Hi Mark, On 24/09/18 14:34, Mark Rutland wrote: > On Mon, Sep 17, 2018 at 12:43:30PM +0800, Jun Yao wrote: >> Create the initial page table in the init_pg_dir. And update the >> init_mm.pgd to make sure that pgd_offset_k() works correctly. When >> the final page table is created, we redirect the init_mm.pgd to the >> swapper_pg_dir.
>> diff --git a/arch/arm64/kernel/head.S b/arch/arm64/kernel/head.S >> index de2aaea00bd2..cf8a58211b80 100644 >> --- a/arch/arm64/kernel/head.S >> +++ b/arch/arm64/kernel/head.S >> @@ -376,7 +376,7 @@ __create_page_tables: >> /* >> * Map the kernel image (starting with PHYS_OFFSET). >> */ >> - adrp x0, swapper_pg_dir >> + adrp x0, init_pg_dir >> mov_q x5, KIMAGE_VADDR + TEXT_OFFSET // compile time __va(_text) >> add x5, x5, x23 // add KASLR displacement >> mov x4, PTRS_PER_PGD >> @@ -439,6 +439,11 @@ __primary_switched: >> bl __pi_memset >> dsb ishst // Make zero page visible to PTW >> >> + // Update init_mm.pgd >> + adrp x0, init_pg_dir >> + adr_l x1, init_mm >> + str x0, [x1, #MM_PGD] >> + > > We should be able to set this up statically with INIT_MM_CONTEXT(). i.e. > in <asm/mmu.h> have: > > #define INIT_MM_CONTEXT(name) \ > .pgd = init_pg_dir Aha, I didn't think this would be possible because we end up setting .pgd twice, but gcc says that's fine, it gets the last value: https://gcc.gnu.org/onlinedocs/gcc/Designated-Inits.html (and it looks like clang has a test for this) Thanks, James
