Merge tag 'arc-4.4-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/vgupta/arc

Pull ARC updates from Vineet Gupta:

 - Support for new MM features in ARCv2 cores (THP, PAE40) Some generic
   THP bits are touched - all ACKed by Kirill

 - Platform framework updates to prepare for EZChip arrival (still in works)

 - ARC Public Mailing list setup finally (linux-snps-arc@lists.infraded.org)

* tag 'arc-4.4-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/vgupta/arc: (42 commits)
  ARC: mm: PAE40 support
  ARC: mm: PAE40: tlbex.S: Explicitify the size of pte_t
  ARC: mm: PAE40: switch to using phys_addr_t for physical addresses
  ARC: mm: HIGHMEM: populate high memory from DT
  ARC: mm: HIGHMEM: kmap API implementation
  ARC: mm: preps ahead of HIGHMEM support #2
  ARC: mm: preps ahead of HIGHMEM support
  ARC: mm: use generic macros _BITUL()/_AC()
  ARC: mm: Improve Duplicate PD Fault handler
  MAINTAINERS: Add public mailing list for ARC
  ARC: Ensure DT mem base is same as what kernel is built with
  ARC: boot: Non Master cpus only need to call EARLY_CPU_SETUP once
  ARCv2: smp: [plat-*]: No need to explicitly call mcip_init_smp()
  ARC: smp: Introduce smp hook @init_irq_cpu called for all cores
  ARC: smp: Rename platform hook @init_smp -> @init_cpu_smp
  ARCv2: smp: [plat-*]: No need to explicitly call mcip_init_early_smp()
  ARC: smp: Introduce smp hook @init_early_smp for Master core
  ARC: remove @init_time, @init_irq platform callbacks
  ARC: smp: irqchip: handle IPI as percpu irq like timer
  ARC: boot: Support Halt-on-reset and Run-on-reset SMP booting modes
  ...

Documentation/features/vm/THP/arch-support.txt

@@ -7,7 +7,7 @@
     |         arch |status|
     -----------------------
     |       alpha: | TODO |
-    |         arc: |  ..  |
+    |         arc: |  ok  |
     |         arm: |  ok  |
     |       arm64: |  ok  |
     |       avr32: |  ..  |

Documentation/features/vm/pte_special/arch-support.txt

@@ -7,7 +7,7 @@
     |         arch |status|
     -----------------------
     |       alpha: | TODO |
-    |         arc: | TODO |
+    |         arc: |  ok  |
     |         arm: |  ok  |
     |       arm64: |  ok  |
     |       avr32: | TODO |

MAINTAINERS

@@ -10121,6 +10121,7 @@ F:	include/net/switchdev.h
 
 SYNOPSYS ARC ARCHITECTURE
 M:	Vineet Gupta <vgupta@synopsys.com>
+L:	linux-snps-arc@lists.infraded.org
 S:	Supported
 F:	arch/arc/
 F:	Documentation/devicetree/bindings/arc/*

arch/arc/Kconfig

@@ -76,6 +76,10 @@ config STACKTRACE_SUPPORT
 config HAVE_LATENCYTOP_SUPPORT
 	def_bool y
 
+config HAVE_ARCH_TRANSPARENT_HUGEPAGE
+	def_bool y
+	depends on ARC_MMU_V4
+
 source "init/Kconfig"
 source "kernel/Kconfig.freezer"
 
@@ -190,6 +194,16 @@ config NR_CPUS
 	range 2 4096
 	default "4"
 
+config ARC_SMP_HALT_ON_RESET
+	bool "Enable Halt-on-reset boot mode"
+	default y if ARC_UBOOT_SUPPORT
+	help
+	  In SMP configuration cores can be configured as Halt-on-reset
+	  or they could all start at same time. For Halt-on-reset, non
+	  masters are parked until Master kicks them so they can start of
+	  at designated entry point. For other case, all jump to common
+	  entry point and spin wait for Master's signal.
+
 endif	#SMP
 
 menuconfig ARC_CACHE
@@ -278,6 +292,8 @@ choice
 	default ARC_MMU_V2 if ARC_CPU_750D
 	default ARC_MMU_V4 if ARC_CPU_HS
 
+if ISA_ARCOMPACT
+
 config ARC_MMU_V1
 	bool "MMU v1"
 	help
@@ -297,6 +313,8 @@ config ARC_MMU_V3
 	  Variable Page size (1k-16k), var JTLB size 128 x (2 or 4)
 	  Shared Address Spaces (SASID)
 
+endif
+
 config ARC_MMU_V4
 	bool "MMU v4"
 	depends on ISA_ARCV2
@@ -428,6 +446,28 @@ config LINUX_LINK_BASE
 	  Linux needs to be scooted a bit.
 	  If you don't know what the above means, leave this setting alone.
 
+config HIGHMEM
+	bool "High Memory Support"
+	help
+	  With ARC 2G:2G address split, only upper 2G is directly addressable by
+	  kernel. Enable this to potentially allow access to rest of 2G and PAE
+	  in future
+
+config ARC_HAS_PAE40
+	bool "Support for the 40-bit Physical Address Extension"
+	default n
+	depends on ISA_ARCV2
+	select HIGHMEM
+	help
+	  Enable access to physical memory beyond 4G, only supported on
+	  ARC cores with 40 bit Physical Addressing support
+
+config ARCH_PHYS_ADDR_T_64BIT
+	def_bool ARC_HAS_PAE40
+
+config ARCH_DMA_ADDR_T_64BIT
+	bool
+
 config ARC_CURR_IN_REG
 	bool "Dedicate Register r25 for current_task pointer"
 	default y

arch/arc/boot/dts/axc001.dtsi

@@ -95,6 +95,6 @@ memory {
 		#size-cells = <1>;
 		ranges = <0x00000000 0x80000000 0x40000000>;
 		device_type = "memory";
-		reg = <0x00000000 0x20000000>;	/* 512MiB */
+		reg = <0x80000000 0x20000000>;	/* 512MiB */
 	};
 };

arch/arc/boot/dts/axc003.dtsi

@@ -98,6 +98,6 @@ memory {
 		#size-cells = <1>;
 		ranges = <0x00000000 0x80000000 0x40000000>;
 		device_type = "memory";
-		reg = <0x00000000 0x20000000>;	/* 512MiB */
+		reg = <0x80000000 0x20000000>;	/* 512MiB */
 	};
 };

arch/arc/boot/dts/axc003_idu.dtsi

@@ -121,6 +121,6 @@ memory {
 		#size-cells = <1>;
 		ranges = <0x00000000 0x80000000 0x40000000>;
 		device_type = "memory";
-		reg = <0x00000000 0x20000000>;	/* 512MiB */
+		reg = <0x80000000 0x20000000>;	/* 512MiB */
 	};
 };

arch/arc/boot/dts/nsim_hs.dts

@@ -11,8 +11,16 @@
 
 / {
 	compatible = "snps,nsim_hs";
+	#address-cells = <2>;
+	#size-cells = <2>;
 	interrupt-parent = <&core_intc>;
 
+	memory {
+		device_type = "memory";
+		reg = <0x0 0x80000000 0x0 0x40000000	/* 1 GB low mem */
+		       0x1 0x00000000 0x0 0x40000000>;	/* 1 GB highmem */
+	};
+
 	chosen {
 		bootargs = "earlycon=arc_uart,mmio32,0xc0fc1000,115200n8 console=ttyARC0,115200n8";
 	};
@@ -26,8 +34,8 @@ fpga {
 		#address-cells = <1>;
 		#size-cells = <1>;
 
-		/* child and parent address space 1:1 mapped */
-		ranges;
+		/* only perip space at end of low mem accessible */
+		ranges = <0x80000000 0x0 0x80000000 0x80000000>;
 
 		core_intc: core-interrupt-controller {
 			compatible = "snps,archs-intc";

arch/arc/boot/dts/skeleton.dtsi

@@ -32,6 +32,6 @@ cpu@0 {
 
 	memory {
 		device_type = "memory";
-		reg = <0x00000000 0x10000000>;	/* 256M */
+		reg = <0x80000000 0x10000000>;	/* 256M */
 	};
 };

arch/arc/boot/dts/vdk_axc003.dtsi

@@ -56,6 +56,6 @@ memory {
 		#size-cells = <1>;
 		ranges = <0x00000000 0x80000000 0x40000000>;
 		device_type = "memory";
-		reg = <0x00000000 0x20000000>;	/* 512MiB */
+		reg = <0x80000000 0x20000000>;	/* 512MiB */
 	};
 };

arch/arc/boot/dts/vdk_axc003_idu.dtsi

@@ -71,6 +71,6 @@ memory {
 		#size-cells = <1>;
 		ranges = <0x00000000 0x80000000 0x40000000>;
 		device_type = "memory";
-		reg = <0x00000000 0x20000000>;	/* 512MiB */
+		reg = <0x80000000 0x20000000>;	/* 512MiB */
 	};
 };

arch/arc/include/asm/arcregs.h

@@ -120,7 +120,7 @@
 
 /* gcc builtin sr needs reg param to be long immediate */
 #define write_aux_reg(reg_immed, val)		\
-		__builtin_arc_sr((unsigned int)val, reg_immed)
+		__builtin_arc_sr((unsigned int)(val), reg_immed)
 
 #else
 
@@ -327,8 +327,8 @@ struct bcr_generic {
  */
 
 struct cpuinfo_arc_mmu {
-	unsigned int ver:4, pg_sz_k:8, s_pg_sz_m:8, u_dtlb:6, u_itlb:6;
-	unsigned int num_tlb:16, sets:12, ways:4;
+	unsigned int ver:4, pg_sz_k:8, s_pg_sz_m:8, pad:10, sasid:1, pae:1;
+	unsigned int sets:12, ways:4, u_dtlb:8, u_itlb:8;
 };
 
 struct cpuinfo_arc_cache {

arch/arc/include/asm/cache.h

@@ -65,6 +65,7 @@ extern int ioc_exists;
 #if defined(CONFIG_ARC_MMU_V3) || defined(CONFIG_ARC_MMU_V4)
 #define ARC_REG_IC_PTAG		0x1E
 #endif
+#define ARC_REG_IC_PTAG_HI	0x1F
 
 /* Bit val in IC_CTRL */
 #define IC_CTRL_CACHE_DISABLE   0x1
@@ -77,6 +78,7 @@ extern int ioc_exists;
 #define ARC_REG_DC_FLSH		0x4B
 #define ARC_REG_DC_FLDL		0x4C
 #define ARC_REG_DC_PTAG		0x5C
+#define ARC_REG_DC_PTAG_HI	0x5F
 
 /* Bit val in DC_CTRL */
 #define DC_CTRL_INV_MODE_FLUSH  0x40

arch/arc/include/asm/cacheflush.h

@@ -31,10 +31,10 @@
 
 void flush_cache_all(void);
 
-void flush_icache_range(unsigned long start, unsigned long end);
-void __sync_icache_dcache(unsigned long paddr, unsigned long vaddr, int len);
-void __inv_icache_page(unsigned long paddr, unsigned long vaddr);
-void __flush_dcache_page(unsigned long paddr, unsigned long vaddr);
+void flush_icache_range(unsigned long kstart, unsigned long kend);
+void __sync_icache_dcache(phys_addr_t paddr, unsigned long vaddr, int len);
+void __inv_icache_page(phys_addr_t paddr, unsigned long vaddr);
+void __flush_dcache_page(phys_addr_t paddr, unsigned long vaddr);
 
 #define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE 1
 

arch/arc/include/asm/entry-compact.h

@@ -110,13 +110,12 @@
 
 .macro FAKE_RET_FROM_EXCPN
 
-	ld  r9, [sp, PT_status32]
-	bic r9, r9, (STATUS_U_MASK|STATUS_DE_MASK)
-	bset  r9, r9, STATUS_L_BIT
-	sr  r9, [erstatus]
-	mov r9, 55f
-	sr  r9, [eret]
-
+	lr	r9, [status32]
+	bclr	r9, r9, STATUS_AE_BIT
+	or	r9, r9, (STATUS_E1_MASK|STATUS_E2_MASK)
+	sr	r9, [erstatus]
+	mov	r9, 55f
+	sr	r9, [eret]
 	rtie
 55:
 .endm

arch/arc/include/asm/highmem.h

+/*
+ * Copyright (C) 2015 Synopsys, Inc. (www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ */
+
+#ifndef _ASM_HIGHMEM_H
+#define _ASM_HIGHMEM_H
+
+#ifdef CONFIG_HIGHMEM
+
+#include <uapi/asm/page.h>
+#include <asm/kmap_types.h>
+
+/* start after vmalloc area */
+#define FIXMAP_BASE		(PAGE_OFFSET - FIXMAP_SIZE - PKMAP_SIZE)
+#define FIXMAP_SIZE		PGDIR_SIZE	/* only 1 PGD worth */
+#define KM_TYPE_NR		((FIXMAP_SIZE >> PAGE_SHIFT)/NR_CPUS)
+#define FIXMAP_ADDR(nr)		(FIXMAP_BASE + ((nr) << PAGE_SHIFT))
+
+/* start after fixmap area */
+#define PKMAP_BASE		(FIXMAP_BASE + FIXMAP_SIZE)
+#define PKMAP_SIZE		PGDIR_SIZE
+#define LAST_PKMAP		(PKMAP_SIZE >> PAGE_SHIFT)
+#define LAST_PKMAP_MASK		(LAST_PKMAP - 1)
+#define PKMAP_ADDR(nr)		(PKMAP_BASE + ((nr) << PAGE_SHIFT))
+#define PKMAP_NR(virt)		(((virt) - PKMAP_BASE) >> PAGE_SHIFT)
+
+#define kmap_prot		PAGE_KERNEL
+
+
+#include <asm/cacheflush.h>
+
+extern void *kmap(struct page *page);
+extern void *kmap_high(struct page *page);
+extern void *kmap_atomic(struct page *page);
+extern void __kunmap_atomic(void *kvaddr);
+extern void kunmap_high(struct page *page);
+
+extern void kmap_init(void);
+
+static inline void flush_cache_kmaps(void)
+{
+	flush_cache_all();
+}
+
+static inline void kunmap(struct page *page)
+{
+	BUG_ON(in_interrupt());
+	if (!PageHighMem(page))
+		return;
+	kunmap_high(page);
+}
+
+
+#endif
+
+#endif

