Merge branch 'linus' into sched/core, to pick up fixes and refresh the branch

Signed-off-by: Ingo Molnar <mingo@kernel.org>

16 files changed, 206 insertions, 93 deletions

diff --git a/arch/arc/Kconfig b/arch/arc/Kconfig
index c75d29077e4a..283099c9560a 100644
--- a/arch/arc/Kconfig
+++ b/arch/arc/Kconfig
@@ -29,7 +29,7 @@ config ARC
 	select HAVE_KPROBES
 	select HAVE_KRETPROBES
 	select HAVE_MEMBLOCK
-	select HAVE_MOD_ARCH_SPECIFIC if ARC_DW2_UNWIND
+	select HAVE_MOD_ARCH_SPECIFIC
 	select HAVE_OPROFILE
 	select HAVE_PERF_EVENTS
 	select HANDLE_DOMAIN_IRQ
diff --git a/arch/arc/include/asm/cache.h b/arch/arc/include/asm/cache.h
index b3410ff6a62d..5008021fba98 100644
--- a/arch/arc/include/asm/cache.h
+++ b/arch/arc/include/asm/cache.h
@@ -67,7 +67,7 @@ extern unsigned long perip_base, perip_end;
 #define ARC_REG_IC_PTAG_HI	0x1F
 
 /* Bit val in IC_CTRL */
-#define IC_CTRL_CACHE_DISABLE   0x1
+#define IC_CTRL_DIS		0x1
 
 /* Data cache related Auxiliary registers */
 #define ARC_REG_DC_BCR		0x72	/* Build Config reg */
@@ -80,8 +80,9 @@ extern unsigned long perip_base, perip_end;
 #define ARC_REG_DC_PTAG_HI	0x5F
 
 /* Bit val in DC_CTRL */
-#define DC_CTRL_INV_MODE_FLUSH  0x40
-#define DC_CTRL_FLUSH_STATUS    0x100
+#define DC_CTRL_DIS		0x001
+#define DC_CTRL_INV_MODE_FLUSH	0x040
+#define DC_CTRL_FLUSH_STATUS	0x100
 
 /*System-level cache (L2 cache) related Auxiliary registers */
 #define ARC_REG_SLC_CFG		0x901
@@ -92,8 +93,8 @@ extern unsigned long perip_base, perip_end;
 #define ARC_REG_SLC_RGN_END	0x916
 
 /* Bit val in SLC_CONTROL */
+#define SLC_CTRL_DIS		0x001
 #define SLC_CTRL_IM		0x040
-#define SLC_CTRL_DISABLE	0x001
 #define SLC_CTRL_BUSY		0x100
 #define SLC_CTRL_RGN_OP_INV	0x200
 
diff --git a/arch/arc/include/asm/delay.h b/arch/arc/include/asm/delay.h
index a36e8601114d..d5da2115d78a 100644
--- a/arch/arc/include/asm/delay.h
+++ b/arch/arc/include/asm/delay.h
@@ -26,7 +26,9 @@ static inline void __delay(unsigned long loops)
 	"	lp  1f			\n"
 	"	nop			\n"
 	"1:				\n"
-	: : "r"(loops));
+	:
+        : "r"(loops)
+        : "lp_count");
 }
 
 extern void __bad_udelay(void);
diff --git a/arch/arc/include/asm/entry-arcv2.h b/arch/arc/include/asm/entry-arcv2.h
index b5ff87e6f4b7..aee1a77934cf 100644
--- a/arch/arc/include/asm/entry-arcv2.h
+++ b/arch/arc/include/asm/entry-arcv2.h
@@ -16,6 +16,7 @@
 	;
 	; Now manually save: r12, sp, fp, gp, r25
 
+	PUSH	r30
 	PUSH	r12
 
 	; Saving pt_regs->sp correctly requires some extra work due to the way
@@ -72,6 +73,7 @@
 	POPAX	AUX_USER_SP
 1:
 	POP	r12
+	POP	r30
 
