Date: Thu, 5 Jul 2001 18:28:24 -0600 From: "Kenneth D. Merry" <ken@kdm.org> To: Joerg Schilling <schilling@fokus.gmd.de> Cc: freebsd-scsi@freebsd.org, gibbs@scsiguy.com, mckay@thehub.com.au Subject: Re: Problems reading burned CDs Message-ID: <20010705182824.A54608@panzer.kdm.org> In-Reply-To: <200107060003.CAA04785@fokus.gmd.de>; from schilling@fokus.gmd.de on Fri, Jul 06, 2001 at 02:03:53AM %2B0200 References: <200107060003.CAA04785@fokus.gmd.de>
next in thread | previous in thread | raw e-mail | index | archive | help
On Fri, Jul 06, 2001 at 02:03:53 +0200, Joerg Schilling wrote: > > >From ken@panzer.kdm.org Fri Jul 6 00:45:37 2001 > > >> > >> >Sending a Bus Device Reset message to a target when it doesn't respond > >> >in the expected amount of time is a bug? Perhaps the driver could try > >> >an abort message first, but the behavior is not completely unreasonable. > >> >Your other complaints seem to be in regard to a "bus reset", which never > >> >occurred in this situation. > >> > >> OK, if it is no bus device reset, it should be OK. > > >Eh? It is a bus device reset... > > I believe you are talking about a target reset which does not reset the whole > bus but a specific target. The SCSI-2 terminology is "Bus Device Reset", SCSI-3 terminology is "Target Reset". You said above "if it is no bus device reset, it should be OK.". It is a bus device reset, a.k.a. a target reset. > >> However, 5 seconds is a too short timeout. > > >For what? The 5 second timeout is for a generic SCSI command facility, > >not for any particular command. 5 seconds is plenty of time for some > >commands, and way too short for others. A user savvy enough to compose > >his own CDBs should also be knowledgeable enough to specify a suitable > >timeout. > > Believe that _any_ SCSI device I know will take much longer if there is some > problem to read the media. > > For a hard disk you should wait at least 20 seconds for error recovery > (this means read retries _and_ eventually error correction). > > If a CD-ROM uses error correction, you should give it ~ 100 seconds. > > If you don't follow these rules, you will end up believing that there are > completely _unreadable_ blocks where the media is only _very hard_ to read. True enough, there are certainly situations where that timeout is too short. It may be a good idea to raise it, but I'll never be able to set it to the "just right" value for any command the user might send. e.g., a 3 hour timeout might be sufficient for most modern hard disks to complete a format unit command, but I don't want a user to have to wait 3 hours to figure out his drive is dead. So in any case, it's still up to the user to set the right timeout for the command if he is specifying the CDB. For the pre-configured commands in camcontrol, like 'format', 'inquiry', 'start', 'stop', etc., the timeout is already set to something appropriate for the command in question. Ken -- Kenneth Merry ken@kdm.org To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-scsi" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20010705182824.A54608>