staging: zsmalloc: add mapping modes

This patch improves mapping performance in zsmalloc by getting
usage information from the user in the form of a "mapping mode"
and using it to avoid unnecessary copying for objects that span
pages.

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

drivers/staging/zcache/zcache-main.c

@@ -716,7 +716,7 @@ static unsigned long zv_create(struct zs_pool *pool, uint32_t pool_id,
 		goto out;
 	atomic_inc(&zv_curr_dist_counts[chunks]);
 	atomic_inc(&zv_cumul_dist_counts[chunks]);
-	zv = zs_map_object(pool, handle);
+	zv = zs_map_object(pool, handle, ZS_MM_WO);
 	zv->index = index;
 	zv->oid = *oid;
 	zv->pool_id = pool_id;
@@ -735,7 +735,7 @@ static void zv_free(struct zs_pool *pool, unsigned long handle)
 	uint16_t size;
 	int chunks;
 
-	zv = zs_map_object(pool, handle);
+	zv = zs_map_object(pool, handle, ZS_MM_RW);
 	ASSERT_SENTINEL(zv, ZVH);
 	size = zv->size + sizeof(struct zv_hdr);
 	INVERT_SENTINEL(zv, ZVH);
@@ -757,7 +757,7 @@ static void zv_decompress(struct page *page, unsigned long handle)
 	int ret;
 	struct zv_hdr *zv;
 
-	zv = zs_map_object(zcache_host.zspool, handle);
+	zv = zs_map_object(zcache_host.zspool, handle, ZS_MM_RO);
 	BUG_ON(zv->size == 0);
 	ASSERT_SENTINEL(zv, ZVH);
 	to_va = kmap_atomic(page);

drivers/staging/zram/zram_drv.c

@@ -220,7 +220,8 @@ static int zram_bvec_read(struct zram *zram, struct bio_vec *bvec,
 		uncmem = user_mem;
 	clen = PAGE_SIZE;
 
-	cmem = zs_map_object(zram->mem_pool, zram->table[index].handle);
+	cmem = zs_map_object(zram->mem_pool, zram->table[index].handle,
+				ZS_MM_RO);
 
 	ret = lzo1x_decompress_safe(cmem, zram->table[index].size,
 				    uncmem, &clen);
@@ -258,7 +259,7 @@ static int zram_read_before_write(struct zram *zram, char *mem, u32 index)
 		return 0;
 	}
 
-	cmem = zs_map_object(zram->mem_pool, handle);
+	cmem = zs_map_object(zram->mem_pool, handle, ZS_MM_RO);
 	ret = lzo1x_decompress_safe(cmem, zram->table[index].size,
 				    mem, &clen);
 	zs_unmap_object(zram->mem_pool, handle);
@@ -351,7 +352,7 @@ static int zram_bvec_write(struct zram *zram, struct bio_vec *bvec, u32 index,
 		ret = -ENOMEM;
 		goto out;
 	}
-	cmem = zs_map_object(zram->mem_pool, handle);
+	cmem = zs_map_object(zram->mem_pool, handle, ZS_MM_WO);
 
 	memcpy(cmem, src, clen);
 

drivers/staging/zsmalloc/zsmalloc-main.c

@@ -484,9 +484,6 @@ static void zs_copy_map_object(char *buf, struct page *firstpage,
 	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]);
@@ -517,9 +514,6 @@ static void zs_copy_unmap_object(char *buf, struct page *firstpage,
 	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,
@@ -754,7 +748,8 @@ EXPORT_SYMBOL_GPL(zs_free);
  *
  * This function returns with preemption and page faults disabled.
 */
-void *zs_map_object(struct zs_pool *pool, unsigned long handle)
+void *zs_map_object(struct zs_pool *pool, unsigned long handle,
+			enum zs_mapmode mm)
 {
 	struct page *page;
 	unsigned long obj_idx, off;
@@ -778,7 +773,11 @@ void *zs_map_object(struct zs_pool *pool, unsigned long handle)
 		return area->vm_addr + off;
 	}
 
-	zs_copy_map_object(area->vm_buf, page, off, class->size);
+	/* disable page faults to match kmap_atomic() return conditions */
+	pagefault_disable();
+
+	if (mm != ZS_MM_WO)
+		zs_copy_map_object(area->vm_buf, page, off, class->size);
 	area->vm_addr = NULL;
 	return area->vm_buf;
 }
@@ -795,13 +794,16 @@ void zs_unmap_object(struct zs_pool *pool, unsigned long handle)
 	struct mapping_area *area;
 
 	area = &__get_cpu_var(zs_map_area);
+	/* single-page object fastpath */
 	if (area->vm_addr) {
-		/* single-page object fastpath */
 		kunmap_atomic(area->vm_addr);
-		put_cpu_var(zs_map_area);
+		goto out;
-		return;
 	}
 
+	/* no write fastpath */
+	if (area->vm_mm == ZS_MM_RO)
+		goto pfenable;
+
 	BUG_ON(!handle);
 
 	obj_handle_to_location(handle, &page, &obj_idx);
@@ -810,6 +812,11 @@ void zs_unmap_object(struct zs_pool *pool, unsigned long handle)
 	off = obj_idx_to_offset(page, obj_idx, class->size);
 
 	zs_copy_unmap_object(area->vm_buf, page, off, class->size);
+
+pfenable:
+	/* enable page faults to match kunmap_atomic() return conditions */
+	pagefault_enable();
+out:
 	put_cpu_var(zs_map_area);
 }
 EXPORT_SYMBOL_GPL(zs_unmap_object);

drivers/staging/zsmalloc/zsmalloc.h

@@ -15,6 +15,17 @@
 
 #include <linux/types.h>
 
+/*
+ * zsmalloc mapping modes
+ *
+ * NOTE: These only make a difference when a mapped object spans pages
+*/
+enum zs_mapmode {
+	ZS_MM_RW, /* normal read-write mapping */
+	ZS_MM_RO, /* read-only (no copy-out at unmap time) */
+	ZS_MM_WO /* write-only (no copy-in at map time) */
+};
+
 struct zs_pool;
 
 struct zs_pool *zs_create_pool(const char *name, gfp_t flags);
@@ -23,7 +34,8 @@ void zs_destroy_pool(struct zs_pool *pool);
 unsigned long zs_malloc(struct zs_pool *pool, size_t size);
 void zs_free(struct zs_pool *pool, unsigned long obj);
 
-void *zs_map_object(struct zs_pool *pool, unsigned long handle);
+void *zs_map_object(struct zs_pool *pool, unsigned long handle,
+			enum zs_mapmode mm);
 void zs_unmap_object(struct zs_pool *pool, unsigned long handle);
 
 u64 zs_get_total_size_bytes(struct zs_pool *pool);

drivers/staging/zsmalloc/zsmalloc_int.h

@@ -112,6 +112,7 @@ static const int fullness_threshold_frac = 4;
 struct mapping_area {
 	char *vm_buf; /* copy buffer for objects that span pages */
 	char *vm_addr; /* address of kmap_atomic()'ed pages */
+	enum zs_mapmode vm_mm; /* mapping mode */
 };
 
 struct size_class {