Merge branch 'master'

Documentation/cachetlb.txt

@@ -49,9 +49,6 @@ changes occur:
 	page table operations such as what happens during
 	fork, and exec.
 
-	Platform developers note that generic code will always
-	invoke this interface without mm->page_table_lock held.
-
 3) void flush_tlb_range(struct vm_area_struct *vma,
 			unsigned long start, unsigned long end)
 
@@ -72,9 +69,6 @@ changes occur:
 	call flush_tlb_page (see below) for each entry which may be
 	modified.
 
-	Platform developers note that generic code will always
-	invoke this interface with mm->page_table_lock held.
-
 4) void flush_tlb_page(struct vm_area_struct *vma, unsigned long addr)
 
 	This time we need to remove the PAGE_SIZE sized translation
@@ -93,9 +87,6 @@ changes occur:
 
 	This is used primarily during fault processing.
 
-	Platform developers note that generic code will always
-	invoke this interface with mm->page_table_lock held.
-
 5) void flush_tlb_pgtables(struct mm_struct *mm,
 			   unsigned long start, unsigned long end)
 

Documentation/kernel-parameters.txt

@@ -1460,8 +1460,6 @@ running once the system is up.
 	stifb=		[HW]
 			Format: bpp:<bpp1>[:<bpp2>[:<bpp3>...]]
 
-	stram_swap=	[HW,M68k]
-
 	swiotlb=	[IA-64] Number of I/O TLB slabs
 
 	switches=	[HW,M68k]

Documentation/m68k/kernel-options.txt

@@ -626,7 +626,7 @@ ignored (others aren't affected).
     can be performed in optimal order. Not all SCSI devices support
     tagged queuing (:-().
 
-4.6 switches=
+4.5 switches=
 -------------
 
 Syntax: switches=<list of switches>
@@ -661,28 +661,6 @@ correctly.
 earlier initialization ("ov_"-less) takes precedence. But the
 switching-off on reset still happens in this case.
 
-4.5) stram_swap=
-----------------
-
-Syntax: stram_swap=<do_swap>[,<max_swap>]
-
-  This option is available only if the kernel has been compiled with
-CONFIG_STRAM_SWAP enabled. Normally, the kernel then determines
-dynamically whether to actually use ST-RAM as swap space. (Currently,
-the fraction of ST-RAM must be less or equal 1/3 of total memory to
-enable this swapping.) You can override the kernel's decision by
-specifying this option. 1 for <do_swap> means always enable the swap,
-even if you have less alternate RAM. 0 stands for never swap to
-ST-RAM, even if it's small enough compared to the rest of memory.
-
-  If ST-RAM swapping is enabled, the kernel usually uses all free
-ST-RAM as swap "device". If the kernel resides in ST-RAM, the region
-allocated by it is obviously never used for swapping :-) You can also
-limit this amount by specifying the second parameter, <max_swap>, if
-you want to use parts of ST-RAM as normal system memory. <max_swap> is
-in kBytes and the number should be a multiple of 4 (otherwise: rounded
-down).
-
 5) Options for Amiga Only:
 ==========================
 

arch/alpha/mm/numa.c

@@ -371,6 +371,8 @@ show_mem(void)
 	show_free_areas();
 	printk("Free swap:       %6ldkB\n", nr_swap_pages<<(PAGE_SHIFT-10));
 	for_each_online_node(nid) {
+		unsigned long flags;
+		pgdat_resize_lock(NODE_DATA(nid), &flags);
 		i = node_spanned_pages(nid);
 		while (i-- > 0) {
 			struct page *page = nid_page_nr(nid, i);
@@ -384,6 +386,7 @@ show_mem(void)
 			else
 				shared += page_count(page) - 1;
 		}
+		pgdat_resize_unlock(NODE_DATA(nid), &flags);
 	}
 	printk("%ld pages of RAM\n",total);
 	printk("%ld free pages\n",free);

arch/alpha/mm/remap.c

@@ -2,7 +2,6 @@
 #include <asm/pgalloc.h>
 #include <asm/cacheflush.h>
 
-/* called with the page_table_lock held */
 static inline void 
 remap_area_pte(pte_t * pte, unsigned long address, unsigned long size, 
 	       unsigned long phys_addr, unsigned long flags)
@@ -31,7 +30,6 @@ remap_area_pte(pte_t * pte, unsigned long address, unsigned long size,
 	} while (address && (address < end));
 }
 