 .endm
 
diff --git a/arch/arc/include/asm/module.h b/arch/arc/include/asm/module.h
index 6e91d8b339c3..567590ea8f6c 100644
--- a/arch/arc/include/asm/module.h
+++ b/arch/arc/include/asm/module.h
@@ -14,13 +14,13 @@
 
 #include <asm-generic/module.h>
 
-#ifdef CONFIG_ARC_DW2_UNWIND
 struct mod_arch_specific {
+#ifdef CONFIG_ARC_DW2_UNWIND
 	void *unw_info;
 	int unw_sec_idx;
+#endif
 	const char *secstr;
 };
-#endif
 
 #define MODULE_PROC_FAMILY "ARC700"
 
diff --git a/arch/arc/include/asm/ptrace.h b/arch/arc/include/asm/ptrace.h
index 69095da1fcfd..47111d565a95 100644
--- a/arch/arc/include/asm/ptrace.h
+++ b/arch/arc/include/asm/ptrace.h
@@ -84,7 +84,7 @@ struct pt_regs {
 	unsigned long fp;
 	unsigned long sp;	/* user/kernel sp depending on where we came from  */
 
-	unsigned long r12;
+	unsigned long r12, r30;
 
 	/*------- Below list auto saved by h/w -----------*/
 	unsigned long r0, r1, r2, r3, r4, r5, r6, r7, r8, r9, r10, r11;
diff --git a/arch/arc/include/asm/setup.h b/arch/arc/include/asm/setup.h
index cb954cdab070..c568a9df82b1 100644
--- a/arch/arc/include/asm/setup.h
+++ b/arch/arc/include/asm/setup.h
@@ -31,6 +31,7 @@ extern int root_mountflags, end_mem;
 
 void setup_processor(void);
 void __init setup_arch_memory(void);
+long __init arc_get_mem_sz(void);
 
 /* Helpers used in arc_*_mumbojumbo routines */
 #define IS_AVAIL1(v, s)		((v) ? s : "")
diff --git a/arch/arc/kernel/head.S b/arch/arc/kernel/head.S
index 689dd867fdff..8b90d25a15cc 100644
--- a/arch/arc/kernel/head.S
+++ b/arch/arc/kernel/head.S
@@ -71,14 +71,14 @@ ENTRY(stext)
 	GET_CPU_ID  r5
 	cmp	r5, 0
 	mov.nz	r0, r5
-#ifdef CONFIG_ARC_SMP_HALT_ON_RESET
-	; Non-Master can proceed as system would be booted sufficiently
-	jnz	first_lines_of_secondary
-#else
+	bz	.Lmaster_proceed
+
 	; Non-Masters wait for Master to boot enough and bring them up
-	jnz	arc_platform_smp_wait_to_boot
-#endif
-	; Master falls thru
+	; when they resume, tail-call to entry point
+	mov	blink, @first_lines_of_secondary
+	j	arc_platform_smp_wait_to_boot
+
+.Lmaster_proceed:
 #endif
 
 	; Clear BSS before updating any globals
diff --git a/arch/arc/kernel/intc-arcv2.c b/arch/arc/kernel/intc-arcv2.c
index 994dca7014db..ecef0fb0b66c 100644
--- a/arch/arc/kernel/intc-arcv2.c
+++ b/arch/arc/kernel/intc-arcv2.c
@@ -77,20 +77,20 @@ void arc_init_IRQ(void)
 
 static void arcv2_irq_mask(struct irq_data *data)
 {
-	write_aux_reg(AUX_IRQ_SELECT, data->irq);
+	write_aux_reg(AUX_IRQ_SELECT, data->hwirq);
 	write_aux_reg(AUX_IRQ_ENABLE, 0);
 }
 
 static void arcv2_irq_unmask(struct irq_data *data)
 {
-	write_aux_reg(AUX_IRQ_SELECT, data->irq);
+	write_aux_reg(AUX_IRQ_SELECT, data->hwirq);
 	write_aux_reg(AUX_IRQ_ENABLE, 1);
 }
 
 void arcv2_irq_enable(struct irq_data *data)
 {
 	/* set default priority */
-	write_aux_reg(AUX_IRQ_SELECT, data->irq);
+	write_aux_reg(AUX_IRQ_SELECT, data->hwirq);
 	write_aux_reg(AUX_IRQ_PRIORITY, ARCV2_IRQ_DEF_PRIO);
 
