, though defined, is not checked in both scsi middle later and block layer.
Regulation of BLK_EH_RESET_TIMER could help LLD fiddle timer in reasonable manner. Signed-off-by: Hillf Danton <[email protected]> --- --- a/drivers/scsi/scsi_error.c 2010-11-01 19:54:12.000000000 +0800 +++ b/drivers/scsi/scsi_error.c 2010-12-01 22:26:36.000000000 +0800 @@ -138,8 +138,15 @@ enum blk_eh_timer_return scsi_times_out( else if (scmd->device->host->hostt->eh_timed_out) rtn = scmd->device->host->hostt->eh_timed_out(scmd); + if (rtn == BLK_EH_RESET_TIMER) + if (req->retries-- > 0) + return rtn; + else + goto handled; + if (unlikely(rtn == BLK_EH_NOT_HANDLED && !scsi_eh_scmd_add(scmd, SCSI_EH_CANCEL_CMD))) { + handled: scmd->result |= DID_TIME_OUT << 16; rtn = BLK_EH_HANDLED; } _______________________________________________ devel mailing list [email protected] http://www.open-fcoe.org/mailman/listinfo/devel
