Date: Tue, 20 Aug 1996 15:51:22 -0500 From: Nathan Denny <SCHCATS@siu.edu> To: questions@freebsd.org Cc: hackers@freebsd.org, install@freebsd.org, bugs@freebsd.org Subject: Disk geometry problems. Message-ID: <321A254A.414@siu.edu>
next in thread | raw e-mail | index | archive | help
I tried a floppy and DOS partition installation of FreeBSD 2.1.0-RELEASE and 2.1.5-RELEASE to each of a 386-40 w/2 IDE 115MB disks, a 486DLC-40 w/1 850MB IDE disk, and a Pentium-100 w/1 1.2GB IDE disk. Each time, I got the warning Calculated sectors percyclinder (xxxx) does not agree ... or something and the installation crashed with a written -1 of 512 bytes, invalid gzip, etc. It is not the data or the media. I've tried two versions, downloaded 10 times, from 3 sites, so at least one of those installations should have worked. In each case, I was dedicating the entire disk(s) to FreeBSD (except the DOS partition installation!). It seems to me that the installation program reads the data, and caches it in core memory. When the buffer is full it flushes it to the disk. However, since the calculated geometry is wrong, it tries to write it to some unknown destination and thppt...crash! In each case the calculated geometry was GROSSLY miscalculated. (My 115MB disk under the calculated geometry of 4096 sectors/track would be like a 2GB+ disk!) Is there any way that I can FORCE the installation to the correct geometry. I've entered the CORRECT geometry under the partition part, with and without LBA (ie. the TRUE geometry and the translated geometry) and it seems that in both cases the net effect is the same. How does FreeBSD get such a wild geometry? It seems to detect the correct geometry at boot and partition parts, but when it creates the file system it's 100%+ wrong. Please help! Nate.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?321A254A.414>