arch/arc/include/asm/hugepage.h

+/*
+ * Copyright (C) 2013-15 Synopsys, Inc. (www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+
+#ifndef _ASM_ARC_HUGEPAGE_H
+#define _ASM_ARC_HUGEPAGE_H
+
+#include <linux/types.h>
+#include <asm-generic/pgtable-nopmd.h>
+
+static inline pte_t pmd_pte(pmd_t pmd)
+{
+	return __pte(pmd_val(pmd));
+}
+
+static inline pmd_t pte_pmd(pte_t pte)
+{
+	return __pmd(pte_val(pte));
+}
+
+#define pmd_wrprotect(pmd)	pte_pmd(pte_wrprotect(pmd_pte(pmd)))
+#define pmd_mkwrite(pmd)	pte_pmd(pte_mkwrite(pmd_pte(pmd)))
+#define pmd_mkdirty(pmd)	pte_pmd(pte_mkdirty(pmd_pte(pmd)))
+#define pmd_mkold(pmd)		pte_pmd(pte_mkold(pmd_pte(pmd)))
+#define pmd_mkyoung(pmd)	pte_pmd(pte_mkyoung(pmd_pte(pmd)))
+#define pmd_mkhuge(pmd)		pte_pmd(pte_mkhuge(pmd_pte(pmd)))
+#define pmd_mknotpresent(pmd)	pte_pmd(pte_mknotpresent(pmd_pte(pmd)))
+#define pmd_mksplitting(pmd)	pte_pmd(pte_mkspecial(pmd_pte(pmd)))
+#define pmd_mkclean(pmd)	pte_pmd(pte_mkclean(pmd_pte(pmd)))
+
+#define pmd_write(pmd)		pte_write(pmd_pte(pmd))
+#define pmd_young(pmd)		pte_young(pmd_pte(pmd))
+#define pmd_pfn(pmd)		pte_pfn(pmd_pte(pmd))
+#define pmd_dirty(pmd)		pte_dirty(pmd_pte(pmd))
+#define pmd_special(pmd)	pte_special(pmd_pte(pmd))
+
+#define mk_pmd(page, prot)	pte_pmd(mk_pte(page, prot))
+
+#define pmd_trans_huge(pmd)	(pmd_val(pmd) & _PAGE_HW_SZ)
+#define pmd_trans_splitting(pmd)	(pmd_trans_huge(pmd) && pmd_special(pmd))
+
+#define pfn_pmd(pfn, prot)	(__pmd(((pfn) << PAGE_SHIFT) | pgprot_val(prot)))
+
+static inline pmd_t pmd_modify(pmd_t pmd, pgprot_t newprot)
+{
+        /*
+         * open-coded pte_modify() with additional retaining of HW_SZ bit
+         * so that pmd_trans_huge() remains true for this PMD
+         */
+        return __pmd((pmd_val(pmd) & (_PAGE_CHG_MASK | _PAGE_HW_SZ)) | pgprot_val(newprot));
+}
+
+static inline void set_pmd_at(struct mm_struct *mm, unsigned long addr,
+			      pmd_t *pmdp, pmd_t pmd)
+{
+	*pmdp = pmd;
+}
+
+extern void update_mmu_cache_pmd(struct vm_area_struct *vma, unsigned long addr,
+				 pmd_t *pmd);
+
+#define has_transparent_hugepage() 1
+
+/* Generic variants assume pgtable_t is struct page *, hence need for these */
+#define __HAVE_ARCH_PGTABLE_DEPOSIT
+extern void pgtable_trans_huge_deposit(struct mm_struct *mm, pmd_t *pmdp,
+				       pgtable_t pgtable);
+
+#define __HAVE_ARCH_PGTABLE_WITHDRAW
+extern pgtable_t pgtable_trans_huge_withdraw(struct mm_struct *mm, pmd_t *pmdp);
+
+#define __HAVE_ARCH_FLUSH_PMD_TLB_RANGE
+extern void flush_pmd_tlb_range(struct vm_area_struct *vma, unsigned long start,
+				unsigned long end);
+
+#endif

arch/arc/include/asm/irq.h

@@ -16,6 +16,7 @@
 #ifdef CONFIG_ISA_ARCOMPACT
 #define TIMER0_IRQ      3
 #define TIMER1_IRQ      4
+#define IPI_IRQ		(NR_CPU_IRQS-1) /* dummy to enable SMP build for up hardware */
 #else
 #define TIMER0_IRQ      16
 #define TIMER1_IRQ      17

arch/arc/include/asm/irqflags-compact.h

@@ -23,11 +23,13 @@
 #define STATUS_E2_BIT		2	/* Int 2 enable */
 #define STATUS_A1_BIT		3	/* Int 1 active */
 #define STATUS_A2_BIT		4	/* Int 2 active */
+#define STATUS_AE_BIT		5	/* Exception active */
 
 #define STATUS_E1_MASK		(1<<STATUS_E1_BIT)
 #define STATUS_E2_MASK		(1<<STATUS_E2_BIT)
 #define STATUS_A1_MASK		(1<<STATUS_A1_BIT)
 #define STATUS_A2_MASK		(1<<STATUS_A2_BIT)
+#define STATUS_AE_MASK		(1<<STATUS_AE_BIT)
 #define STATUS_IE_MASK		(STATUS_E1_MASK | STATUS_E2_MASK)
 
 /* Other Interrupt Handling related Aux regs */
@@ -91,7 +93,19 @@ static inline void arch_local_irq_restore(unsigned long flags)
 /*
  * Unconditionally Enable IRQs
  */
-extern void arch_local_irq_enable(void);
+static inline void arch_local_irq_enable(void)
+{
+	unsigned long temp;
+
+	__asm__ __volatile__(
+	"	lr   %0, [status32]	\n"
+	"	or   %0, %0, %1		\n"
+	"	flag %0			\n"
+	: "=&r"(temp)
+	: "n"((STATUS_E1_MASK | STATUS_E2_MASK))
+	: "cc", "memory");
+}
+
 
 /*
  * Unconditionally Disable IRQs

arch/arc/include/asm/kmap_types.h

+/*
+ * Copyright (C) 2015 Synopsys, Inc. (www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ */
+
+#ifndef _ASM_KMAP_TYPES_H
+#define _ASM_KMAP_TYPES_H
+
+/*
+ * We primarily need to define KM_TYPE_NR here but that in turn
+ * is a function of PGDIR_SIZE etc.
+ * To avoid circular deps issue, put everything in asm/highmem.h
+ */
+#endif

arch/arc/include/asm/mach_desc.h

@@ -23,11 +23,8 @@
  * @dt_compat:		Array of device tree 'compatible' strings
  * 			(XXX: although only 1st entry is looked at)
  * @init_early:		Very early callback [called from setup_arch()]
- * @init_irq:		setup external IRQ controllers [called from init_IRQ()]
- * @init_smp:		for each CPU (e.g. setup IPI)
+ * @init_cpu_smp:	for each CPU as it is coming up (SMP as well as UP)
  * 			[(M):init_IRQ(), (o):start_kernel_secondary()]
- * @init_time:		platform specific clocksource/clockevent registration
- * 			[called from time_init()]
  * @init_machine:	arch initcall level callback (e.g. populate static
  * 			platform devices or parse Devicetree)
  * @init_late:		Late initcall level callback
@@ -36,13 +33,10 @@
 struct machine_desc {
 	const char		*name;
 	const char		**dt_compat;
-
 	void			(*init_early)(void);
-	void			(*init_irq)(void);
 #ifdef CONFIG_SMP
-	void			(*init_smp)(unsigned int);
+	void			(*init_cpu_smp)(unsigned int);
 #endif
-	void			(*init_time)(void);
 	void			(*init_machine)(void);
 	void			(*init_late)(void);
 

arch/arc/include/asm/mcip.h

@@ -86,9 +86,6 @@ static inline void __mcip_cmd_data(unsigned int cmd, unsigned int param,
 	__mcip_cmd(cmd, param);
 }
 
-extern void mcip_init_early_smp(void);
-extern void mcip_init_smp(unsigned int cpu);
-
 #endif
 
 #endif

arch/arc/include/asm/mmu.h

@@ -24,6 +24,7 @@
 #if (CONFIG_ARC_MMU_VER < 4)
 #define ARC_REG_TLBPD0		0x405
 #define ARC_REG_TLBPD1		0x406
+#define ARC_REG_TLBPD1HI	0	/* Dummy: allows code sharing with ARC700 */
 #define ARC_REG_TLBINDEX	0x407
 #define ARC_REG_TLBCOMMAND	0x408
 #define ARC_REG_PID		0x409
@@ -31,6 +32,7 @@
 #else
 #define ARC_REG_TLBPD0		0x460
 #define ARC_REG_TLBPD1		0x461
+#define ARC_REG_TLBPD1HI	0x463
 #define ARC_REG_TLBINDEX	0x464
 #define ARC_REG_TLBCOMMAND	0x465
 #define ARC_REG_PID		0x468
@@ -83,6 +85,11 @@ void arc_mmu_init(void);
 extern char *arc_mmu_mumbojumbo(int cpu_id, char *buf, int len);
 void read_decode_mmu_bcr(void);
 
+static inline int is_pae40_enabled(void)
+{
+	return IS_ENABLED(CONFIG_ARC_HAS_PAE40);
+}
+
 #endif	/* !__ASSEMBLY__ */
 
 #endif

arch/arc/include/asm/page.h

@@ -43,7 +43,6 @@ typedef struct {
 typedef struct {
 	unsigned long pgprot;
 } pgprot_t;
-typedef unsigned long pgtable_t;
 
 #define pte_val(x)      ((x).pte)
 #define pgd_val(x)      ((x).pgd)
@@ -57,20 +56,26 @@ typedef unsigned long pgtable_t;
 
 #else /* !STRICT_MM_TYPECHECKS */
 
+#ifdef CONFIG_ARC_HAS_PAE40
+typedef unsigned long long pte_t;
+#else
 typedef unsigned long pte_t;
+#endif
 typedef unsigned long pgd_t;
 typedef unsigned long pgprot_t;
-typedef unsigned long pgtable_t;
 
 #define pte_val(x)	(x)
 #define pgd_val(x)	(x)
 #define pgprot_val(x)	(x)
 #define __pte(x)	(x)
+#define __pgd(x)	(x)
 #define __pgprot(x)	(x)
 #define pte_pgprot(x)	(x)
 
 #endif
 
+typedef pte_t * pgtable_t;
+
 #define ARCH_PFN_OFFSET     (CONFIG_LINUX_LINK_BASE >> PAGE_SHIFT)
 
 #define pfn_valid(pfn)      (((pfn) - ARCH_PFN_OFFSET) < max_mapnr)

arch/arc/include/asm/pgalloc.h

@@ -49,7 +49,7 @@ pmd_populate(struct mm_struct *mm, pmd_t *pmd, pgtable_t ptep)
 
 static inline int __get_order_pgd(void)
 {
-	return get_order(PTRS_PER_PGD * 4);
+	return get_order(PTRS_PER_PGD * sizeof(pgd_t));
 }
 
 static inline pgd_t *pgd_alloc(struct mm_struct *mm)
@@ -87,7 +87,7 @@ static inline void pgd_free(struct mm_struct *mm, pgd_t *pgd)
 
 static inline int __get_order_pte(void)
 {
-	return get_order(PTRS_PER_PTE * 4);
+	return get_order(PTRS_PER_PTE * sizeof(pte_t));
 }
 
 static inline pte_t *pte_alloc_one_kernel(struct mm_struct *mm,
@@ -107,10 +107,10 @@ pte_alloc_one(struct mm_struct *mm, unsigned long address)
 	pgtable_t pte_pg;
 	struct page *page;
 
-	pte_pg = __get_free_pages(GFP_KERNEL | __GFP_REPEAT, __get_order_pte());
+	pte_pg = (pgtable_t)__get_free_pages(GFP_KERNEL | __GFP_REPEAT, __get_order_pte());
 	if (!pte_pg)
 		return 0;
-	memzero((void *)pte_pg, PTRS_PER_PTE * 4);
+	memzero((void *)pte_pg, PTRS_PER_PTE * sizeof(pte_t));
 	page = virt_to_page(pte_pg);
 	if (!pgtable_page_ctor(page)) {
 		__free_page(page);
@@ -128,12 +128,12 @@ static inline void pte_free_kernel(struct mm_struct *mm, pte_t *pte)
 static inline void pte_free(struct mm_struct *mm, pgtable_t ptep)
 {
 	pgtable_page_dtor(virt_to_page(ptep));
-	free_pages(ptep, __get_order_pte());
+	free_pages((unsigned long)ptep, __get_order_pte());
 }
 
 #define __pte_free_tlb(tlb, pte, addr)  pte_free((tlb)->mm, pte)
 
 #define check_pgt_cache()   do { } while (0)
-#define pmd_pgtable(pmd) pmd_page_vaddr(pmd)
+#define pmd_pgtable(pmd)	((pgtable_t) pmd_page_vaddr(pmd))
 
 #endif /* _ASM_ARC_PGALLOC_H */

arch/arc/include/asm/pgtable.h

@@ -38,6 +38,7 @@
 #include <asm/page.h>
 #include <asm/mmu.h>
 #include <asm-generic/pgtable-nopmd.h>
+#include <linux/const.h>
 
