Latest working kernel version:? Earliest failing kernel version:2.6.27 Distribution:Gentoo 2008.0/desktop Hardware Environment: Athlon XP 2600+, 1GiB DDR RAM Software Environment: gcc-4.1.2, glibc-2.6.1-r0, linux 2.6.27.7, linux-headers- 2.6.27-r2, libtool-1.5.26, bash-3.2_p33, python-2.5.2-r7 Problem Description: this warning, when compiled with make CONFIG_DEBUG_SECTION_MISMATCH=y [...] MODPOST vmlinux.o WARNING: vmlinux.o(.cpuinit.data+0x0): Section mismatch in reference from the variable initial_code to the function .init.text:i386_start_kernel() The variable __cpuinitdata initial_code references a function __init i386_start_kernel(). If i386_start_kernel is only used by initial_code then annotate i386_start_kernel with a matching annotation. GEN .version [...] Steps to reproduce: 1) Enter a vanilla source directory of version 2.6.27, 2.6.27.6 or 2.6.27.7 (these were the versions I tested, but obviously others may be affected) 2) generate configuration. Optionally, use the attached stripped-to-the-minimum .config 3) make CONFIG_DEBUG_SECTION_MISMATCH=y I will attach a .config that triggers this bug (named minimal-config), and also the config of the running kernel (named as host-config)
Created attachment 18968 [details] A stripped-to-the-minimal config that triggers this bug
Created attachment 18969 [details] config of the kernel running on the system host-config
I have this warning on gentoo-sources kernel 2.6.28-r1
This warning happens with 2.6.30-r1 too... (Gentoo-Sources)
Happens too on 2.6.31-rc6-git6 MODPOST vmlinux.o WARNING: vmlinux.o(.cpuinit.data+0x0): Section mismatch in reference from the variable initial_code to the function .init.text:i386_start_kernel() The variable __cpuinitdata initial_code references a function __init i386_start_kernel(). If i386_start_kernel is only used by initial_code then annotate i386_start_kernel with a matching annotation.
Created attachment 22802 [details] linux-2.6.31-rc6-git6/.config
Please, do not file compile failures, link failures or section mismatches into bugzilla. Such errors come and go, fixed pretty quickly and bugzilla is simply to far away from people who will fix them. Believe me, developers know about them and they know about i386_start_kernel(). It's likely a false positive.