Merge branch 'parisc-5.2-1' of git://git.kernel.org/pub/scm/linux/kernel/git/deller/parisc-linux

Pull parisc updates from Helge Deller:
 "Many great new features, fixes and optimizations, including:

   - Convert page table updates to use per-pagetable spinlocks which
     overall improves performance on SMP machines a lot, by Mikulas
     Patocka

   - Kernel debugger (KGDB) support, by Sven Schnelle

   - KPROBES support, by Sven Schnelle

   - Lots of TLB lock/flush improvements, by Dave Anglin

   - Drop DISCONTIGMEM and switch to SPARSEMEM

   - Added JUMP_LABEL, branch runtime-patching support

   - Lots of other small speedups and cleanups, e.g. for QEMU, stack
     randomization, avoidance of name clashes, documentation updates,
     etc ..."

* 'parisc-5.2-1' of git://git.kernel.org/pub/scm/linux/kernel/git/deller/parisc-linux: (28 commits)
  parisc: Add static branch and JUMP_LABEL feature
  parisc: Use PA_ASM_LEVEL in boot code
  parisc: Rename LEVEL to PA_ASM_LEVEL to avoid name clash with DRBD code
  parisc: Update huge TLB page support to use per-pagetable spinlock
  parisc: Use per-pagetable spinlock
  parisc: Allow live-patching of __meminit functions
  parisc: Add memory barrier to asm pdc and sync instructions
  parisc: Add memory clobber to TLB purges
  parisc: Use ldcw instruction for SMP spinlock release barrier
  parisc: Remove lock code to serialize TLB operations in pacache.S
  parisc: Switch from DISCONTIGMEM to SPARSEMEM
  parisc: enable wide mode early
  parisc: update feature lists
  parisc: Show n/a if product number not available
  parisc: remove unused flags parameter in __patch_text()
  doc: update kprobes supported architecture list
  parisc: Implement kretprobes
  parisc: remove kprobes.h from generic-y
  parisc: Implement kprobes
  parisc: add functions required by KPROBE_EVENTS
  ...

Documentation/features/debug/kgdb/arch-support.txt

@@ -21,7 +21,7 @@
     |       nds32: | TODO |
     |       nios2: |  ok  |
     |    openrisc: | TODO |
-    |      parisc: | TODO |
+    |      parisc: |  ok  |
     |     powerpc: |  ok  |
     |       riscv: | TODO |
     |        s390: | TODO |

Documentation/features/debug/kprobes/arch-support.txt

@@ -21,7 +21,7 @@
     |       nds32: | TODO |
     |       nios2: | TODO |
     |    openrisc: | TODO |
-    |      parisc: | TODO |
+    |      parisc: |  ok  |
     |     powerpc: |  ok  |
     |       riscv: |  ok  |
     |        s390: |  ok  |

Documentation/features/debug/kretprobes/arch-support.txt

@@ -21,7 +21,7 @@
     |       nds32: | TODO |
     |       nios2: | TODO |
     |    openrisc: | TODO |
-    |      parisc: | TODO |
+    |      parisc: |  ok  |
     |     powerpc: |  ok  |
     |       riscv: | TODO |
     |        s390: |  ok  |

Documentation/kprobes.txt

@@ -321,6 +321,7 @@ architectures:
 - ppc
 - mips
 - s390
+- parisc
 
 Configuring Kprobes
 ===================

arch/parisc/Kconfig

@@ -36,6 +36,7 @@ config PARISC
 	select GENERIC_STRNCPY_FROM_USER
 	select SYSCTL_ARCH_UNALIGN_ALLOW
 	select SYSCTL_EXCEPTION_TRACE
+	select ARCH_DISCARD_MEMBLOCK
 	select HAVE_MOD_ARCH_SPECIFIC
 	select VIRT_TO_BUS
 	select MODULES_USE_ELF_RELA
@@ -44,6 +45,8 @@ config PARISC
 	select HAVE_DEBUG_STACKOVERFLOW
 	select HAVE_ARCH_AUDITSYSCALL
 	select HAVE_ARCH_HASH
+	select HAVE_ARCH_JUMP_LABEL
+	select HAVE_ARCH_JUMP_LABEL_RELATIVE
 	select HAVE_ARCH_SECCOMP_FILTER
 	select HAVE_ARCH_TRACEHOOK
 	select HAVE_REGS_AND_STACK_ACCESS_API
@@ -54,6 +57,9 @@ config PARISC
 	select CPU_NO_EFFICIENT_FFS
 	select NEED_DMA_MAP_STATE
 	select NEED_SG_DMA_LENGTH
+	select HAVE_ARCH_KGDB
+	select HAVE_KPROBES
+	select HAVE_KRETPROBES
 
 	help
 	  The PA-RISC microprocessor is designed by Hewlett-Packard and used
@@ -305,21 +311,16 @@ config ARCH_SELECT_MEMORY_MODEL
 	def_bool y
 	depends on 64BIT
 