 /**************************************************************************
  * Page Table Flags
@@ -60,7 +61,8 @@
 #define _PAGE_EXECUTE       (1<<3)	/* Page has user execute perm (H) */
 #define _PAGE_WRITE         (1<<4)	/* Page has user write perm (H) */
 #define _PAGE_READ          (1<<5)	/* Page has user read perm (H) */
-#define _PAGE_MODIFIED      (1<<6)	/* Page modified (dirty) (S) */
+#define _PAGE_DIRTY         (1<<6)	/* Page modified (dirty) (S) */
+#define _PAGE_SPECIAL       (1<<7)
 #define _PAGE_GLOBAL        (1<<8)	/* Page is global (H) */
 #define _PAGE_PRESENT       (1<<10)	/* TLB entry is valid (H) */
 
@@ -71,7 +73,8 @@
 #define _PAGE_WRITE         (1<<2)	/* Page has user write perm (H) */
 #define _PAGE_READ          (1<<3)	/* Page has user read perm (H) */
 #define _PAGE_ACCESSED      (1<<4)	/* Page is accessed (S) */
-#define _PAGE_MODIFIED      (1<<5)	/* Page modified (dirty) (S) */
+#define _PAGE_DIRTY         (1<<5)	/* Page modified (dirty) (S) */
+#define _PAGE_SPECIAL       (1<<6)
 
 #if (CONFIG_ARC_MMU_VER >= 4)
 #define _PAGE_WTHRU         (1<<7)	/* Page cache mode write-thru (H) */
@@ -81,32 +84,33 @@
 #define _PAGE_PRESENT       (1<<9)	/* TLB entry is valid (H) */
 
 #if (CONFIG_ARC_MMU_VER >= 4)
-#define _PAGE_SZ            (1<<10)	/* Page Size indicator (H) */
+#define _PAGE_HW_SZ         (1<<10)	/* Page Size indicator (H): 0 normal, 1 super */
 #endif
 
 #define _PAGE_SHARED_CODE   (1<<11)	/* Shared Code page with cmn vaddr
 					   usable for shared TLB entries (H) */
+
+#define _PAGE_UNUSED_BIT    (1<<12)
 #endif
 
 /* vmalloc permissions */
 #define _K_PAGE_PERMS  (_PAGE_EXECUTE | _PAGE_WRITE | _PAGE_READ | \
 			_PAGE_GLOBAL | _PAGE_PRESENT)
 
-#ifdef CONFIG_ARC_CACHE_PAGES
-#define _PAGE_DEF_CACHEABLE _PAGE_CACHEABLE
-#else
-#define _PAGE_DEF_CACHEABLE (0)
+#ifndef CONFIG_ARC_CACHE_PAGES
+#undef _PAGE_CACHEABLE
+#define _PAGE_CACHEABLE 0
 #endif
 
-/* Helper for every "user" page
- * -kernel can R/W/X
- * -by default cached, unless config otherwise
- * -present in memory
- */
-#define ___DEF (_PAGE_PRESENT | _PAGE_DEF_CACHEABLE)
+#ifndef _PAGE_HW_SZ
+#define _PAGE_HW_SZ	0
+#endif
+
+/* Defaults for every user page */
+#define ___DEF (_PAGE_PRESENT | _PAGE_CACHEABLE)
 
 /* Set of bits not changed in pte_modify */
-#define _PAGE_CHG_MASK	(PAGE_MASK | _PAGE_ACCESSED | _PAGE_MODIFIED)
+#define _PAGE_CHG_MASK	(PAGE_MASK | _PAGE_ACCESSED | _PAGE_DIRTY)
 
 /* More Abbrevaited helpers */
 #define PAGE_U_NONE     __pgprot(___DEF)
@@ -122,15 +126,20 @@
  * user vaddr space - visible in all addr spaces, but kernel mode only
  * Thus Global, all-kernel-access, no-user-access, cached
  */
-#define PAGE_KERNEL          __pgprot(_K_PAGE_PERMS | _PAGE_DEF_CACHEABLE)
+#define PAGE_KERNEL          __pgprot(_K_PAGE_PERMS | _PAGE_CACHEABLE)
 
 /* ioremap */
 #define PAGE_KERNEL_NO_CACHE __pgprot(_K_PAGE_PERMS)
 
 /* Masks for actual TLB "PD"s */
-#define PTE_BITS_IN_PD0		(_PAGE_GLOBAL | _PAGE_PRESENT)
+#define PTE_BITS_IN_PD0		(_PAGE_GLOBAL | _PAGE_PRESENT | _PAGE_HW_SZ)
 #define PTE_BITS_RWX		(_PAGE_EXECUTE | _PAGE_WRITE | _PAGE_READ)
+
+#ifdef CONFIG_ARC_HAS_PAE40
+#define PTE_BITS_NON_RWX_IN_PD1	(0xff00000000 | PAGE_MASK | _PAGE_CACHEABLE)
+#else
 #define PTE_BITS_NON_RWX_IN_PD1	(PAGE_MASK | _PAGE_CACHEABLE)
+#endif
 
 /**************************************************************************
  * Mapping of vm_flags (Generic VM) to PTE flags (arch specific)
@@ -191,26 +200,22 @@
 
 /* Optimal Sizing of Pg Tbl - based on MMU page size */
 #if defined(CONFIG_ARC_PAGE_SIZE_8K)
-#define BITS_FOR_PTE	8
+#define BITS_FOR_PTE	8		/* 11:8:13 */
 #elif defined(CONFIG_ARC_PAGE_SIZE_16K)
-#define BITS_FOR_PTE	8
+#define BITS_FOR_PTE	8		/* 10:8:14 */
 #elif defined(CONFIG_ARC_PAGE_SIZE_4K)
-#define BITS_FOR_PTE	9
+#define BITS_FOR_PTE	9		/* 11:9:12 */
 #endif
 
 #define BITS_FOR_PGD	(32 - BITS_FOR_PTE - BITS_IN_PAGE)
 
-#define PGDIR_SHIFT	(BITS_FOR_PTE + BITS_IN_PAGE)
+#define PGDIR_SHIFT	(32 - BITS_FOR_PGD)
 #define PGDIR_SIZE	(1UL << PGDIR_SHIFT)	/* vaddr span, not PDG sz */
 #define PGDIR_MASK	(~(PGDIR_SIZE-1))
 
-#ifdef __ASSEMBLY__
-#define	PTRS_PER_PTE	(1 << BITS_FOR_PTE)
-#define	PTRS_PER_PGD	(1 << BITS_FOR_PGD)
-#else
-#define	PTRS_PER_PTE	(1UL << BITS_FOR_PTE)
-#define	PTRS_PER_PGD	(1UL << BITS_FOR_PGD)
-#endif
+#define	PTRS_PER_PTE	_BITUL(BITS_FOR_PTE)
+#define	PTRS_PER_PGD	_BITUL(BITS_FOR_PGD)
+
 /*
  * Number of entries a user land program use.
  * TASK_SIZE is the maximum vaddr that can be used by a userland program.
@@ -270,15 +275,10 @@ static inline void pmd_set(pmd_t *pmdp, pte_t *ptep)
 		(unsigned long)(((pte_val(x) - CONFIG_LINUX_LINK_BASE) >> \
 				PAGE_SHIFT)))
 
-#define mk_pte(page, pgprot)						\
-({									\
-	pte_t pte;							\
-	pte_val(pte) = __pa(page_address(page)) + pgprot_val(pgprot);	\
-	pte;								\
-})
-
+#define mk_pte(page, prot)	pfn_pte(page_to_pfn(page), prot)
 #define pte_pfn(pte)		(pte_val(pte) >> PAGE_SHIFT)
-#define pfn_pte(pfn, prot)	(__pte(((pfn) << PAGE_SHIFT) | pgprot_val(prot)))
+#define pfn_pte(pfn, prot)	(__pte(((pte_t)(pfn) << PAGE_SHIFT) | \
+				 pgprot_val(prot)))
 #define __pte_index(addr)	(((addr) >> PAGE_SHIFT) & (PTRS_PER_PTE - 1))
 
 /*
@@ -295,23 +295,26 @@ static inline void pmd_set(pmd_t *pmdp, pte_t *ptep)
 /* Zoo of pte_xxx function */
 #define pte_read(pte)		(pte_val(pte) & _PAGE_READ)
 #define pte_write(pte)		(pte_val(pte) & _PAGE_WRITE)
-#define pte_dirty(pte)		(pte_val(pte) & _PAGE_MODIFIED)
+#define pte_dirty(pte)		(pte_val(pte) & _PAGE_DIRTY)
 #define pte_young(pte)		(pte_val(pte) & _PAGE_ACCESSED)
-#define pte_special(pte)	(0)
+#define pte_special(pte)	(pte_val(pte) & _PAGE_SPECIAL)
 
 #define PTE_BIT_FUNC(fn, op) \
 	static inline pte_t pte_##fn(pte_t pte) { pte_val(pte) op; return pte; }
 
+PTE_BIT_FUNC(mknotpresent,	&= ~(_PAGE_PRESENT));
 PTE_BIT_FUNC(wrprotect,	&= ~(_PAGE_WRITE));
 PTE_BIT_FUNC(mkwrite,	|= (_PAGE_WRITE));
-PTE_BIT_FUNC(mkclean,	&= ~(_PAGE_MODIFIED));
-PTE_BIT_FUNC(mkdirty,	|= (_PAGE_MODIFIED));
+PTE_BIT_FUNC(mkclean,	&= ~(_PAGE_DIRTY));
+PTE_BIT_FUNC(mkdirty,	|= (_PAGE_DIRTY));
 PTE_BIT_FUNC(mkold,	&= ~(_PAGE_ACCESSED));
 PTE_BIT_FUNC(mkyoung,	|= (_PAGE_ACCESSED));
 PTE_BIT_FUNC(exprotect,	&= ~(_PAGE_EXECUTE));
 PTE_BIT_FUNC(mkexec,	|= (_PAGE_EXECUTE));
+PTE_BIT_FUNC(mkspecial,	|= (_PAGE_SPECIAL));
+PTE_BIT_FUNC(mkhuge,	|= (_PAGE_HW_SZ));
 
-static inline pte_t pte_mkspecial(pte_t pte) { return pte; }
+#define __HAVE_ARCH_PTE_SPECIAL
 
 static inline pte_t pte_modify(pte_t pte, pgprot_t newprot)
 {
@@ -357,7 +360,6 @@ static inline void set_pte_at(struct mm_struct *mm, unsigned long addr,
 #define pgd_offset_fast(mm, addr)	pgd_offset(mm, addr)
 #endif
 
-extern void paging_init(void);
 extern pgd_t swapper_pg_dir[] __aligned(PAGE_SIZE);
 void update_mmu_cache(struct vm_area_struct *vma, unsigned long address,
 		      pte_t *ptep);
@@ -383,6 +385,10 @@ void update_mmu_cache(struct vm_area_struct *vma, unsigned long address,
  * remap a physical page `pfn' of size `size' with page protection `prot'
  * into virtual address `from'
  */
+#ifdef CONFIG_TRANSPARENT_HUGEPAGE
+#include <asm/hugepage.h>
+#endif
+
 #include <asm-generic/pgtable.h>
 
 /* to cope with aliasing VIPT cache */

arch/arc/include/asm/processor.h

@@ -114,7 +114,12 @@ extern unsigned int get_wchan(struct task_struct *p);
  * -----------------------------------------------------------------------------
  */
 #define VMALLOC_START	0x70000000
-#define VMALLOC_SIZE	(PAGE_OFFSET - VMALLOC_START)
+
+/*
+ * 1 PGDIR_SIZE each for fixmap/pkmap, 2 PGDIR_SIZE gutter
+ * See asm/highmem.h for details
+ */
+#define VMALLOC_SIZE	(PAGE_OFFSET - VMALLOC_START - PGDIR_SIZE * 4)
 #define VMALLOC_END	(VMALLOC_START + VMALLOC_SIZE)
 
 #define USER_KERNEL_GUTTER    0x10000000

arch/arc/include/asm/setup.h

@@ -33,4 +33,11 @@ extern int root_mountflags, end_mem;
 void setup_processor(void);
 void __init setup_arch_memory(void);
 
+/* Helpers used in arc_*_mumbojumbo routines */
+#define IS_AVAIL1(v, s)		((v) ? s : "")
+#define IS_DISABLED_RUN(v)	((v) ? "" : "(disabled) ")
+#define IS_USED_RUN(v)		((v) ? "" : "(not used) ")
+#define IS_USED_CFG(cfg)	IS_USED_RUN(IS_ENABLED(cfg))
+#define IS_AVAIL2(v, s, cfg)	IS_AVAIL1(v, s), IS_AVAIL1(v, IS_USED_CFG(cfg))
+
 #endif /* __ASMARC_SETUP_H */

arch/arc/include/asm/smp.h

@@ -45,12 +45,19 @@ extern int smp_ipi_irq_setup(int cpu, int irq);
  * struct plat_smp_ops	- SMP callbacks provided by platform to ARC SMP
  *
  * @info:		SoC SMP specific info for /proc/cpuinfo etc
+ * @init_early_smp:	A SMP specific h/w block can init itself
+ * 			Could be common across platforms so not covered by
+ * 			mach_desc->init_early()
+ * @init_irq_cpu:	Called for each core so SMP h/w block driver can do
+ * 			any needed setup per cpu (e.g. IPI request)
  * @cpu_kick:		For Master to kickstart a cpu (optionally at a PC)
  * @ipi_send:		To send IPI to a @cpu
  * @ips_clear:		To clear IPI received at @irq
  */
 struct plat_smp_ops {
 	const char 	*info;
+	void		(*init_early_smp)(void);
+	void		(*init_irq_cpu)(int cpu);
 	void		(*cpu_kick)(int cpu, unsigned long pc);
 	void		(*ipi_send)(int cpu);
 	void		(*ipi_clear)(int irq);

arch/arc/include/asm/tlbflush.h

@@ -17,6 +17,8 @@ void local_flush_tlb_page(struct vm_area_struct *vma, unsigned long page);
 void local_flush_tlb_kernel_range(unsigned long start, unsigned long end);
 void local_flush_tlb_range(struct vm_area_struct *vma,
 			   unsigned long start, unsigned long end);
+void local_flush_pmd_tlb_range(struct vm_area_struct *vma, unsigned long start,
+			       unsigned long end);
 
