MIPS64: Support of at least 48 bits of SEGBITS

SEGBITS is 40 bits or more, depending on CPU type.  Introduces optional
support for 48 bits of application virtual address space.  Only 16K and
64K pages are supported.

Enabling will result in a memory overhead of a small number of pages for
small applications.  For 64K pages a 3rd level of page tables is required
which has some impact during software TLB refill.

[ralf@linux-mips.org: Fixed things raised in the review of the version
posted and changed kconfig to be a bit more userfriendly.]
Signed-off-by: Leonid Yegoshin <Leonid.Yegoshin@imgtec.com>
Cc: aleksey.makarov@auriga.com
Cc: james.hogan@imgtec.com
Cc: paul.burton@imgtec.com
Cc: david.daney@cavium.com
Cc: peterz@infradead.org
Cc: linux-mips@linux-mips.org
Cc: linux-kernel@vger.kernel.org
Cc: davidlohr@hp.com
Cc: kirill@shutemov.name
Cc: akpm@linux-foundation.org
Cc: mingo@kernel.org
Patchwork: https://patchwork.linux-mips.org/patch/10051/Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

arch/mips/Kconfig

@@ -2076,6 +2076,16 @@ config KVM_GUEST_TIMER_FREQ
 	  emulation when determining guest CPU Frequency. Instead, the guest's
 	  timer frequency is specified directly.
 
+config MIPS_VA_BITS_48
+	bool "48 bits virtual memory"
+	depends on 64BIT
+	help
+	  Support a maximum at least 48 bits of application virtual memory.
+	  Default is 40 bits or less, depending on the CPU.
+	  This option result in a small memory overhead for page tables.
+	  This option is only supported with 16k and 64k page sizes.
+	  If unsure, say N.
+
 choice
 	prompt "Kernel page size"
 	default PAGE_SIZE_4KB
@@ -2083,6 +2093,7 @@ choice
 config PAGE_SIZE_4KB
 	bool "4kB"
 	depends on !CPU_LOONGSON2 && !CPU_LOONGSON3
+	depends on !MIPS_VA_BITS_48
 	help
 	 This option select the standard 4kB Linux page size.  On some
 	 R3000-family processors this is the only available page size.  Using
@@ -2092,6 +2103,7 @@ config PAGE_SIZE_4KB
 config PAGE_SIZE_8KB
 	bool "8kB"
 	depends on CPU_R8000 || CPU_CAVIUM_OCTEON
+	depends on !MIPS_VA_BITS_48
 	help
 	  Using 8kB page size will result in higher performance kernel at
 	  the price of higher memory consumption.  This option is available
@@ -2110,6 +2122,7 @@ config PAGE_SIZE_16KB
 config PAGE_SIZE_32KB
 	bool "32kB"
 	depends on CPU_CAVIUM_OCTEON
+	depends on !MIPS_VA_BITS_48
 	help
 	  Using 32kB page size will result in higher performance kernel at
 	  the price of higher memory consumption.  This option is available

arch/mips/include/asm/pgtable-64.h

@@ -17,7 +17,7 @@
 #include <asm/cachectl.h>
 #include <asm/fixmap.h>
 
-#ifdef CONFIG_PAGE_SIZE_64KB
+#if defined(CONFIG_PAGE_SIZE_64KB) && !defined(CONFIG_MIPS_VA_BITS_48)
 #include <asm-generic/pgtable-nopmd.h>
 #else
 #include <asm-generic/pgtable-nopud.h>
@@ -90,7 +90,11 @@
 #define PTE_ORDER		0
 #endif
 #ifdef CONFIG_PAGE_SIZE_16KB
-#define PGD_ORDER		0
+#ifdef CONFIG_MIPS_VA_BITS_48
+#define PGD_ORDER               1
+#else
+#define PGD_ORDER               0
+#endif
 #define PUD_ORDER		aieeee_attempt_to_allocate_pud
 #define PMD_ORDER		0
 #define PTE_ORDER		0
@@ -104,7 +108,11 @@
 #ifdef CONFIG_PAGE_SIZE_64KB
 #define PGD_ORDER		0
 #define PUD_ORDER		aieeee_attempt_to_allocate_pud
+#ifdef CONFIG_MIPS_VA_BITS_48
+#define PMD_ORDER		0
+#else
 #define PMD_ORDER		aieeee_attempt_to_allocate_pmd
+#endif
 #define PTE_ORDER		0
 #endif
 
@@ -114,11 +122,7 @@
 #endif
 #define PTRS_PER_PTE	((PAGE_SIZE << PTE_ORDER) / sizeof(pte_t))
 
-#if PGDIR_SIZE >= TASK_SIZE64
-#define USER_PTRS_PER_PGD	(1)
-#else
-#define USER_PTRS_PER_PGD	(TASK_SIZE64 / PGDIR_SIZE)
-#endif
+#define USER_PTRS_PER_PGD       ((TASK_SIZE64 / PGDIR_SIZE)?(TASK_SIZE64 / PGDIR_SIZE):1)
 #define FIRST_USER_ADDRESS	0UL
 
 /*

arch/mips/include/asm/processor.h

@@ -63,7 +63,11 @@ extern unsigned int vced_count, vcei_count;
  * 8192EB ...
  */
 #define TASK_SIZE32	0x7fff8000UL
-#define TASK_SIZE64	0x10000000000UL
+#ifdef CONFIG_MIPS_VA_BITS_48
+#define TASK_SIZE64     (0x1UL << ((cpu_data[0].vmbits>48)?48:cpu_data[0].vmbits))
+#else
+#define TASK_SIZE64     0x10000000000UL
+#endif
 #define TASK_SIZE (test_thread_flag(TIF_32BIT_ADDR) ? TASK_SIZE32 : TASK_SIZE64)
 #define STACK_TOP_MAX	TASK_SIZE64