-/* called with the page_table_lock held */
 static inline int 
 remap_area_pmd(pmd_t * pmd, unsigned long address, unsigned long size, 
 	       unsigned long phys_addr, unsigned long flags)
@@ -46,7 +44,7 @@ remap_area_pmd(pmd_t * pmd, unsigned long address, unsigned long size,
 	if (address >= end)
 		BUG();
 	do {
-		pte_t * pte = pte_alloc_kernel(&init_mm, pmd, address);
+		pte_t * pte = pte_alloc_kernel(pmd, address);
 		if (!pte)
 			return -ENOMEM;
 		remap_area_pte(pte, address, end - address, 
@@ -70,7 +68,6 @@ __alpha_remap_area_pages(unsigned long address, unsigned long phys_addr,
 	flush_cache_all();
 	if (address >= end)
 		BUG();
-	spin_lock(&init_mm.page_table_lock);
 	do {
 		pmd_t *pmd;
 		pmd = pmd_alloc(&init_mm, dir, address);
@@ -84,7 +81,6 @@ __alpha_remap_area_pages(unsigned long address, unsigned long phys_addr,
 		address = (address + PGDIR_SIZE) & PGDIR_MASK;
 		dir++;
 	} while (address && (address < end));
-	spin_unlock(&init_mm.page_table_lock);
 	return error;
 }
 

arch/arm/kernel/signal.c

@@ -139,93 +139,33 @@ struct iwmmxt_sigframe {
 	unsigned long	storage[0x98/4];
 };
 
-static int page_present(struct mm_struct *mm, void __user *uptr, int wr)
-{
-	unsigned long addr = (unsigned long)uptr;
-	pgd_t *pgd = pgd_offset(mm, addr);
-	if (pgd_present(*pgd)) {
-		pmd_t *pmd = pmd_offset(pgd, addr);
-		if (pmd_present(*pmd)) {
-			pte_t *pte = pte_offset_map(pmd, addr);
-			return (pte_present(*pte) && (!wr || pte_write(*pte)));
-		}
-	}
-	return 0;
-}
-
-static int copy_locked(void __user *uptr, void *kptr, size_t size, int write,
-		       void (*copyfn)(void *, void __user *))
-{
-	unsigned char v, __user *userptr = uptr;
-	int err = 0;
-
-	do {
-		struct mm_struct *mm;
-
-		if (write) {
-			__put_user_error(0, userptr, err);
-			__put_user_error(0, userptr + size - 1, err);
-		} else {
-			__get_user_error(v, userptr, err);
-			__get_user_error(v, userptr + size - 1, err);
-		}
-
-		if (err)
-			break;
-
-		mm = current->mm;
-		spin_lock(&mm->page_table_lock);
-		if (page_present(mm, userptr, write) &&
-		    page_present(mm, userptr + size - 1, write)) {
-		    	copyfn(kptr, uptr);
-		} else
-			err = 1;
-		spin_unlock(&mm->page_table_lock);
-	} while (err);
-
-	return err;
-}
-
 static int preserve_iwmmxt_context(struct iwmmxt_sigframe *frame)
 {
-	int err = 0;
+	char kbuf[sizeof(*frame) + 8];
+	struct iwmmxt_sigframe *kframe;
 
