avr32: Store virtual addresses in the PGD

Instead of storing physical addresses along with page flags in the
PGD, store virtual addresses and use NULL to indicate a not present
second-level page table. A non-page-aligned page table indicates a bad
PMD.

This simplifies the TLB miss handler since it no longer has to check
the Present bit and no longer has to convert the PGD entry from
physical to virtual address. Instead, it has to check for a NULL
entry, which is slightly cheaper than either.

Signed-off-by: Haavard Skinnemoen <haavard.skinnemoen@atmel.com>

3 files changed, 27 insertions, 26 deletions

diff --git a/arch/avr32/kernel/entry-avr32b.S b/arch/avr32/kernel/entry-avr32b.S
index 050c006..3cdd707 100644
--- a/arch/avr32/kernel/entry-avr32b.S
+++ b/arch/avr32/kernel/entry-avr32b.S
@@ -74,12 +74,6 @@ exception_vectors:
 	.align	2
 	bral	do_dtlb_modified
 
-	/*
-	 * r0 :	PGD/PT/PTE
-	 * r1 : Offending address
-	 * r2 : Scratch register
-	 * r3 : Cause (5, 12 or 13)
-	 */
 #define	tlbmiss_save	pushm	r0-r3
 #define tlbmiss_restore	popm	r0-r3
 
@@ -108,17 +102,17 @@ tlb_miss_common:
 	bld	r0, 31
 	brcs	handle_vmalloc_miss
 
-	/* First level lookup */
+	/*
+	 * First level lookup: The PGD contains virtual pointers to
+	 * the second-level page tables, but they may be NULL if not
+	 * present.
+	 */
 pgtbl_lookup:
 	lsr	r2, r0, PGDIR_SHIFT
 	ld.w	r3, r1[r2 << 2]
 	bfextu	r1, r0, PAGE_SHIFT, PGDIR_SHIFT - PAGE_SHIFT
-	bld	r3, _PAGE_BIT_PRESENT
-	brcc	page_table_not_present
-
-	/* Translate to virtual address in P1. */
-	andl	r3, 0xf000
-	sbr	r3, 31
+	cp.w	r3, 0
+	breq	page_table_not_present
 
 	/* Second level lookup */
 	ld.w	r2, r3[r1 << 2]
@@ -155,6 +149,19 @@ handle_vmalloc_miss:
 	orh	r1, hi(swapper_pg_dir)
 	rjmp	pgtbl_lookup
 
+	/* The slow path of the TLB miss handler */
+	.align	2
+page_table_not_present:
+page_not_present:
+	tlbmiss_restore
+	sub	sp, 4
+	stmts	--sp, r0-lr
+	rcall	save_full_context_ex
+	mfsr	r12, SYSREG_ECR
+	mov	r11, sp
+	rcall	do_page_fault
+	rjmp	ret_from_exception
+
 
 	/* ---                    System Call                    --- */
 
@@ -267,18 +274,6 @@ syscall_exit_work:
 	brcc	syscall_exit_cont
 	rjmp	enter_monitor_mode
 
-	/* The slow path of the TLB miss handler */
-page_table_not_present:
-page_not_present:
-	tlbmiss_restore
-	sub	sp, 4
-	stmts	--sp, r0-lr
-	rcall	save_full_context_ex
-	mfsr	r12, SYSREG_ECR
-	mov	r11, sp
-	rcall	do_page_fault
-	rjmp	ret_from_exception
-
 	/* This function expects to find offending PC in SYSREG_RAR_EX */
 	.type	save_full_context_ex, @function
 	.align	2
diff --git a/arch/avr32/kernel/vmlinux.lds.S b/arch/avr32/kernel/vmlinux.lds.S
index 033dd46..5d25d8e 100644
--- a/arch/avr32/kernel/vmlinux.lds.S
+++ b/arch/avr32/kernel/vmlinux.lds.S
@@ -99,6 +99,10 @@ SECTIONS
 		 */
 		*(.data.init_task)
 
+		/* Then, the page-aligned data */
+		. = ALIGN(PAGE_SIZE);
+		*(.data.page_aligned)
+
 		/* Then, the cacheline aligned data */
 		. = ALIGN(L1_CACHE_BYTES);
 		*(.data.cacheline_aligned)
diff --git a/arch/avr32/mm/init.c b/arch/avr32/mm/init.c
index 0e77578..3f90a87 100644
--- a/arch/avr32/mm/init.c
+++ b/arch/avr32/mm/init.c
@@ -24,9 +24,11 @@
 #include <asm/setup.h>
 #include <asm/sections.h>
 
+#define __page_aligned	__attribute__((section(".data.page_aligned")))
+
 DEFINE_PER_CPU(struct mmu_gather, mmu_gathers);
 
-pgd_t swapper_pg_dir[PTRS_PER_PGD];
+pgd_t swapper_pg_dir[PTRS_PER_PGD] __page_aligned;
 
 struct page *empty_zero_page;
 EXPORT_SYMBOL(empty_zero_page);