5 files changed, 154 insertions, 111 deletions

diff --git a/arch/arc/mm/cache_arc700.c b/arch/arc/mm/cache_arc700.c
index 5a1259cd948..6b58c1de757 100644
--- a/arch/arc/mm/cache_arc700.c
+++ b/arch/arc/mm/cache_arc700.c
@@ -182,7 +182,7 @@ void arc_cache_init(void)
 
 #ifdef CONFIG_ARC_HAS_ICACHE
 	/* 1. Confirm some of I-cache params which Linux assumes */
-	if (ic->line_len != ARC_ICACHE_LINE_LEN)
+	if (ic->line_len != L1_CACHE_BYTES)
 		panic("Cache H/W doesn't match kernel Config");
 
 	if (ic->ver != CONFIG_ARC_MMU_VER)
@@ -205,7 +205,7 @@ chk_dc:
 		return;
 
 #ifdef CONFIG_ARC_HAS_DCACHE
-	if (dc->line_len != ARC_DCACHE_LINE_LEN)
+	if (dc->line_len != L1_CACHE_BYTES)
 		panic("Cache H/W doesn't match kernel Config");
 
 	/* check for D-Cache aliasing */
@@ -240,6 +240,67 @@ chk_dc:
 #define OP_INV		0x1
 #define OP_FLUSH	0x2
 #define OP_FLUSH_N_INV	0x3
+#define OP_INV_IC	0x4
+
+/*
+ * Common Helper for Line Operations on {I,D}-Cache
+ */
+static inline void __cache_line_loop(unsigned long paddr, unsigned long vaddr,
+				     unsigned long sz, const int cacheop)
+{
+	unsigned int aux_cmd, aux_tag;
+	int num_lines;
+	const int full_page_op = __builtin_constant_p(sz) && sz == PAGE_SIZE;
+
+	if (cacheop == OP_INV_IC) {
+		aux_cmd = ARC_REG_IC_IVIL;
+		aux_tag = ARC_REG_IC_PTAG;
+	}
+	else {
+		/* d$ cmd: INV (discard or wback-n-discard) OR FLUSH (wback) */
+		aux_cmd = cacheop & OP_INV ? ARC_REG_DC_IVDL : ARC_REG_DC_FLDL;
+		aux_tag = ARC_REG_DC_PTAG;
+	}
+
+	/* Ensure we properly floor/ceil the non-line aligned/sized requests
+	 * and have @paddr - aligned to cache line and integral @num_lines.
+	 * This however can be avoided for page sized since:
+	 *  -@paddr will be cache-line aligned already (being page aligned)
+	 *  -@sz will be integral multiple of line size (being page sized).
+	 */
+	if (!full_page_op) {
+		sz += paddr & ~CACHE_LINE_MASK;
+		paddr &= CACHE_LINE_MASK;
+		vaddr &= CACHE_LINE_MASK;
+	}
+
+	num_lines = DIV_ROUND_UP(sz, L1_CACHE_BYTES);
+
+#if (CONFIG_ARC_MMU_VER <= 2)
+	/* MMUv2 and before: paddr contains stuffed vaddrs bits */
+	paddr |= (vaddr >> PAGE_SHIFT) & 0x1F;
+#else
+	/* if V-P const for loop, PTAG can be written once outside loop */
+	if (full_page_op)
+		write_aux_reg(ARC_REG_DC_PTAG, paddr);
+#endif
+
+	while (num_lines-- > 0) {
+#if (CONFIG_ARC_MMU_VER > 2)
+		/* MMUv3, cache ops require paddr seperately */
+		if (!full_page_op) {
+			write_aux_reg(aux_tag, paddr);
+			paddr += L1_CACHE_BYTES;
+		}
+
+		write_aux_reg(aux_cmd, vaddr);
+		vaddr += L1_CACHE_BYTES;
+#else
+		write_aux_reg(aux, paddr);
+		paddr += L1_CACHE_BYTES;
+#endif
+	}
+}
 
 #ifdef CONFIG_ARC_HAS_DCACHE
 
@@ -289,53 +350,6 @@ static inline void __dc_entire_op(const int cacheop)
 		write_aux_reg(ARC_REG_DC_CTRL, tmp & ~DC_CTRL_INV_MODE_FLUSH);
 }
 
