[PATCH v3 3/4 -fix] zram: make deduplication feature optional

From: Joonsoo Kim
Date: Thu Apr 20 2017 - 20:50:42 EST

Benefit of deduplication is dependent on the workload so it's not
preferable to always enable. Therefore, make it optional in Kconfig
and device param. Default is 'off'. This option will be beneficial
for users who use the zram as blockdev and stores build output to it.

Signed-off-by: Joonsoo Kim <iamjoonsoo.kim@xxxxxxx>
---
Documentation/ABI/testing/sysfs-block-zram | 10 ++++
Documentation/blockdev/zram.txt | 1 +
drivers/block/zram/Kconfig | 14 ++++++
drivers/block/zram/Makefile | 5 +-
drivers/block/zram/zram_dedup.c | 15 ++++++
drivers/block/zram/zram_dedup.h | 23 +++++++++
drivers/block/zram/zram_drv.c | 81 ++++++++++++++++++++++++++----
drivers/block/zram/zram_drv.h | 9 ++++
8 files changed, 145 insertions(+), 13 deletions(-)

diff --git a/Documentation/ABI/testing/sysfs-block-zram b/Documentation/ABI/testing/sysfs-block-zram
index 451b6d8..3c1945f 100644
--- a/Documentation/ABI/testing/sysfs-block-zram
+++ b/Documentation/ABI/testing/sysfs-block-zram
@@ -90,3 +90,13 @@ Description:
device's debugging info useful for kernel developers. Its
format is not documented intentionally and may change
anytime without any notice.
+
+What: /sys/block/zram<id>/use_dedup
+Date: March 2017
+Contact: Joonsoo Kim <iamjoonsoo.kim@xxxxxxx>
+Description:
+ The use_dedup file is read-write and specifies deduplication
+ feature is used or not. If enabled, duplicated data is
+ managed by reference count and will not be stored in memory
+ twice. Benefit of this feature largely depends on the workload
+ so keep attention when use.
diff --git a/Documentation/blockdev/zram.txt b/Documentation/blockdev/zram.txt
index 2cdc303..cbbe39b 100644
--- a/Documentation/blockdev/zram.txt
+++ b/Documentation/blockdev/zram.txt
@@ -168,6 +168,7 @@ max_comp_streams RW the number of possible concurrent compress operations
comp_algorithm RW show and change the compression algorithm
compact WO trigger memory compaction
debug_stat RO this file is used for zram debugging purposes
+use_dedup RW show and set deduplication feature


User space is advised to use the following files to read the device statistics.
diff --git a/drivers/block/zram/Kconfig b/drivers/block/zram/Kconfig
index b8ecba6..2f3dd1f 100644
--- a/drivers/block/zram/Kconfig
+++ b/drivers/block/zram/Kconfig
@@ -13,3 +13,17 @@ config ZRAM
disks and maybe many more.

See zram.txt for more information.
+
+config ZRAM_DEDUP
+ bool "Deduplication support for ZRAM data"
+ depends on ZRAM
+ default n
+ help
+ Deduplicate ZRAM data to reduce amount of memory consumption.
+ Advantage largely depends on the workload. In some cases, this
+ option reduces memory usage to the half. However, if there is no
+ duplicated data, the amount of memory consumption would be
+ increased due to additional metadata usage. And, there is
+ computation time trade-off. Please check the benefit before
+ enabling this option. Experiment shows the positive effect when
+ the zram is used as blockdev and is used to store build output.
diff --git a/drivers/block/zram/Makefile b/drivers/block/zram/Makefile
index 29cb008..1f6fecd 100644
--- a/drivers/block/zram/Makefile
+++ b/drivers/block/zram/Makefile
@@ -1,3 +1,4 @@
-zram-y := zcomp.o zram_drv.o zram_dedup.o
+zram-y := zcomp.o zram_drv.o

-obj-$(CONFIG_ZRAM) += zram.o
+obj-$(CONFIG_ZRAM) += zram.o
+obj-$(CONFIG_ZRAM_DEDUP) += zram_dedup.o
diff --git a/drivers/block/zram/zram_dedup.c b/drivers/block/zram/zram_dedup.c
index a8427f7..560b1f5 100644
--- a/drivers/block/zram/zram_dedup.c
+++ b/drivers/block/zram/zram_dedup.c
@@ -41,6 +41,9 @@ void zram_dedup_insert(struct zram *zram, struct zram_entry *new,
struct rb_node **rb_node, *parent = NULL;
struct zram_entry *entry;

+ if (!zram_dedup_enabled(zram))
+ return;
+
new->checksum = checksum;
hash = &zram->hash[checksum % zram->hash_size];
rb_root = &hash->rb_root;
@@ -148,6 +151,9 @@ struct zram_entry *zram_dedup_find(struct zram *zram, struct page *page,
void *mem;
struct zram_entry *entry;

+ if (!zram_dedup_enabled(zram))
+ return NULL;
+
mem = kmap_atomic(page);
*checksum = zram_dedup_checksum(mem);

@@ -160,6 +166,9 @@ struct zram_entry *zram_dedup_find(struct zram *zram, struct page *page,
void zram_dedup_init_entry(struct zram *zram, struct zram_entry *entry,
unsigned long handle, unsigned int len)
{
+ if (!zram_dedup_enabled(zram))
+ return;
+
entry->handle = handle;
entry->refcount = 1;
entry->len = len;
@@ -167,6 +176,9 @@ void zram_dedup_init_entry(struct zram *zram, struct zram_entry *entry,

bool zram_dedup_put_entry(struct zram *zram, struct zram_entry *entry)
{
+ if (!zram_dedup_enabled(zram))
+ return true;
+
if (zram_dedup_put(zram, entry))
return false;

