From Bugzilla Helper: User-Agent: Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.0) Description of problem: In a Primergy RX800/IBM x445 with 2 CECs, 8 CPUs and 8 GB memory running Red Hat Enterprise AS 2.1 with latest errata kernel 2.4.9-e.25summit the file /boot/kernel.h has wrong defines. __BOOT_KERNEL_SMP is 0, should be 1 __BOOT_KERNEL_UP is 1, should be 0 Packages that need to be compiled during installation (e.g. FSC's ServerView eecd daemon srvmagt-eecd-3.00-14.redhat.rpm) and that rely on correct defines in /boot/kernel.h fail to compile. The same incorrect defines were also seen in e.16summit. /boot/kernel.h: #ifndef __BOOT_KERNEL_H_ #define __BOOT_KERNEL_H_ /* Kernel type i686 */ #ifndef __MODULE_KERNEL_i686 #define __MODULE_KERNEL_i686 1 #endif #ifndef __BOOT_KERNEL_ENTERPRISE #define __BOOT_KERNEL_ENTERPRISE 0 #endif #ifndef __BOOT_KERNEL_SMP #define __BOOT_KERNEL_SMP 0 #endif #ifndef __BOOT_KERNEL_UP #define __BOOT_KERNEL_UP 1 #endif #endif Version-Release number of selected component (if applicable): How reproducible: Always Steps to Reproduce: 1. Boot your x445/RX800 2. Have a look at /boot/kernel.h 3. --- Actual Results: /boot/kernel.h has wrong defines. __BOOT_KERNEL_UP 1 tells us we have a uniprocessor machine which is definitely wrong. Expected Results: __BOOT_KERNEL_UP should be 0 __BOOT_KERNEL_SMP should be 1 Additional info:
Actual Results: /boot/kernel.h has wrong defines. __BOOT_KERNEL_UP 1 tells us we have a uniprocessor machine which is definitely wrong. __BOOT_KERNEL_UP does NOT say you have an uniprocessor machine, just as __BOOT_KERNEL_SMP doesn't tell you you have a smp machine. __BOOT_KERNEL_SMP should ABSOLUTELY not be 1 for the summit kernel, since it means you are running the kernel-smp kernel, which you are not. For example, if you run the enterprise kernel (which is SMP) this is also 0. What the real bug is is that there is no __BOOT_KERNEL_SUMMIT in this file, while there should be one. (not that you can build external modules against the summit kernel right now so I fail to see how this can actually break something).
Some software use the kernel.h file to install modules in the appropriate places, such as Intel's profiling software (vtune). J
1) kernel.h is generated by initscripts not the kernel 2) summit kernels you can't build external modules against directly, esp not in the e.25 era (old!)
Closing; re-running mkkerneldoth should solve the issue.