-/*
- * Per Line Operation on D-Cache
- * Doesn't deal with type-of-op/IRQ-disabling/waiting-for-flush-to-complete
- * It's sole purpose is to help gcc generate ZOL
- * (aliasing VIPT dcache flushing needs both vaddr and paddr)
- */
-static inline void __dc_line_loop(unsigned long paddr, unsigned long vaddr,
-				  unsigned long sz, const int aux_reg)
-{
-	int num_lines;
-
-	/* Ensure we properly floor/ceil the non-line aligned/sized requests
-	 * and have @paddr - aligned to cache line and integral @num_lines.
-	 * This however can be avoided for page sized since:
-	 *  -@paddr will be cache-line aligned already (being page aligned)
-	 *  -@sz will be integral multiple of line size (being page sized).
-	 */
-	if (!(__builtin_constant_p(sz) && sz == PAGE_SIZE)) {
-		sz += paddr & ~DCACHE_LINE_MASK;
-		paddr &= DCACHE_LINE_MASK;
-		vaddr &= DCACHE_LINE_MASK;
-	}
-
-	num_lines = DIV_ROUND_UP(sz, ARC_DCACHE_LINE_LEN);
-
-#if (CONFIG_ARC_MMU_VER <= 2)
-	paddr |= (vaddr >> PAGE_SHIFT) & 0x1F;
-#endif
-
-	while (num_lines-- > 0) {
-#if (CONFIG_ARC_MMU_VER > 2)
-		/*
-		 * Just as for I$, in MMU v3, D$ ops also require
-		 * "tag" bits in DC_PTAG, "index" bits in FLDL,IVDL ops
-		 */
-		write_aux_reg(ARC_REG_DC_PTAG, paddr);
-
-		write_aux_reg(aux_reg, vaddr);
-		vaddr += ARC_DCACHE_LINE_LEN;
-#else
-		/* paddr contains stuffed vaddrs bits */
-		write_aux_reg(aux_reg, paddr);
-#endif
-		paddr += ARC_DCACHE_LINE_LEN;
-	}
-}
-
 /* For kernel mappings cache operation: index is same as paddr */
 #define __dc_line_op_k(p, sz, op)	__dc_line_op(p, p, sz, op)
 
@@ -346,7 +360,6 @@ static inline void __dc_line_op(unsigned long paddr, unsigned long vaddr,
 				unsigned long sz, const int cacheop)
 {
 	unsigned long flags, tmp = tmp;
-	int aux;
 
 	local_irq_save(flags);
 
@@ -361,12 +374,7 @@ static inline void __dc_line_op(unsigned long paddr, unsigned long vaddr,
 		write_aux_reg(ARC_REG_DC_CTRL, tmp | DC_CTRL_INV_MODE_FLUSH);
 	}
 
-	if (cacheop & OP_INV)	/* Inv / flush-n-inv use same cmd reg */
-		aux = ARC_REG_DC_IVDL;
-	else
-		aux = ARC_REG_DC_FLDL;
-
-	__dc_line_loop(paddr, vaddr, sz, aux);
+	__cache_line_loop(paddr, vaddr, sz, cacheop);
 
 	if (cacheop & OP_FLUSH)	/* flush / flush-n-inv both wait */
 		wait_for_flush();
@@ -438,42 +446,9 @@ static void __ic_line_inv_vaddr(unsigned long paddr, unsigned long vaddr,
 				unsigned long sz)
 {
 	unsigned long flags;
-	int num_lines;
-
-	/*
-	 * Ensure we properly floor/ceil the non-line aligned/sized requests:
-	 * However page sized flushes can be compile time optimised.
-	 *  -@paddr will be cache-line aligned already (being page aligned)
-	 *  -@sz will be integral multiple of line size (being page sized).
-	 */
-	if (!(__builtin_constant_p(sz) && sz == PAGE_SIZE)) {
-		sz += paddr & ~ICACHE_LINE_MASK;
-		paddr &= ICACHE_LINE_MASK;
-		vaddr &= ICACHE_LINE_MASK;
-	}
-
-	num_lines = DIV_ROUND_UP(sz, ARC_ICACHE_LINE_LEN);
-
-#if (CONFIG_ARC_MMU_VER <= 2)
-	/* bits 17:13 of vaddr go as bits 4:0 of paddr */
-	paddr |= (vaddr >> PAGE_SHIFT) & 0x1F;
-#endif
 
 	local_irq_save(flags);
-	while (num_lines-- > 0) {
-#if (CONFIG_ARC_MMU_VER > 2)
-		/* tag comes from phy addr */
-		write_aux_reg(ARC_REG_IC_PTAG, paddr);
-
-		/* index bits come from vaddr */
-		write_aux_reg(ARC_REG_IC_IVIL, vaddr);
-		vaddr += ARC_ICACHE_LINE_LEN;
-#else
-		/* paddr contains stuffed vaddrs bits */
-		write_aux_reg(ARC_REG_IC_IVIL, paddr);
-#endif
-		paddr += ARC_ICACHE_LINE_LEN;
-	}
+	__cache_line_loop(paddr, vaddr, sz, OP_INV_IC);
 	local_irq_restore(flags);
 }
 
diff --git a/arch/arc/mm/fault.c b/arch/arc/mm/fault.c
index d63f3de0cd5..9c69552350c 100644
--- a/arch/arc/mm/fault.c
+++ b/arch/arc/mm/fault.c
@@ -17,7 +17,7 @@
 #include <asm/pgalloc.h>
 #include <asm/mmu.h>
 