 	/* the iWMMXt context must be 64 bit aligned */
-	WARN_ON((unsigned long)frame & 7);
-
-	__put_user_error(IWMMXT_MAGIC0, &frame->magic0, err);
-	__put_user_error(IWMMXT_MAGIC1, &frame->magic1, err);
-
-	/*
-	 * iwmmxt_task_copy() doesn't check user permissions.
-	 * Let's do a dummy write on the upper boundary to ensure
-	 * access to user mem is OK all way up.
-	 */
-	err |= copy_locked(&frame->storage, current_thread_info(),
-			   sizeof(frame->storage), 1, iwmmxt_task_copy);
-	return err;
+	kframe = (struct iwmmxt_sigframe *)((unsigned long)(kbuf + 8) & ~7);
+	kframe->magic0 = IWMMXT_MAGIC0;
+	kframe->magic1 = IWMMXT_MAGIC1;
+	iwmmxt_task_copy(current_thread_info(), &kframe->storage);
+	return __copy_to_user(frame, kframe, sizeof(*frame));
 }
 
 static int restore_iwmmxt_context(struct iwmmxt_sigframe *frame)
 {
-	unsigned long magic0, magic1;
-	int err = 0;
+	char kbuf[sizeof(*frame) + 8];
+	struct iwmmxt_sigframe *kframe;
 
-	/* the iWMMXt context is 64 bit aligned */
-	WARN_ON((unsigned long)frame & 7);
-
-	/*
-	 * Validate iWMMXt context signature.
-	 * Also, iwmmxt_task_restore() doesn't check user permissions.
-	 * Let's do a dummy write on the upper boundary to ensure
-	 * access to user mem is OK all way up.
-	 */
-	__get_user_error(magic0, &frame->magic0, err);
-	__get_user_error(magic1, &frame->magic1, err);
-	if (!err && magic0 == IWMMXT_MAGIC0 && magic1 == IWMMXT_MAGIC1)
-		err = copy_locked(&frame->storage, current_thread_info(),
-				  sizeof(frame->storage), 0, iwmmxt_task_restore);
-	return err;
+	/* the iWMMXt context must be 64 bit aligned */
+	kframe = (struct iwmmxt_sigframe *)((unsigned long)(kbuf + 8) & ~7);
+	if (__copy_from_user(kframe, frame, sizeof(*frame)))
+		return -1;
+	if (kframe->magic0 != IWMMXT_MAGIC0 ||
+	    kframe->magic1 != IWMMXT_MAGIC1)
+		return -1;
+	iwmmxt_task_restore(current_thread_info(), &kframe->storage);
+	return 0;
 }
 
 #endif

arch/arm/kernel/traps.c

@@ -483,29 +483,33 @@ asmlinkage int arm_syscall(int no, struct pt_regs *regs)
 		unsigned long addr = regs->ARM_r2;
 		struct mm_struct *mm = current->mm;
 		pgd_t *pgd; pmd_t *pmd; pte_t *pte;
+		spinlock_t *ptl;
 
 		regs->ARM_cpsr &= ~PSR_C_BIT;
-		spin_lock(&mm->page_table_lock);
+		down_read(&mm->mmap_sem);
 		pgd = pgd_offset(mm, addr);
 		if (!pgd_present(*pgd))
 			goto bad_access;
 		pmd = pmd_offset(pgd, addr);
 		if (!pmd_present(*pmd))
 			goto bad_access;
-		pte = pte_offset_map(pmd, addr);
-		if (!pte_present(*pte) || !pte_write(*pte))
+		pte = pte_offset_map_lock(mm, pmd, addr, &ptl);
+		if (!pte_present(*pte) || !pte_write(*pte)) {
+			pte_unmap_unlock(pte, ptl);
 			goto bad_access;
+		}
 		val = *(unsigned long *)addr;
 		val -= regs->ARM_r0;
 		if (val == 0) {
 			*(unsigned long *)addr = regs->ARM_r1;
 			regs->ARM_cpsr |= PSR_C_BIT;
 		}
-		spin_unlock(&mm->page_table_lock);
+		pte_unmap_unlock(pte, ptl);
+		up_read(&mm->mmap_sem);
 		return val;
 
 		bad_access:
-		spin_unlock(&mm->page_table_lock);
+		up_read(&mm->mmap_sem);
 		/* simulate a write access fault */
 		do_DataAbort(addr, 15 + (1 << 11), regs);
 		return -1;

arch/arm/mm/consistent.c

@@ -397,8 +397,6 @@ static int __init consistent_init(void)
 	pte_t *pte;
 	int ret = 0;
 