 	/*
diff --git a/arch/arc/kernel/intc-compact.c b/arch/arc/kernel/intc-compact.c
index ce9deb953ca9..8c1fd5c00782 100644
--- a/arch/arc/kernel/intc-compact.c
+++ b/arch/arc/kernel/intc-compact.c
@@ -57,7 +57,7 @@ static void arc_irq_mask(struct irq_data *data)
 	unsigned int ienb;
 
 	ienb = read_aux_reg(AUX_IENABLE);
-	ienb &= ~(1 << data->irq);
+	ienb &= ~(1 << data->hwirq);
 	write_aux_reg(AUX_IENABLE, ienb);
 }
 
@@ -66,7 +66,7 @@ static void arc_irq_unmask(struct irq_data *data)
 	unsigned int ienb;
 
 	ienb = read_aux_reg(AUX_IENABLE);
-	ienb |= (1 << data->irq);
+	ienb |= (1 << data->hwirq);
 	write_aux_reg(AUX_IENABLE, ienb);
 }
 
diff --git a/arch/arc/kernel/mcip.c b/arch/arc/kernel/mcip.c
index 560c4afc2af4..9f6b68fd4f3b 100644
--- a/arch/arc/kernel/mcip.c
+++ b/arch/arc/kernel/mcip.c
@@ -10,6 +10,7 @@
 
 #include <linux/smp.h>
 #include <linux/irq.h>
+#include <linux/irqchip/chained_irq.h>
 #include <linux/spinlock.h>
 #include <soc/arc/mcip.h>
 #include <asm/irqflags-arcv2.h>
@@ -92,11 +93,10 @@ static void mcip_probe_n_setup(void)
 	READ_BCR(ARC_REG_MCIP_BCR, mp);
 
 	sprintf(smp_cpuinfo_buf,
-		"Extn [SMP]\t: ARConnect (v%d): %d cores with %s%s%s%s%s\n",
+		"Extn [SMP]\t: ARConnect (v%d): %d cores with %s%s%s%s\n",
 		mp.ver, mp.num_cores,
 		IS_AVAIL1(mp.ipi, "IPI "),
 		IS_AVAIL1(mp.idu, "IDU "),
-		IS_AVAIL1(mp.llm, "LLM "),
 		IS_AVAIL1(mp.dbg, "DEBUG "),
 		IS_AVAIL1(mp.gfrc, "GFRC"));
 
@@ -174,7 +174,6 @@ static void idu_irq_unmask(struct irq_data *data)
 	raw_spin_unlock_irqrestore(&mcip_lock, flags);
 }
 
-#ifdef CONFIG_SMP
 static int
 idu_irq_set_affinity(struct irq_data *data, const struct cpumask *cpumask,
 		     bool force)
@@ -204,12 +203,27 @@ idu_irq_set_affinity(struct irq_data *data, const struct cpumask *cpumask,
 
 	return IRQ_SET_MASK_OK;
 }
-#endif
+
+static void idu_irq_enable(struct irq_data *data)
+{
+	/*
+	 * By default send all common interrupts to all available online CPUs.
+	 * The affinity of common interrupts in IDU must be set manually since
+	 * in some cases the kernel will not call irq_set_affinity() by itself:
+	 *   1. When the kernel is not configured with support of SMP.
+	 *   2. When the kernel is configured with support of SMP but upper
+	 *      interrupt controllers does not support setting of the affinity
+	 *      and cannot propagate it to IDU.
+	 */
+	idu_irq_set_affinity(data, cpu_online_mask, false);
+	idu_irq_unmask(data);
+}
 
 static struct irq_chip idu_irq_chip = {
 	.name			= "MCIP IDU Intc",
 	.irq_mask		= idu_irq_mask,
 	.irq_unmask		= idu_irq_unmask,
+	.irq_enable		= idu_irq_enable,
 #ifdef CONFIG_SMP
 	.irq_set_affinity       = idu_irq_set_affinity,
 #endif
@@ -221,10 +235,13 @@ static irq_hw_number_t idu_first_hwirq;
 static void idu_cascade_isr(struct irq_desc *desc)
 {
 	struct irq_domain *idu_domain = irq_desc_get_handler_data(desc);
+	struct irq_chip *core_chip = irq_desc_get_chip(desc);
 	irq_hw_number_t core_hwirq = irqd_to_hwirq(irq_desc_get_irq_data(desc));
 	irq_hw_number_t idu_hwirq = core_hwirq - idu_first_hwirq;
 