-static int handle_vmalloc_fault(struct mm_struct *mm, unsigned long address)
+static int handle_vmalloc_fault(unsigned long address)
 {
 	/*
 	 * Synchronize this task's top level page-table
@@ -27,7 +27,7 @@ static int handle_vmalloc_fault(struct mm_struct *mm, unsigned long address)
 	pud_t *pud, *pud_k;
 	pmd_t *pmd, *pmd_k;
 
-	pgd = pgd_offset_fast(mm, address);
+	pgd = pgd_offset_fast(current->active_mm, address);
 	pgd_k = pgd_offset_k(address);
 
 	if (!pgd_present(*pgd_k))
@@ -52,7 +52,7 @@ bad_area:
 	return 1;
 }
 
-void do_page_fault(struct pt_regs *regs, unsigned long address)
+void do_page_fault(unsigned long address, struct pt_regs *regs)
 {
 	struct vm_area_struct *vma = NULL;
 	struct task_struct *tsk = current;
@@ -72,7 +72,7 @@ void do_page_fault(struct pt_regs *regs, unsigned long address)
 	 * nothing more.
 	 */
 	if (address >= VMALLOC_START && address <= VMALLOC_END) {
-		ret = handle_vmalloc_fault(mm, address);
+		ret = handle_vmalloc_fault(address);
 		if (unlikely(ret))
 			goto bad_area_nosemaphore;
 		else
diff --git a/arch/arc/mm/init.c b/arch/arc/mm/init.c
index 81279ec73a6..55e0a85bea7 100644
--- a/arch/arc/mm/init.c
+++ b/arch/arc/mm/init.c
@@ -125,10 +125,3 @@ void __init free_initrd_mem(unsigned long start, unsigned long end)
 	free_reserved_area((void *)start, (void *)end, -1, "initrd");
 }
 #endif
-
-#ifdef CONFIG_OF_FLATTREE
-void __init early_init_dt_setup_initrd_arch(u64 start, u64 end)
-{
-	pr_err("%s(%llx, %llx)\n", __func__, start, end);
-}
-#endif /* CONFIG_OF_FLATTREE */
diff --git a/arch/arc/mm/tlb.c b/arch/arc/mm/tlb.c
index 71cb26df425..e1acf0ce564 100644
--- a/arch/arc/mm/tlb.c
+++ b/arch/arc/mm/tlb.c
@@ -100,7 +100,7 @@
 
 
 /* A copy of the ASID from the PID reg is kept in asid_cache */
-unsigned int asid_cache = MM_CTXT_FIRST_CYCLE;
+DEFINE_PER_CPU(unsigned int, asid_cache) = MM_CTXT_FIRST_CYCLE;
 
 /*
  * Utility Routine to erase a J-TLB entry
@@ -274,6 +274,7 @@ noinline void local_flush_tlb_mm(struct mm_struct *mm)
 void local_flush_tlb_range(struct vm_area_struct *vma, unsigned long start,
 			   unsigned long end)
 {
+	const unsigned int cpu = smp_processor_id();
 	unsigned long flags;
 
 	/* If range @start to @end is more than 32 TLB entries deep,
@@ -297,9 +298,9 @@ void local_flush_tlb_range(struct vm_area_struct *vma, unsigned long start,
 
 	local_irq_save(flags);
 
-	if (vma->vm_mm->context.asid != MM_CTXT_NO_ASID) {
+	if (asid_mm(vma->vm_mm, cpu) != MM_CTXT_NO_ASID) {
 		while (start < end) {
-			tlb_entry_erase(start | hw_pid(vma->vm_mm));
+			tlb_entry_erase(start | hw_pid(vma->vm_mm, cpu));
 			start += PAGE_SIZE;
 		}
 	}
@@ -346,6 +347,7 @@ void local_flush_tlb_kernel_range(unsigned long start, unsigned long end)
 
 void local_flush_tlb_page(struct vm_area_struct *vma, unsigned long page)
 {
+	const unsigned int cpu = smp_processor_id();
 	unsigned long flags;
 
 	/* Note that it is critical that interrupts are DISABLED between
@@ -353,14 +355,87 @@ void local_flush_tlb_page(struct vm_area_struct *vma, unsigned long page)
 	 */
 	local_irq_save(flags);
 
