Re: problems with 2.4 kernel on oldworld mac
okay, i rebuit for the right processor this time :) still hangs at the same place. worse off than i was with the mandrake kernel and initrd image. i've been thinking about moving /boot over to the ide drive. is this all i would have to do (besides telling bootx where to boot from) to bypass this whole scsi issue? that way i could tell if it's really a scsi issue or something else with my machine. thanks again. -c - Original Message - From: Stew Benedict [EMAIL PROTECTED] To: [EMAIL PROTECTED] Sent: Wednesday, September 26, 2001 10:15 AM Subject: Re: problems with 2.4 kernel on oldworld mac On Wed, 26 Sep 2001, Chris Mann wrote: okay, i tried building my own kernel from the 2.4.4-6.4mdk source rpms. first i did a make config. i left everything as default except for the following scsi settings, which i changed to Y: SCSI disk support SCSI generic support MESH (Power Mac internal SCSI) support 53C94 (Power Mac external SCSI) support one thing i wasn't sure about was the Processor Type. You've got a choice between 6xx/7xx/74xx/8260, POWER3, POWR4, and 8xx. I went with POWER3, cause I'm assuming that means G3 and this machine has a G3 upgrade, but I'm not sure. Nope - should have checked the help: CONFIG_6xx: x x x x There are four types of PowerPC chips supported. The more common x x types (601, 603, 604, 740, 750, 7400), the Motorola embedded versions x x (821, 823, 850, 855, 860, 8260), the IBM embedded versions (403 and x x 405) and the high end 64 bit Power processors (Power 3, Power 4). x x Unless you are building a kernel for one of the embedded processor x x systems, or a 64 bit IBM RS/6000, choose 6xx. Note that the kernel x x runs in 32-bit mode even on 64-bit chips. Stick with 6xx/7xx/74xx/8260 Then make mrproper, make dep make clean, make vmlinux, make modules, make modules-install. (i also changed the extra version number for the kernel so it woulnd't overwrite teh mandrake version) moved the kernel and system map over to /boot (vmlinx-2.4.4-6.4scsi and System.map-2.4.4-6.4scsi) and copied the kernel over to the System Folder:Linux Kernels folder on the mac side. Select that kernel in BootX when rebooting and the machine hangs at this point: Welcome to Linux, kernel 2.4.4-6.4scsi started at: 0x linked at: 0xc000 frame buffer at: 0x81801400 (phys), 0x81801400 (log) klimit : 0xc02cbb30 MSR: 0x0042 HID0 :0x8c204 ICTC: 0x booting... which is worse than what i was getting with the default mandrake 2.4 kernel, but about what i got with the benh 2.4 kernel. does BootX take care of changing the symbolic links for System.map and vmlinux or should i do that myself. and also what should the file persmissions be on the System.map files. some are read write and others are just read. Well nothing should need to write to System.map, only read. The init scripts setup the symlink on a successful boot - rc.sysinit Try with the right processor type and see what happens. Stew Benedict -- MandrakeSoft OH/TN, USA http://perso.mandrakesoft.com/~sbenedict/ PPC Faq: http://marc.theaimsgroup.com/?l=mandrake-cooker-ppcm=99441208917647w=
Re: problems with 2.4 kernel on oldworld mac
yup, i tried both the bitkeeper and the benh kernels. both freeze after the booting... message. which makes me wonder if it's something with my machine. although as far as i can remember it's basically a 9500 or 9600 motherboard with umax written on the front. but since i have that ide card in there with a 27 and 80gb drive, the 4gb scsi drive isn't really necessary. -c
Re: problems with 2.4 kernel on oldworld mac
