From owner-freebsd-bugs Sat Nov 11 11:42:43 1995 Return-Path: owner-bugs Received: (from root@localhost) by freefall.freebsd.org (8.6.12/8.6.6) id LAA24852 for bugs-outgoing; Sat, 11 Nov 1995 11:42:43 -0800 Received: from godzilla.zeta.org.au (godzilla.zeta.org.au [203.2.228.19]) by freefall.freebsd.org (8.6.12/8.6.6) with ESMTP id LAA24847 ; Sat, 11 Nov 1995 11:42:36 -0800 Received: (from bde@localhost) by godzilla.zeta.org.au (8.6.9/8.6.9) id GAA07419; Sun, 12 Nov 1995 06:41:44 +1100 Date: Sun, 12 Nov 1995 06:41:44 +1100 From: Bruce Evans Message-Id: <199511111941.GAA07419@godzilla.zeta.org.au> To: bugs@freebsd.org, gtc@aloft.att.com, jkh@freebsd.org Subject: Re: panic: vm_bounce_alloc during initial installation (writing disk info) Sender: owner-bugs@freebsd.org Precedence: bulk >(10/26 if I recall). When trying to perform the initial installation on my >machine, after going through all the menus to partition and label the disk, >when I go to commit the changes to disk, I immediately get the following error >message (as seen on the debug screen): >vm_bounce_alloc: b_bufsize(0x200) < b_bcount(0x1ff0a400) !! This might be caused by a negative block number somehow being passed to dscheck(). If this happens, then "truncating" the i/o would actually expand the i/o to the size of the partition plus the absolute value of the negative block number. Do you have a (BSD) partition of size approximately 0x1ff0a400 bytes (500+ MB)? Negative block numbers are used by the file system code but are not supposed to reach device drivers. >My environment is: > Pentium 90 (AMI Titan II EISA/PCI motherboard, one CPU installed) > BT946C BusLogic PCI bus-mastering SCSI controller >... >One of the problems may be that bounce buffers shouldn't really be used on >my machine, given that I've got a 32-bit bus-mastering SCSI controller (this >is probably related to kernel problem #378). But even if it does want to Bounce buffers are used because the GENERIC config specifies them and the `bt' driver handles all BusLogic devices from isa through pci using much the same code - it requests bounce buffer support for all buses. Bruce