Return to BSD News archive
Newsgroups: comp.os.386bsd.bugs Path: sserve!newshost.anu.edu.au!munnari.oz.au!news.Hawaii.Edu!ames!saimiri.primate.wisc.edu!usenet.coe.montana.edu!nate From: nate@cs.montana.edu (Nate Williams) Subject: Re: patch 0.2.3 BUGS ? Message-ID: <1993Apr29.050748.12854@coe.montana.edu> Sender: usenet@coe.montana.edu (USENET News System) Organization: CS References: <1993Apr29.031107.3815@sparc4.ncu.edu.tw> Date: Thu, 29 Apr 1993 05:07:48 GMT Lines: 39 > The Patchkit 0.2.3 exist 2 serious bugs : > > 1. It eat too much memory, I have tried to do the "buildworld.sh" in many > machines, but if the RAM is less than 16MB, the process will hang > forever, even block the system. I have tried it on 3 386 machine (-25, > -33) with 4 or 8 MB ram, no one survived. But if I add the RAM to > 16MB, all successful!!! > > because the the process hang in making the lib or nroff the man page. > I guess the compiler or nroff catch too much memory Nroff needs to be compiled, and then it will work better. Groff had a bug in it that never allowed ANY programs to free up memory that were compiled by it. In the patchkit (0.2), this was fixed. Try it again, after everything has been done. > > 2. the new /dev/MAKEFILE seems buggy, if old device exists, for example > com0, the new will be com00. And the minor numvber for VCONSOLE > start from com0, not com1, we should change if by mknod. > That is where they should be. :-) That is the standard way to do things under Unix. Bill changed it to be more 'dos-like', and the reason for com00 is so that you can have more than 10 com ports (multi-port cards). Both of these are from stock 0.1, and have nothing to do with the patchkit. As a matter of fact, the patchkit is fixing at least one of the problems you reported. Nate -- osynw@terra.oscs.montana.edu | Still trying to find a good reason for nate@cs.montana.edu | these 'computer' things. Personally, work #: (406) 994-4836 | I don't think they'll catch on - home #: (406) 586-0579 | Don Hammerstrom