-config ARCH_DISCONTIGMEM_ENABLE
+config ARCH_SPARSEMEM_ENABLE
 	def_bool y
 	depends on 64BIT
 
 config ARCH_FLATMEM_ENABLE
 	def_bool y
 
-config ARCH_DISCONTIGMEM_DEFAULT
+config ARCH_SPARSEMEM_DEFAULT
 	def_bool y
-	depends on ARCH_DISCONTIGMEM_ENABLE
-
-config NODES_SHIFT
-	int
-	default "3"
-	depends on NEED_MULTIPLE_NODES
+	depends on ARCH_SPARSEMEM_ENABLE
 
 source "kernel/Kconfig.hz"
 

arch/parisc/boot/compressed/head.S

@@ -22,7 +22,7 @@
 	__HEAD
 
 ENTRY(startup)
-	 .level LEVEL
+	 .level PA_ASM_LEVEL
 
 #define PSW_W_SM	0x200
 #define PSW_W_BIT       36
@@ -63,7 +63,7 @@ $bss_loop:
 	load32	BOOTADDR(decompress_kernel),%r3
 
 #ifdef CONFIG_64BIT
-	.level LEVEL
+	.level PA_ASM_LEVEL
 	ssm	PSW_W_SM, %r0		/* set W-bit */
 	depdi	0, 31, 32, %r3
 #endif
@@ -72,7 +72,7 @@ $bss_loop:
 
 startup_continue:
 #ifdef CONFIG_64BIT
-	.level LEVEL
+	.level PA_ASM_LEVEL
 	rsm	PSW_W_SM, %r0		/* clear W-bit */
 #endif
 

arch/parisc/boot/compressed/misc.c

@@ -145,14 +145,13 @@ static int putchar(int c)
 
 void __noreturn error(char *x)
 {
-	puts("\n\n");
-	puts(x);
-	puts("\n\n -- System halted");
+	if (x) puts(x);
+	puts("\n -- System halted\n");
 	while (1)	/* wait forever */
 		;
 }
 
-static int print_hex(unsigned long num)
+static int print_num(unsigned long num, int base)
 {
 	const char hex[] = "0123456789abcdef";
 	char str[40];
@@ -160,12 +159,14 @@ static int print_hex(unsigned long num)
 
 	str[i--] = '\0';
 	do {
-		str[i--] = hex[num & 0x0f];
-		num >>= 4;
+		str[i--] = hex[num % base];
+		num = num / base;
 	} while (num);
 
-	str[i--] = 'x';
-	str[i] = '0';
+	if (base == 16) {
+		str[i--] = 'x';
+		str[i] = '0';
+	} else i++;
 	puts(&str[i]);
 
 	return 0;
@@ -187,8 +188,9 @@ int printf(const char *fmt, ...)
 
 		if (fmt[++i] == '%')
 			goto put;
+		print_num(va_arg(args, unsigned long),
+			fmt[i] == 'x' ? 16:10);
 		++i;
-		print_hex(va_arg(args, unsigned long));
 	}
 
 	va_end(args);
@@ -327,8 +329,15 @@ unsigned long decompress_kernel(unsigned int started_wide,
 		free_mem_end_ptr = rd_start;
 #endif
 
-	if (free_mem_ptr >= free_mem_end_ptr)
-		error("Kernel too big for machine.");
+	if (free_mem_ptr >= free_mem_end_ptr) {
+		int free_ram;
+		free_ram = (free_mem_ptr >> 20) + 1;
+		if (free_ram < 32)
+			free_ram = 32;
+		printf("\nKernel requires at least %d MB RAM.\n",
+			free_ram);
+		error(NULL);
+	}
 