 #ifndef CONFIG_SMP
 #define flush_tlb_range(vma, s, e)	local_flush_tlb_range(vma, s, e)
@@ -24,6 +26,7 @@ void local_flush_tlb_range(struct vm_area_struct *vma,
 #define flush_tlb_kernel_range(s, e)	local_flush_tlb_kernel_range(s, e)
 #define flush_tlb_all()			local_flush_tlb_all()
 #define flush_tlb_mm(mm)		local_flush_tlb_mm(mm)
+#define flush_pmd_tlb_range(vma, s, e)	local_flush_pmd_tlb_range(vma, s, e)
 #else
 extern void flush_tlb_range(struct vm_area_struct *vma, unsigned long start,
 							 unsigned long end);
@@ -31,5 +34,7 @@ extern void flush_tlb_page(struct vm_area_struct *vma, unsigned long page);
 extern void flush_tlb_kernel_range(unsigned long start, unsigned long end);
 extern void flush_tlb_all(void);
 extern void flush_tlb_mm(struct mm_struct *mm);
+extern void flush_pmd_tlb_range(struct vm_area_struct *vma, unsigned long start, unsigned long end);
+
 #endif /* CONFIG_SMP */
 #endif

arch/arc/include/uapi/asm/page.h

@@ -9,6 +9,8 @@
 #ifndef _UAPI__ASM_ARC_PAGE_H
 #define _UAPI__ASM_ARC_PAGE_H
 
+#include <linux/const.h>
+
 /* PAGE_SHIFT determines the page size */
 #if defined(CONFIG_ARC_PAGE_SIZE_16K)
 #define PAGE_SHIFT 14
@@ -25,13 +27,8 @@
 #define PAGE_SHIFT 13
 #endif
 
-#ifdef __ASSEMBLY__
-#define PAGE_SIZE	(1 << PAGE_SHIFT)
-#define PAGE_OFFSET	(0x80000000)
-#else
-#define PAGE_SIZE	(1UL << PAGE_SHIFT)	/* Default 8K */
-#define PAGE_OFFSET	(0x80000000UL)		/* Kernel starts at 2G onwards */
-#endif
+#define PAGE_SIZE	_BITUL(PAGE_SHIFT)	/* Default 8K */
+#define PAGE_OFFSET	_AC(0x80000000, UL)	/* Kernel starts at 2G onwrds */
 
 #define PAGE_MASK	(~(PAGE_SIZE-1))
 

arch/arc/kernel/entry-arcv2.S

@@ -24,7 +24,7 @@
 	.align 4
 
 # Initial 16 slots are Exception Vectors
-VECTOR	stext			; Restart Vector (jump to entry point)
+VECTOR	res_service		; Reset Vector
 VECTOR	mem_service		; Mem exception
 VECTOR	instr_service		; Instrn Error
 VECTOR	EV_MachineCheck		; Fatal Machine check

arch/arc/kernel/entry-compact.S

@@ -86,7 +86,7 @@
  */
 
 ; ********* Critical System Events **********************
-VECTOR   res_service             ; 0x0, Restart Vector  (0x0)
+VECTOR   res_service             ; 0x0, Reset Vector	(0x0)
 VECTOR   mem_service             ; 0x8, Mem exception   (0x1)
 VECTOR   instr_service           ; 0x10, Instrn Error   (0x2)
 
@@ -155,13 +155,9 @@ int2_saved_reg:
 ; ---------------------------------------------
 	.section .text, "ax",@progbits
 
-res_service:		; processor restart
-	flag    0x1     ; not implemented
-	nop
-	nop
 
-reserved:		; processor restart
-	rtie            ; jump to processor initializations
+reserved:
+	flag 1		; Unexpected event, halt
 
 ;##################### Interrupt Handling ##############################
 
@@ -175,12 +171,25 @@ ENTRY(handle_interrupt_level2)
 
 	;------------------------------------------------------
 	; if L2 IRQ interrupted a L1 ISR, disable preemption
+	;
+	; This is to avoid a potential L1-L2-L1 scenario
+	;  -L1 IRQ taken
+	;  -L2 interrupts L1 (before L1 ISR could run)
+	;  -preemption off IRQ, user task in syscall picked to run
+	;  -RTIE to userspace
+	;	Returns from L2 context fine
+	;	But both L1 and L2 re-enabled, so another L1 can be taken
+	;	while prev L1 is still unserviced
+	;
 	;------------------------------------------------------
 
+	; L2 interrupting L1 implies both L2 and L1 active
+	; However both A2 and A1 are NOT set in STATUS32, thus
+	; need to check STATUS32_L2 to determine if L1 was active
+
 	ld r9, [sp, PT_status32]        ; get statu32_l2 (saved in pt_regs)
 	bbit0 r9, STATUS_A1_BIT, 1f     ; L1 not active when L2 IRQ, so normal
 
-	; A1 is set in status32_l2
 	; bump thread_info->preempt_count (Disable preemption)
 	GET_CURR_THR_INFO_FROM_SP   r10
 	ld      r9, [r10, THREAD_INFO_PREEMPT_COUNT]
@@ -320,11 +329,10 @@ END(call_do_page_fault)
 	; Note that we use realtime STATUS32 (not pt_regs->status32) to
 	; decide that.
 
-	; if Returning from Exception
-	btst   r10, STATUS_AE_BIT
-	bnz    .Lexcep_ret
+	and.f	0, r10, (STATUS_A1_MASK|STATUS_A2_MASK)
+	bz	.Lexcep_or_pure_K_ret
 
-	; Not Exception so maybe Interrupts (Level 1 or 2)
+	; Returning from Interrupts (Level 1 or 2)
 
 #ifdef CONFIG_ARC_COMPACT_IRQ_LEVELS
 
@@ -365,8 +373,7 @@ END(call_do_page_fault)
 	st      r9, [r10, THREAD_INFO_PREEMPT_COUNT]
 
 149:
-	;return from level 2
-	INTERRUPT_EPILOGUE 2
+	INTERRUPT_EPILOGUE 2	; return from level 2 interrupt
 debug_marker_l2:
 	rtie
 
@@ -374,15 +381,11 @@ not_level2_interrupt:
 
 #endif
 
-	bbit0  r10, STATUS_A1_BIT, .Lpure_k_mode_ret
-
-	;return from level 1
-	INTERRUPT_EPILOGUE 1
+	INTERRUPT_EPILOGUE 1	; return from level 1 interrupt
 debug_marker_l1:
 	rtie
 
-.Lexcep_ret:
-.Lpure_k_mode_ret:
+.Lexcep_or_pure_K_ret:
 
 	;this case is for syscalls or Exceptions or pure kernel mode
 

arch/arc/kernel/head.S

@@ -50,28 +50,37 @@
 .endm
 
 	.section .init.text, "ax",@progbits
-	.type stext, @function
-	.globl stext
-stext:
-	;-------------------------------------------------------------------
-	; Don't clobber r0-r2 yet. It might have bootloader provided info
-	;-------------------------------------------------------------------
+
+;----------------------------------------------------------------
+; Default Reset Handler (jumped into from Reset vector)
+; - Don't clobber r0,r1,r2 as they might have u-boot provided args
+; - Platforms can override this weak version if needed
+;----------------------------------------------------------------
+WEAK(res_service)
+	j	stext
+END(res_service)
+
+;----------------------------------------------------------------
+; Kernel Entry point
+;----------------------------------------------------------------
+ENTRY(stext)
 
 	CPU_EARLY_SETUP
 
 #ifdef CONFIG_SMP
-	; Ensure Boot (Master) proceeds. Others wait in platform dependent way
-	;	IDENTITY Reg [ 3  2  1  0 ]
-	;	(cpu-id)             ^^^	=> Zero for UP ARC700
-	;					=> #Core-ID if SMP (Master 0)
-	; Note that non-boot CPUs might not land here if halt-on-reset and
-	; instead breath life from @first_lines_of_secondary, but we still
-	; need to make sure only boot cpu takes this path.
 	GET_CPU_ID  r5
 	cmp	r5, 0
-	mov.ne	r0, r5
-	jne	arc_platform_smp_wait_to_boot
+	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
+	; Non-Masters wait for Master to boot enough and bring them up
+	jnz	arc_platform_smp_wait_to_boot
 #endif
+	; Master falls thru
+#endif
+
 	; Clear BSS before updating any globals
 	; XXX: use ZOL here
 	mov	r5, __bss_start
@@ -102,18 +111,14 @@ stext:
 	GET_TSK_STACK_BASE r9, sp	; r9 = tsk, sp = stack base(output)
 
 	j	start_kernel	; "C" entry point
+END(stext)
 
 #ifdef CONFIG_SMP
 ;----------------------------------------------------------------
 ;     First lines of code run by secondary before jumping to 'C'
 ;----------------------------------------------------------------
 	.section .text, "ax",@progbits
-	.type first_lines_of_secondary, @function
-	.globl first_lines_of_secondary
-
-first_lines_of_secondary:
-
-	CPU_EARLY_SETUP
+ENTRY(first_lines_of_secondary)
 
 	; setup per-cpu idle task as "current" on this CPU
 	ld	r0, [@secondary_idle_tsk]
@@ -126,5 +131,5 @@ first_lines_of_secondary:
 	GET_TSK_STACK_BASE r0, sp
 
 	j	start_kernel_secondary
-
+END(first_lines_of_secondary)
 #endif

arch/arc/kernel/intc-compact.c

@@ -79,17 +79,16 @@ static struct irq_chip onchip_intc = {
 static int arc_intc_domain_map(struct irq_domain *d, unsigned int irq,
 			       irq_hw_number_t hw)
 {
-	/*
-	 * XXX: the IPI IRQ needs to be handled like TIMER too. However ARC core
-	 *      code doesn't own it (like TIMER0). ISS IDU / ezchip define it
-	 *      in platform header which can't be included here as it goes
-	 *      against multi-platform image philisophy
-	 */
-	if (irq == TIMER0_IRQ)
+	switch (irq) {
+	case TIMER0_IRQ:
+#ifdef CONFIG_SMP
+	case IPI_IRQ:
+#endif
 		irq_set_chip_and_handler(irq, &onchip_intc, handle_percpu_irq);
-	else
+		break;
+	default:
 		irq_set_chip_and_handler(irq, &onchip_intc, handle_level_irq);
-
+	}
 	return 0;
 }
 
@@ -148,78 +147,15 @@ IRQCHIP_DECLARE(arc_intc, "snps,arc700-intc", init_onchip_IRQ);
 
 void arch_local_irq_enable(void)
 {
-
 	unsigned long flags = arch_local_save_flags();
 
-	/* Allow both L1 and L2 at the onset */
-	flags |= (STATUS_E1_MASK | STATUS_E2_MASK);
-
-	/* Called from hard ISR (between irq_enter and irq_exit) */
-	if (in_irq()) {
-
-		/* If in L2 ISR, don't re-enable any further IRQs as this can
-		 * cause IRQ priorities to get upside down. e.g. it could allow
-		 * L1 be taken while in L2 hard ISR which is wrong not only in
-		 * theory, it can also cause the dreaded L1-L2-L1 scenario
-		 */
-		if (flags & STATUS_A2_MASK)
-			flags &= ~(STATUS_E1_MASK | STATUS_E2_MASK);
-
-		/* Even if in L1 ISR, allowe Higher prio L2 IRQs */
-		else if (flags & STATUS_A1_MASK)
-			flags &= ~(STATUS_E1_MASK);
-	}
-
-	/* called from soft IRQ, ideally we want to re-enable all levels */
-
-	else if (in_softirq()) {
-
-		/* However if this is case of L1 interrupted by L2,
-		 * re-enabling both may cause whaco L1-L2-L1 scenario
-		 * because ARC700 allows level 1 to interrupt an active L2 ISR
-		 * Thus we disable both
-		 * However some code, executing in soft ISR wants some IRQs
-		 * to be enabled so we re-enable L2 only
-		 *
-		 * How do we determine L1 intr by L2
-		 *  -A2 is set (means in L2 ISR)
-		 *  -E1 is set in this ISR's pt_regs->status32 which is
-		 *      saved copy of status32_l2 when l2 ISR happened
-		 */
-		struct pt_regs *pt = get_irq_regs();
-
-		if ((flags & STATUS_A2_MASK) && pt &&
-		    (pt->status32 & STATUS_A1_MASK)) {
-			/*flags &= ~(STATUS_E1_MASK | STATUS_E2_MASK); */
-			flags &= ~(STATUS_E1_MASK);
-		}
-	}
+	if (flags & STATUS_A2_MASK)
+		flags |= STATUS_E2_MASK;
+	else if (flags & STATUS_A1_MASK)
+		flags |= STATUS_E1_MASK;
 
 	arch_local_irq_restore(flags);
 }
 
-#else /* ! CONFIG_ARC_COMPACT_IRQ_LEVELS */
-
-/*
- * Simpler version for only 1 level of interrupt
- * Here we only Worry about Level 1 Bits
- */
-void arch_local_irq_enable(void)
-{
-	unsigned long flags;
-
-	/*
-	 * ARC IDE Drivers tries to re-enable interrupts from hard-isr
-	 * context which is simply wrong
-	 */
-	if (in_irq()) {
-		WARN_ONCE(1, "IRQ enabled from hard-isr");
-		return;
-	}
-
-	flags = arch_local_save_flags();
-	flags |= (STATUS_E1_MASK | STATUS_E2_MASK);
-	arch_local_irq_restore(flags);
-}
-#endif
 EXPORT_SYMBOL(arch_local_irq_enable);
+#endif

arch/arc/kernel/irq.c

@@ -10,6 +10,7 @@
 #include <linux/interrupt.h>
 #include <linux/irqchip.h>
 #include <asm/mach_desc.h>
+#include <asm/smp.h>
 
