Hi Akashi, On 07/02/17 08:08, AKASHI Takahiro wrote: > arch_kexec_protect_crashkres() and arch_kexec_unprotect_crashkres() > are meant to be called by kexec_load() in order to protect the memory > allocated for crash dump kernel once it's loaded. > > Here, the protection is implemented by unmapping the relevant range > of memory, rather than making it read-only, to prevent any corruption > due to potential cache alias (with different attributes) problem. > > To make the things work correctly, we have to > - use page-level mappings entirely > - have the mappings isolated from the other normal memory > - move copying kexec's control_code_page to machine_kexec_prepare() > > Note that page-level mappings are required to allow shrinking the region, > through /sys/kernel/kexec_crash_size, to the size of any number of pages > and putting the freed memory back to buddy system. This shrinking means memory marked memblock:reserve gets used by the slab allocator. This makes me feel uneasy, but I agree its not going to break anything, and we can't easily un-reserve it. The temporary no-map when building the linear map is a neat trick! Reviewed-by: James Morse <james.morse at arm.com> This patch will conflict with Ard's 'arm64: mmu: avoid writeable-executable mappings' series[0], but they may be complimentary as he adds a update_mapping_prot() call in patch 2 [1] which has a similar use-case. Thanks, James [0] https://www.spinics.net/lists/arm-kernel/msg562724.html [1] https://www.spinics.net/lists/arm-kernel/msg562726.html
