arm64: Use 42-bit address space with 64K pages

This patch expands the VA_BITS to 42 when the 64K page configuration is
enabled allowing 2TB kernel linear mapping. Linux still uses 2 levels of
page tables in this configuration with pgd now being a full page.

Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Acked-by: Will Deacon <will.deacon@arm.com>
Acked-by: Marc Zyngier <marc.zyngier@arm.com>

Documentation/arm64/memory.txt

@@ -21,7 +21,7 @@ The swapper_pgd_dir address is written to TTBR1 and never written to
 TTBR0.
 
 
-AArch64 Linux memory layout:
+AArch64 Linux memory layout with 4KB pages:
 
 Start			End			Size		Use
 -----------------------------------------------------------------------
@@ -46,6 +46,31 @@ ffffffbffc000000	ffffffbfffffffff	  64MB		modules
 ffffffc000000000	ffffffffffffffff	 256GB		kernel logical memory map
 
 
+AArch64 Linux memory layout with 64KB pages:
+
+Start			End			Size		Use
+-----------------------------------------------------------------------
+0000000000000000	000003ffffffffff	   4TB		user
+
+fffffc0000000000	fffffdfbfffeffff	  ~2TB		vmalloc
+
+fffffdfbffff0000	fffffdfbffffffff	  64KB		[guard page]
+
+fffffdfc00000000	fffffdfdffffffff	   8GB		vmemmap
+
+fffffdfe00000000	fffffdfffbbfffff	  ~8GB		[guard, future vmmemap]
+
+fffffdfffbc00000	fffffdfffbdfffff	   2MB		earlyprintk device
+
+fffffdfffbe00000	fffffdfffbe0ffff	  64KB		PCI I/O space
+
+fffffdfffbe10000	fffffdfffbffffff	  ~2MB		[guard]
+
+fffffdfffc000000	fffffdffffffffff	  64MB		modules
+
+fffffe0000000000	ffffffffffffffff	   2TB		kernel logical memory map
+
+
 Translation table lookup with 4KB pages:
 
 +--------+--------+--------+--------+--------+--------+--------+--------+

arch/arm64/include/asm/memory.h

@@ -33,18 +33,23 @@
 #define UL(x) _AC(x, UL)
 
 /*
- * PAGE_OFFSET - the virtual address of the start of the kernel image.
+ * PAGE_OFFSET - the virtual address of the start of the kernel image (top
+ *		 (VA_BITS - 1))
  * VA_BITS - the maximum number of bits for virtual addresses.
  * TASK_SIZE - the maximum size of a user space task.
  * TASK_UNMAPPED_BASE - the lower boundary of the mmap VM area.
  * The module space lives between the addresses given by TASK_SIZE
  * and PAGE_OFFSET - it must be within 128MB of the kernel text.
  */
-#define PAGE_OFFSET		UL(0xffffffc000000000)
+#ifdef CONFIG_ARM64_64K_PAGES
+#define VA_BITS			(42)
+#else
+#define VA_BITS			(39)
+#endif
+#define PAGE_OFFSET		(UL(0xffffffffffffffff) << (VA_BITS - 1))
 #define MODULES_END		(PAGE_OFFSET)
 #define MODULES_VADDR		(MODULES_END - SZ_64M)
 #define EARLYCON_IOBASE		(MODULES_VADDR - SZ_4M)
-#define VA_BITS			(39)
 #define TASK_SIZE_64		(UL(1) << VA_BITS)
 
 #ifdef CONFIG_COMPAT

arch/arm64/include/asm/pgtable-2level-hwdef.h

@@ -21,10 +21,10 @@
  * 8192 entries of 8 bytes each, occupying a 64KB page. Levels 0 and 1 are not
  * used. The 2nd level table (PGD for Linux) can cover a range of 4TB, each
  * entry representing 512MB. The user and kernel address spaces are limited to
- * 512GB and therefore we only use 1024 entries in the PGD.
+ * 4TB in the 64KB page configuration.
  */
 #define PTRS_PER_PTE		8192
-#define PTRS_PER_PGD		1024
+#define PTRS_PER_PGD		8192
 
 /*
  * PGDIR_SHIFT determines the size a top-level page table entry can map.

arch/arm64/include/asm/pgtable.h

@@ -33,7 +33,7 @@
 /*
  * VMALLOC and SPARSEMEM_VMEMMAP ranges.
  */
-#define VMALLOC_START		UL(0xffffff8000000000)
+#define VMALLOC_START		(UL(0xffffffffffffffff) << VA_BITS)
 #define VMALLOC_END		(PAGE_OFFSET - UL(0x400000000) - SZ_64K)
 
 #define vmemmap			((struct page *)(VMALLOC_END + SZ_64K))