 #ifdef DEBUG
 	printf("\n");

arch/parisc/include/asm/Kbuild

@@ -10,7 +10,6 @@ generic-y += hw_irq.h
 generic-y += irq_regs.h
 generic-y += irq_work.h
 generic-y += kdebug.h
-generic-y += kprobes.h
 generic-y += kvm_para.h
 generic-y += local.h
 generic-y += local64.h

arch/parisc/include/asm/assembly.h

@@ -61,14 +61,14 @@
 #define LDCW		ldcw,co
 #define BL		b,l
 # ifdef CONFIG_64BIT
-#  define LEVEL		2.0w
+#  define PA_ASM_LEVEL	2.0w
 # else
-#  define LEVEL		2.0
+#  define PA_ASM_LEVEL	2.0
 # endif
 #else
 #define LDCW		ldcw
 #define BL		bl
-#define LEVEL		1.1
+#define PA_ASM_LEVEL	1.1
 #endif
 
 #ifdef __ASSEMBLY__

arch/parisc/include/asm/cache.h

@@ -44,22 +44,22 @@ void parisc_setup_cache_timing(void);
 
 #define pdtlb(addr)	asm volatile("pdtlb 0(%%sr1,%0)" \
 			ALTERNATIVE(ALT_COND_NO_SMP, INSN_PxTLB) \
-			: : "r" (addr))
+			: : "r" (addr) : "memory")
 #define pitlb(addr)	asm volatile("pitlb 0(%%sr1,%0)" \
 			ALTERNATIVE(ALT_COND_NO_SMP, INSN_PxTLB) \
 			ALTERNATIVE(ALT_COND_NO_SPLIT_TLB, INSN_NOP) \
-			: : "r" (addr))
+			: : "r" (addr) : "memory")
 #define pdtlb_kernel(addr)  asm volatile("pdtlb 0(%0)"   \
 			ALTERNATIVE(ALT_COND_NO_SMP, INSN_PxTLB) \
-			: : "r" (addr))
+			: : "r" (addr) : "memory")
 
 #define asm_io_fdc(addr) asm volatile("fdc %%r0(%0)" \
 			ALTERNATIVE(ALT_COND_NO_DCACHE, INSN_NOP) \
 			ALTERNATIVE(ALT_COND_NO_IOC_FDC, INSN_NOP) \
-			: : "r" (addr))
+			: : "r" (addr) : "memory")
 #define asm_io_sync()	asm volatile("sync" \
 			ALTERNATIVE(ALT_COND_NO_DCACHE, INSN_NOP) \
-			ALTERNATIVE(ALT_COND_NO_IOC_FDC, INSN_NOP) :: )
+			ALTERNATIVE(ALT_COND_NO_IOC_FDC, INSN_NOP) :::"memory")
 
 #endif /* ! __ASSEMBLY__ */
 

arch/parisc/include/asm/fixmap.h

@@ -15,17 +15,34 @@
  * from areas congruently mapped with user space.  It is 8MB large
  * and must be 16MB aligned */
 #define TMPALIAS_MAP_START	((__PAGE_OFFSET) - 16*1024*1024)
+
+#define FIXMAP_SIZE		(FIX_BITMAP_COUNT << PAGE_SHIFT)
+#define FIXMAP_START		(TMPALIAS_MAP_START - FIXMAP_SIZE)
 /* This is the kernel area for all maps (vmalloc, dma etc.)  most
  * usually, it extends up to TMPALIAS_MAP_START.  Virtual addresses
  * 0..GATEWAY_PAGE_SIZE are reserved for the gateway page */
 #define KERNEL_MAP_START	(GATEWAY_PAGE_SIZE)
-#define KERNEL_MAP_END		(TMPALIAS_MAP_START)
+#define KERNEL_MAP_END		(FIXMAP_START)
 
 #ifndef __ASSEMBLY__
+
+
+enum fixed_addresses {
+	/* Support writing RO kernel text via kprobes, jump labels, etc. */
+	FIX_TEXT_POKE0,
+	FIX_BITMAP_COUNT
+};
+
 extern void *parisc_vmalloc_start;
 #define PCXL_DMA_MAP_SIZE	(8*1024*1024)
 #define VMALLOC_START		((unsigned long)parisc_vmalloc_start)
 #define VMALLOC_END		(KERNEL_MAP_END)
+
+#define __fix_to_virt(_x) (FIXMAP_START + ((_x) << PAGE_SHIFT))
+
+void set_fixmap(enum fixed_addresses idx, phys_addr_t phys);
+void clear_fixmap(enum fixed_addresses idx);
+
 #endif /*__ASSEMBLY__*/
 
