Return to BSD News archive
Path: sserve!manuel!munnari.oz.au!samsung!sdd.hp.com!mips!newsun!gateway.novell.com!terry From: terry@npd.Novell.COM (Terry Lambert) Newsgroups: comp.unix.bsd Subject: Re: HELP!!! Keywords: Keyboard, 386BSD, help, problem, boot Message-ID: <1992Jun23.224426.27527@gateway.novell.com> Date: 23 Jun 92 22:44:26 GMT References: <1992Jun23.171404.6463@lut.fi> Sender: news@gateway.novell.com (NetNews) Organization: Novell NPD -- Sandy, UT Lines: 48 Nntp-Posting-Host: thisbe.eng.sandy.novell.com In article <1992Jun23.171404.6463@lut.fi> ats@lut.fi (Ari Suutari) writes: >Hi everybody! > >I'm trying to get 386BSD work on my computer, but I can't get past >the boot stage. >Problem: the system apparently reads my keyboard wrong and decides that >I want to do an 'Operator abort -- booting 386bsd.alt' and because none >of the other files are found, it goes aroung in circles for a while and >then dies. >My system is a 33-MHz 486DX with 4MB of RAM and 129MB of ATID hard disk. >I have a standard keyboard (at least I thought so) and the controller >is fairly standard -- it has not given any trouble in the DOS world. >I'm booting from a 1.44Mb diskette with 386BSD v0.0, not patched >in any way (I think). This problem is *common*, from what I've seen; therefore, against the advice of the poster, I'm going to discuss it here, in the hopes that the soloution will be incorporated into the next release. I will, of course, mail this off to Tuomas, as well. The problem is the time latency on a keyboard controller reset. The boot code resets the keyboard, even though the ROM bios has already done so. This reset makes absolutely no sense, as one must turn ones computer on in order to boot an OS, and the BIOS reset runs prior to the disk bootstrap; is there an issue of which I am unaware, requiring a boot-code reset here? I have personally gotten around the problem by cross-compiling all of the BSD sources on a Sun SparcServer 2, and then binarily patching the offending code to NOP's before installing (to get around the problem without changing sources; as I said, I might be missing the "why" for this code). I have been holding off ftp'ing up my boot disks in anticipation of 0.1, and because the way it's set up, the source does not match the objects generated. Is this issue resolved in 0.1? I hate to duplicate real bugfix effort, and I hate to kill code which must have been provided for some reason, even if I can't see it, probably to support Bob's weird hardware from Europe... can't go breaking someone elses box to fix my own. There is a lot of VFS and other work I have done that I would like to roll in, including an implementation of a pseudo-/proc file system and batch support in the scheduler, as well as some minor but rather architecture dependant utilities; is 0.1 still immanently pending? Terry Lambert terry_lambert@gateway.novell.com terry@icarus.weber.edu --- Disclaimer: Any opinions in this posting are my own and not those of my present or previous employers.