Bug 10679
| Summary: | Module dependencies fail after recompilation of kernel | ||
|---|---|---|---|
| Product: | [Retired] Red Hat Linux | Reporter: | boost |
| Component: | kernel | Assignee: | Michael K. Johnson <johnsonm> |
| Status: | CLOSED NOTABUG | QA Contact: | |
| Severity: | medium | Docs Contact: | |
| Priority: | medium | ||
| Version: | 6.2 | CC: | liondatasystems, sabin |
| Target Milestone: | --- | ||
| Target Release: | --- | ||
| Hardware: | i386 | ||
| OS: | Linux | ||
| Whiteboard: | |||
| Fixed In Version: | Doc Type: | Bug Fix | |
| Doc Text: | Story Points: | --- | |
| Clone Of: | Environment: | ||
| Last Closed: | 2000-04-22 06:46:31 UTC | Type: | --- |
| Regression: | --- | Mount Type: | --- |
| Documentation: | --- | CRM: | |
| Verified Versions: | Category: | --- | |
| oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |
| Cloudforms Team: | --- | Target Upstream Version: | |
| Embargoed: | |||
|
Description
boost
2000-04-09 13:22:37 UTC
I too have had a problem with this. I had to eventually reformat and reinstall redhat 6.1 I can confirm this problem. I figured I goofed so very carefully deleted the 2.2.14-6.0.1 source and /lib/modules trees, reloaded the source, did the usual make mrproper xconfig dep clean bzImage modules modules_install. Then I copied the new kernel and System.map to /boot with appropriate names, ran lilo and rebooted. Same problem. I can confirm this problem. I figured I goofed so very carefully deleted the 2.2.14-6.0.1 source and /lib/modules trees, reloaded the source, did the usual make mrproper xconfig dep clean bzImage modules modules_install. Then I copied the new kernel and System.map to /boot with appropriate names, ran lilo and rebooted. Same problem. Did you remember to remake the initrd image in the /boot directory with the new modules before rerunning lilo and rebooting the system? If you don't perform that step, then the initrd image will attempt to load modules from the stock kernel instead of your newly compiled kernel and they will fail as you describe. As an additional item, when changing kernel options and recompiling, you very well may need to run the command: rm /usr/src/linux/include/linux/modules/* in order to force the make process to rebuild the symbol versions with the new values. I'm closing this bug since if you remove the files as indicated above and add the remaking of the initrd image to the other steps listed above then you should have a working kernel. I would also suggest that you make sure to use a different kernel version string than the default one. I typically add -custom to the end of the EXTRAVERSION string in the file /usr/src/linux/Makefile I didn't USE an initrd image...disabled the initial ramdisk support, deleted /usr/src/linux, untarred the kernel source there, moved modules 2.2.14-whatever to 2.2.14-whatever-old and then recompiled with version symbols. I can do this in redhat 6.1 with absolutely no problem, BUT on RedHat 6.2 it would fail not on compile, not with running lilo, but on boot. I've ran the EXACT same kernel config script and exact same steps in redhat 6.1 and produced a perfect working kernel. I have had the same experience. When compiling the kernel from the RedHat 6.2 installation CD, I cannot compile a kernel without initrd support anymore, whereas this is perfectly possible with the standard 2.2.14 kernel. What did RedHat do that initrd is now mandatory? Man, I wonder that the problems really were. If anyone is still looking at this thread, what errors do you get on modprobe/insmod? I get unresolved symbols for best_memcopy, best_memset, and a couple of others - all the same .c file - best_config.c (I think - I'm not at home just at the moment). Seems to me the problem is in make xconfig not setting CONFIG_X86_CPU_OPTIMIZATIONS. I anyone knows whether initrd is really required for correct module symbol resolution, I'd like to hear about it. (e-mail OK) Thanks, jed |