+	chained_irq_enter(core_chip, desc);
 	generic_handle_irq(irq_find_mapping(idu_domain, idu_hwirq));
+	chained_irq_exit(core_chip, desc);
 }
 
 static int idu_irq_map(struct irq_domain *d, unsigned int virq, irq_hw_number_t hwirq)
@@ -239,36 +256,14 @@ static int idu_irq_xlate(struct irq_domain *d, struct device_node *n,
 			 const u32 *intspec, unsigned int intsize,
 			 irq_hw_number_t *out_hwirq, unsigned int *out_type)
 {
-	irq_hw_number_t hwirq = *out_hwirq = intspec[0];
-	int distri = intspec[1];
-	unsigned long flags;
-
+	/*
+	 * Ignore value of interrupt distribution mode for common interrupts in
+	 * IDU which resides in intspec[1] since setting an affinity using value
+	 * from Device Tree is deprecated in ARC.
+	 */
+	*out_hwirq = intspec[0];
 	*out_type = IRQ_TYPE_NONE;
 
-	/* XXX: validate distribution scheme again online cpu mask */
-	if (distri == 0) {
-		/* 0 - Round Robin to all cpus, otherwise 1 bit per core */
-		raw_spin_lock_irqsave(&mcip_lock, flags);
-		idu_set_dest(hwirq, BIT(num_online_cpus()) - 1);
-		idu_set_mode(hwirq, IDU_M_TRIG_LEVEL, IDU_M_DISTRI_RR);
-		raw_spin_unlock_irqrestore(&mcip_lock, flags);
-	} else {
-		/*
-		 * DEST based distribution for Level Triggered intr can only
-		 * have 1 CPU, so generalize it to always contain 1 cpu
-		 */
-		int cpu = ffs(distri);
-
-		if (cpu != fls(distri))
-			pr_warn("IDU irq %lx distri mode set to cpu %x\n",
-				hwirq, cpu);
-
-		raw_spin_lock_irqsave(&mcip_lock, flags);
-		idu_set_dest(hwirq, cpu);
-		idu_set_mode(hwirq, IDU_M_TRIG_LEVEL, IDU_M_DISTRI_DEST);
-		raw_spin_unlock_irqrestore(&mcip_lock, flags);
-	}
-
 	return 0;
 }
 
diff --git a/arch/arc/kernel/module.c b/arch/arc/kernel/module.c
index 42e964db2967..3d99a6091332 100644
--- a/arch/arc/kernel/module.c
+++ b/arch/arc/kernel/module.c
@@ -32,8 +32,8 @@ int module_frob_arch_sections(Elf_Ehdr *hdr, Elf_Shdr *sechdrs,
 #ifdef CONFIG_ARC_DW2_UNWIND
 	mod->arch.unw_sec_idx = 0;
 	mod->arch.unw_info = NULL;
-	mod->arch.secstr = secstr;
 #endif
+	mod->arch.secstr = secstr;
 	return 0;
 }
 
@@ -113,8 +113,10 @@ int apply_relocate_add(Elf32_Shdr *sechdrs,
 
 	}
 
+#ifdef CONFIG_ARC_DW2_UNWIND
 	if (strcmp(module->arch.secstr+sechdrs[tgtsec].sh_name, ".eh_frame") == 0)
 		module->arch.unw_sec_idx = tgtsec;
+#endif
 
 	return 0;
 
diff --git a/arch/arc/kernel/smp.c b/arch/arc/kernel/smp.c
index 88674d972c9d..2afbafadb6ab 100644
--- a/arch/arc/kernel/smp.c
+++ b/arch/arc/kernel/smp.c
@@ -90,22 +90,37 @@ void __init smp_cpus_done(unsigned int max_cpus)
  */
 static volatile int wake_flag;
 
