On Mon, Sep 10, 2012 at 01:05:50AM +0100, Simon Horman wrote: > On Fri, Sep 07, 2012 at 10:11:19AM +0100, Matthew Leach wrote: > > Hi all, > > > > Running a kexec() on ARM platforms has become an issue as information > > about the platform that is presented via device-tree is required to > > boot newer kernels. > > > > The ATAGs kernel code is re-used, replacing the ATAGs with a > > device-tree blob so that only kexec-tools needs to be changed. There > > are two options for generating a dtb to be passed to the kernel from > > kexec-tools: 1) Read in the device-tree that is exposed in > > /proc/device-tree and flatten this out into a dtb. 2) Allow the user > > to specify a dtb file on the command line. These blobs are then > > loaded into memory before the kernel image and the address passed > > through to r2, as stated in the ARM booting documentation. > > > > These patches depend upon a kernel patch, namely: > > http://lists.infradead.org/pipermail/linux-arm-kernel/2012-September/118275.html > > > > Simon - you are welcome to add your signed-off-by to my final patch as > > I incorporated bits of your final patch into mine. > > Sure, I'll add the sign-off as part of merging the changes. > > What I have in mind is to to merge the non ppc64/powerpc patches, > that is 5 of the 7 patches. I'm a but unsure how the kernel-side of things > will pan out, but an incremental patch can be made to kexec-tools > if needed and this series shouldn't break any existing support > Does that work for you? Sounds good to me, I'll pursue the kernel-side changes for ARM. Cheers, Will
