Hi all, I numbered this v4 because this is still revolving around the same initial desire to cut the build time of an ARM64 kernel when toggling CONFIG_BLK_DEV_INITRD. This 4th version is basically the 3rd possible way to just get rid of __early_init_dt_declare_initrd() for ARM64. I previously mentioned that I was working on making phys_initrd_start/phys_initrd_size generic across architectures that make use of it, which would cover ARM (32-bit), unicore32 and possibly arm64, but upon second glance, this does not necessarily help, or rather this patch series actually helps make a smoother conversion in the future. Comments welcome, sorry for making so many iterations, it's Friday. Previous discussions/submissions list here: v3: https://www.spinics.net/lists/arm-kernel/msg683566.html v2: https://lkml.org/lkml/2018/10/25/4 Changes in v4: - perform the physical to virtual initrd conversion straight within arm64_memblock_init() to get the correct memblock reservation to occur Changes in v3: - attempt to change drivers/of/fdt.c to absorb ARM64's specific behavior Changes in v2: - put an /* empty */ comment in the asm-generic/initrd.h file - trim down the CC list to maximize the chances of people receiving this Florian Fainelli (2): arm64: Get rid of __early_init_dt_declare_initrd() of/fdt: Remove definition check for __early_init_dt_declare_initrd() arch/arm64/include/asm/memory.h | 8 -------- arch/arm64/mm/init.c | 26 ++++++++++++++++---------- drivers/of/fdt.c | 2 -- 3 files changed, 16 insertions(+), 20 deletions(-) -- 2.17.1