Do you see anything in the kernel messages referring to loading the ramdisk? You should not need anything too big, but probably double your actual ramdisk image wouldn't hurt. yeah, i see something about the ramdisk being initialized, but it scrolls off the screen pretty quick (is there anyway to turn off that big purple graphic?). it seems to be going thru the hardware initialization stage, and then hangs after recognizing the floppy driver. which is where it would normally see the scsi disks. Are you booting the 2.2 kernel with an initrd also? (I would think so since SCSI is not built in) nope, it's always worked without it. which is why i had no idea about the 2.4 kernel needing it. It's also possible that things are getting corrupted moving them from Linux to MacOS. You might want to consider building a kernel with integrated SCSI, rather than modules. this could make your life a little easier. i could try that. does that just involve downloading the header and source rpms and following the normal kernel building howto's? Can you show us: ls -l /boot total 20480 lrwxrwxrwx1 root root 21 Sep 25 07:36 System.map - System.map-2.2.20pre9* -rw-r--r--1 root root 246824 May 22 17:32 System.map-2.2.19-14mdk -rwxr-xr-x1 root root 268553 Aug 21 12:04 System.map-2.2.20pre9* -rwxr-xr-x1 root root 503902 Aug 29 01:01 System.map-2.4.10-pre2* -rw-r--r--1 root root 455451 Jun 27 21:32 System.map-2.4.4-6.2mdk -rw-r--r--1 root root 455451 Sep 17 10:58 System.map-2.4.4-6.4mdk -rwxrwxr-x1 root root 502484 Jul 2 07:46 System.map-2.4.5-pre5* lrwxrwxrwx1 root root 21 Sep 10 00:20 System.map.old - System.map-2.2.20pre9* -rw-r--r--1 root root28844 Jun 27 21:32 config-2.4.4-6.2mdk -rw-r--r--1 root root28948 Sep 17 10:58 config-2.4.4-6.4mdk -rw-r--r--1 root root 498121 Sep 24 15:39 initrd-2.4.4-6.4mdk -rw-r--r--1 root root 441 Sep 25 07:36 kernel.h -rwxr-xr-x1 root root 2631512 Aug 21 12:04 vmlinux-2.2.20pre9* -rwxr-xr-x1 root root 3685270 Sep 9 23:27 vmlinux-2.4.10-pre2* -rwxrwxr-x1 root root 3780829 Jul 2 07:46 vmlinux-2.4.5-pre5* lrwxrwxrwx1 root root 20 Sep 21 17:05 vmlinuz - vmlinuz-2.4.4-6.4mdk* lrwxrwxrwx1 root root 20 Sep 9 21:57 vmlinuz-2.2 - vmlinuz-2.2.19-14mdk* -rwxr-xr-x1 root root 2343956 May 22 17:39 vmlinuz-2.2.19-14mdk* -rwxr-xr-x1 root root 2720888 Sep 21 15:44 vmlinuz-2.4.4-6.2mdk* -rwxr-xr-x1 root root 2720888 Sep 17 11:20 vmlinuz-2.4.4-6.4mdk* A screen shot of the BootX setup. (some common format - I'd rather not have to boot MacOS just to see it ;^)) http://mrmann.com/linux/bootx-main.gif http://mrmann.com/linux/bootx-opt.gif Finally - have you tried the kernel/initrd image for 2.4 on my web page? yup. with that kernel and initrd i only get the following before the machine hangs: Welcome to Linux, kernel 2.4.4-6.2mdk started at: 0x linked at: 0xc000 frame buffer at: 0x81801400 (phys), 0x81801400 (log) klimit : 0xc02cbb30 MSR: 0x0042 HID0 :0x8c204 ICTC: 0x booting... pmac_init(): exit id mach():done MMU:enter MMU : hash init hash: enter hash: patch hash: done MMU:mapin MMU:setbat MMU:exit setup_arch: enter setup_arch: bootmem arch: exit with the rebuilt 2.4.4-6.4 kernel and my own initrd i get past this, but it hangs when it should be initializing the scsi disks. also the initrd on your webpage is only 4k whereas the initrd i made is 488k. which i thought was kinda odd. One last Q - do you recall if you were able to use the 2.4 kernel on the CD when you installed? i didn't try the 2.4 kernel when i installed. i had to use a different kernel than the one on the cd to get the installation to work the first time around. thanks -chris
Re: problems with 2.4 kernel on oldworld mac