 /*
  * Late Interrupt system init called from start_kernel for Boot CPU only
@@ -19,17 +20,20 @@
  */
 void __init init_IRQ(void)
 {
-	/* Any external intc can be setup here */
-	if (machine_desc->init_irq)
-		machine_desc->init_irq();
-
-	/* process the entire interrupt tree in one go */
+	/*
+	 * process the entire interrupt tree in one go
+	 * Any external intc will be setup provided DT chains them
+	 * properly
+	 */
 	irqchip_init();
 
 #ifdef CONFIG_SMP
-	/* Master CPU can initialize it's side of IPI */
-	if (machine_desc->init_smp)
-		machine_desc->init_smp(smp_processor_id());
+	/* a SMP H/w block could do IPI IRQ request here */
+	if (plat_smp_ops.init_irq_cpu)
+		plat_smp_ops.init_irq_cpu(smp_processor_id());
+
+	if (machine_desc->init_cpu_smp)
+		machine_desc->init_cpu_smp(smp_processor_id());
 #endif
 }
 

arch/arc/kernel/mcip.c

@@ -12,20 +12,14 @@
 #include <linux/irq.h>
 #include <linux/spinlock.h>
 #include <asm/mcip.h>
+#include <asm/setup.h>
 
 static char smp_cpuinfo_buf[128];
 static int idu_detected;
 
 static DEFINE_RAW_SPINLOCK(mcip_lock);
 
-/*
- * Any SMP specific init any CPU does when it comes up.
- * Here we setup the CPU to enable Inter-Processor-Interrupts
- * Called for each CPU
- * -Master      : init_IRQ()
- * -Other(s)    : start_kernel_secondary()
- */
-void mcip_init_smp(unsigned int cpu)
+static void mcip_setup_per_cpu(int cpu)
 {
 	smp_ipi_irq_setup(cpu, IPI_IRQ);
 }
@@ -96,34 +90,8 @@ static void mcip_ipi_clear(int irq)
 #endif
 }
 
-volatile int wake_flag;
-
-static void mcip_wakeup_cpu(int cpu, unsigned long pc)
-{
-	BUG_ON(cpu == 0);
-	wake_flag = cpu;
-}
-
-void arc_platform_smp_wait_to_boot(int cpu)
+static void mcip_probe_n_setup(void)
 {
-	while (wake_flag != cpu)
-		;
-
-	wake_flag = 0;
-	__asm__ __volatile__("j @first_lines_of_secondary	\n");
-}
-
-struct plat_smp_ops plat_smp_ops = {
-	.info		= smp_cpuinfo_buf,
-	.cpu_kick	= mcip_wakeup_cpu,
-	.ipi_send	= mcip_ipi_send,
-	.ipi_clear	= mcip_ipi_clear,
-};
-
-void mcip_init_early_smp(void)
-{
-#define IS_AVAIL1(var, str)    ((var) ? str : "")
-
 	struct mcip_bcr {
 #ifdef CONFIG_CPU_BIG_ENDIAN
 		unsigned int pad3:8,
@@ -161,6 +129,14 @@ void mcip_init_early_smp(void)
 		panic("kernel trying to use non-existent GRTC\n");
 }
 
+struct plat_smp_ops plat_smp_ops = {
+	.info		= smp_cpuinfo_buf,
+	.init_early_smp	= mcip_probe_n_setup,
+	.init_irq_cpu	= mcip_setup_per_cpu,
+	.ipi_send	= mcip_ipi_send,
+	.ipi_clear	= mcip_ipi_clear,
+};
+
 /***************************************************************************
  * ARCv2 Interrupt Distribution Unit (IDU)
  *

arch/arc/kernel/setup.c

@@ -160,10 +160,6 @@ static const struct cpuinfo_data arc_cpu_tbl[] = {
 	{ {0x00, NULL		} }
 };
 
-#define IS_AVAIL1(v, s)		((v) ? s : "")
-#define IS_USED_RUN(v)		((v) ? "" : "(not used) ")
-#define IS_USED_CFG(cfg)	IS_USED_RUN(IS_ENABLED(cfg))
-#define IS_AVAIL2(v, s, cfg)	IS_AVAIL1(v, s), IS_AVAIL1(v, IS_USED_CFG(cfg))
 
 static char *arc_cpu_mumbojumbo(int cpu_id, char *buf, int len)
 {
@@ -415,8 +411,9 @@ void __init setup_arch(char **cmdline_p)
 	if (machine_desc->init_early)
 		machine_desc->init_early();
 
-	setup_processor();
 	smp_init_cpus();
+
+	setup_processor();
 	setup_arch_memory();
 
 	/* copy flat DT out of .init and then unflatten it */

arch/arc/kernel/smp.c

@@ -42,8 +42,13 @@ void __init smp_prepare_boot_cpu(void)
 }
 
 /*
- * Initialise the CPU possible map early - this describes the CPUs
- * which may be present or become present in the system.
+ * Called from setup_arch() before calling setup_processor()
+ *
+ * - Initialise the CPU possible map early - this describes the CPUs
+ *   which may be present or become present in the system.
+ * - Call early smp init hook. This can initialize a specific multi-core
+ *   IP which is say common to several platforms (hence not part of
+ *   platform specific int_early() hook)
  */
 void __init smp_init_cpus(void)
 {
@@ -51,6 +56,9 @@ void __init smp_init_cpus(void)
 
 	for (i = 0; i < NR_CPUS; i++)
 		set_cpu_possible(i, true);
+
+	if (plat_smp_ops.init_early_smp)
+		plat_smp_ops.init_early_smp();
 }
 
 /* called from init ( ) =>  process 1 */
@@ -72,35 +80,29 @@ void __init smp_cpus_done(unsigned int max_cpus)
 }
 
 /*
- * After power-up, a non Master CPU needs to wait for Master to kick start it
- *
- * The default implementation halts
- *
- * This relies on platform specific support allowing Master to directly set
- * this CPU's PC (to be @first_lines_of_secondary() and kick start it.
- *
- * In lack of such h/w assist, platforms can override this function
- *   - make this function busy-spin on a token, eventually set by Master
- *     (from arc_platform_smp_wakeup_cpu())
- *   - Once token is available, jump to @first_lines_of_secondary
- *     (using inline asm).
- *
- * Alert: can NOT use stack here as it has not been determined/setup for CPU.
- *        If it turns out to be elaborate, it's better to code it in assembly
- *
+ * Default smp boot helper for Run-on-reset case where all cores start off
+ * together. Non-masters need to wait for Master to start running.
+ * This is implemented using a flag in memory, which Non-masters spin-wait on.
+ * Master sets it to cpu-id of core to "ungate" it.
  */
-void __weak arc_platform_smp_wait_to_boot(int cpu)
+static volatile int wake_flag;
+
+static void arc_default_smp_cpu_kick(int cpu, unsigned long pc)
 {
-	/*
-	 * As a hack for debugging - since debugger will single-step over the
-	 * FLAG insn - wrap the halt itself it in a self loop
-	 */
-	__asm__ __volatile__(
-	"1:		\n"
-	"	flag 1	\n"
-	"	b 1b	\n");
+	BUG_ON(cpu == 0);
+	wake_flag = cpu;
+}
+
+void arc_platform_smp_wait_to_boot(int cpu)
+{
+	while (wake_flag != cpu)
+		;
+
+	wake_flag = 0;
+	__asm__ __volatile__("j @first_lines_of_secondary	\n");
 }
 
+
 const char *arc_platform_smp_cpuinfo(void)
 {
 	return plat_smp_ops.info ? : "";
@@ -129,8 +131,12 @@ void start_kernel_secondary(void)
 
 	pr_info("## CPU%u LIVE ##: Executing Code...\n", cpu);
 
-	if (machine_desc->init_smp)
-		machine_desc->init_smp(cpu);
+	/* Some SMP H/w setup - for each cpu */
+	if (plat_smp_ops.init_irq_cpu)
+		plat_smp_ops.init_irq_cpu(cpu);
+
+	if (machine_desc->init_cpu_smp)
+		machine_desc->init_cpu_smp(cpu);
 
 	arc_local_timer_setup();
 
@@ -161,6 +167,8 @@ int __cpu_up(unsigned int cpu, struct task_struct *idle)
 	if (plat_smp_ops.cpu_kick)
 		plat_smp_ops.cpu_kick(cpu,
 				(unsigned long)first_lines_of_secondary);
+	else
+		arc_default_smp_cpu_kick(cpu, (unsigned long)NULL);
 
 	/* wait for 1 sec after kicking the secondary */
 	wait_till = jiffies + HZ;

arch/arc/kernel/time.c

@@ -285,7 +285,4 @@ void __init time_init(void)
 
 	/* sets up the periodic event timer */
 	arc_local_timer_setup();
-
-	if (machine_desc->init_time)
-		machine_desc->init_time();
 }

arch/arc/kernel/vmlinux.lds.S

@@ -12,7 +12,7 @@
 #include <asm/thread_info.h>
 
 OUTPUT_ARCH(arc)
-ENTRY(_stext)
+ENTRY(res_service)
 
 #ifdef CONFIG_CPU_BIG_ENDIAN
 jiffies = jiffies_64 + 4;

arch/arc/mm/Makefile

@@ -8,3 +8,4 @@
 
 obj-y	:= extable.o ioremap.o dma.o fault.o init.o
 obj-y	+= tlb.o tlbex.o cache.o mmap.o
+obj-$(CONFIG_HIGHMEM)	+= highmem.o

arch/arc/mm/cache.c

@@ -25,7 +25,7 @@ static int l2_line_sz;
 int ioc_exists;
 volatile int slc_enable = 1, ioc_enable = 1;
 
-void (*_cache_line_loop_ic_fn)(unsigned long paddr, unsigned long vaddr,
+void (*_cache_line_loop_ic_fn)(phys_addr_t paddr, unsigned long vaddr,
 			       unsigned long sz, const int cacheop);
 
 void (*__dma_cache_wback_inv)(unsigned long start, unsigned long sz);
@@ -37,7 +37,6 @@ char *arc_cache_mumbojumbo(int c, char *buf, int len)
 	int n = 0;
 	struct cpuinfo_arc_cache *p;
 
-#define IS_USED_RUN(v)		((v) ? "" : "(disabled) ")
 #define PR_CACHE(p, cfg, str)						\
 	if (!(p)->ver)							\
 		n += scnprintf(buf + n, len - n, str"\t\t: N/A\n");	\
@@ -47,7 +46,7 @@ char *arc_cache_mumbojumbo(int c, char *buf, int len)
 			(p)->sz_k, (p)->assoc, (p)->line_len,		\
 			(p)->vipt ? "VIPT" : "PIPT",			\
 			(p)->alias ? " aliasing" : "",			\
-			IS_ENABLED(cfg) ? "" : " (not used)");
+			IS_USED_CFG(cfg));
 
 	PR_CACHE(&cpuinfo_arc700[c].icache, CONFIG_ARC_HAS_ICACHE, "I-Cache");
 	PR_CACHE(&cpuinfo_arc700[c].dcache, CONFIG_ARC_HAS_DCACHE, "D-Cache");
@@ -63,7 +62,7 @@ char *arc_cache_mumbojumbo(int c, char *buf, int len)
 
 	if (ioc_exists)
 		n += scnprintf(buf + n, len - n, "IOC\t\t:%s\n",
-				IS_USED_RUN(ioc_enable));
+				IS_DISABLED_RUN(ioc_enable));
 
 	return buf;
 }
@@ -217,7 +216,7 @@ void read_decode_cache_bcr(void)
  */
 
 static inline
-void __cache_line_loop_v2(unsigned long paddr, unsigned long vaddr,
+void __cache_line_loop_v2(phys_addr_t paddr, unsigned long vaddr,
 			  unsigned long sz, const int op)
 {
 	unsigned int aux_cmd;
@@ -254,8 +253,12 @@ void __cache_line_loop_v2(unsigned long paddr, unsigned long vaddr,
 	}
 }
 
+/*
+ * For ARC700 MMUv3 I-cache and D-cache flushes
+ * Also reused for HS38 aliasing I-cache configuration
+ */
 static inline
-void __cache_line_loop_v3(unsigned long paddr, unsigned long vaddr,
+void __cache_line_loop_v3(phys_addr_t paddr, unsigned long vaddr,
 			  unsigned long sz, const int op)
 {
 	unsigned int aux_cmd, aux_tag;
@@ -290,6 +293,16 @@ void __cache_line_loop_v3(unsigned long paddr, unsigned long vaddr,
 	if (full_page)
 		write_aux_reg(aux_tag, paddr);
 
+	/*
+	 * This is technically for MMU v4, using the MMU v3 programming model
+	 * Special work for HS38 aliasing I-cache configuratino with PAE40
+	 *   - upper 8 bits of paddr need to be written into PTAG_HI
+	 *   - (and needs to be written before the lower 32 bits)
+	 * Note that PTAG_HI is hoisted outside the line loop
+	 */
+	if (is_pae40_enabled() && op == OP_INV_IC)
+		write_aux_reg(ARC_REG_IC_PTAG_HI, (u64)paddr >> 32);
+
 	while (num_lines-- > 0) {
 		if (!full_page) {
 			write_aux_reg(aux_tag, paddr);
@@ -302,14 +315,20 @@ void __cache_line_loop_v3(unsigned long paddr, unsigned long vaddr,
 }
 