 #endif /*_ASM_FIXMAP_H*/

arch/parisc/include/asm/hardware.h

@@ -120,7 +120,7 @@ extern void get_pci_node_path(struct pci_dev *dev, struct hardware_path *path);
 extern void init_parisc_bus(void);
 extern struct device *hwpath_to_device(struct hardware_path *modpath);
 extern void device_to_hwpath(struct device *dev, struct hardware_path *path);
-
+extern int machine_has_merced_bus(void);
 
 /* inventory.c: */
 extern void do_memory_inventory(void);

arch/parisc/include/asm/jump_label.h

+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef _ASM_PARISC_JUMP_LABEL_H
+#define _ASM_PARISC_JUMP_LABEL_H
+
+#ifndef __ASSEMBLY__
+
+#include <linux/types.h>
+#include <asm/assembly.h>
+
+#define JUMP_LABEL_NOP_SIZE 4
+
+static __always_inline bool arch_static_branch(struct static_key *key, bool branch)
+{
+	asm_volatile_goto("1:\n\t"
+		 "nop\n\t"
+		 ".pushsection __jump_table,  \"aw\"\n\t"
+		 ".word 1b - ., %l[l_yes] - .\n\t"
+		 __stringify(ASM_ULONG_INSN) " %c0 - .\n\t"
+		 ".popsection\n\t"
+		 : :  "i" (&((char *)key)[branch]) :  : l_yes);
+
+	return false;
+l_yes:
+	return true;
+}
+
+static __always_inline bool arch_static_branch_jump(struct static_key *key, bool branch)
+{
+	asm_volatile_goto("1:\n\t"
+		 "b,n %l[l_yes]\n\t"
+		 ".pushsection __jump_table,  \"aw\"\n\t"
+		 ".word 1b - ., %l[l_yes] - .\n\t"
+		 __stringify(ASM_ULONG_INSN) " %c0 - .\n\t"
+		 ".popsection\n\t"
+		 : :  "i" (&((char *)key)[branch]) :  : l_yes);
+
+	return false;
+l_yes:
+	return true;
+}
+
+#endif  /* __ASSEMBLY__ */
+#endif

arch/parisc/include/asm/kgdb.h

+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * PA-RISC KGDB support
+ *
+ * Copyright (c) 2019 Sven Schnelle <svens@stackframe.org>
+ *
+ */
+
+#ifndef __PARISC_KGDB_H__
+#define __PARISC_KGDB_H__
+
+#define BREAK_INSTR_SIZE		4
+#define PARISC_KGDB_COMPILED_BREAK_INSN	0x3ffc01f
+#define PARISC_KGDB_BREAK_INSN		0x3ffa01f
+
+
+#define NUMREGBYTES			sizeof(struct parisc_gdb_regs)
+#define BUFMAX				4096
+
+#define CACHE_FLUSH_IS_SAFE		1
+
+#ifndef __ASSEMBLY__
+
+static inline void arch_kgdb_breakpoint(void)
+{
+	asm(".word %0" : : "i"(PARISC_KGDB_COMPILED_BREAK_INSN) : "memory");
+}
+
+struct parisc_gdb_regs {
+	unsigned long gpr[32];
+	unsigned long sar;
+	unsigned long iaoq_f;
+	unsigned long iasq_f;
+	unsigned long iaoq_b;
+	unsigned long iasq_b;
+	unsigned long eiem;
+	unsigned long iir;
+	unsigned long isr;
+	unsigned long ior;
+	unsigned long ipsw;
+	unsigned long __unused0;
+	unsigned long sr4;
+	unsigned long sr0;
+	unsigned long sr1;
+	unsigned long sr2;
+	unsigned long sr3;
+	unsigned long sr5;
+	unsigned long sr6;
+	unsigned long sr7;
+	unsigned long cr0;
+	unsigned long pid1;
+	unsigned long pid2;
+	unsigned long scrccr;
+	unsigned long pid3;
+	unsigned long pid4;
+	unsigned long cr24;
+	unsigned long cr25;
+	unsigned long cr26;
+	unsigned long cr27;
+	unsigned long cr28;
+	unsigned long cr29;
+	unsigned long cr30;
+
+	u64 fr[32];
+};
+
+#endif
+#endif