On Tue, 25 Sep 2001, Stew Benedict wrote: I'm seemingly stumped then. I'm 99% sure there is no integrated SCSI support in the Mandrake 2.2 kernel, as I can see all the modules in /lib/modules. So somehow you are booting a SCSI / without the benefit of SCSI drivers. When you boot on 2.2 can you send the output of: okay, i rebooted into the 2.2.19 mandrake kernel. cat /proc/version Linux version 2.2.19-1k ([EMAIL PROTECTED]) (gcc version 2.95.3 20010111 (prerelease/franzo/20010111)) #1 Thu May 31 14:57:36 MDT 2001 df FilesystemSize Used Avail Use% Mounted on /dev/sda8 1.7G 533M 1.1G 32% / /dev/hde6 25G 83M 24G 0% /home /dev/hdg1 75G 43G 28G 60% /home/mp3 /sbin/lsmod Module Size Used by dmesg | grep scsi scsi0 : MESH scsi1 : 53C94 scsi : 2 hosts. Detected scsi CD-ROM sr0 at scsi0, channel 0, id 3, lun 0 Detected scsi disk sda at scsi1, channel 0, id 0, lun 0 scsi : detected 1 SCSI cdrom 1 SCSI disk total. I see lot of kernels there, not all mdk. Is the 2.2 you are successfully using a Mandrake one? If not, we can discard most of the previous discussion about initrd's. That only applies if you are using SCSI as a module. I do this because I'm trying to accomodate a multitude of machines. If I were you, for my personal machine, I would just build into the kernel the modules I needed. yeah, i tried a few to see if i could get them to work. with the 2.2 mandrake kernel i couldn't get sound to work. but i just noticed that when i boot into that i get a message about modules not being able to be loaded cause modules.dep can not be opened for writing. which would explain why /sbin/lsmod shows nothing. huh, i think i would have noticed that to begin with. i looked at that file and can't figure out why it can't be opened. it is owned by root and has read write privs. It could very well be that there is some basic issue with 2.4 and this machine as I see you've got a number of flavors of 2.4 kernels there. If you want to eliminate SCSI from the equation, build a kernel with no SCSI support at all, and then boot the rescue image and see if you can at least get to the command prompt. okay, i just downloaded the headers and the source, and i'll try this later today and let you know what happens. i've never built a kernel from scratch, any advice? In thinking about the installkernel script that should have run when you rpm -i the 2.4 kernel, as far as linux is concerned, you have no bootloader, since it all happens on the MacOS side, and that may be why the initrd was not created, as the script checks for the presence of a bootloader (yaboot, lilo, grub). aah, that makes sense. thanks again -c
Re: problems with 2.4 kernel on oldworld mac
BZZT - I don't think anyone is building Mandrake kernels at terraplex.com. This answers a lot for me - it looks like a YDL kernel with integrated mac53c94. If it were Mandrake - you would see something like grapefruit.mandrakesoft.com. huh, does the installer just install the kernel that you use to boot into it? that could explain it. i guess i just assumed it was the mandrake kernel. Nothing really special to note in building the kernel. The default config should already be there in the source tree, and you can modify it as you need to. okay, i made a new config file to include scsi support. did a make dep and make clean. when i try to make bzImage, i get an error that here are no rules for that target. i guess the README doesn't apply to building for ppc. what steps should i follow? thanks -chris
Re: problems with 2.4 kernel on oldworld mac