 /*
- * In HS38x (MMU v4), although icache is VIPT, only paddr is needed for cache
- * maintenance ops (in IVIL reg), as long as icache doesn't alias.
+ * In HS38x (MMU v4), I-cache is VIPT (can alias), D-cache is PIPT
+ * Here's how cache ops are implemented
+ *
+ *  - D-cache: only paddr needed (in DC_IVDL/DC_FLDL)
+ *  - I-cache Non Aliasing: Despite VIPT, only paddr needed (in IC_IVIL)
+ *  - I-cache Aliasing: Both vaddr and paddr needed (in IC_IVIL, IC_PTAG
+ *    respectively, similar to MMU v3 programming model, hence
+ *    __cache_line_loop_v3() is used)
  *
- * For Aliasing icache, vaddr is also needed (in IVIL), while paddr is
- * specified in PTAG (similar to MMU v3)
+ * If PAE40 is enabled, independent of aliasing considerations, the higher bits
+ * needs to be written into PTAG_HI
  */
 static inline
-void __cache_line_loop_v4(unsigned long paddr, unsigned long vaddr,
+void __cache_line_loop_v4(phys_addr_t paddr, unsigned long vaddr,
 			  unsigned long sz, const int cacheop)
 {
 	unsigned int aux_cmd;
@@ -336,6 +355,22 @@ void __cache_line_loop_v4(unsigned long paddr, unsigned long vaddr,
 
 	num_lines = DIV_ROUND_UP(sz, L1_CACHE_BYTES);
 
+	/*
+	 * For HS38 PAE40 configuration
+	 *   - upper 8 bits of paddr need to be written into PTAG_HI
+	 *   - (and needs to be written before the lower 32 bits)
+	 */
+	if (is_pae40_enabled()) {
+		if (cacheop == OP_INV_IC)
+			/*
+			 * Non aliasing I-cache in HS38,
+			 * aliasing I-cache handled in __cache_line_loop_v3()
+			 */
+			write_aux_reg(ARC_REG_IC_PTAG_HI, (u64)paddr >> 32);
+		else
+			write_aux_reg(ARC_REG_DC_PTAG_HI, (u64)paddr >> 32);
+	}
+
 	while (num_lines-- > 0) {
 		write_aux_reg(aux_cmd, paddr);
 		paddr += L1_CACHE_BYTES;
@@ -413,7 +448,7 @@ static inline void __dc_entire_op(const int op)
 /*
  * D-Cache Line ops: Per Line INV (discard or wback+discard) or FLUSH (wback)
  */
-static inline void __dc_line_op(unsigned long paddr, unsigned long vaddr,
+static inline void __dc_line_op(phys_addr_t paddr, unsigned long vaddr,
 				unsigned long sz, const int op)
 {
 	unsigned long flags;
@@ -446,7 +481,7 @@ static inline void __ic_entire_inv(void)
 }
 
 static inline void
-__ic_line_inv_vaddr_local(unsigned long paddr, unsigned long vaddr,
+__ic_line_inv_vaddr_local(phys_addr_t paddr, unsigned long vaddr,
 			  unsigned long sz)
 {
 	unsigned long flags;
@@ -463,7 +498,7 @@ __ic_line_inv_vaddr_local(unsigned long paddr, unsigned long vaddr,
 #else
 
 struct ic_inv_args {
-	unsigned long paddr, vaddr;
+	phys_addr_t paddr, vaddr;
 	int sz;
 };
 
@@ -474,7 +509,7 @@ static void __ic_line_inv_vaddr_helper(void *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,
+static void __ic_line_inv_vaddr(phys_addr_t paddr, unsigned long vaddr,
 				unsigned long sz)
 {
 	struct ic_inv_args ic_inv = {
@@ -495,7 +530,7 @@ static void __ic_line_inv_vaddr(unsigned long paddr, unsigned long vaddr,
 
 #endif /* CONFIG_ARC_HAS_ICACHE */
 
-noinline void slc_op(unsigned long paddr, unsigned long sz, const int op)
+noinline void slc_op(phys_addr_t paddr, unsigned long sz, const int op)
 {
 #ifdef CONFIG_ISA_ARCV2
 	/*
@@ -585,7 +620,7 @@ void flush_dcache_page(struct page *page)
 	} else if (page_mapped(page)) {
 
 		/* kernel reading from page with U-mapping */
-		unsigned long paddr = (unsigned long)page_address(page);
+		phys_addr_t paddr = (unsigned long)page_address(page);
 		unsigned long vaddr = page->index << PAGE_CACHE_SHIFT;
 
 		if (addr_not_cache_congruent(paddr, vaddr))
@@ -733,14 +768,14 @@ EXPORT_SYMBOL(flush_icache_range);
  *    builtin kernel page will not have any virtual mappings.
  *    kprobe on loadable module will be kernel vaddr.
  */
-void __sync_icache_dcache(unsigned long paddr, unsigned long vaddr, int len)
+void __sync_icache_dcache(phys_addr_t paddr, unsigned long vaddr, int len)
 {
 	__dc_line_op(paddr, vaddr, len, OP_FLUSH_N_INV);
 	__ic_line_inv_vaddr(paddr, vaddr, len);
 }
 
 /* wrapper to compile time eliminate alignment checks in flush loop */
-void __inv_icache_page(unsigned long paddr, unsigned long vaddr)
+void __inv_icache_page(phys_addr_t paddr, unsigned long vaddr)
 {
 	__ic_line_inv_vaddr(paddr, vaddr, PAGE_SIZE);
 }
@@ -749,7 +784,7 @@ void __inv_icache_page(unsigned long paddr, unsigned long vaddr)
  * wrapper to clearout kernel or userspace mappings of a page
  * For kernel mappings @vaddr == @paddr
  */
-void __flush_dcache_page(unsigned long paddr, unsigned long vaddr)
+void __flush_dcache_page(phys_addr_t paddr, unsigned long vaddr)
 {
 	__dc_line_op(paddr, vaddr & PAGE_MASK, PAGE_SIZE, OP_FLUSH_N_INV);
 }
@@ -807,8 +842,8 @@ void flush_anon_page(struct vm_area_struct *vma, struct page *page,
 void copy_user_highpage(struct page *to, struct page *from,
 	unsigned long u_vaddr, struct vm_area_struct *vma)
 {
-	unsigned long kfrom = (unsigned long)page_address(from);
-	unsigned long kto = (unsigned long)page_address(to);
+	void *kfrom = kmap_atomic(from);
+	void *kto = kmap_atomic(to);
 	int clean_src_k_mappings = 0;
 
 	/*
@@ -818,13 +853,16 @@ void copy_user_highpage(struct page *to, struct page *from,
 	 *
 	 * Note that while @u_vaddr refers to DST page's userspace vaddr, it is
 	 * equally valid for SRC page as well
+	 *
+	 * For !VIPT cache, all of this gets compiled out as
+	 * addr_not_cache_congruent() is 0
 	 */
 	if (page_mapped(from) && addr_not_cache_congruent(kfrom, u_vaddr)) {
-		__flush_dcache_page(kfrom, u_vaddr);
+		__flush_dcache_page((unsigned long)kfrom, u_vaddr);
 		clean_src_k_mappings = 1;
 	}
 
-	copy_page((void *)kto, (void *)kfrom);
+	copy_page(kto, kfrom);
 
 	/*
 	 * Mark DST page K-mapping as dirty for a later finalization by
@@ -841,11 +879,14 @@ void copy_user_highpage(struct page *to, struct page *from,
 	 * sync the kernel mapping back to physical page
 	 */
 	if (clean_src_k_mappings) {
-		__flush_dcache_page(kfrom, kfrom);
+		__flush_dcache_page((unsigned long)kfrom, (unsigned long)kfrom);
 		set_bit(PG_dc_clean, &from->flags);
 	} else {
 		clear_bit(PG_dc_clean, &from->flags);
 	}
+
+	kunmap_atomic(kto);
+	kunmap_atomic(kfrom);
 }
 
 void clear_user_page(void *to, unsigned long u_vaddr, struct page *page)

arch/arc/mm/fault.c

@@ -18,7 +18,14 @@
 #include <asm/pgalloc.h>
 #include <asm/mmu.h>
 
-static int handle_vmalloc_fault(unsigned long address)
+/*
+ * kernel virtual address is required to implement vmalloc/pkmap/fixmap
+ * Refer to asm/processor.h for System Memory Map
+ *
+ * It simply copies the PMD entry (pointer to 2nd level page table or hugepage)
+ * from swapper pgdir to task pgdir. The 2nd level table/page is thus shared
+ */
+noinline static int handle_kernel_vaddr_fault(unsigned long address)
 {
 	/*
 	 * Synchronize this task's top level page-table
@@ -72,8 +79,8 @@ void do_page_fault(unsigned long address, struct pt_regs *regs)
 	 * only copy the information from the master page table,
 	 * nothing more.
 	 */
-	if (address >= VMALLOC_START && address <= VMALLOC_END) {
-		ret = handle_vmalloc_fault(address);
+	if (address >= VMALLOC_START) {
+		ret = handle_kernel_vaddr_fault(address);
 		if (unlikely(ret))
 			goto bad_area_nosemaphore;
 		else

arch/arc/mm/highmem.c

+/*
+ * Copyright (C) 2015 Synopsys, Inc. (www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ */
+
+#include <linux/bootmem.h>
+#include <linux/export.h>
+#include <linux/highmem.h>
+#include <asm/processor.h>
+#include <asm/pgtable.h>
+#include <asm/pgalloc.h>
+#include <asm/tlbflush.h>
+
+/*
+ * HIGHMEM API:
+ *
+ * kmap() API provides sleep semantics hence refered to as "permanent maps"
+ * It allows mapping LAST_PKMAP pages, using @last_pkmap_nr as the cursor
+ * for book-keeping
+ *
+ * kmap_atomic() can't sleep (calls pagefault_disable()), thus it provides
+ * shortlived ala "temporary mappings" which historically were implemented as
+ * fixmaps (compile time addr etc). Their book-keeping is done per cpu.
+ *
+ *	Both these facts combined (preemption disabled and per-cpu allocation)
+ *	means the total number of concurrent fixmaps will be limited to max
+ *	such allocations in a single control path. Thus KM_TYPE_NR (another
+ *	historic relic) is a small'ish number which caps max percpu fixmaps
+ *
+ * ARC HIGHMEM Details
+ *
+ * - the kernel vaddr space from 0x7z to 0x8z (currently used by vmalloc/module)
+ *   is now shared between vmalloc and kmap (non overlapping though)
+ *
+ * - Both fixmap/pkmap use a dedicated page table each, hooked up to swapper PGD
+ *   This means each only has 1 PGDIR_SIZE worth of kvaddr mappings, which means
+ *   2M of kvaddr space for typical config (8K page and 11:8:13 traversal split)
+ *
+ * - fixmap anyhow needs a limited number of mappings. So 2M kvaddr == 256 PTE
+ *   slots across NR_CPUS would be more than sufficient (generic code defines
+ *   KM_TYPE_NR as 20).
+ *
+ * - pkmap being preemptible, in theory could do with more than 256 concurrent
+ *   mappings. However, generic pkmap code: map_new_virtual(), doesn't traverse
+ *   the PGD and only works with a single page table @pkmap_page_table, hence
+ *   sets the limit
+ */
+
+extern pte_t * pkmap_page_table;
+static pte_t * fixmap_page_table;
+
+void *kmap(struct page *page)
+{
+	BUG_ON(in_interrupt());
+	if (!PageHighMem(page))
+		return page_address(page);
+
+	return kmap_high(page);
+}
+
+void *kmap_atomic(struct page *page)
+{
+	int idx, cpu_idx;
+	unsigned long vaddr;
+
+	preempt_disable();
+	pagefault_disable();
+	if (!PageHighMem(page))
+		return page_address(page);
+
+	cpu_idx = kmap_atomic_idx_push();
+	idx = cpu_idx + KM_TYPE_NR * smp_processor_id();
+	vaddr = FIXMAP_ADDR(idx);
+
+	set_pte_at(&init_mm, vaddr, fixmap_page_table + idx,
+		   mk_pte(page, kmap_prot));
+
+	return (void *)vaddr;
+}
+EXPORT_SYMBOL(kmap_atomic);
+
+void __kunmap_atomic(void *kv)
+{
+	unsigned long kvaddr = (unsigned long)kv;
+
+	if (kvaddr >= FIXMAP_BASE && kvaddr < (FIXMAP_BASE + FIXMAP_SIZE)) {
+
+		/*
+		 * Because preemption is disabled, this vaddr can be associated
+		 * with the current allocated index.
+		 * But in case of multiple live kmap_atomic(), it still relies on
+		 * callers to unmap in right order.
+		 */
+		int cpu_idx = kmap_atomic_idx();
+		int idx = cpu_idx + KM_TYPE_NR * smp_processor_id();
+
+		WARN_ON(kvaddr != FIXMAP_ADDR(idx));
+
+		pte_clear(&init_mm, kvaddr, fixmap_page_table + idx);
+		local_flush_tlb_kernel_range(kvaddr, kvaddr + PAGE_SIZE);
+
+		kmap_atomic_idx_pop();
+	}
+
+	pagefault_enable();
+	preempt_enable();
+}
+EXPORT_SYMBOL(__kunmap_atomic);
+
+noinline pte_t *alloc_kmap_pgtable(unsigned long kvaddr)
+{
+	pgd_t *pgd_k;
+	pud_t *pud_k;
+	pmd_t *pmd_k;
+	pte_t *pte_k;
+
+	pgd_k = pgd_offset_k(kvaddr);
+	pud_k = pud_offset(pgd_k, kvaddr);
+	pmd_k = pmd_offset(pud_k, kvaddr);
+
+	pte_k = (pte_t *)alloc_bootmem_low_pages(PAGE_SIZE);
+	pmd_populate_kernel(&init_mm, pmd_k, pte_k);
+	return pte_k;
+}
+
+void kmap_init(void)
+{
+	/* Due to recursive include hell, we can't do this in processor.h */
+	BUILD_BUG_ON(PAGE_OFFSET < (VMALLOC_END + FIXMAP_SIZE + PKMAP_SIZE));
+
+	BUILD_BUG_ON(KM_TYPE_NR > PTRS_PER_PTE);
+	pkmap_page_table = alloc_kmap_pgtable(PKMAP_BASE);
+
+	BUILD_BUG_ON(LAST_PKMAP > PTRS_PER_PTE);
+	fixmap_page_table = alloc_kmap_pgtable(FIXMAP_BASE);
+}

arch/arc/mm/init.c

@@ -15,6 +15,7 @@
 #endif
 #include <linux/swap.h>
 #include <linux/module.h>
+#include <linux/highmem.h>
 #include <asm/page.h>
 #include <asm/pgalloc.h>
 #include <asm/sections.h>
@@ -24,16 +25,22 @@ pgd_t swapper_pg_dir[PTRS_PER_PGD] __aligned(PAGE_SIZE);
 char empty_zero_page[PAGE_SIZE] __aligned(PAGE_SIZE);
 EXPORT_SYMBOL(empty_zero_page);
 