arch/parisc/include/asm/kprobes.h

+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * arch/parisc/include/asm/kprobes.h
+ *
+ * PA-RISC kprobes implementation
+ *
+ * Copyright (c) 2019 Sven Schnelle <svens@stackframe.org>
+ */
+
+#ifndef _PARISC_KPROBES_H
+#define _PARISC_KPROBES_H
+
+#ifdef CONFIG_KPROBES
+
+#include <asm-generic/kprobes.h>
+#include <linux/types.h>
+#include <linux/ptrace.h>
+#include <linux/notifier.h>
+
+#define PARISC_KPROBES_BREAK_INSN	0x3ff801f
+#define  __ARCH_WANT_KPROBES_INSN_SLOT
+#define MAX_INSN_SIZE 1
+
+typedef u32 kprobe_opcode_t;
+struct kprobe;
+
+void arch_remove_kprobe(struct kprobe *p);
+
+#define flush_insn_slot(p) \
+	flush_icache_range((unsigned long)&(p)->ainsn.insn[0], \
+			   (unsigned long)&(p)->ainsn.insn[0] + \
+			   sizeof(kprobe_opcode_t))
+
+#define kretprobe_blacklist_size    0
+
+struct arch_specific_insn {
+	kprobe_opcode_t *insn;
+};
+
+struct prev_kprobe {
+	struct kprobe *kp;
+	unsigned long status;
+};
+
+struct kprobe_ctlblk {
+	unsigned int kprobe_status;
+	struct prev_kprobe prev_kprobe;
+	unsigned long iaoq[2];
+};
+
+int __kprobes parisc_kprobe_break_handler(struct pt_regs *regs);
+int __kprobes parisc_kprobe_ss_handler(struct pt_regs *regs);
+
+#endif /* CONFIG_KPROBES */
+#endif /* _PARISC_KPROBES_H */

arch/parisc/include/asm/mmzone.h

@@ -2,62 +2,6 @@
 #ifndef _PARISC_MMZONE_H
 #define _PARISC_MMZONE_H
 
-#define MAX_PHYSMEM_RANGES 8 /* Fix the size for now (current known max is 3) */
+#define MAX_PHYSMEM_RANGES 4 /* Fix the size for now (current known max is 3) */
 
-#ifdef CONFIG_DISCONTIGMEM
-
-extern int npmem_ranges;
-
-struct node_map_data {
-    pg_data_t pg_data;
-};
-
-extern struct node_map_data node_data[];
-
-#define NODE_DATA(nid)          (&node_data[nid].pg_data)
-
-/* We have these possible memory map layouts:
- * Astro: 0-3.75, 67.75-68, 4-64
- * zx1: 0-1, 257-260, 4-256
- * Stretch (N-class): 0-2, 4-32, 34-xxx
- */
-
-/* Since each 1GB can only belong to one region (node), we can create
- * an index table for pfn to nid lookup; each entry in pfnnid_map 
- * represents 1GB, and contains the node that the memory belongs to. */
-
-#define PFNNID_SHIFT (30 - PAGE_SHIFT)
-#define PFNNID_MAP_MAX  512     /* support 512GB */
-extern signed char pfnnid_map[PFNNID_MAP_MAX];
-
-#ifndef CONFIG_64BIT
-#define pfn_is_io(pfn) ((pfn & (0xf0000000UL >> PAGE_SHIFT)) == (0xf0000000UL >> PAGE_SHIFT))
-#else
-/* io can be 0xf0f0f0f0f0xxxxxx or 0xfffffffff0000000 */
-#define pfn_is_io(pfn) ((pfn & (0xf000000000000000UL >> PAGE_SHIFT)) == (0xf000000000000000UL >> PAGE_SHIFT))
-#endif
-
-static inline int pfn_to_nid(unsigned long pfn)
-{
-	unsigned int i;
-
-	if (unlikely(pfn_is_io(pfn)))
-		return 0;
-
-	i = pfn >> PFNNID_SHIFT;
-	BUG_ON(i >= ARRAY_SIZE(pfnnid_map));
-
-	return pfnnid_map[i];
-}
-
-static inline int pfn_valid(int pfn)
-{
-	int nid = pfn_to_nid(pfn);
-
-	if (nid >= 0)
-		return (pfn < node_end_pfn(nid));
-	return 0;
-}
-
-#endif
 #endif /* _PARISC_MMZONE_H */

arch/parisc/include/asm/page.h

@@ -147,9 +147,9 @@ extern int npmem_ranges;
 #define __pa(x)			((unsigned long)(x)-PAGE_OFFSET)
 #define __va(x)			((void *)((unsigned long)(x)+PAGE_OFFSET))
 
