sh: Support for extended ASIDs on PTEAEX-capable SH-X3 cores.

This adds support for extended ASIDs (up to 16-bits) on newer SH-X3 cores
that implement the PTAEX register and respective functionality. Presently
only the 65nm SH7786 (90nm only supports legacy 8-bit ASIDs).

The main change is in how the PTE is written out when loading the entry
in to the TLB, as well as in how the TLB entry is selectively flushed.

While SH-X2 extended mode splits out the memory-mapped U and I-TLB data
arrays for extra bits, extended ASID mode splits out the address arrays.
While we don't use the memory-mapped data array access, the address
array accesses are necessary for selective TLB flushes, so these are
implemented newly and replace the generic SH-4 implementation.

With this, TLB flushes in switch_mm() are almost non-existent on newer
parts.

Signed-off-by: Paul Mundt <lethal@linux-sh.org>

2 files changed, 103 insertions, 2 deletions

diff --git a/arch/sh/mm/Makefile_32 b/arch/sh/mm/Makefile_32
index 469ff167245..986a1e05583 100644
--- a/arch/sh/mm/Makefile_32
+++ b/arch/sh/mm/Makefile_32
@@ -25,8 +25,10 @@ obj-$(CONFIG_CPU_SH4)	+= cache-debugfs.o
 endif
 
 ifdef CONFIG_MMU
-obj-$(CONFIG_CPU_SH3)	+= tlb-sh3.o
-obj-$(CONFIG_CPU_SH4)	+= tlb-sh4.o
+tlb-$(CONFIG_CPU_SH3)		:= tlb-sh3.o
+tlb-$(CONFIG_CPU_SH4)		:= tlb-sh4.o
+tlb-$(CONFIG_CPU_HAS_PTEAEX)	:= tlb-pteaex.o
+obj-y				+= $(tlb-y)
 ifndef CONFIG_CACHE_OFF
 obj-$(CONFIG_CPU_SH4)		+= pg-sh4.o
 obj-$(CONFIG_SH7705_CACHE_32KB)	+= pg-sh7705.o
diff --git a/arch/sh/mm/tlb-pteaex.c b/arch/sh/mm/tlb-pteaex.c
new file mode 100644
index 00000000000..5c9b2d781e0
--- /dev/null
+++ b/arch/sh/mm/tlb-pteaex.c
@@ -0,0 +1,99 @@
+/*
+ * arch/sh/mm/tlb-pteaex.c
+ *
+ * TLB operations for SH-X3 CPUs featuring PTE ASID Extensions.
+ *
+ * Copyright (C) 2009 Paul Mundt
+ *
+ * This file is subject to the terms and conditions of the GNU General Public
+ * License.  See the file "COPYING" in the main directory of this archive
+ * for more details.
+ */
+#include <linux/kernel.h>
+#include <linux/mm.h>
+#include <linux/io.h>
+#include <asm/system.h>
+#include <asm/mmu_context.h>
+#include <asm/cacheflush.h>
+
+void update_mmu_cache(struct vm_area_struct * vma,
+		      unsigned long address, pte_t pte)
+{
+	unsigned long flags;
+	unsigned long pteval;
+	unsigned long vpn;
+
+	/* Ptrace may call this routine. */
+	if (vma && current->active_mm != vma->vm_mm)
+		return;
+
+#ifndef CONFIG_CACHE_OFF
+	{
+		unsigned long pfn = pte_pfn(pte);
+
+		if (pfn_valid(pfn)) {
+			struct page *page = pfn_to_page(pfn);
+
+			if (!test_bit(PG_mapped, &page->flags)) {
+				unsigned long phys = pte_val(pte) & PTE_PHYS_MASK;
+				__flush_wback_region((void *)P1SEGADDR(phys),
+						     PAGE_SIZE);
+				__set_bit(PG_mapped, &page->flags);
+			}
+		}
+	}
+#endif
+
+	local_irq_save(flags);
+
+	/* Set PTEH register */
+	vpn = address & MMU_VPN_MASK;
+	__raw_writel(vpn, MMU_PTEH);
+
+	/* Set PTEAEX */
+	__raw_writel(get_asid(), MMU_PTEAEX);
+
+	pteval = pte.pte_low;
+
+	/* Set PTEA register */
+#ifdef CONFIG_X2TLB
+	/*
+	 * For the extended mode TLB this is trivial, only the ESZ and
+	 * EPR bits need to be written out to PTEA, with the remainder of
+	 * the protection bits (with the exception of the compat-mode SZ
+	 * and PR bits, which are cleared) being written out in PTEL.
+	 */
+	__raw_writel(pte.pte_high, MMU_PTEA);
+#else
+	/* TODO: make this look less hacky */
+	__raw_writel(((pteval >> 28) & 0xe) | (pteval & 0x1), MMU_PTEA);
+#endif
+
+	/* Set PTEL register */
+	pteval &= _PAGE_FLAGS_HARDWARE_MASK; /* drop software flags */
+#ifdef CONFIG_CACHE_WRITETHROUGH
+	pteval |= _PAGE_WT;
+#endif
+	/* conveniently, we want all the software flags to be 0 anyway */
+	__raw_writel(pteval, MMU_PTEL);
+
+	/* Load the TLB */
+	asm volatile("ldtlb": /* no output */ : /* no input */ : "memory");
+	local_irq_restore(flags);
+}
+
+/*
+ * While SH-X2 extended TLB mode splits out the memory-mapped I/UTLB
+ * data arrays, SH-X3 cores with PTEAEX split out the memory-mapped
+ * address arrays. In compat mode the second array is inaccessible, while
+ * in extended mode, the legacy 8-bit ASID field in address array 1 has
+ * undefined behaviour.
+ */
+void __uses_jump_to_uncached local_flush_tlb_one(unsigned long asid,
+						 unsigned long page)
+{
+	jump_to_uncached();
+	__raw_writel(page, MMU_UTLB_ADDRESS_ARRAY | MMU_PAGE_ASSOC_BIT);
+	__raw_writel(asid, MMU_UTLB_ADDRESS_ARRAY2 | MMU_PAGE_ASSOC_BIT);
+	back_to_cached();
+}