-	spin_lock(&init_mm.page_table_lock);
-
 	do {
 		pgd = pgd_offset(&init_mm, CONSISTENT_BASE);
 		pmd = pmd_alloc(&init_mm, pgd, CONSISTENT_BASE);
@@ -409,7 +407,7 @@ static int __init consistent_init(void)
 		}
 		WARN_ON(!pmd_none(*pmd));
 
-		pte = pte_alloc_kernel(&init_mm, pmd, CONSISTENT_BASE);
+		pte = pte_alloc_kernel(pmd, CONSISTENT_BASE);
 		if (!pte) {
 			printk(KERN_ERR "%s: no pte tables\n", __func__);
 			ret = -ENOMEM;
@@ -419,8 +417,6 @@ static int __init consistent_init(void)
 		consistent_pte = pte;
 	} while (0);
 
-	spin_unlock(&init_mm.page_table_lock);
-
 	return ret;
 }
 

arch/arm/mm/fault-armv.c

@@ -26,6 +26,11 @@ static unsigned long shared_pte_mask = L_PTE_CACHEABLE;
 /*
  * We take the easy way out of this problem - we make the
  * PTE uncacheable.  However, we leave the write buffer on.
+ *
+ * Note that the pte lock held when calling update_mmu_cache must also
+ * guard the pte (somewhere else in the same mm) that we modify here.
+ * Therefore those configurations which might call adjust_pte (those
+ * without CONFIG_CPU_CACHE_VIPT) cannot support split page_table_lock.
  */
 static int adjust_pte(struct vm_area_struct *vma, unsigned long address)
 {
@@ -127,7 +132,7 @@ void __flush_dcache_page(struct address_space *mapping, struct page *page);
  *  2. If we have multiple shared mappings of the same space in
  *     an object, we need to deal with the cache aliasing issues.
  *
- * Note that the page_table_lock will be held.
+ * Note that the pte lock will be held.
  */
 void update_mmu_cache(struct vm_area_struct *vma, unsigned long addr, pte_t pte)
 {

arch/arm/mm/ioremap.c

@@ -75,7 +75,7 @@ remap_area_pmd(pmd_t * pmd, unsigned long address, unsigned long size,
 
 	pgprot = __pgprot(L_PTE_PRESENT | L_PTE_YOUNG | L_PTE_DIRTY | L_PTE_WRITE | flags);
 	do {
-		pte_t * pte = pte_alloc_kernel(&init_mm, pmd, address);
+		pte_t * pte = pte_alloc_kernel(pmd, address);
 		if (!pte)
 			return -ENOMEM;
 		remap_area_pte(pte, address, end - address, address + phys_addr, pgprot);
@@ -97,7 +97,6 @@ remap_area_pages(unsigned long start, unsigned long phys_addr,
 	phys_addr -= address;
 	dir = pgd_offset(&init_mm, address);
 	BUG_ON(address >= end);
-	spin_lock(&init_mm.page_table_lock);
 	do {
 		pmd_t *pmd = pmd_alloc(&init_mm, dir, address);
 		if (!pmd) {
@@ -114,7 +113,6 @@ remap_area_pages(unsigned long start, unsigned long phys_addr,
 		dir++;
 	} while (address && (address < end));
 
-	spin_unlock(&init_mm.page_table_lock);
 	flush_cache_vmap(start, end);
 	return err;
 }