On 11/17/06, Dave Anderson <[EMAIL PROTECTED]> wrote: > Dave Anderson wrote: > > > > > > > In any case, *something* is needed to give the crash > > utility the physical base address used for identity-mapped > > kernel virtual addresses in those circumstances where > > it's not zero. > > > > more accurately put: > > In any case, *something* is needed to give the crash > utility the physical base address used for architectures > where special (non-identity-mapped) virtual address > regions are used to map kernel-text/static-data.
I'm with you. Thanks for explaining. My plan was basically that if we don't know certain things when loading the crash kernel with kexec, don't guess using values hardcoded in the kexec tool. I think the crash utility has a much better chance of getting things right using the vmlinux file provided by the user. But that does not help us if the physical address where the kernel is loaded is different from the physical address in the vmlinux headers. So the relocatable kernel is a bit tricky, and so are the special kernel segments. But doesn't this mean that we can treat x86_64 and ia64 in a similar fashion? Maybe just by adding a PT_LOAD program header for the kernel segment? / magnus _______________________________________________ fastboot mailing list fastboot@lists.osdl.org https://lists.osdl.org/mailman/listinfo/fastboot