+#ifdef CONFIG_ISA_ARCOMPACT
+
+#define __boot_read(f)		f
+#define __boot_write(f, v)	f = v
+
+#else
+
+#define __boot_read(f)		arc_read_uncached_32(&f)
+#define __boot_write(f, v)	arc_write_uncached_32(&f, v)
+
+#endif
+
 static void arc_default_smp_cpu_kick(int cpu, unsigned long pc)
 {
 	BUG_ON(cpu == 0);
-	wake_flag = cpu;
+
+	__boot_write(wake_flag, cpu);
 }
 
 void arc_platform_smp_wait_to_boot(int cpu)
 {
-	while (wake_flag != cpu)
+	/* for halt-on-reset, we've waited already */
+	if (IS_ENABLED(CONFIG_ARC_SMP_HALT_ON_RESET))
+		return;
+
+	while (__boot_read(wake_flag) != cpu)
 		;
 
-	wake_flag = 0;
-	__asm__ __volatile__("j @first_lines_of_secondary	\n");
+	__boot_write(wake_flag, 0);
 }
 
-
 const char *arc_platform_smp_cpuinfo(void)
 {
 	return plat_smp_ops.info ? : "";
diff --git a/arch/arc/kernel/unaligned.c b/arch/arc/kernel/unaligned.c
index abd961f3e763..91ebe382147f 100644
--- a/arch/arc/kernel/unaligned.c
+++ b/arch/arc/kernel/unaligned.c
@@ -241,8 +241,9 @@ int misaligned_fixup(unsigned long address, struct pt_regs *regs,
 	if (state.fault)
 		goto fault;
 
+	/* clear any remanants of delay slot */
 	if (delay_mode(regs)) {
-		regs->ret = regs->bta;
+		regs->ret = regs->bta ~1U;
 		regs->status32 &= ~STATUS_DE_MASK;
 	} else {
 		regs->ret += state.instr_len;
diff --git a/arch/arc/mm/cache.c b/arch/arc/mm/cache.c
index ec86ac0e3321..d408fa21a07c 100644
--- a/arch/arc/mm/cache.c
+++ b/arch/arc/mm/cache.c
@@ -23,7 +23,7 @@
 
 static int l2_line_sz;
 static int ioc_exists;
-int slc_enable = 1, ioc_enable = 0;
+int slc_enable = 1, ioc_enable = 1;
 unsigned long perip_base = ARC_UNCACHED_ADDR_SPACE; /* legacy value for boot */
 unsigned long perip_end = 0xFFFFFFFF; /* legacy value */
 
@@ -271,7 +271,11 @@ void __cache_line_loop_v2(phys_addr_t paddr, unsigned long vaddr,
 
 /*
  * For ARC700 MMUv3 I-cache and D-cache flushes
- * Also reused for HS38 aliasing I-cache configuration
+ *  - ARC700 programming model requires paddr and vaddr be passed in seperate
+ *    AUX registers (*_IV*L and *_PTAG respectively) irrespective of whether the
+ *    caches actually alias or not.
+ * -  For HS38, only the aliasing I-cache configuration uses the PTAG reg
+ *    (non aliasing I-cache version doesn't; while D-cache can't possibly alias)
  */
 static inline
 void __cache_line_loop_v3(phys_addr_t paddr, unsigned long vaddr,
@@ -458,6 +462,21 @@ static inline void __dc_entire_op(const int op)
 	__after_dc_op(op);
 }
 
+static inline void __dc_disable(void)
+{
+	const int r = ARC_REG_DC_CTRL;
+
+	__dc_entire_op(OP_FLUSH_N_INV);
+	write_aux_reg(r, read_aux_reg(r) | DC_CTRL_DIS);
+}
+
+static void __dc_enable(void)
+{
+	const int r = ARC_REG_DC_CTRL;
+
+	write_aux_reg(r, read_aux_reg(r) & ~DC_CTRL_DIS);
+}
+
 /* 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)
 
