On Fri, Apr 22, 2005 at 10:20:36AM -0700, Nathanael Noblet wrote: [..FILO..] > >Indeed, this is another option. But a new kernel is still neccessary > >since the one in the ELF image (/boot/initrd-2.4.20-8smp.img) came > >with neither ext3 nor IDE drivers. > > Not so, according to his first email, the linux install (RH 9 I think) > booted fine from the local system prior to LinuxBIOS installation.
I was thinking that Red Hat might always use some kind of initrd that loads modules and without it those drivers had to be in the kernel. I actually don't remember whether FILO can deal with initrd:s as well. > Obviously LinuxBIOS seems to be working, unless his problem is that the > IDE controller is not detected or interrupts are off (not right). This could very well be the problem, Beneo, could you post the complete output you get from starting the system with LinuxBIOS? > Thus the installed kernel should be able to boot again without > modification using FILO. If FILO does initrd:s, yep. On Fri, Apr 22, 2005 at 10:40:32AM -0700, beneo wrote: > I have one more question on the comment Peter made, > >Peter Wrote: > > Indeed, this is another option. But a new kernel is still neccessary > > since the one in the ELF image (/boot/initrd-2.4.20-8smp.img) came > > with neither ext3 nor IDE drivers. > > My question is, if initd-2.4.20-8smp.img doesn't understand ext3 and > don't have IDE driver, why I can boot everything from HD using AMI > BIOS? The initd-2.4.20-8smp.img is the same for both LinuxBIOS boot > and AMI BIOS boot. Sorry, I copied-and-pasted the wrong filename, it should have been the kernel file /boot/vmlinuz-2.4.20-8smp, not the initrd file. The initrd has modules for ext3 and the IDE driver, they aren't in the kernel. I guess both drivers load with AMI BIOS but the IDE driver doesn't load with LinuxBIOS. This actually indicates that the IDE controller wasn't initialized properly by LinuxBIOS, just as Nathanael wrote. Try compiling a fresh kernel with support for your IDE chipset, make an ELF image out of it, boot the target system and send all of the output from the boot to the list. I'm afraid any IDE driver messages that could help find the problem will get lost if an initrd tries to load the driver as a module. > I assume after ramdisk loaded, Kernel would rely on Linux IDE > driver and no longer make any BIOS hard drive service call, right? Yep, that's correct. But the IDE driver can be included in the kernel from the beginning too, that way it will try to load much sooner and you'll be certain not to miss any messages from the driver being caught and hidden by klogd on the initrd. //Peter _______________________________________________ Linuxbios mailing list Linuxbios@clustermatic.org http://www.clustermatic.org/mailman/listinfo/linuxbios
