Red Hat Bugzilla – Bug 220143
kexec installation of kernel fails due to overlapping segment detection
Last modified: 2007-11-30 17:11:51 EST
Description of problem: Forcing dump under kdump failed with explanation that dump kernel wasn't installed, so investigated manually installing dump kernel. Oh, and yes, there was a properly formed crashkernel=128M@16M option on the boot commandline, plus the appropriate region appeared carved out in /proc/iomem. Using the kexec installation line from /etc/init.d/kdump: /sbin/kexec -p --args-linux --command-line="ro root=/dev/VolGroup00/LogVol00 rhgb quiet" --initrd=/boot/initrd-2.6.18-1.2868.fc6kdump.img /boot/vmlinux-2.6.18-1.2868.fc6kdump I got the message: Overlapping memory segments at 0x1460000 sort_segments failed This may be similar or identical to a bug reported against kexec w/ x86-64 on the fastboot mailing list around the beginning of October, and may require tweaking the dump kernel. This was reported against kexec-tools as #216171, and then forwarded to kernel maintainers at that time: ------- Additional Comments From nhorman@redhat.com 2006-11-21 11:44 EST ------- Hey! Thanks for the report. This is actually a kernel bug though, that results from overlapping physical address for program headers getting written into the compiled kdump kernel. you need this patch for your kernel: http://www.kernel.org/git/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=dbaab49f92ff6ae6255762a948375e4036cbdbd2 It just got added to Linus' tree today, so I expect Dave J. will have it in the FC-6 kernel shortly. This fresh _new_ kernel release has _the same problem_ as the previous release! Version-Release number of selected component (if applicable): 2.6.18-1.2868.fc6 How reproducible: always Steps to Reproduce: 1. run kexec line above 2. 3. Actual results: Overlapping memory segments at 0x1460000 sort_segments failed Expected results: an installed crash kernel Additional info:
The overlapping segments issue should be fixed with kexec-tools-1.101-56.fc6, due out in a few days. however the x86_64 rpm still has an issue with loading due to an alternate kernel problem (one in which /proc/kcore returns EPERM for all users universally), blocking the load operation. That is being tracked in bz 241362 *** This bug has been marked as a duplicate of 238768 ***