@@ -483,6 +502,8 @@ static inline void __dc_line_op(phys_addr_t paddr, unsigned long vaddr,
 #else
 
 #define __dc_entire_op(op)
+#define __dc_disable()
+#define __dc_enable()
 #define __dc_line_op(paddr, vaddr, sz, op)
 #define __dc_line_op_k(paddr, sz, op)
 
@@ -597,6 +618,40 @@ noinline void slc_op(phys_addr_t paddr, unsigned long sz, const int op)
 #endif
 }
 
+noinline static void slc_entire_op(const int op)
+{
+	unsigned int ctrl, r = ARC_REG_SLC_CTRL;
+
+	ctrl = read_aux_reg(r);
+
+	if (!(op & OP_FLUSH))		/* i.e. OP_INV */
+		ctrl &= ~SLC_CTRL_IM;	/* clear IM: Disable flush before Inv */
+	else
+		ctrl |= SLC_CTRL_IM;
+
+	write_aux_reg(r, ctrl);
+
+	write_aux_reg(ARC_REG_SLC_INVALIDATE, 1);
+
+	/* Important to wait for flush to complete */
+	while (read_aux_reg(r) & SLC_CTRL_BUSY);
+}
+
+static inline void arc_slc_disable(void)
+{
+	const int r = ARC_REG_SLC_CTRL;
+
+	slc_entire_op(OP_FLUSH_N_INV);
+	write_aux_reg(r, read_aux_reg(r) | SLC_CTRL_DIS);
+}
+
+static inline void arc_slc_enable(void)
+{
+	const int r = ARC_REG_SLC_CTRL;
+
+	write_aux_reg(r, read_aux_reg(r) & ~SLC_CTRL_DIS);
+}
+
 /***********************************************************
  * Exported APIs
  */
@@ -923,21 +978,54 @@ SYSCALL_DEFINE3(cacheflush, uint32_t, start, uint32_t, sz, uint32_t, flags)
 	return 0;
 }
 
-void arc_cache_init(void)
+/*
+ * IO-Coherency (IOC) setup rules:
+ *
+ * 1. Needs to be at system level, so only once by Master core
+ *    Non-Masters need not be accessing caches at that time
+ *    - They are either HALT_ON_RESET and kick started much later or
+ *    - if run on reset, need to ensure that arc_platform_smp_wait_to_boot()
+ *      doesn't perturb caches or coherency unit
+ *
+ * 2. caches (L1 and SLC) need to be purged (flush+inv) before setting up IOC,
+ *    otherwise any straggler data might behave strangely post IOC enabling
+ *
+ * 3. All Caches need to be disabled when setting up IOC to elide any in-flight
+ *    Coherency transactions
+ */
+noinline void __init arc_ioc_setup(void)
 {
-	unsigned int __maybe_unused cpu = smp_processor_id();
-	char str[256];
+	unsigned int ap_sz;
 
-	printk(arc_cache_mumbojumbo(0, str, sizeof(str)));
+	/* Flush + invalidate + disable L1 dcache */
+	__dc_disable();
+
+	/* Flush + invalidate SLC */
+	if (read_aux_reg(ARC_REG_SLC_BCR))
+		slc_entire_op(OP_FLUSH_N_INV);
+
+	/* IOC Aperture start: TDB: handle non default CONFIG_LINUX_LINK_BASE */
+	write_aux_reg(ARC_REG_IO_COH_AP0_BASE, 0x80000);
 
