I have been wrestling with this same "Invalid memory segment 0x92000 - 95000" error Using kboot-10; kexec-tools-1.101, 2.6.18, bzImage native built on x86_64, cmd-line 'kexec -t bzImage -d --initrd=/boot/initrd-2.6.18.img --command-line="ro root=/dev/sda" /boot/vmlinuz-2.6.18'
Debug so far has the bzImage segment[0].bufsz being 10 bytes less than what the same code compiled/running on x86(i686) reports. bzImage segment[1].bufsz is 64 bytes greater than what x86 reports. Initially suspected uClibc problem, when uClibc was replaced with GNU libc, same failure? Using native built (x86_64) kexec-tools-1.101, 2.6.18 kernel, REL4.u4 the problem does not occur? Stan. Vivek Goyal wrote: > On Thu, Nov 09, 2006 at 12:06:38PM -0800, Jay Lan wrote: >> Don Zickus wrote: >>> On Wed, Nov 08, 2006 at 08:07:22PM -0800, Lu, Yinghai wrote: >>>> Eric, >>>> >>>> I got "Invalid memory segment 0x100000 - ..." >>>> using kexec latest kernel... >>> >>> I usually see this when people forget to add the "[EMAIL PROTECTED]" >>> into their /etc/grub.conf kernel command line. Where X and Y are >>> arch specific. >> >> I have had "Invalid memory segment 0x4000000 - 0x4997fff" problem >> with '-l' option _always_. Since my priority was on '-p' i did not >> spent time on debugging this problem yet... >> >> Maybe this "[EMAIL PROTECTED]" was the cause of my problem? Some >> platform can not specify a location to load so that it is legal to >> only specify "crashkernel=X" now. Is it possible '-l' code path >> still expect to >> see Y? >> > > kexec -l patch does not worry about crashkernel=. Only kexec -p path > does. So this is something else. > > Thanks > Vivek > _______________________________________________ > fastboot mailing list > [email protected] > https://lists.osdl.org/mailman/listinfo/fastboot _______________________________________________ fastboot mailing list [email protected] https://lists.osdl.org/mailman/listinfo/fastboot
