Date: Tue, 04 Feb 2014 12:03:17 +0000 From: Frank Leonhardt <frank2@fjl.co.uk> To: freebsd-doc@freebsd.org Subject: Re: GEOM meaning Message-ID: <52F0D705.1090402@fjl.co.uk> In-Reply-To: <52F0D58F.4040703@fjl.co.uk> References: <52F0C512.8080909@online.de> <52F0D58F.4040703@fjl.co.uk>
next in thread | previous in thread | raw e-mail | index | archive | help
On 04/02/2014 11:57, Frank Leonhardt wrote: > On 04/02/2014 10:46, Dieter Lange wrote: >> Hi, >> >> having looked at quite a few sites now, not just >> <http://www.freebsd.org/doc/en_US.ISO8859-1/books/handbook/geom.html>; >> I still cannot find out what "GEOM" means (wrt disks etc., not >> geography or so). It probably does not mean "Modular Disk >> Transformation Framework". I am not talking of its use and/or >> definitions, just the meaning of the abbreviation or word... >> >> Thanks+kind regards from the only person on the WWW who doesn't know... >> DL > > I've always assumed it was short for (disk) geometry - i.e. converting > logical requests to match the disk geometry. Eh? Well, back in my > youth we did talk about the "geometr"y of DASD (disk!). For example, > how many platters (heads), cylinders (tracks) and sectors/track were > present. With ATA and SCSI this has become less relevant as you only > get to see the logical structure of a disk (a load of blocks > sequentially numbered 0...n). You may well ask why anyone would call > these parameters "geometry", but I can't think of any other better > name for it, nor any other word in common use for referring to them > (other than CHT). But a disk's geometry was highly relevant because > you (the programmer) would either be responsible for moving the head > (via a stepper motor) to the correct track, or at the very least, you > had to be sensitive to where the head was on the disk when optimising > your code. > > I've no proof whatsoever that this is why the geom library is so > called - it could all be a complete coincidence. I don't remember > hearing about "geom" on System V, nor on BSD until recently (late 1990s). > > Regards, Frank. > P.S. The reason why I'm not 100% happy with the above theory is that the geom library sits between DEVFS and the device driver (pretending to be a device driver to DEVFS). This doesn't seem to me the logical place for geometry translations, but it wouldn't be the first time a name has ended up migrating to another purpose.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?52F0D705.1090402>