From owner-freebsd-questions Wed Oct 4 13:46:37 2000 Delivered-To: freebsd-questions@freebsd.org Received: from femail2.sdc1.sfba.home.com (femail2.sdc1.sfba.home.com [24.0.95.82]) by hub.freebsd.org (Postfix) with ESMTP id 0D01537B66D for ; Wed, 4 Oct 2000 13:46:30 -0700 (PDT) Received: from home.com ([24.20.70.64]) by femail2.sdc1.sfba.home.com (InterMail vM.4.01.03.00 201-229-121) with ESMTP id <20001004204622.RIJY27630.femail2.sdc1.sfba.home.com@home.com>; Wed, 4 Oct 2000 13:46:22 -0700 Message-ID: <39DB97C3.81D20E85@home.com> Date: Wed, 04 Oct 2000 13:49:07 -0700 From: xavian anderson macpherson Organization: http://www.professional3d.com X-Mailer: Mozilla 4.72 [en] (X11; I; Linux 2.2.14-SMP i586) X-Accept-Language: en MIME-Version: 1.0 To: "freebsd-questions@FreeBSD.ORG" , tagdot57@aol.com, mongor@mail.com Subject: is the loader process of the 5.0-current toolkit different from the 4.0 four-disk set? Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-freebsd-questions@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG i have tried booting using the kern.flp and mfsroot.flp from the 5.0-current (june 2000) on the first cd of the 6-cd toolkit set. i am not even able to get beyond the device selection in the installation process. i cannot boot from my scsi-cdrom, the controller is non-bootable. the program leaves the selection point and tries to initialize the aic0 modules for my SOUNDBLASTER 16 (CT 1770 from compaq) SCSI card. it just loops (aimlessly spinning the scsi-cdrom which i need to load the system; i geuss) trying to determine the parameters for the device. (there was a box in the configuration for that device that was labeled `flag'; i set the entry there to 0x340, which is the address of my scsi card; there was no entry for irq; if this `flag' is not for the address [why the 0x0 then], what is it for?). even prior to this (the point where it is attempting to read the cdrom), i noticed that it was unable to ascertain the parameters of the pnp-devices. there is a module (under miscellaneous) called pc-card controller. it conflicts with the irq (10) of my NE2000 (REALTEK RTL-8029) ethernet card. i thought i needed this to read my pci+isa cards; i just found this is really for pcmcia. but i do remember trying to continue without this, but it still didn't work! so i removed the ethernet card module in desparation, thinking that if i could only get the system installed, then i could then go back later and deal with web communications. is ex0 the module for my intel ethernet express card? nice plan! except that it does not work! i chose the 5.0-current, because according to someone from this list, it is the only version that allows me to have the kernel (with vinum support) and all of it's components in the /boot directory. i felt i would need this ability to mount /boot with the kernel and all of it's modules, configs, (etc) onto it's own partition as i do with linux. yes, to the person that commented about the 1024 cylinder limit; i am using lilo, because my current `/' is on hdd1 (or wd3s1). in fact i had two (2) `/'s. one on wd3s1 (for linux-mandrake-7.0 with it's 1750 rpms) and the other for suse-6.4 (and it's 465 rpms; which i am still using on wd3s1). i will quote that message below. as a last resort, can i just copy sysinstall(which may very well run in linux) to my harddisk, to format my disks and load freebsd from linux? i just tried and succeeded in formatting and mounting /dev/hdd4(linux) or /dev/wd3s4(freebsd), using YaST(a suse-linux program) to type b7(which is BSDI fs). i have several options (using YaST) which i can format my disks as; 86 NTFS Volume set 87 NTFS Volume set 8E Linux LVM a5* OpenBSD b7* BSDI fs b8* BSDI swap fd Linux raid auto 63 GNU HURD o SysV 64* Novell Netware 65* Novell Netware 3c Partition Magic 7 HPFS/NTFS the partition types followed by an asterisk seem like they would be the most compatable of the group. i am geussing that types 63 and 64 are the NFS types mentioned in ======================================================================= JONATHAN WRITES, No. The location of the kernel is arbitrary, but must be located where the loader can find it. Since the loader doesn't understand vinum filesystems, this is why you can't put the kernel on a vinum device. ----------------------------------------------------------------------- XAVIAN WRITES; lilo (linux loader) for linux does not have this problem (it also supports booting any number of os's; and with VMware, you can run them concurrently). with lilo i can mount the `/' anywhere i want it (as long as /boot is it's own partition). furthermore, i do not want to mount the kernel on vinum. i want to put the kernel in /boot which (again) would then be mounted on it's own partition. that partition would not be on a vinum device. if the