-#ifndef CONFIG_DISCONTIGMEM
+#ifndef CONFIG_SPARSEMEM
 #define pfn_valid(pfn)		((pfn) < max_mapnr)
-#endif /* CONFIG_DISCONTIGMEM */
+#endif
 
 #ifdef CONFIG_HUGETLB_PAGE
 #define HPAGE_SHIFT		PMD_SHIFT /* fixed for transparent huge pages */

arch/parisc/include/asm/patch.h

+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef _PARISC_KERNEL_PATCH_H
+#define _PARISC_KERNEL_PATCH_H
+
+/* stop machine and patch kernel text */
+void patch_text(void *addr, unsigned int insn);
+
+/* patch kernel text with machine already stopped (e.g. in kgdb) */
+void __patch_text(void *addr, unsigned int insn);
+
+#endif

arch/parisc/include/asm/pgalloc.h

@@ -41,6 +41,7 @@ static inline pgd_t *pgd_alloc(struct mm_struct *mm)
 		__pgd_val_set(*pgd, PxD_FLAG_ATTACHED);
 #endif
 	}
+	spin_lock_init(pgd_spinlock(actual_pgd));
 	return actual_pgd;
 }
 

arch/parisc/include/asm/pgtable.h

@@ -17,7 +17,7 @@
 #include <asm/processor.h>
 #include <asm/cache.h>
 
-extern spinlock_t pa_tlb_lock;
+static inline spinlock_t *pgd_spinlock(pgd_t *);
 
 /*
  * kern_addr_valid(ADDR) tests if ADDR is pointing to valid kernel
@@ -34,16 +34,46 @@ extern spinlock_t pa_tlb_lock;
  */
 #define kern_addr_valid(addr)	(1)
 
-/* Purge data and instruction TLB entries.  Must be called holding
- * the pa_tlb_lock.  The TLB purge instructions are slow on SMP
- * machines since the purge must be broadcast to all CPUs.
+/* This is for the serialization of PxTLB broadcasts. At least on the N class
+ * systems, only one PxTLB inter processor broadcast can be active at any one
+ * time on the Merced bus.
+
+ * PTE updates are protected by locks in the PMD.
+ */
+extern spinlock_t pa_tlb_flush_lock;
+extern spinlock_t pa_swapper_pg_lock;
+#if defined(CONFIG_64BIT) && defined(CONFIG_SMP)
+extern int pa_serialize_tlb_flushes;
+#else
+#define pa_serialize_tlb_flushes        (0)
+#endif
+
+#define purge_tlb_start(flags)  do { \
+	if (pa_serialize_tlb_flushes)	\
+		spin_lock_irqsave(&pa_tlb_flush_lock, flags); \
+	else \
+		local_irq_save(flags);	\
+	} while (0)
+#define purge_tlb_end(flags)	do { \
+	if (pa_serialize_tlb_flushes)	\
+		spin_unlock_irqrestore(&pa_tlb_flush_lock, flags); \
+	else \
+		local_irq_restore(flags); \
+	} while (0)
+
+/* Purge data and instruction TLB entries. The TLB purge instructions
+ * are slow on SMP machines since the purge must be broadcast to all CPUs.
  */
 
 static inline void purge_tlb_entries(struct mm_struct *mm, unsigned long addr)
 {
+	unsigned long flags;
+
+	purge_tlb_start(flags);
 	mtsp(mm->context, 1);
 	pdtlb(addr);
 	pitlb(addr);
+	purge_tlb_end(flags);
 }
 
 /* Certain architectures need to do special things when PTEs
@@ -59,11 +89,11 @@ static inline void purge_tlb_entries(struct mm_struct *mm, unsigned long addr)
 	do {							\
 		pte_t old_pte;					\
 		unsigned long flags;				\
-		spin_lock_irqsave(&pa_tlb_lock, flags);		\
+		spin_lock_irqsave(pgd_spinlock((mm)->pgd), flags);\
 		old_pte = *ptep;				\
 		set_pte(ptep, pteval);				\
 		purge_tlb_entries(mm, addr);			\
-		spin_unlock_irqrestore(&pa_tlb_lock, flags);	\
+		spin_unlock_irqrestore(pgd_spinlock((mm)->pgd), flags);\
 	} while (0)
 
