sparc64: Make PAGE_OFFSET variable.

Choose PAGE_OFFSET dynamically based upon cpu type.

Original UltraSPARC-I (spitfire) chips only supported a 44-bit
virtual address space.

Newer chips (T4 and later) support 52-bit virtual addresses
and up to 47-bits of physical memory space.

Therefore we have to adjust PAGE_SIZE dynamically based upon
the capabilities of the chip.

Note that this change alone does not allow us to support > 43-bit
physical memory, to do that we need to re-arrange our page table
support.  The current encodings of the pmd_t and pgd_t pointers
restricts us to "32 + 11" == 43 bits.

This change can waste quite a bit of memory for the various tables.
In particular, a future change should work to size and allocate
kern_linear_bitmap[] and sparc64_valid_addr_bitmap[] dynamically.
This isn't easy as we really cannot take a TLB miss when accessing
kern_linear_bitmap[].  We'd have to lock it into the TLB or similar.

Signed-off-by: David S. Miller <davem@davemloft.net>
Acked-by: Bob Picco <bob.picco@oracle.com>

arch/sparc/include/asm/page_64.h

@@ -112,24 +112,16 @@ typedef pte_t *pgtable_t;
 
 #include <asm-generic/memory_model.h>
 
-#endif /* !(__ASSEMBLY__) */
-
-/* We used to stick this into a hard-coded global register (%g4)
- * but that does not make sense anymore.
- */
-#define MAX_SUPPORTED_PA_BITS	43
 #define PAGE_OFFSET_BY_BITS(X)	(-(_AC(1,UL) << (X)))
-#define PAGE_OFFSET		PAGE_OFFSET_BY_BITS(MAX_SUPPORTED_PA_BITS)
+extern unsigned long PAGE_OFFSET;
 
-/* The "virtual" portion of PAGE_OFFSET, used to clip off the non-physical
- * bits of a linear kernel address.
- */
-#define PAGE_OFFSET_VA_BITS	(64 - MAX_SUPPORTED_PA_BITS)
+#endif /* !(__ASSEMBLY__) */
 
-/* The actual number of physical memory address bits we support, this is
- * used to size various tables used to manage kernel TLB misses.
+/* The maximum number of physical memory address bits we support, this
+ * is used to size various tables used to manage kernel TLB misses and
+ * also the sparsemem code.
  */
-#define MAX_PHYS_ADDRESS_BITS	41
+#define MAX_PHYS_ADDRESS_BITS	47
 
 /* These two shift counts are used when indexing sparc64_valid_addr_bitmap
  * and kpte_linear_bitmap.

arch/sparc/kernel/ktlb.S

@@ -153,12 +153,19 @@ kvmap_dtlb_tsb4m_miss:
 	/* Clear the PAGE_OFFSET top virtual bits, shift
 	 * down to get PFN, and make sure PFN is in range.
 	 */
-	sllx		%g4, PAGE_OFFSET_VA_BITS, %g5
+661:	sllx		%g4, 0, %g5
+	.section	.page_offset_shift_patch, "ax"
+	.word		661b
+	.previous
 
 	/* Check to see if we know about valid memory at the 4MB
 	 * chunk this physical address will reside within.
 	 */
-	srlx		%g5, PAGE_OFFSET_VA_BITS + MAX_PHYS_ADDRESS_BITS, %g2
+661:	srlx		%g5, MAX_PHYS_ADDRESS_BITS, %g2
+	.section	.page_offset_shift_patch, "ax"
+	.word		661b
+	.previous
+
 	brnz,pn		%g2, kvmap_dtlb_longpath
 	 nop
 
@@ -176,7 +183,11 @@ valid_addr_bitmap_patch:
 	or		%g7, %lo(sparc64_valid_addr_bitmap), %g7
 	.previous
 
-	srlx		%g5, PAGE_OFFSET_VA_BITS + ILOG2_4MB, %g2
+661:	srlx		%g5, ILOG2_4MB, %g2
+	.section	.page_offset_shift_patch, "ax"
+	.word		661b
+	.previous
+
 	srlx		%g2, 6, %g5
 	and		%g2, 63, %g2
 	sllx		%g5, 3, %g5
@@ -189,9 +200,18 @@ valid_addr_bitmap_patch:
 2:	 sethi		%hi(kpte_linear_bitmap), %g2
 
