Date: Sun, 7 May 1995 06:14:18 -0400 (EDT) From: Peter Dufault <dufault@hda.com> To: imb@scgt.oz.au Cc: hackers@FreeBSD.org Subject: Re: SCSI drive auto error reallocation Message-ID: <199505071014.GAA17539@hda.com> In-Reply-To: <199505070937.TAA28473@asstdc.scgt.oz.au> from "michael butler" at May 7, 95 07:37:44 pm
next in thread | previous in thread | raw e-mail | index | archive | help
(I'm moving the discussion to -hackers. The announcement of the new software is pertinent to -current, but the technical discussion probably should go on in -hackers). michael butler writes: > > > The newest version of "scsi" lets you edit your mode pages. Now that I've > > added that you may want to check to see if you have automatic read and > > write reallocation enabled. I have three SCSI drives and none did. > > Nor did mine .. however, I can understand wishing to dynamically reallocate > on write failures, but reads ? The drive will only do the reallocation if it can successfully recover the data. > > > The command is "scsi -f /dev/rsd?c -m 1" to dump mode page 1 .. > > On an ISA machine (running -current, 7/7/95, BusLogic 542B, OLD firmware) > this can cause .. > > May 7 17:46:50 asstdc /kernel: aha0: DMA beyond end Of ISA > May 7 17:46:50 asstdc /kernel: biodone: buffer already done > > .. needless to say the ioctl fails. Not fatal, but interesting. BUG: The user mode ioctl doesn't use bounce buffers. Do you have more than 16 MB of memory? How did you manage to do the ioctl? I assume I just have to add > /* > * Use a bounce buffer if necessary > */ >#ifdef BOUNCE_BUFFERS > if (sc_link->flags & SDEV_BOUNCE) > vm_bounce_alloc(bp); >#endif before the scsi_scsi_cmd in scsistrategy. > > > hda# scsi -f /dev/rsd2c -m 1 > > > AWRE (Auto Write Reallocation Enbld): 1 > > > ARRE (Auto Read Reallocation Enbld): 1 > > > TB (Transfer Block): 1 > > > RC (Read Continuous): 0 > > > EER (Enable Early Recovery): 1 > > > PER (Post Error): 0 > > > DTE (Disable Transfer on Error): 0 > > > DCR (Disable Correction): 0 > > > Read Retry Count: 18 > > > Correction Span: 11 > > > Head Offset Count: 0 > > > Data Strobe Offset Count: 0 > > > Write Retry Count: 18 > > > Recovery Time Limit: 0 > > > If AWRE and ARRE are 0 you want to run the command with the "-e -P 3" > > switches to permanently enable read and write reallocation. > > Could someone please explain 'TB' and 'EER' which seem to differ from teh > above on a drive here ? You seem to be setup properly and this disk is setup improperly. TB: "A transfer block (TB) bit of one indicates that a data block that is not recovered within the recovery limits specified shall be transferred to the initiator before CHECK CONDITION status is returned. A TB bit of zero indicates that such a data block shall not be transferred to the initiator". It is probably better to change this to 0, though the driver should properly detect this as an error. EER: "An enable early recovery bit of one indicates that the target shall use the most expedient form of error recovery first. An EER bit of zero indicates that the target shall use an error recovery procedure that minimizes the risk of mis-detection or mis-correction". A note says "An EER bit of zero typically means that the specified retry limit is exhausted prior to using error correction codes". It is probably better to change this to 0 also. PER means to report recovered errors and should be left at 0, since I bet an error report would be interpreted as an error. DTE of 1 indicates the target should terminate the data phase upon detection of a recovered error. It should be left at 0. RC means return the right amount of data as fast as possible regardless of whether or not it is correct, even if it needs to be fabricated. Probably best left at 0. -- Peter Dufault Real Time Machine Control and Simulation HD Associates, Inc. Voice: 508 433 6936 dufault@hda.com Fax: 508 433 5267
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199505071014.GAA17539>