This is an alternative approach to the series I posted in November 2014: thread.gmane.org/gmane.linux.kernel.efi/5133 For arm64, we need to keep track of which parts of the physical memory space are backed by normal memory, since device mappings and memory mappings are mutually incompatible (device mappings don't allow unaligned accesses which may occur when parsing ACPI tables, for instance, and mapping devices as memory is also not allowed) Instead of adding a physmem memblock table that contains all of memory, including the memory that should not be covered by the linear mapping, this series introduces a MEMBLOCK_NOMAP attribute that allows us to carry the same information in the ordinary 'memory' memblock table. Patch #1 introduces the attribute and the associated plumbing to interpret the attribute and to ensure MEMBLOCK_NOMAP regions are not considered for allocations. Patch #2 makes the arm64 core mapping code aware of MEMBLOCK_NOMAP, by honoring it when setting up the linear mapping, and by making pfn_valid() aware of it. Patch #3 updates the UEFI memory map handling logic to mark reserved regions as MEMBLOCK_NOMAP. Notable about this series is that it does not require any changes to the iomem resource table handling or the definition of page_is_ram(), since the nomap regions are still registered as 'System RAM'. Ard Biesheuvel (3): mm/memblock: add MEMBLOCK_NOMAP attribute to memblock memory table arm64: only consider memblocks with NOMAP cleared for linear mapping arm64/efi: mark UEFI reserved regions as MEMBLOCK_NOMAP arch/arm64/kernel/efi.c | 5 ++-- arch/arm64/mm/init.c | 2 +- arch/arm64/mm/mmu.c | 2 ++ include/linux/memblock.h | 8 ++++++ mm/memblock.c | 28 ++++++++++++++++++++ 5 files changed, 41 insertions(+), 4 deletions(-) -- 1.9.1 -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@xxxxxxxxx. For more info on Linux MM, see: http://www.linux-mm.org/ . Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>