On 3/3/07, Greg Kurtzer <[EMAIL PROTECTED]> wrote: > On testing of RHEL5 beta with Perceus (uses kexec), people are having > problems with the kexec. I can't replicate locally but here is the > info that I got: > > # kexec --initrd=/vnfs.img --force --append="masterip=192.168.1.254 > console=ttyS0,115200 noht nousb ramdisk_blocksize=1024 noapic" /vmlinuz > Ramdisks not supported with generic elf arguments > # file /vmlinuz > vmlinuz: ELF 64-bit LSB shared object, AMD x86-64, version 1, stripped > > Red Hat says this: > > The reason that the file shows up this way is because the RHEL5 > kernel is now relocatable. Relocatable kernels are required for use > with the kdump / kexec facility. Kdump, of course, will be replacing > diskdump and netdump for capturing vmcore files after a crash. > > Has anyone come across this yet or have ideas for a solution?
Not sure about RHEL5 and which version of kexec-tools you are using, but I pass "--args-linux" to the kexec binary from kexec-tools-testing. This on i386 and x86_64 - but not on ia64. If you load a bzImage you should pass "-t bzImage" instead. I assume you are talking about plain kexec - ie rebooting the system - and for that vmlinux images should work just fine. If you load a relocatable kernel image for kdump use I think you need to use bzImage. I don't know the relocatable code that well though. / magnus _______________________________________________ fastboot mailing list [email protected] https://lists.osdl.org/mailman/listinfo/fastboot