kernel is in /boot and mounted on it's own (slice wd0s1) partition, the kernel would (then) mount first (along with it's vinum support components) before the `/' device is even mounted. thereby removing the `chicken and the egg' situation that someone else refered to by the `/' not being able to mount, because it is on a device that cannot be (defined, and subsequently) accessed by a kernel that has not yet been loaded. let me also state that i need this installation to be a shared os environment. if i had ERASED! my disk, so that i would have a freebsd-only environment, i would have had to RELOAD! all of my linux AGAIN!, just so that i could send this email to get help. just imagine how any of you would have felt under the same circumstances (and how much SHOUTING! you would be doing). as i stated (very strongly) to the people in linux. it seems very ludicrous to talk about superiority, when you can't even load the new environment without absolutely ANY! dependency on microsoft or any other operating system to get (environment parameters) information on the current environment to load the new one. dependency is not superiority! if you think that is unrealistic, how UNREALISTIC! is it to declare anything other than MICROSOFT superior? and don't bother talking about my tone. if all i had to do was insert the (boot) disks (which you don't provide, but linux does) and start the system (as i do with linux), there would be no tone! you can't even start the system without first depending on MICROSOFT (which i don't have and don't want; although now, i am beginning to wonder why?) or someone else to make the boot floppies. how's that for a `chicken and the egg" convolution? ----------------------------------------------------------------------- HE CONTINUES, In 5.x, the kernel and modules have been moved into a subdirectory of /boot, so on -current, we now have the following structure: /boot/kernel.GENERIC/kernel /vesa.ko /kernel.MACHINE/kernel /vinum.ko If you load the kernel + vinum module at boot time, I don't see any technical reason why you can't mount a vinum partition as /, and then overlay your /boot partition on top of that. ------------------------------------------------------------------------ XAVIAN WRITES, that's what i was trying to say. he has essentially quoted (verbatum) the very essense of my proposal. but he (or anyone else) still does not see the solution in front of him. mount /boot with the kernel on it's own slice (in `wd or sd0s1'). i am beginning to feel as though i brought someone else's idea of a great car(aka, ferrari). only to find out that i actually joined a garage of mechanics (however well equiped or intentioned), who tell me that i first have to install the engine (with chevy's tools) before i can drive it! not only that, but i have to search around for (rare, and often incomplete) components to keep it on the road once(if) i actually get it running! ferrari's are great. but at least you can drive them off the showroom floor (without borrowing parts from a chevy). ------------------------------------------------------------------------ HE CONTINUES, However, in practical terms, mounting root is a "special" operation, and must have per-filesystem support, so the actual code to support mounting the vinum partition as / may not be present. -- Jonathan (ends) -------------------------------------------------------------------- XAVIAN WRITES, JONATHAN, I AM SORRY. this is not directed at you personally. i am just experiencing de je vu, all over again. i also went through similar troubles with linux. however, if you use my (linux) approach, you don't need any "special" operations or `actual code' to mount the vinum device/partition/slice as `/'. but linux is (still a `bleeding edge developement environment; which is) inferior and freebsd isn't. some(any)thing so essential (as installetion) to an advanced system like freebsd, should not be so difficult to use! if you want to get a real sense of the simplicity that i LOATH! departing from, download a copy of YaST from ftp://ftp.suse.com/pub/suse/i386/update/7.0/a1/yast-1.07.1-11.i386.rpm [for the rpm] ftp://ftp.suse.com/pub/suse/i386/update/7.0/zq1/yast-1.07.1-11.src.rpm [for the source] and see for yourselves how SIMPLE! and comprehensive it is to use! if you think it is so beneath you to do so, you have no basis to criticize me! at least take a look at it to get an idea of what i am talking about! i was able to run this on my linux-mandrake-7.0 `/', which is almost completely incompatable with the suse-6.4 `/'. so it should (hopefully run in any current elf `/' that supports redhat rpms. ==================================================================================== To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-questions" in the body of the message