-/* Default tot mem from .config */
-static unsigned long arc_mem_sz = 0x20000000;  /* some default */
+static const unsigned long low_mem_start = CONFIG_LINUX_LINK_BASE;
+static unsigned long low_mem_sz;
+
+#ifdef CONFIG_HIGHMEM
+static unsigned long min_high_pfn;
+static u64 high_mem_start;
+static u64 high_mem_sz;
+#endif
 
 /* User can over-ride above with "mem=nnn[KkMm]" in cmdline */
 static int __init setup_mem_sz(char *str)
 {
-	arc_mem_sz = memparse(str, NULL) & PAGE_MASK;
+	low_mem_sz = memparse(str, NULL) & PAGE_MASK;
 
 	/* early console might not be setup yet - it will show up later */
-	pr_info("\"mem=%s\": mem sz set to %ldM\n", str, TO_MB(arc_mem_sz));
+	pr_info("\"mem=%s\": mem sz set to %ldM\n", str, TO_MB(low_mem_sz));
 
 	return 0;
 }
@@ -41,8 +48,22 @@ early_param("mem", setup_mem_sz);
 
 void __init early_init_dt_add_memory_arch(u64 base, u64 size)
 {
-	arc_mem_sz = size & PAGE_MASK;
-	pr_info("Memory size set via devicetree %ldM\n", TO_MB(arc_mem_sz));
+	int in_use = 0;
+
+	if (!low_mem_sz) {
+		BUG_ON(base != low_mem_start);
+		low_mem_sz = size;
+		in_use = 1;
+	} else {
+#ifdef CONFIG_HIGHMEM
+		high_mem_start = base;
+		high_mem_sz = size;
+		in_use = 1;
+#endif
+	}
+
+	pr_info("Memory @ %llx [%lldM] %s\n",
+		base, TO_MB(size), !in_use ? "Not used":"");
 }
 
 #ifdef CONFIG_BLK_DEV_INITRD
@@ -72,46 +93,62 @@ early_param("initrd", early_initrd);
 void __init setup_arch_memory(void)
 {
 	unsigned long zones_size[MAX_NR_ZONES];
-	unsigned long end_mem = CONFIG_LINUX_LINK_BASE + arc_mem_sz;
+	unsigned long zones_holes[MAX_NR_ZONES];
 
 	init_mm.start_code = (unsigned long)_text;
 	init_mm.end_code = (unsigned long)_etext;
 	init_mm.end_data = (unsigned long)_edata;
 	init_mm.brk = (unsigned long)_end;
 
-	/*
-	 * We do it here, so that memory is correctly instantiated
-	 * even if "mem=xxx" cmline over-ride is given and/or
-	 * DT has memory node. Each causes an update to @arc_mem_sz
-	 * and we finally add memory one here
-	 */
-	memblock_add(CONFIG_LINUX_LINK_BASE, arc_mem_sz);
-
-	/*------------- externs in mm need setting up ---------------*/
-
 	/* first page of system - kernel .vector starts here */
 	min_low_pfn = ARCH_PFN_OFFSET;
 
-	/* Last usable page of low mem (no HIGHMEM yet for ARC port) */
-	max_low_pfn = max_pfn = PFN_DOWN(end_mem);
+	/* Last usable page of low mem */
+	max_low_pfn = max_pfn = PFN_DOWN(low_mem_start + low_mem_sz);
 
-	max_mapnr = max_low_pfn - min_low_pfn;
+#ifdef CONFIG_HIGHMEM
+	min_high_pfn = PFN_DOWN(high_mem_start);
+	max_pfn = PFN_DOWN(high_mem_start + high_mem_sz);
+#endif
+
+	max_mapnr = max_pfn - min_low_pfn;
 
-	/*------------- reserve kernel image -----------------------*/
-	memblock_reserve(CONFIG_LINUX_LINK_BASE,
-			 __pa(_end) - CONFIG_LINUX_LINK_BASE);
+	/*------------- bootmem allocator setup -----------------------*/
+
+	/*
+	 * seed the bootmem allocator after any DT memory node parsing or
+	 * "mem=xxx" cmdline overrides have potentially updated @arc_mem_sz
+	 *
+	 * Only low mem is added, otherwise we have crashes when allocating
+	 * mem_map[] itself. NO_BOOTMEM allocates mem_map[] at the end of
+	 * avail memory, ending in highmem with a > 32-bit address. However
+	 * it then tries to memset it with a truncaed 32-bit handle, causing
+	 * the crash
+	 */
+
+	memblock_add(low_mem_start, low_mem_sz);
+	memblock_reserve(low_mem_start, __pa(_end) - low_mem_start);
 
 #ifdef CONFIG_BLK_DEV_INITRD
-	/*------------- reserve initrd image -----------------------*/
 	if (initrd_start)
 		memblock_reserve(__pa(initrd_start), initrd_end - initrd_start);
 #endif
 
 	memblock_dump_all();
 
-	/*-------------- node setup --------------------------------*/
+	/*----------------- node/zones setup --------------------------*/
 	memset(zones_size, 0, sizeof(zones_size));
-	zones_size[ZONE_NORMAL] = max_mapnr;
+	memset(zones_holes, 0, sizeof(zones_holes));
+
+	zones_size[ZONE_NORMAL] = max_low_pfn - min_low_pfn;
+	zones_holes[ZONE_NORMAL] = 0;
+
+#ifdef CONFIG_HIGHMEM
+	zones_size[ZONE_HIGHMEM] = max_pfn - max_low_pfn;
+
+	/* This handles the peripheral address space hole */
+	zones_holes[ZONE_HIGHMEM] = min_high_pfn - max_low_pfn;
+#endif
 
 	/*
 	 * We can't use the helper free_area_init(zones[]) because it uses
@@ -122,9 +159,12 @@ void __init setup_arch_memory(void)
 	free_area_init_node(0,			/* node-id */
 			    zones_size,		/* num pages per zone */
 			    min_low_pfn,	/* first pfn of node */
-			    NULL);		/* NO holes */
+			    zones_holes);	/* holes */
 
-	high_memory = (void *)end_mem;
+#ifdef CONFIG_HIGHMEM
+	high_memory = (void *)(min_high_pfn << PAGE_SHIFT);
+	kmap_init();
+#endif
 }
 
 /*
@@ -135,6 +175,14 @@ void __init setup_arch_memory(void)
  */
 void __init mem_init(void)
 {
+#ifdef CONFIG_HIGHMEM
+	unsigned long tmp;
+
+	reset_all_zones_managed_pages();
+	for (tmp = min_high_pfn; tmp < max_pfn; tmp++)
+		free_highmem_page(pfn_to_page(tmp));
+#endif
+
 	free_all_bootmem();
 	mem_init_print_info(NULL);
 }

arch/arc/mm/tlbex.S

@@ -205,20 +205,38 @@ ex_saved_reg1:
 #endif
 
 	lsr     r0, r2, PGDIR_SHIFT     ; Bits for indexing into PGD
-	ld.as   r1, [r1, r0]            ; PGD entry corresp to faulting addr
-	and.f   r1, r1, PAGE_MASK       ; Ignoring protection and other flags
-	;   contains Ptr to Page Table
-	bz.d    do_slow_path_pf         ; if no Page Table, do page fault
+	ld.as   r3, [r1, r0]            ; PGD entry corresp to faulting addr
+	tst	r3, r3
+	bz	do_slow_path_pf         ; if no Page Table, do page fault
+
+#ifdef CONFIG_TRANSPARENT_HUGEPAGE
+	and.f	0, r3, _PAGE_HW_SZ	; Is this Huge PMD (thp)
+	add2.nz	r1, r1, r0
+	bnz.d	2f		; YES: PGD == PMD has THP PTE: stop pgd walk
+	mov.nz	r0, r3
+
+#endif
+	and	r1, r3, PAGE_MASK
 
 	; Get the PTE entry: The idea is
 	; (1) x = addr >> PAGE_SHIFT 	-> masks page-off bits from @fault-addr
 	; (2) y = x & (PTRS_PER_PTE - 1) -> to get index
-	; (3) z = pgtbl[y]
-	; To avoid the multiply by in end, we do the -2, <<2 below
+	; (3) z = (pgtbl + y * 4)
+
+#ifdef CONFIG_ARC_HAS_PAE40
+#define PTE_SIZE_LOG	3	/* 8 == 2 ^ 3 */
+#else
+#define PTE_SIZE_LOG	2	/* 4 == 2 ^ 2 */
+#endif
+
+	; multiply in step (3) above avoided by shifting lesser in step (1)
+	lsr     r0, r2, ( PAGE_SHIFT - PTE_SIZE_LOG )
+	and     r0, r0, ( (PTRS_PER_PTE - 1) << PTE_SIZE_LOG )
+	ld.aw   r0, [r1, r0]            ; r0: PTE (lower word only for PAE40)
+					; r1: PTE ptr
+
+2:
 
-	lsr     r0, r2, (PAGE_SHIFT - 2)
-	and     r0, r0, ( (PTRS_PER_PTE - 1) << 2)
-	ld.aw   r0, [r1, r0]            ; get PTE and PTE ptr for fault addr
 #ifdef CONFIG_ARC_DBG_TLB_MISS_COUNT
 	and.f 0, r0, _PAGE_PRESENT
 	bz   1f
@@ -233,18 +251,23 @@ ex_saved_reg1:
 ;-----------------------------------------------------------------
 ; Convert Linux PTE entry into TLB entry
 ; A one-word PTE entry is programmed as two-word TLB Entry [PD0:PD1] in mmu
+;    (for PAE40, two-words PTE, while three-word TLB Entry [PD0:PD1:PD1HI])
 ; IN: r0 = PTE, r1 = ptr to PTE
 
 .macro CONV_PTE_TO_TLB
-	and    r3, r0, PTE_BITS_RWX	;       r w x
-	lsl    r2, r3, 3		; r w x 0 0 0 (GLOBAL, kernel only)
+	and    r3, r0, PTE_BITS_RWX	;          r  w  x
+	lsl    r2, r3, 3		; Kr Kw Kx 0  0  0 (GLOBAL, kernel only)
 	and.f  0,  r0, _PAGE_GLOBAL
-	or.z   r2, r2, r3		; r w x r w x (!GLOBAL, user page)
+	or.z   r2, r2, r3		; Kr Kw Kx Ur Uw Ux (!GLOBAL, user page)
 
 	and r3, r0, PTE_BITS_NON_RWX_IN_PD1 ; Extract PFN+cache bits from PTE
 	or  r3, r3, r2
 
-	sr  r3, [ARC_REG_TLBPD1]    	; these go in PD1
+	sr  r3, [ARC_REG_TLBPD1]    	; paddr[31..13] | Kr Kw Kx Ur Uw Ux | C
+#ifdef	CONFIG_ARC_HAS_PAE40
+	ld	r3, [r1, 4]		; paddr[39..32]
+	sr	r3, [ARC_REG_TLBPD1HI]
+#endif
 
 	and r2, r0, PTE_BITS_IN_PD0 ; Extract other PTE flags: (V)alid, (G)lb
 
@@ -365,7 +388,7 @@ ENTRY(EV_TLBMissD)
 	lr      r3, [ecr]
 	or      r0, r0, _PAGE_ACCESSED        ; Accessed bit always
 	btst_s  r3,  ECR_C_BIT_DTLB_ST_MISS   ; See if it was a Write Access ?
-	or.nz   r0, r0, _PAGE_MODIFIED        ; if Write, set Dirty bit as well
+	or.nz   r0, r0, _PAGE_DIRTY           ; if Write, set Dirty bit as well
 	st_s    r0, [r1]                      ; Write back PTE
 
 	CONV_PTE_TO_TLB

arch/arc/plat-axs10x/axs10x.c

@@ -455,11 +455,6 @@ static void __init axs103_early_init(void)
 	axs10x_print_board_ver(AXC003_CREG + 4088, "AXC003 CPU Card");
 
 	axs10x_early_init();
-
-#ifdef CONFIG_ARC_MCIP
-	/* No Hardware init, but filling the smp ops callbacks */
-	mcip_init_early_smp();
-#endif
 }
 #endif
 
@@ -487,9 +482,6 @@ static const char *axs103_compat[] __initconst = {
 MACHINE_START(AXS103, "axs103")
 	.dt_compat	= axs103_compat,
 	.init_early	= axs103_early_init,
-#ifdef CONFIG_ARC_MCIP
-	.init_smp	= mcip_init_smp,
-#endif
 MACHINE_END
 
