1 files changed, 108 insertions, 173 deletions

diff --git a/arch/arc/mm/cache_arc700.c b/arch/arc/mm/cache_arc700.c
index 5a1259cd948..353b202c37c 100644
--- a/arch/arc/mm/cache_arc700.c
+++ b/arch/arc/mm/cache_arc700.c
@@ -73,37 +73,9 @@
 #include <asm/cachectl.h>
 #include <asm/setup.h>
 
-/* Instruction cache related Auxiliary registers */
-#define ARC_REG_IC_BCR		0x77	/* Build Config reg */
-#define ARC_REG_IC_IVIC		0x10
-#define ARC_REG_IC_CTRL		0x11
-#define ARC_REG_IC_IVIL		0x19
-#if (CONFIG_ARC_MMU_VER > 2)
-#define ARC_REG_IC_PTAG		0x1E
-#endif
-
-/* Bit val in IC_CTRL */
-#define IC_CTRL_CACHE_DISABLE   0x1
-
-/* Data cache related Auxiliary registers */
-#define ARC_REG_DC_BCR		0x72	/* Build Config reg */
-#define ARC_REG_DC_IVDC		0x47
-#define ARC_REG_DC_CTRL		0x48
-#define ARC_REG_DC_IVDL		0x4A
-#define ARC_REG_DC_FLSH		0x4B
-#define ARC_REG_DC_FLDL		0x4C
-#if (CONFIG_ARC_MMU_VER > 2)
-#define ARC_REG_DC_PTAG		0x5C
-#endif
-
-/* Bit val in DC_CTRL */
-#define DC_CTRL_INV_MODE_FLUSH  0x40
-#define DC_CTRL_FLUSH_STATUS    0x100
-
-char *arc_cache_mumbojumbo(int cpu_id, char *buf, int len)
+char *arc_cache_mumbojumbo(int c, char *buf, int len)
 {
 	int n = 0;
-	unsigned int c = smp_processor_id();
 
 #define PR_CACHE(p, enb, str)						\
 {									\
@@ -169,78 +141,114 @@ void read_decode_cache_bcr(void)
  */
 void arc_cache_init(void)
 {
-	unsigned int cpu = smp_processor_id();
-	struct cpuinfo_arc_cache *ic = &cpuinfo_arc700[cpu].icache;
-	struct cpuinfo_arc_cache *dc = &cpuinfo_arc700[cpu].dcache;
-	unsigned int dcache_does_alias, temp;
+	unsigned int __maybe_unused cpu = smp_processor_id();
+	struct cpuinfo_arc_cache __maybe_unused *ic, __maybe_unused *dc;
 	char str[256];
 
 	printk(arc_cache_mumbojumbo(0, str, sizeof(str)));
 
-	if (!ic->ver)
-		goto chk_dc;
-
 #ifdef CONFIG_ARC_HAS_ICACHE
-	/* 1. Confirm some of I-cache params which Linux assumes */
-	if (ic->line_len != ARC_ICACHE_LINE_LEN)
-		panic("Cache H/W doesn't match kernel Config");
-
-	if (ic->ver != CONFIG_ARC_MMU_VER)
-		panic("Cache ver doesn't match MMU ver\n");
+	ic = &cpuinfo_arc700[cpu].icache;
+	if (ic->ver) {
+		if (ic->line_len != L1_CACHE_BYTES)
+			panic("ICache line [%d] != kernel Config [%d]",
+			      ic->line_len, L1_CACHE_BYTES);
+
+		if (ic->ver != CONFIG_ARC_MMU_VER)
+			panic("Cache ver [%d] doesn't match MMU ver [%d]\n",
+			      ic->ver, CONFIG_ARC_MMU_VER);
+	}
 #endif
 
-	/* Enable/disable I-Cache */
-	temp = read_aux_reg(ARC_REG_IC_CTRL);
+#ifdef CONFIG_ARC_HAS_DCACHE
+	dc = &cpuinfo_arc700[cpu].dcache;
+	if (dc->ver) {
+		unsigned int dcache_does_alias;
 
-#ifdef CONFIG_ARC_HAS_ICACHE
-	temp &= ~IC_CTRL_CACHE_DISABLE;
-#else
-	temp |= IC_CTRL_CACHE_DISABLE;
-#endif
+		if (dc->line_len != L1_CACHE_BYTES)
+			panic("DCache line [%d] != kernel Config [%d]",
+			      dc->line_len, L1_CACHE_BYTES);
 
-	write_aux_reg(ARC_REG_IC_CTRL, temp);
+		/* check for D-Cache aliasing */
+		dcache_does_alias = (dc->sz / dc->assoc) > PAGE_SIZE;
 
-chk_dc:
-	if (!dc->ver)
-		return;
+		if (dcache_does_alias && !cache_is_vipt_aliasing())
+			panic("Enable CONFIG_ARC_CACHE_VIPT_ALIASING\n");
+		else if (!dcache_does_alias && cache_is_vipt_aliasing())
+			panic("Don't need CONFIG_ARC_CACHE_VIPT_ALIASING\n");
+	}
+#endif
+}
 