 	/* Get the 256MB physical address index. */
-	sllx		%g4, PAGE_OFFSET_VA_BITS, %g5
+661:	sllx		%g4, 0, %g5
+	.section	.page_offset_shift_patch, "ax"
+	.word		661b
+	.previous
+
 	or		%g2, %lo(kpte_linear_bitmap), %g2
-	srlx		%g5, PAGE_OFFSET_VA_BITS + ILOG2_256MB, %g5
+
+661:	srlx		%g5, ILOG2_256MB, %g5
+	.section	.page_offset_shift_patch, "ax"
+	.word		661b
+	.previous
+
 	and		%g5, (32 - 1), %g7
 
 	/* Divide by 32 to get the offset into the bitmask.  */

arch/sparc/kernel/vmlinux.lds.S

@@ -122,6 +122,11 @@ SECTIONS
 		*(.swapper_4m_tsb_phys_patch)
 		__swapper_4m_tsb_phys_patch_end = .;
 	}
+	.page_offset_shift_patch : {
+		__page_offset_shift_patch = .;
+		*(.page_offset_shift_patch)
+		__page_offset_shift_patch_end = .;
+	}
 	.popc_3insn_patch : {
 		__popc_3insn_patch = .;
 		*(.popc_3insn_patch)

arch/sparc/lib/clear_page.S

@@ -37,10 +37,10 @@ _clear_page:		/* %o0=dest */
 	.globl		clear_user_page
 clear_user_page:	/* %o0=dest, %o1=vaddr */
 	lduw		[%g6 + TI_PRE_COUNT], %o2
-	sethi		%uhi(PAGE_OFFSET), %g2
+	sethi		%hi(PAGE_OFFSET), %g2
 	sethi		%hi(PAGE_SIZE), %o4
 
-	sllx		%g2, 32, %g2
+	ldx		[%g2 + %lo(PAGE_OFFSET)], %g2
 	sethi		%hi(PAGE_KERNEL_LOCKED), %g3
 
 	ldx		[%g3 + %lo(PAGE_KERNEL_LOCKED)], %g3

arch/sparc/lib/copy_page.S

@@ -46,10 +46,10 @@
 	.type		copy_user_page,#function
 copy_user_page:		/* %o0=dest, %o1=src, %o2=vaddr */
 	lduw		[%g6 + TI_PRE_COUNT], %o4
-	sethi		%uhi(PAGE_OFFSET), %g2
+	sethi		%hi(PAGE_OFFSET), %g2
 	sethi		%hi(PAGE_SIZE), %o3
 
-	sllx		%g2, 32, %g2
+	ldx		[%g2 + %lo(PAGE_OFFSET)], %g2
 	sethi		%hi(PAGE_KERNEL_LOCKED), %g3
 
 	ldx		[%g3 + %lo(PAGE_KERNEL_LOCKED)], %g3

arch/sparc/mm/init_64.c

@@ -1557,6 +1557,96 @@ unsigned long __init find_ecache_flush_span(unsigned long size)
 	return ~0UL;
 }
 