-	if (vma->vm_mm->context.asid != MM_CTXT_NO_ASID) {
-		tlb_entry_erase((page & PAGE_MASK) | hw_pid(vma->vm_mm));
+	if (asid_mm(vma->vm_mm, cpu) != MM_CTXT_NO_ASID) {
+		tlb_entry_erase((page & PAGE_MASK) | hw_pid(vma->vm_mm, cpu));
 		utlb_invalidate();
 	}
 
 	local_irq_restore(flags);
 }
 
+#ifdef CONFIG_SMP
+
+struct tlb_args {
+	struct vm_area_struct *ta_vma;
+	unsigned long ta_start;
+	unsigned long ta_end;
+};
+
+static inline void ipi_flush_tlb_page(void *arg)
+{
+	struct tlb_args *ta = arg;
+
+	local_flush_tlb_page(ta->ta_vma, ta->ta_start);
+}
+
+static inline void ipi_flush_tlb_range(void *arg)
+{
+	struct tlb_args *ta = arg;
+
+	local_flush_tlb_range(ta->ta_vma, ta->ta_start, ta->ta_end);
+}
+
+static inline void ipi_flush_tlb_kernel_range(void *arg)
+{
+	struct tlb_args *ta = (struct tlb_args *)arg;
+
+	local_flush_tlb_kernel_range(ta->ta_start, ta->ta_end);
+}
+
+void flush_tlb_all(void)
+{
+	on_each_cpu((smp_call_func_t)local_flush_tlb_all, NULL, 1);
+}
+
+void flush_tlb_mm(struct mm_struct *mm)
+{
+	on_each_cpu_mask(mm_cpumask(mm), (smp_call_func_t)local_flush_tlb_mm,
+			 mm, 1);
+}
+
+void flush_tlb_page(struct vm_area_struct *vma, unsigned long uaddr)
+{
+	struct tlb_args ta = {
+		.ta_vma = vma,
+		.ta_start = uaddr
+	};
+
+	on_each_cpu_mask(mm_cpumask(vma->vm_mm), ipi_flush_tlb_page, &ta, 1);
+}
+
+void flush_tlb_range(struct vm_area_struct *vma, unsigned long start,
+		     unsigned long end)
+{
+	struct tlb_args ta = {
+		.ta_vma = vma,
+		.ta_start = start,
+		.ta_end = end
+	};
+
+	on_each_cpu_mask(mm_cpumask(vma->vm_mm), ipi_flush_tlb_range, &ta, 1);
+}
+
+void flush_tlb_kernel_range(unsigned long start, unsigned long end)
+{
+	struct tlb_args ta = {
+		.ta_start = start,
+		.ta_end = end
+	};
+
+	on_each_cpu(ipi_flush_tlb_kernel_range, &ta, 1);
+}
+#endif
+
 /*
  * Routine to create a TLB entry
  */
@@ -400,7 +475,7 @@ void create_tlb(struct vm_area_struct *vma, unsigned long address, pte_t *ptep)
 
 	local_irq_save(flags);
 
-	tlb_paranoid_check(vma->vm_mm->context.asid, address);
+	tlb_paranoid_check(asid_mm(vma->vm_mm, smp_processor_id()), address);
 
 	address &= PAGE_MASK;
 
@@ -610,9 +685,9 @@ void do_tlb_overlap_fault(unsigned long cause, unsigned long address,
 			  struct pt_regs *regs)
 {
 	int set, way, n;
-	unsigned int pd0[4], pd1[4];	/* assume max 4 ways */
 	unsigned long flags, is_valid;
 	struct cpuinfo_arc_mmu *mmu = &cpuinfo_arc700[smp_processor_id()].mmu;
+	unsigned int pd0[mmu->ways], pd1[mmu->ways];
 
 	local_irq_save(flags);
 
@@ -637,7 +712,7 @@ void do_tlb_overlap_fault(unsigned long cause, unsigned long address,
 			continue;
 
 		/* Scan the set for duplicate ways: needs a nested loop */
-		for (way = 0; way < mmu->ways; way++) {
+		for (way = 0; way < mmu->ways - 1; way++) {
 			if (!pd0[way])
 				continue;
 
diff --git a/arch/arc/mm/tlbex.S b/arch/arc/mm/tlbex.S
index cf7d7d9ad69..3fcfdb38d24 100644
--- a/arch/arc/mm/tlbex.S
+++ b/arch/arc/mm/tlbex.S
@@ -369,8 +369,8 @@ do_slow_path_pf:
 	EXCEPTION_PROLOGUE
 
 	; ------- setup args for Linux Page fault Hanlder ---------
-	mov_s r0, sp
-	lr  r1, [efa]
+	mov_s r1, sp
+	lr    r0, [efa]
 
 	; We don't want exceptions to be disabled while the fault is handled.
 	; Now that we have saved the context we return from exception hence