Date: Tue, 24 Dec 1996 20:06:23 -0500 (EST) From: "John S. Dyson" <toor@dyson.iquest.net> To: se@freebsd.org (Stefan Esser) Cc: joerg_wunsch@uriah.heep.sax.de, freebsd-current@freebsd.org, cg@archimedia.khs-linz.ac.at Subject: Re: DAT: reading with blocksize=256K Message-ID: <199612250106.UAA01427@dyson.iquest.net> In-Reply-To: <Mutt.19961225014401.se@x14.mi.uni-koeln.de> from "Stefan Esser" at Dec 25, 96 01:44:01 am
next in thread | previous in thread | raw e-mail | index | archive | help
> On Dec 24, j@uriah.heep.sax.de (J Wunsch) wrote: > > As Christian Gusenbauer wrote: > > > > > I just got a DAT cartridge with a tar backup. It seems that the backup > > > was made with a blocksize of 256K. Isn't it possible to get the data into > > > my pc with current (it looks like there's a limit of 64K)?? > > > > This has been discussed at lenth already: it's currently limited by > > physio(9) to chunks of at most 64 KB size, due to the limitations in > > the scatter/gather list of some SCSI controllers that don't allow for > > more than 16 scatter/gather segments. > > I can't talk for any other driver, but in case of the NCR driver, > a simple rebuild of the kernel with MAX_SCATTER (in /sys/pci/ncr.c) > set to 129 will allow to read 256KB tape blocks. > Since it is that simple in the driver level for at least one of our SCSI interfaces, then it might be desirable to start supporting larger physio's. I have a bit of technology in mind to support bigger I/O. Since I use NCR controllers, I'll look into it soon. It would be neat to also support larger than 64K clustered reads/writes. Thanks for the info!!! John dyson@freebsd.org
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199612250106.UAA01427>