 	/*
-	 * Only master CPU needs to execute rest of function:
-	 *  - Assume SMP so all cores will have same cache config so
-	 *    any geomtry checks will be same for all
-	 *  - IOC setup / dma callbacks only need to be setup once
+	 * IOC Aperture size:
+	 *   decoded as 2 ^ (SIZE + 2) KB: so setting 0x11 implies 512M
+	 * TBD: fix for PGU + 1GB of low mem
+	 * TBD: fix for PAE
 	 */
-	if (cpu)
-		return;
+	ap_sz = order_base_2(arc_get_mem_sz()/1024) - 2;
+	write_aux_reg(ARC_REG_IO_COH_AP0_SIZE, ap_sz);
+
+	write_aux_reg(ARC_REG_IO_COH_PARTIAL, 1);
+	write_aux_reg(ARC_REG_IO_COH_ENABLE, 1);
+
+	/* Re-enable L1 dcache */
+	__dc_enable();
+}
+
+void __init arc_cache_init_master(void)
+{
+	unsigned int __maybe_unused cpu = smp_processor_id();
 
 	if (IS_ENABLED(CONFIG_ARC_HAS_ICACHE)) {
 		struct cpuinfo_arc_cache *ic = &cpuinfo_arc700[cpu].icache;
@@ -985,30 +1073,14 @@ void arc_cache_init(void)
 		}
 	}
 
-	if (is_isa_arcv2() && l2_line_sz && !slc_enable) {
-
-		/* IM set : flush before invalidate */
-		write_aux_reg(ARC_REG_SLC_CTRL,
-			read_aux_reg(ARC_REG_SLC_CTRL) | SLC_CTRL_IM);
+	/* Note that SLC disable not formally supported till HS 3.0 */
+	if (is_isa_arcv2() && l2_line_sz && !slc_enable)
+		arc_slc_disable();
 
-		write_aux_reg(ARC_REG_SLC_INVALIDATE, 1);
-
-		/* Important to wait for flush to complete */
-		while (read_aux_reg(ARC_REG_SLC_CTRL) & SLC_CTRL_BUSY);
-		write_aux_reg(ARC_REG_SLC_CTRL,
-			read_aux_reg(ARC_REG_SLC_CTRL) | SLC_CTRL_DISABLE);
-	}
+	if (is_isa_arcv2() && ioc_enable)
+		arc_ioc_setup();
 
 	if (is_isa_arcv2() && ioc_enable) {
-		/* IO coherency base - 0x8z */
-		write_aux_reg(ARC_REG_IO_COH_AP0_BASE, 0x80000);
-		/* IO coherency aperture size - 512Mb: 0x8z-0xAz */
-		write_aux_reg(ARC_REG_IO_COH_AP0_SIZE, 0x11);
-		/* Enable partial writes */
-		write_aux_reg(ARC_REG_IO_COH_PARTIAL, 1);
-		/* Enable IO coherency */
-		write_aux_reg(ARC_REG_IO_COH_ENABLE, 1);
-
 		__dma_cache_wback_inv = __dma_cache_wback_inv_ioc;
 		__dma_cache_inv = __dma_cache_inv_ioc;
 		__dma_cache_wback = __dma_cache_wback_ioc;
@@ -1022,3 +1094,20 @@ void arc_cache_init(void)
 		__dma_cache_wback = __dma_cache_wback_l1;
 	}
 }
+
+void __ref arc_cache_init(void)
+{
+	unsigned int __maybe_unused cpu = smp_processor_id();
+	char str[256];
+
+	printk(arc_cache_mumbojumbo(0, str, sizeof(str)));
+
+	/*
+	 * Only master CPU needs to execute rest of function:
+	 *  - Assume SMP so all cores will have same cache config so
+	 *    any geomtry checks will be same for all
+	 *  - IOC setup / dma callbacks only need to be setup once
+	 */
+	if (!cpu)
+		arc_cache_init_master();
+}
diff --git a/arch/arc/mm/init.c b/arch/arc/mm/init.c
index 399e2f223d25..8c9415ed6280 100644
--- a/arch/arc/mm/init.c
+++ b/arch/arc/mm/init.c
@@ -40,6 +40,11 @@ struct pglist_data node_data[MAX_NUMNODES] __read_mostly;
 EXPORT_SYMBOL(node_data);
 #endif
 
+long __init arc_get_mem_sz(void)
+{
+	return low_mem_sz;
+}
+
 /* User can over-ride above with "mem=nnn[KkMm]" in cmdline */
 static int __init setup_mem_sz(char *str)
 {