staging: zsmalloc: remove x86 dependency

This patch replaces the page table assisted object mapping
method, which has x86 dependencies, with a arch-independent
method that does a simple copy into a temporary per-cpu
buffer.

While a copy seems like it would be worse than mapping the pages,
tests demonstrate the copying is always faster and, in the case of
running inside a KVM guest, roughly 4x faster.

Signed-off-by: Seth Jennings <sjenning@linux.vnet.ibm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

drivers/staging/zsmalloc/Kconfig

@@ -1,9 +1,5 @@
 config ZSMALLOC
 	tristate "Memory allocator for compressed pages"
-	# X86 dependency is because of the use of __flush_tlb_one and set_pte
-	# in zsmalloc-main.c.
-	# TODO: convert these to portable functions
-	depends on X86
 	default n
 	help
 	  zsmalloc is a slab-based memory allocator designed to store

drivers/staging/zsmalloc/zsmalloc-main.c

@@ -470,6 +470,57 @@ static struct page *find_get_zspage(struct size_class *class)
 	return page;
 }
 
+static void zs_copy_map_object(char *buf, struct page *firstpage,
+				int off, int size)
+{
+	struct page *pages[2];
+	int sizes[2];
+	void *addr;
+
+	pages[0] = firstpage;
+	pages[1] = get_next_page(firstpage);
+	BUG_ON(!pages[1]);
+
+	sizes[0] = PAGE_SIZE - off;
+	sizes[1] = size - sizes[0];
+
+	/* disable page faults to match kmap_atomic() return conditions */
+	pagefault_disable();
+
+	/* copy object to per-cpu buffer */
+	addr = kmap_atomic(pages[0]);
+	memcpy(buf, addr + off, sizes[0]);
+	kunmap_atomic(addr);
+	addr = kmap_atomic(pages[1]);
+	memcpy(buf + sizes[0], addr, sizes[1]);
+	kunmap_atomic(addr);
+}
+
+static void zs_copy_unmap_object(char *buf, struct page *firstpage,
+				int off, int size)
+{
+	struct page *pages[2];
+	int sizes[2];
+	void *addr;
+
+	pages[0] = firstpage;
+	pages[1] = get_next_page(firstpage);
+	BUG_ON(!pages[1]);
+
+	sizes[0] = PAGE_SIZE - off;
+	sizes[1] = size - sizes[0];
+
+	/* copy per-cpu buffer to object */
+	addr = kmap_atomic(pages[0]);
+	memcpy(addr + off, buf, sizes[0]);
+	kunmap_atomic(addr);
+	addr = kmap_atomic(pages[1]);
+	memcpy(addr, buf + sizes[0], sizes[1]);
+	kunmap_atomic(addr);
+
+	/* enable page faults to match kunmap_atomic() return conditions */
+	pagefault_enable();
+}
 
 static int zs_cpu_notifier(struct notifier_block *nb, unsigned long action,
 				void *pcpu)
@@ -480,18 +531,23 @@ static int zs_cpu_notifier(struct notifier_block *nb, unsigned long action,
 	switch (action) {
 	case CPU_UP_PREPARE:
 		area = &per_cpu(zs_map_area, cpu);
-		if (area->vm)
+		/*
-			break;
+		 * Make sure we don't leak memory if a cpu UP notification
-		area->vm = alloc_vm_area(2 * PAGE_SIZE, area->vm_ptes);
+		 * and zs_init() race and both call zs_cpu_up() on the same cpu
-		if (!area->vm)
+		 */
-			return notifier_from_errno(-ENOMEM);
+		if (area->vm_buf)
+			return 0;
+		area->vm_buf = (char *)__get_free_page(GFP_KERNEL);
+		if (!area->vm_buf)
+			return -ENOMEM;
+		return 0;
 		break;
 	case CPU_DEAD:
 	case CPU_UP_CANCELED:
 		area = &per_cpu(zs_map_area, cpu);
-		if (area->vm)
+		if (area->vm_buf)
-			free_vm_area(area->vm);
+			free_page((unsigned long)area->vm_buf);
-		area->vm = NULL;
+		area->vm_buf = NULL;
 		break;
 	}
 
@@ -714,22 +770,11 @@ void *zs_map_object(struct zs_pool *pool, unsigned long handle)
 	if (off + class->size <= PAGE_SIZE) {
 		/* this object is contained entirely within a page */
 		area->vm_addr = kmap_atomic(page);
-	} else {
+		return area->vm_addr + off;
-		/* this object spans two pages */
-		struct page *nextp;
-
-		nextp = get_next_page(page);
-		BUG_ON(!nextp);
-
-
-		set_pte(area->vm_ptes[0], mk_pte(page, PAGE_KERNEL));
-		set_pte(area->vm_ptes[1], mk_pte(nextp, PAGE_KERNEL));
-
-		/* We pre-allocated VM area so mapping can never fail */
-		area->vm_addr = area->vm->addr;
 	}
 
-	return area->vm_addr + off;
+	zs_copy_map_object(area->vm_buf, page, off, class->size);
+	return area->vm_buf;
 }
 EXPORT_SYMBOL_GPL(zs_map_object);
 
@@ -751,14 +796,10 @@ void zs_unmap_object(struct zs_pool *pool, unsigned long handle)
 	off = obj_idx_to_offset(page, obj_idx, class->size);
 
 	area = &__get_cpu_var(zs_map_area);
-	if (off + class->size <= PAGE_SIZE) {
+	if (off + class->size <= PAGE_SIZE)
 		kunmap_atomic(area->vm_addr);
-	} else {
+	else
-		set_pte(area->vm_ptes[0], __pte(0));
+		zs_copy_unmap_object(area->vm_buf, page, off, class->size);
-		set_pte(area->vm_ptes[1], __pte(0));
-		__flush_tlb_one((unsigned long)area->vm_addr);
-		__flush_tlb_one((unsigned long)area->vm_addr + PAGE_SIZE);
-	}
 	put_cpu_var(zs_map_area);
 }
 EXPORT_SYMBOL_GPL(zs_unmap_object);

drivers/staging/zsmalloc/zsmalloc_int.h

@@ -110,9 +110,8 @@ enum fullness_group {
 static const int fullness_threshold_frac = 4;
 
 struct mapping_area {
-	struct vm_struct *vm;
+	char *vm_buf; /* copy buffer for objects that span pages */
-	pte_t *vm_ptes[2];
+	char *vm_addr; /* address of kmap_atomic()'ed pages */
-	char *vm_addr;
 };
 
 struct size_class {