-#ifdef CONFIG_ARC_HAS_DCACHE
-	if (dc->line_len != ARC_DCACHE_LINE_LEN)
-		panic("Cache H/W doesn't match kernel Config");
+#define OP_INV		0x1
+#define OP_FLUSH	0x2
+#define OP_FLUSH_N_INV	0x3
+#define OP_INV_IC	0x4
 
-	/* check for D-Cache aliasing */
-	dcache_does_alias = (dc->sz / dc->assoc) > PAGE_SIZE;
+/*
+ * 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 (dcache_does_alias && !cache_is_vipt_aliasing())
-		panic("Enable CONFIG_ARC_CACHE_VIPT_ALIASING\n");
-	else if (!dcache_does_alias && cache_is_vipt_aliasing())
-		panic("Don't need CONFIG_ARC_CACHE_VIPT_ALIASING\n");
+	if (cacheop == OP_INV_IC) {
+		aux_cmd = ARC_REG_IC_IVIL;
+#if (CONFIG_ARC_MMU_VER > 2)
+		aux_tag = ARC_REG_IC_PTAG;
 #endif
+	}
+	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;
+#if (CONFIG_ARC_MMU_VER > 2)
+		aux_tag = ARC_REG_DC_PTAG;
+#endif
+	}
 
-	/* Set the default Invalidate Mode to "simpy discard dirty lines"
-	 *  as this is more frequent then flush before invalidate
-	 * Ofcourse we toggle this default behviour when desired
+	/* 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).
 	 */
-	temp = read_aux_reg(ARC_REG_DC_CTRL);
-	temp &= ~DC_CTRL_INV_MODE_FLUSH;
+	if (!full_page_op) {
+		sz += paddr & ~CACHE_LINE_MASK;
+		paddr &= CACHE_LINE_MASK;
+		vaddr &= CACHE_LINE_MASK;
+	}
 
-#ifdef CONFIG_ARC_HAS_DCACHE
-	/* Enable D-Cache: Clear Bit 0 */
-	write_aux_reg(ARC_REG_DC_CTRL, temp & ~IC_CTRL_CACHE_DISABLE);
+	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
-	/* Flush D cache */
-	write_aux_reg(ARC_REG_DC_FLSH, 0x1);
-	/* Disable D cache */
-	write_aux_reg(ARC_REG_DC_CTRL, temp | IC_CTRL_CACHE_DISABLE);
+	/* if V-P const for loop, PTAG can be written once outside loop */
+	if (full_page_op)
+		write_aux_reg(aux_tag, paddr);
 #endif
 
-	return;
+	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_cmd, paddr);
+		paddr += L1_CACHE_BYTES;
+#endif
+	}
 }
 
-#define OP_INV		0x1
-#define OP_FLUSH	0x2
-#define OP_FLUSH_N_INV	0x3
-
 #ifdef CONFIG_ARC_HAS_DCACHE
 
 /***************************************************************
@@ -289,53 +297,6 @@ 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 +307,6 @@ 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 +321,7 @@ 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();
@@ -434,46 +389,13 @@ static inline void __dc_line_op(unsigned long paddr, unsigned long vaddr,
 /***********************************************************
  * Machine specific helper for per line I-Cache invalidate.
  */
-static void __ic_line_inv_vaddr(unsigned long paddr, unsigned long vaddr,
+static void __ic_line_inv_vaddr_local(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);
 }
 
@@ -483,6 +405,23 @@ static inline void __ic_entire_inv(void)
 	read_aux_reg(ARC_REG_IC_CTRL);	/* blocks */
 }
 
+struct ic_line_inv_vaddr_ipi {
+	unsigned long paddr, vaddr;
+	int sz;
+};
+
+static void __ic_line_inv_vaddr_helper(void *info)
+{
+        struct ic_line_inv_vaddr_ipi *ic_inv = (struct ic_line_inv_vaddr_ipi*) info;
+        __ic_line_inv_vaddr_local(ic_inv->paddr, ic_inv->vaddr, ic_inv->sz);
+}
+
+static void __ic_line_inv_vaddr(unsigned long paddr, unsigned long vaddr,
+				unsigned long sz)
+{
+	struct ic_line_inv_vaddr_ipi ic_inv = { paddr, vaddr , sz};
+	on_each_cpu(__ic_line_inv_vaddr_helper, &ic_inv, 1);
+}
 #else
 
 #define __ic_entire_inv()
@@ -631,12 +570,8 @@ void flush_icache_range(unsigned long kstart, unsigned long kend)
  */
 void __sync_icache_dcache(unsigned long paddr, unsigned long vaddr, int len)
 {
-	unsigned long flags;
-
-	local_irq_save(flags);
-	__ic_line_inv_vaddr(paddr, vaddr, len);
 	__dc_line_op(paddr, vaddr, len, OP_FLUSH_N_INV);
-	local_irq_restore(flags);
+	__ic_line_inv_vaddr(paddr, vaddr, len);
 }
 
 /* wrapper to compile time eliminate alignment checks in flush loop */