+unsigned long PAGE_OFFSET;
+EXPORT_SYMBOL(PAGE_OFFSET);
+
+static void __init page_offset_shift_patch_one(unsigned int *insn, unsigned long phys_bits)
+{
+	unsigned long final_shift;
+	unsigned int val = *insn;
+	unsigned int cnt;
+
+	/* We are patching in ilog2(max_supported_phys_address), and
+	 * we are doing so in a manner similar to a relocation addend.
+	 * That is, we are adding the shift value to whatever value
+	 * is in the shift instruction count field already.
+	 */
+	cnt = (val & 0x3f);
+	val &= ~0x3f;
+
+	/* If we are trying to shift >= 64 bits, clear the destination
+	 * register.  This can happen when phys_bits ends up being equal
+	 * to MAX_PHYS_ADDRESS_BITS.
+	 */
+	final_shift = (cnt + (64 - phys_bits));
+	if (final_shift >= 64) {
+		unsigned int rd = (val >> 25) & 0x1f;
+
+		val = 0x80100000 | (rd << 25);
+	} else {
+		val |= final_shift;
+	}
+	*insn = val;
+
+	__asm__ __volatile__("flush	%0"
+			     : /* no outputs */
+			     : "r" (insn));
+}
+
+static void __init page_offset_shift_patch(unsigned long phys_bits)
+{
+	extern unsigned int __page_offset_shift_patch;
+	extern unsigned int __page_offset_shift_patch_end;
+	unsigned int *p;
+
+	p = &__page_offset_shift_patch;
+	while (p < &__page_offset_shift_patch_end) {
+		unsigned int *insn = (unsigned int *)(unsigned long)*p;
+
+		page_offset_shift_patch_one(insn, phys_bits);
+
+		p++;
+	}
+}
+
+static void __init setup_page_offset(void)
+{
+	unsigned long max_phys_bits = 40;
+
+	if (tlb_type == cheetah || tlb_type == cheetah_plus) {
+		max_phys_bits = 42;
+	} else if (tlb_type == hypervisor) {
+		switch (sun4v_chip_type) {
+		case SUN4V_CHIP_NIAGARA1:
+		case SUN4V_CHIP_NIAGARA2:
+			max_phys_bits = 39;
+			break;
+		case SUN4V_CHIP_NIAGARA3:
+			max_phys_bits = 43;
+			break;
+		case SUN4V_CHIP_NIAGARA4:
+		case SUN4V_CHIP_NIAGARA5:
+		case SUN4V_CHIP_SPARC64X:
+		default:
+			max_phys_bits = 47;
+			break;
+		}
+	}
+
+	if (max_phys_bits > MAX_PHYS_ADDRESS_BITS) {
+		prom_printf("MAX_PHYS_ADDRESS_BITS is too small, need %lu\n",
+			    max_phys_bits);
+		prom_halt();
+	}
+
+	PAGE_OFFSET = PAGE_OFFSET_BY_BITS(max_phys_bits);
+
+	pr_info("PAGE_OFFSET is 0x%016lx (max_phys_bits == %lu)\n",
+		PAGE_OFFSET, max_phys_bits);
+
+	page_offset_shift_patch(max_phys_bits);
+}
+
 static void __init tsb_phys_patch(void)
 {
 	struct tsb_ldquad_phys_patch_entry *pquad;
@@ -1763,6 +1853,8 @@ void __init paging_init(void)
 	unsigned long real_end, i;
 	int node;
 
+	setup_page_offset();
+
 	/* These build time checkes make sure that the dcache_dirty_cpu()
 	 * page->flags usage will work.
 	 *

arch/sparc/mm/ultra.S

@@ -153,10 +153,10 @@ __spitfire_flush_tlb_mm_slow:
 	.globl		__flush_icache_page
 __flush_icache_page:	/* %o0 = phys_page */
 	srlx		%o0, PAGE_SHIFT, %o0
-	sethi		%uhi(PAGE_OFFSET), %g1
+	sethi		%hi(PAGE_OFFSET), %g1
 	sllx		%o0, PAGE_SHIFT, %o0
 	sethi		%hi(PAGE_SIZE), %g2
-	sllx		%g1, 32, %g1
+	ldx		[%g1 + %lo(PAGE_OFFSET)], %g1
 	add		%o0, %g1, %o0
 1:	subcc		%g2, 32, %g2
 	bne,pt		%icc, 1b
@@ -178,8 +178,8 @@ __flush_icache_page:	/* %o0 = phys_page */
 	.align		64
 	.globl		__flush_dcache_page
 __flush_dcache_page:	/* %o0=kaddr, %o1=flush_icache */
-	sethi		%uhi(PAGE_OFFSET), %g1
-	sllx		%g1, 32, %g1
+	sethi		%hi(PAGE_OFFSET), %g1
+	ldx		[%g1 + %lo(PAGE_OFFSET)], %g1
 	sub		%o0, %g1, %o0			! physical address
 	srlx		%o0, 11, %o0			! make D-cache TAG
 	sethi		%hi(1 << 14), %o2		! D-cache size
@@ -287,8 +287,8 @@ __cheetah_flush_tlb_pending:	/* 27 insns */
 
 #ifdef DCACHE_ALIASING_POSSIBLE
 __cheetah_flush_dcache_page: /* 11 insns */
-	sethi		%uhi(PAGE_OFFSET), %g1
-	sllx		%g1, 32, %g1
+	sethi		%hi(PAGE_OFFSET), %g1
+	ldx		[%g1 + %lo(PAGE_OFFSET)], %g1
 	sub		%o0, %g1, %o0
 	sethi		%hi(PAGE_SIZE), %o4
 1:	subcc		%o4, (1 << 5), %o4