As near as I have been able to figure out, anaconda detects the hardware on which it is loading, and looks for rpms of kernels which it can load, and determines the most advanced kernel it can use and loads it. The distribution CD contains several rpm's with pre-compiled kernels which anaconda can choose from. I _think_ that the way to make those rpm's is to unpack the kernel source rpm, run rpm -ba on it, and look for the completed rpm's in /usr/src/redhat/RPMS. I tried this with the 2.4.2 kernel source rpm from RH 7.1, and it seemed to work. Except, it only made 386 kernels, not 586 or 686. (The source kernel rpm I started with was kernel-2.4.2-2.src.rpm from the RH 7.1 source CD.) Yes, I know, I can look through the kernel source spec file and find the answer, but does anyone know how to get that spec file to also build 586 or 686 kernels? Thanks, --Seth sethal@xxxxxxxxx --- Dan Peterson <pete@xxxxxxx> wrote: > How/when is the post-install kernel generated? Or, > is there a generic > pre-compiled kernel somewhere on the installation > that's used for all > platforms? > > If it's a generic kernel used for all platforms, is > there a .config > file somewhere that was used to generate this > kernel? > > I have a special box with limited devices and low > memory: > > Cyrix CPU (pentium 166) > 32MB memory (some models only 16MB or 8MB) > 3-5GB disk > ethernet card with boot ROM > VGA > no CD > no floppy > no mouse > no battery for clock > > I've converted the RH7.1 bootnet image into a bootp > image and I'm able to > start the installation process using bootp. My > ks.cfg file does a clean > install using nfs and everything works fine. > > The problem comes when I reboot the box for the > first time (at the end of > the install) and boot the kernel. It get's to the > point where > /etc/rc.d/rc.sysinit is running, just after mounting > /proc then hangs. > I've tracked it down to where it's running > /sbin/hwclock. > > Rebooting with "linux init=/bin/bash" let me get in > and manually run > /sbin/hwclock with --debug and it appears the RTC > never responds. > > I've discovered that if I rebuild the kernel with > just the right settings > (and I haven't tracked down all the right settings > yet), I can get it to > boot just fine. > > When I started to compare stuff between my custom > kernel and the > post-install kernel, I noticed the RTC driver is > missing in my custom > kernel. Running /sbin/hwclock ... --debug shows: > > hwclock 2.4c/util-linux-2.10s > hwclock: Open of /dev/rtc failed, errno=19: No > such device. > Using direct I/O instructions to ISA clock. > Assuming hardware clock is kept in UTC time. > Waiting for clock tick... > ... got clock tick > <and continues fine and does it's job> > > Doing "cat /proc/driver/rtc" fails with: > > cat: /proc/driver/rtc: No such file or directory > > Which makes sense since the driver is not available. > > Running /sbin/hwclock ... --debug on the > post-install kernel shows: > > hwclock 2.4c/util-linux-2.10s > Using /dev/rtc interface to clock. > Assuming hardware clock is kept in UTC time. > Waiting for clock tick... > > At this point it just hangs. > > Doing "cat /proc/driver/rtc" shows: > > rtc_time : 00:03:09 > rtc_date : 1988-01-01 > rtc_epoch : 1900 > alarm : 00:00:00 > DST_enable : yes > BCD : yes > 24hr : yes > square_wave : no > alarm_IRQ : no > update_IRQ : no > periodic_IRQ : no > periodic_freq : 1024 > batt_status : dead > > All this is a bit beside the point, which is, how > can I generate a custom > kernel which is used the first time the box is > booted after the install?? > > And, is there a .config file associated with the > post-install kernel? > > > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list@xxxxxxxxxx > https://listman.redhat.com/mailman/listinfo/kickstart-list __________________________________________________ Do You Yahoo!? Get personalized email addresses from Yahoo! Mail - only $35 a year! http://personal.mail.yahoo.com/
