Merge http://test1.bkbits.net/system-2.5

into penguin.transmeta.com:/home/penguin/torvalds/repositories/kernel/linux

Documentation/cachetlb.txt

@@ -49,6 +49,9 @@ 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)
 
@@ -69,6 +72,9 @@ 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 page)
 
 	This time we need to remove the PAGE_SIZE sized translation
@@ -87,6 +93,9 @@ 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/vm/locking

@@ -77,6 +77,13 @@ The page_table_lock is grabbed while holding the kernel_lock spinning monitor.
 
 The page_table_lock is a spin lock.
 
+Note: PTL can also be used to guarantee that no new clones using the
+mm start up ... this is a loose form of stability on mm_users. For
+example, it is used in copy_mm to protect against a racing tlb_gather_mmu
+single address space optimization, so that the zap_page_range (from
+vmtruncate) does not loose sending ipi's to cloned threads that might 
+be spawned underneath it and go to user mode to drag in pte's into tlbs.
+
 swap_list_lock/swap_device_lock
 -------------------------------
 The swap devices are chained in priority order from the "swap_list" header. 

kernel/fork.c

@@ -367,6 +367,13 @@ static int copy_mm(unsigned long clone_flags, struct task_struct * tsk)
 	if (clone_flags & CLONE_VM) {
 		atomic_inc(&oldmm->mm_users);
 		mm = oldmm;
+		/*
+		 * There are cases where the PTL is held to ensure no
+		 * new threads start up in user mode using an mm, which
+		 * allows optimizing out ipis; the tlb_gather_mmu code
+		 * is an example.
+		 */
+		spin_unlock_wait(&oldmm->page_table_lock);
 		goto good_mm;
 	}
 

mm/memory.c

@@ -839,8 +839,8 @@ int zeromap_page_range(struct vm_area_struct *vma, unsigned long address, unsign
 		address = (address + PGDIR_SIZE) & PGDIR_MASK;
 		dir++;
 	} while (address && (address < end));
-	spin_unlock(&mm->page_table_lock);
 	flush_tlb_range(vma, beg, end);
+	spin_unlock(&mm->page_table_lock);
 	return error;
 }
 
@@ -922,8 +922,8 @@ int remap_page_range(struct vm_area_struct *vma, unsigned long from, unsigned lo
 		from = (from + PGDIR_SIZE) & PGDIR_MASK;
 		dir++;
 	} while (from && (from < end));
-	spin_unlock(&mm->page_table_lock);
 	flush_tlb_range(vma, beg, end);
+	spin_unlock(&mm->page_table_lock);
 	return error;
 }
 

mm/mprotect.c

@@ -92,8 +92,8 @@ static void change_protection(struct vm_area_struct *vma, unsigned long start, u
 		start = (start + PGDIR_SIZE) & PGDIR_MASK;
 		dir++;
 	} while (start && (start < end));
-	spin_unlock(&current->mm->page_table_lock);
 	flush_tlb_range(vma, beg, end);
+	spin_unlock(&current->mm->page_table_lock);
 	return;
 }
 

mm/mremap.c

@@ -81,8 +81,9 @@ static inline int copy_one_pte(struct mm_struct *mm, pte_t * src, pte_t * dst)
 	return error;
 }
 
-static int move_one_page(struct mm_struct *mm, unsigned long old_addr, unsigned long new_addr)
+static int move_one_page(struct vm_area_struct *vma, unsigned long old_addr, unsigned long new_addr)
 {
+	struct mm_struct *mm = vma->vm_mm;
 	int error = 0;
 	pte_t *src, *dst;
 
@@ -94,6 +95,7 @@ static int move_one_page(struct mm_struct *mm, unsigned long old_addr, unsigned
 		pte_unmap_nested(src);
 		pte_unmap(dst);
 	}
+	flush_tlb_page(vma, old_addr);
 	spin_unlock(&mm->page_table_lock);
 	return error;
 }
@@ -113,10 +115,9 @@ static int move_page_tables(struct vm_area_struct *vma,
 	 */
 	while (offset) {
 		offset -= PAGE_SIZE;
-		if (move_one_page(mm, old_addr + offset, new_addr + offset))
+		if (move_one_page(vma, old_addr + offset, new_addr + offset))
 			goto oops_we_failed;
 	}
-	flush_tlb_range(vma, old_addr, old_addr + len);
 	return 0;
 
 	/*
@@ -129,7 +130,7 @@ static int move_page_tables(struct vm_area_struct *vma,
 oops_we_failed:
 	flush_cache_range(vma, new_addr, new_addr + len);
 	while ((offset += PAGE_SIZE) < len)
-		move_one_page(mm, new_addr + offset, old_addr + offset);
+		move_one_page(vma, new_addr + offset, old_addr + offset);
 	zap_page_range(vma, new_addr, len);
 	return -1;
 }