On Tue, Sep 26, 2006 at 11:20:13AM -0400, Don Zickus wrote: > Pardon my ignorance, but why is there arch dependent functions to begin > with? This seems like it should be generic code? Yes, I read through the > ppc64 and i386/x86_64 patches and see both implementations are different, > but I also didn't see anything arch specific in either of those patches. > > Isn't the /proc/iomem trick you are using for i386/x86_64 also usable for > every arch? >
In PPC64, /proc/iomem does not inclue the entry for "Crash kernel" region as in i386/x86_64. PPC64 code checks for the presence of linux,crashkernel-base file to check kdump kernel reservation, while i386/x86_64 checks for /proc/iomem By looking through /proc/iomem is the only way to find whether memory for crashkernel is reserved successfully or not. Even if the kernel is not able to reserve the memory for kdump kernel because of some other reasons, /proc/cmdline will not reflect this (it can not). So in my opinion checking /proc/iomem is the right option. Vivek, parsing /proc/cmdline may not be a solution for this. Because COMMAND_LINE_SIZE differs for each architecure. So parsing /proc/cmdline may also involve some architecture dependent macro expansion. > Cheers, > Don _______________________________________________ fastboot mailing list [email protected] https://lists.osdl.org/mailman/listinfo/fastboot