 #endif /* !__ASSEMBLY__ */
@@ -88,10 +118,10 @@ static inline void purge_tlb_entries(struct mm_struct *mm, unsigned long addr)
 #if CONFIG_PGTABLE_LEVELS == 3
 #define PGD_ORDER	1 /* Number of pages per pgd */
 #define PMD_ORDER	1 /* Number of pages per pmd */
-#define PGD_ALLOC_ORDER	2 /* first pgd contains pmd */
+#define PGD_ALLOC_ORDER	(2 + 1) /* first pgd contains pmd */
 #else
 #define PGD_ORDER	1 /* Number of pages per pgd */
-#define PGD_ALLOC_ORDER	PGD_ORDER
+#define PGD_ALLOC_ORDER	(PGD_ORDER + 1)
 #endif
 
 /* Definitions for 3rd level (we use PLD here for Page Lower directory
@@ -459,6 +489,15 @@ extern void update_mmu_cache(struct vm_area_struct *, unsigned long, pte_t *);
 #define __pte_to_swp_entry(pte)		((swp_entry_t) { pte_val(pte) })
 #define __swp_entry_to_pte(x)		((pte_t) { (x).val })
 
+
+static inline spinlock_t *pgd_spinlock(pgd_t *pgd)
+{
+	if (unlikely(pgd == swapper_pg_dir))
+		return &pa_swapper_pg_lock;
+	return (spinlock_t *)((char *)pgd + (PAGE_SIZE << (PGD_ALLOC_ORDER - 1)));
+}
+
+
 static inline int ptep_test_and_clear_young(struct vm_area_struct *vma, unsigned long addr, pte_t *ptep)
 {
 	pte_t pte;
@@ -467,15 +506,15 @@ static inline int ptep_test_and_clear_young(struct vm_area_struct *vma, unsigned
 	if (!pte_young(*ptep))
 		return 0;
 
-	spin_lock_irqsave(&pa_tlb_lock, flags);
+	spin_lock_irqsave(pgd_spinlock(vma->vm_mm->pgd), flags);
 	pte = *ptep;
 	if (!pte_young(pte)) {
-		spin_unlock_irqrestore(&pa_tlb_lock, flags);
+		spin_unlock_irqrestore(pgd_spinlock(vma->vm_mm->pgd), flags);
 		return 0;
 	}
 	set_pte(ptep, pte_mkold(pte));
 	purge_tlb_entries(vma->vm_mm, addr);
-	spin_unlock_irqrestore(&pa_tlb_lock, flags);
+	spin_unlock_irqrestore(pgd_spinlock(vma->vm_mm->pgd), flags);
 	return 1;
 }
 
@@ -485,11 +524,11 @@ static inline pte_t ptep_get_and_clear(struct mm_struct *mm, unsigned long addr,
 	pte_t old_pte;
 	unsigned long flags;
 
-	spin_lock_irqsave(&pa_tlb_lock, flags);
+	spin_lock_irqsave(pgd_spinlock(mm->pgd), flags);
 	old_pte = *ptep;
 	set_pte(ptep, __pte(0));
 	purge_tlb_entries(mm, addr);
-	spin_unlock_irqrestore(&pa_tlb_lock, flags);
+	spin_unlock_irqrestore(pgd_spinlock(mm->pgd), flags);
 
 	return old_pte;
 }
@@ -497,10 +536,10 @@ static inline pte_t ptep_get_and_clear(struct mm_struct *mm, unsigned long addr,
 static inline void ptep_set_wrprotect(struct mm_struct *mm, unsigned long addr, pte_t *ptep)
 {
 	unsigned long flags;
-	spin_lock_irqsave(&pa_tlb_lock, flags);
+	spin_lock_irqsave(pgd_spinlock(mm->pgd), flags);
 	set_pte(ptep, pte_wrprotect(*ptep));
 	purge_tlb_entries(mm, addr);
-	spin_unlock_irqrestore(&pa_tlb_lock, flags);
+	spin_unlock_irqrestore(pgd_spinlock(mm->pgd), flags);
 }
 
 #define pte_same(A,B)	(pte_val(A) == pte_val(B))