 /*

arch/arc/plat-sim/platform.c

@@ -30,8 +30,4 @@ static const char *simulation_compat[] __initconst = {
 
 MACHINE_START(SIMULATION, "simulation")
 	.dt_compat	= simulation_compat,
-#ifdef CONFIG_ARC_MCIP
-	.init_early	= mcip_init_early_smp,
-	.init_smp	= mcip_init_smp,
-#endif
 MACHINE_END

include/asm-generic/pgtable.h

@@ -30,9 +30,19 @@ extern int ptep_set_access_flags(struct vm_area_struct *vma,
 #endif
 
 #ifndef __HAVE_ARCH_PMDP_SET_ACCESS_FLAGS
+#ifdef CONFIG_TRANSPARENT_HUGEPAGE
 extern int pmdp_set_access_flags(struct vm_area_struct *vma,
 				 unsigned long address, pmd_t *pmdp,
 				 pmd_t entry, int dirty);
+#else
+static inline int pmdp_set_access_flags(struct vm_area_struct *vma,
+					unsigned long address, pmd_t *pmdp,
+					pmd_t entry, int dirty)
+{
+	BUILD_BUG();
+	return 0;
+}
+#endif /* CONFIG_TRANSPARENT_HUGEPAGE */
 #endif
 
 #ifndef __HAVE_ARCH_PTEP_TEST_AND_CLEAR_YOUNG
@@ -64,12 +74,12 @@ static inline int pmdp_test_and_clear_young(struct vm_area_struct *vma,
 		set_pmd_at(vma->vm_mm, address, pmdp, pmd_mkold(pmd));
 	return r;
 }
-#else /* CONFIG_TRANSPARENT_HUGEPAGE */
+#else
 static inline int pmdp_test_and_clear_young(struct vm_area_struct *vma,
 					    unsigned long address,
 					    pmd_t *pmdp)
 {
-	BUG();
+	BUILD_BUG();
 	return 0;
 }
 #endif /* CONFIG_TRANSPARENT_HUGEPAGE */
@@ -81,8 +91,21 @@ int ptep_clear_flush_young(struct vm_area_struct *vma,
 #endif
 
 #ifndef __HAVE_ARCH_PMDP_CLEAR_YOUNG_FLUSH
-int pmdp_clear_flush_young(struct vm_area_struct *vma,
-			   unsigned long address, pmd_t *pmdp);
+#ifdef CONFIG_TRANSPARENT_HUGEPAGE
+extern int pmdp_clear_flush_young(struct vm_area_struct *vma,
+				  unsigned long address, pmd_t *pmdp);
+#else
+/*
+ * Despite relevant to THP only, this API is called from generic rmap code
+ * under PageTransHuge(), hence needs a dummy implementation for !THP
+ */
+static inline int pmdp_clear_flush_young(struct vm_area_struct *vma,
+					 unsigned long address, pmd_t *pmdp)
+{
+	BUILD_BUG();
+	return 0;
+}
+#endif /* CONFIG_TRANSPARENT_HUGEPAGE */
 #endif
 
 #ifndef __HAVE_ARCH_PTEP_GET_AND_CLEAR
@@ -175,11 +198,11 @@ static inline void pmdp_set_wrprotect(struct mm_struct *mm,
 	pmd_t old_pmd = *pmdp;
 	set_pmd_at(mm, address, pmdp, pmd_wrprotect(old_pmd));
 }
-#else /* CONFIG_TRANSPARENT_HUGEPAGE */
+#else
 static inline void pmdp_set_wrprotect(struct mm_struct *mm,
 				      unsigned long address, pmd_t *pmdp)
 {
-	BUG();
+	BUILD_BUG();
 }
 #endif /* CONFIG_TRANSPARENT_HUGEPAGE */
 #endif
@@ -248,7 +271,7 @@ static inline int pmd_same(pmd_t pmd_a, pmd_t pmd_b)
 #else /* CONFIG_TRANSPARENT_HUGEPAGE */
 static inline int pmd_same(pmd_t pmd_a, pmd_t pmd_b)
 {
-	BUG();
+	BUILD_BUG();
 	return 0;
 }
 #endif /* CONFIG_TRANSPARENT_HUGEPAGE */

mm/huge_memory.c

@@ -1880,7 +1880,7 @@ static int __split_huge_page_map(struct page *page,
 		 * here). But it is generally safer to never allow
 		 * small and huge TLB entries for the same virtual
 		 * address to be loaded simultaneously. So instead of
-		 * doing "pmd_populate(); flush_tlb_range();" we first
+		 * doing "pmd_populate(); flush_pmd_tlb_range();" we first
 		 * mark the current pmd notpresent (atomically because
 		 * here the pmd_trans_huge and pmd_trans_splitting
 		 * must remain set at all times on the pmd until the

mm/pgtable-generic.c

@@ -57,35 +57,59 @@ int ptep_set_access_flags(struct vm_area_struct *vma,
 }
 #endif
 
+#ifndef __HAVE_ARCH_PTEP_CLEAR_YOUNG_FLUSH
+int ptep_clear_flush_young(struct vm_area_struct *vma,
+			   unsigned long address, pte_t *ptep)
+{
+	int young;
+	young = ptep_test_and_clear_young(vma, address, ptep);
+	if (young)
+		flush_tlb_page(vma, address);
+	return young;
+}
+#endif
+
+#ifndef __HAVE_ARCH_PTEP_CLEAR_FLUSH
+pte_t ptep_clear_flush(struct vm_area_struct *vma, unsigned long address,
+		       pte_t *ptep)
+{
+	struct mm_struct *mm = (vma)->vm_mm;
+	pte_t pte;
+	pte = ptep_get_and_clear(mm, address, ptep);
+	if (pte_accessible(mm, pte))
+		flush_tlb_page(vma, address);
+	return pte;
+}
+#endif
+
+#ifdef CONFIG_TRANSPARENT_HUGEPAGE
+
+#ifndef __HAVE_ARCH_FLUSH_PMD_TLB_RANGE
+
+/*
+ * ARCHes with special requirements for evicting THP backing TLB entries can
+ * implement this. Otherwise also, it can help optimize normal TLB flush in
+ * THP regime. stock flush_tlb_range() typically has optimization to nuke the
+ * entire TLB TLB if flush span is greater than a threshhold, which will
+ * likely be true for a single huge page. Thus a single thp flush will
+ * invalidate the entire TLB which is not desitable.
+ * e.g. see arch/arc: flush_pmd_tlb_range
+ */
+#define flush_pmd_tlb_range(vma, addr, end)	flush_tlb_range(vma, addr, end)
+#endif
+
 #ifndef __HAVE_ARCH_PMDP_SET_ACCESS_FLAGS
 int pmdp_set_access_flags(struct vm_area_struct *vma,
 			  unsigned long address, pmd_t *pmdp,
 			  pmd_t entry, int dirty)
 {
-#ifdef CONFIG_TRANSPARENT_HUGEPAGE
 	int changed = !pmd_same(*pmdp, entry);
 	VM_BUG_ON(address & ~HPAGE_PMD_MASK);
 	if (changed) {
 		set_pmd_at(vma->vm_mm, address, pmdp, entry);
-		flush_tlb_range(vma, address, address + HPAGE_PMD_SIZE);
+		flush_pmd_tlb_range(vma, address, address + HPAGE_PMD_SIZE);
 	}
 	return changed;
-#else /* CONFIG_TRANSPARENT_HUGEPAGE */
-	BUG();
-	return 0;
-#endif /* CONFIG_TRANSPARENT_HUGEPAGE */
-}
-#endif
-
-#ifndef __HAVE_ARCH_PTEP_CLEAR_YOUNG_FLUSH
-int ptep_clear_flush_young(struct vm_area_struct *vma,
-			   unsigned long address, pte_t *ptep)
-{
-	int young;
-	young = ptep_test_and_clear_young(vma, address, ptep);
-	if (young)
-		flush_tlb_page(vma, address);
-	return young;
 }
 #endif
 
@@ -94,33 +118,15 @@ int pmdp_clear_flush_young(struct vm_area_struct *vma,
 			   unsigned long address, pmd_t *pmdp)
 {
 	int young;
-#ifdef CONFIG_TRANSPARENT_HUGEPAGE
 	VM_BUG_ON(address & ~HPAGE_PMD_MASK);
-#else
-	BUG();
-#endif /* CONFIG_TRANSPARENT_HUGEPAGE */
 	young = pmdp_test_and_clear_young(vma, address, pmdp);
 	if (young)
-		flush_tlb_range(vma, address, address + HPAGE_PMD_SIZE);
+		flush_pmd_tlb_range(vma, address, address + HPAGE_PMD_SIZE);
 	return young;
 }
 #endif
 
-#ifndef __HAVE_ARCH_PTEP_CLEAR_FLUSH
-pte_t ptep_clear_flush(struct vm_area_struct *vma, unsigned long address,
-		       pte_t *ptep)
-{
-	struct mm_struct *mm = (vma)->vm_mm;
-	pte_t pte;
-	pte = ptep_get_and_clear(mm, address, ptep);
-	if (pte_accessible(mm, pte))
-		flush_tlb_page(vma, address);
-	return pte;
-}
-#endif
-
 #ifndef __HAVE_ARCH_PMDP_HUGE_CLEAR_FLUSH
-#ifdef CONFIG_TRANSPARENT_HUGEPAGE
 pmd_t pmdp_huge_clear_flush(struct vm_area_struct *vma, unsigned long address,
 			    pmd_t *pmdp)
 {
@@ -128,14 +134,12 @@ pmd_t pmdp_huge_clear_flush(struct vm_area_struct *vma, unsigned long address,
 	VM_BUG_ON(address & ~HPAGE_PMD_MASK);
 	VM_BUG_ON(!pmd_trans_huge(*pmdp));
 	pmd = pmdp_huge_get_and_clear(vma->vm_mm, address, pmdp);
-	flush_tlb_range(vma, address, address + HPAGE_PMD_SIZE);
+	flush_pmd_tlb_range(vma, address, address + HPAGE_PMD_SIZE);
 	return pmd;
 }
-#endif /* CONFIG_TRANSPARENT_HUGEPAGE */
 #endif
 
 #ifndef __HAVE_ARCH_PMDP_SPLITTING_FLUSH
-#ifdef CONFIG_TRANSPARENT_HUGEPAGE
 void pmdp_splitting_flush(struct vm_area_struct *vma, unsigned long address,
 			  pmd_t *pmdp)
 {
@@ -143,13 +147,11 @@ void pmdp_splitting_flush(struct vm_area_struct *vma, unsigned long address,
 	VM_BUG_ON(address & ~HPAGE_PMD_MASK);
 	set_pmd_at(vma->vm_mm, address, pmdp, pmd);
 	/* tlb flush only to serialize against gup-fast */
-	flush_tlb_range(vma, address, address + HPAGE_PMD_SIZE);
+	flush_pmd_tlb_range(vma, address, address + HPAGE_PMD_SIZE);
 }
-#endif /* CONFIG_TRANSPARENT_HUGEPAGE */
 #endif
 
 #ifndef __HAVE_ARCH_PGTABLE_DEPOSIT
-#ifdef CONFIG_TRANSPARENT_HUGEPAGE
 void pgtable_trans_huge_deposit(struct mm_struct *mm, pmd_t *pmdp,
 				pgtable_t pgtable)
 {
@@ -162,11 +164,9 @@ void pgtable_trans_huge_deposit(struct mm_struct *mm, pmd_t *pmdp,
 		list_add(&pgtable->lru, &pmd_huge_pte(mm, pmdp)->lru);
 	pmd_huge_pte(mm, pmdp) = pgtable;
 }
-#endif /* CONFIG_TRANSPARENT_HUGEPAGE */
 #endif
 
 #ifndef __HAVE_ARCH_PGTABLE_WITHDRAW
-#ifdef CONFIG_TRANSPARENT_HUGEPAGE
 /* no "address" argument so destroys page coloring of some arch */
 pgtable_t pgtable_trans_huge_withdraw(struct mm_struct *mm, pmd_t *pmdp)
 {
@@ -185,23 +185,19 @@ pgtable_t pgtable_trans_huge_withdraw(struct mm_struct *mm, pmd_t *pmdp)
 	}
 	return pgtable;
 }
-#endif /* CONFIG_TRANSPARENT_HUGEPAGE */
 #endif
 
 #ifndef __HAVE_ARCH_PMDP_INVALIDATE
-#ifdef CONFIG_TRANSPARENT_HUGEPAGE
 void pmdp_invalidate(struct vm_area_struct *vma, unsigned long address,
 		     pmd_t *pmdp)
 {
 	pmd_t entry = *pmdp;
 	set_pmd_at(vma->vm_mm, address, pmdp, pmd_mknotpresent(entry));
-	flush_tlb_range(vma, address, address + HPAGE_PMD_SIZE);
+	flush_pmd_tlb_range(vma, address, address + HPAGE_PMD_SIZE);
 }
-#endif /* CONFIG_TRANSPARENT_HUGEPAGE */
 #endif
 
 #ifndef pmdp_collapse_flush
-#ifdef CONFIG_TRANSPARENT_HUGEPAGE
 pmd_t pmdp_collapse_flush(struct vm_area_struct *vma, unsigned long address,
 			  pmd_t *pmdp)
 {
@@ -214,8 +210,8 @@ pmd_t pmdp_collapse_flush(struct vm_area_struct *vma, unsigned long address,
 	VM_BUG_ON(address & ~HPAGE_PMD_MASK);
 	VM_BUG_ON(pmd_trans_huge(*pmdp));
 	pmd = pmdp_huge_get_and_clear(vma->vm_mm, address, pmdp);
-	flush_tlb_range(vma, address, address + HPAGE_PMD_SIZE);
+	flush_pmd_tlb_range(vma, address, address + HPAGE_PMD_SIZE);
 	return pmd;
 }
-#endif /* CONFIG_TRANSPARENT_HUGEPAGE */
 #endif
+#endif /* CONFIG_TRANSPARENT_HUGEPAGE */