okay, i built the initrd and then transferred it over to the mac os partition. selected it with bootx, set video driver to off. the boot process halts right after detecting the floppy disk (where it would usually then detect the scsi bus). i get the error, Oops: kernel access of bad area. sig: 11 then a few lines of more info (mainly looks like machine code to me, except for Task=c0466000 [1] 'swapper' last syscall: 120 kernel panic: attempted to kill init! then i have to reboot. i left the ramdisk size as the default in bootx, should i increase that? thanks for all the help. -chris - Original Message - From: Stew Benedict [EMAIL PROTECTED] To: [EMAIL PROTECTED] Sent: Monday, September 24, 2001 4:02 PM Subject: Re: problems with 2.4 kernel on oldworld mac On Mon, 24 Sep 2001, Chris Mann wrote: Okay, I see. But I'm on an oldworld mac, so I'm using BootX. Do i need to make this ram disk and then move it over to the mac side of the world so i can then tell bootx to load from this ram disk initially? oh, and the rpm -i command didn't seem to make a initrd for me. Yes you do - set it up in BootX like a ramdisk for the installer. Stew Benedict -- MandrakeSoft OH/TN, USA http://perso.mandrakesoft.com/~sbenedict/ PPC Faq: http://marc.theaimsgroup.com/?l=mandrake-cooker-ppcm=99441208917647w=
weird disk problems
Hi, I'm trying to install a new maxtor 80gig drive in my umax j700 with a turbomax card runnning mandrake 8.0, but i'm having some problems that i can't figure out. was hoping somewhere here might have some ideas. first i tried diskdrake, but i got an illegal division by zero error whenever i tried to format that drive. so i switched to fdisk and besides the fact that it mentioned something about 9964 cylinders instead of 4096, everything seemed to work. set one primary partition and that was it. formatted with mke2fs and everything seemed fine (although i wonder how necessary it is to reserve 5% of the disk space for the superuser. on an 80GB drive, this is a big chunk?). but disk performance seemed fairly slow, so i ran hdparm to see what the settings on the disk where and this is what i get: []# hdparm /dev/hdg /dev/hdg1: multcount = 0 (off) I/O support = 0 (default 16-bit) unmaskirq = 0 (off) using_dma = 1 (on) keepsettings = 0 (off) nowerr = 0 (off) readonly = 0 (off) readahead = 8 (on) HDIO_GETGEO_BIG failed: Input/output error []# hdparm -i /dev/hdg /dev/hdg: Model=Maxtor 4W080H6, FwRev=AAH41310, SerialNo-W6043SXC Config={ Fixed } RawCHS=16383/16/63, TrkSize=0, SectSize=0, ECCbytes=57 BuffType=DualPortCache, BuffSize=2048kB, MaxMultSec=16, MultSect=off CurCHS=16383/16/63, CurSects=-66060037, LBA=yes, LBAsects=160086528 IORDY=on/off, tPIO={min:120,w/IORDY:120}, tDMA={min:120,rec:120} PIO modes: pio0 pio1 pio2 pio3 pio4 DMA modes: mdma0 mdma1 mdma2 udma0 udma1 *udma2 udma3 udma4 udma5 AdvancedPM=yes: disabled (255) Drive Supports: ATA/ATAPI-6 T13 1410D revision 0 : ATA-1 ATA-2 ATA-3 ATA-4 ATA-5 ATA-6 i'm not quite sure why it can't get the geometry of the drive. when i run this on the other drive attached to the turbomax card, i have the same problem. trying to run it on the scsi disk is even worse. so i'm wondering if there's something wrong with hdparm or what? also, now when i run diskdrake just to check out the drive, it says that i have two paritions, one 530 GB and one 1TB. which is obviously not right. this is the partition table from fdisk Disk /dev/hdg: 255 heads, 63 sectors, 9964 cylinders Units = cylinders of 16065 * 512 bytes Device Boot Start End Blocks Id System /dev/hdg1 1 9964 80035798+ 83 Linux like i said, the drive seems to be working okay, just slow. but all this weird info kinda worries me. should i just ignore it all? thanks -chris