Guys, I believe this is caused by kexec which changes the init calls. I fixed this in our tree (which had kexec) around 2.6.10 for i386 but kexec didn't go in kernel.org until 2.6.12 or so..
Take a look at this patch for an idea to solve it. Source: MontaVista Software, Inc. MR: 12703 Type: Defect Fix Disposition: needs submitting to [email protected] (mailing list) Signed-off-by: Steven Dake <[EMAIL PROTECTED]> Description: The KDB init call section is not properly setup in the linker script. This patch sets up the init call linker start location so that the object tools can build the kernel with KDB modules turned on. Index: linux-2.6.10/arch/i386/kernel/vmlinux.lds.S =================================================================== --- linux-2.6.10.orig/arch/i386/kernel/vmlinux.lds.S +++ linux-2.6.10/arch/i386/kernel/vmlinux.lds.S @@ -91,7 +91,9 @@ } __con_initcall_end = .; __kdb_initcall_start = .; - .kdb_initcall.init : { *(.kdb_initcall.init) } + .kdb_initcall.init : AT(ADDR(.kdb_initcall.init) - LOAD_OFFSET) { + *(.kdb_initcall.init) + } __kdb_initcall_end = .; SECURITY_INIT On Wed, 2005-08-31 at 11:33 -0700, Jack F Vogel wrote: > On Wed, Aug 31, 2005 at 03:48:13PM -0500, Ray Bryant wrote: > > > > If I turn off kdb_modules, 2.6.13 plus your patch and the common kdb patch > > compiles, boots, and passes an initial sniff test. This is on a 4 socket, > > dual core Opteron. > > > > I haven't investigated further as to why turning on kdb_modules results in > > something that fails during link with the messages: > > > > BFD: Warning: Writing section `.kdb_initcall.init' to huge (ie negative) > > file > > offset 0xffffffff8058efc0. > > objcopy: arch/x86_64/boot/compressed/vmlinux.bin: File truncated > > make[2]: *** [arch/x86_64/boot/compressed/vmlinux.bin] Error 1 > > make[1]: *** [arch/x86_64/boot/compressed/vmlinux] Error 2 > > make: *** [bzImage] Error 2 > > Hmm, I dont normally have modules defined anyway, but that is > odd. So this didnt happen with the 2.6.11 kernel? > > If no one here has any ideas about this we should ask in LKML > I think. > > Thanks for the testing Ray :) > > Jack > > --------------------------- > Use http://oss.sgi.com/ecartis to modify your settings or to unsubscribe. --------------------------- Use http://oss.sgi.com/ecartis to modify your settings or to unsubscribe.