@@ -178,6 +190,9 @@ int zram_dedup_init(struct zram *zram, size_t num_pages)
int i;
struct zram_hash *hash;

+ if (!zram_dedup_enabled(zram))
+ return 0;
+
zram->hash_size = num_pages >> ZRAM_HASH_SHIFT;
zram->hash_size = min_t(size_t, ZRAM_HASH_SIZE_MAX, zram->hash_size);
zram->hash_size = max_t(size_t, ZRAM_HASH_SIZE_MIN, zram->hash_size);
diff --git a/drivers/block/zram/zram_dedup.h b/drivers/block/zram/zram_dedup.h
index ebe6bff..8ab267b 100644
--- a/drivers/block/zram/zram_dedup.h
+++ b/drivers/block/zram/zram_dedup.h
@@ -4,6 +4,8 @@
struct zram;
struct zram_entry;

+#ifdef CONFIG_ZRAM_DEDUP
+
u64 zram_dedup_dup_size(struct zram *zram);
u64 zram_dedup_meta_size(struct zram *zram);

@@ -18,5 +20,26 @@ bool zram_dedup_put_entry(struct zram *zram, struct zram_entry *entry);

int zram_dedup_init(struct zram *zram, size_t num_pages);
void zram_dedup_fini(struct zram *zram);
+#else
+
+static inline u64 zram_dedup_dup_size(struct zram *zram) { return 0; }
+static inline u64 zram_dedup_meta_size(struct zram *zram) { return 0; }
+
+static inline void zram_dedup_insert(struct zram *zram, struct zram_entry *new,
+ u32 checksum) { }
+static inline struct zram_entry *zram_dedup_find(struct zram *zram,
+ struct page *page, u32 *checksum) { return NULL; }
+
+static inline void zram_dedup_init_entry(struct zram *zram,
+ struct zram_entry *entry, unsigned long handle,
+ unsigned int len) { }
+static inline bool zram_dedup_put_entry(struct zram *zram,
+ struct zram_entry *entry) { return true; }
+
+static inline int zram_dedup_init(struct zram *zram,
+ size_t num_pages) { return 0; }
+static inline void zram_dedup_fini(struct zram *zram) { }
+
+#endif

#endif /* _ZRAM_DEDUP_H_ */
diff --git a/drivers/block/zram/zram_drv.c b/drivers/block/zram/zram_drv.c
index 8eab8a0..372602c 100644
--- a/drivers/block/zram/zram_drv.c
+++ b/drivers/block/zram/zram_drv.c
@@ -333,6 +333,41 @@ static ssize_t comp_algorithm_store(struct device *dev,
return len;
}

+static ssize_t use_dedup_show(struct device *dev,
+ struct device_attribute *attr, char *buf)
+{
+ bool val;
+ struct zram *zram = dev_to_zram(dev);
+
+ down_read(&zram->init_lock);
+ val = zram->use_dedup;
+ up_read(&zram->init_lock);
+
+ return scnprintf(buf, PAGE_SIZE, "%d\n", (int)val);
+}
+
+#ifdef CONFIG_ZRAM_DEDUP
+static ssize_t use_dedup_store(struct device *dev,
+ struct device_attribute *attr, const char *buf, size_t len)
+{
+ int val;
+ struct zram *zram = dev_to_zram(dev);
+
+ if (kstrtoint(buf, 10, &val) || (val != 0 && val != 1))
+ return -EINVAL;
+
+ down_write(&zram->init_lock);
+ if (init_done(zram)) {
+ up_write(&zram->init_lock);
+ pr_info("Can't change dedup usage for initialized device\n");
+ return -EBUSY;
+ }
+ zram->use_dedup = val;
+ up_write(&zram->init_lock);
+ return len;
+}
+#endif
+
static ssize_t compact_store(struct device *dev,
struct device_attribute *attr, const char *buf, size_t len)
{
@@ -425,6 +460,15 @@ static DEVICE_ATTR_RO(io_stat);
static DEVICE_ATTR_RO(mm_stat);
static DEVICE_ATTR_RO(debug_stat);

+static unsigned long zram_entry_handle(struct zram *zram,
+ struct zram_entry *entry)
+{
+ if (zram_dedup_enabled(zram))
+ return entry->handle;
+ else
+ return (unsigned long)entry;
+}
+
static void zram_slot_lock(struct zram *zram, u32 index)
{
bit_spin_lock(ZRAM_ACCESS, &zram->table[index].value);
@@ -485,14 +529,17 @@ static struct zram_entry *zram_entry_alloc(struct zram *zram,
struct zram_entry *entry;
unsigned long handle;

- entry = kzalloc(sizeof(*entry),
- flags & ~(__GFP_HIGHMEM|__GFP_MOVABLE));
- if (!entry)
+ handle = zs_malloc(zram->mem_pool, len, flags);
+ if (!handle)
return NULL;

- handle = zs_malloc(zram->mem_pool, len, flags);
- if (!handle) {
- kfree(entry);
+ if (!zram_dedup_enabled(zram))
+ return (struct zram_entry *)handle;
+
+ entry = kzalloc(sizeof(*entry),
+ flags & ~(__GFP_HIGHMEM|__GFP_MOVABLE));
+ if (!entry) {
+ zs_free(zram->mem_pool, handle);
return NULL;
}

@@ -507,7 +554,11 @@ void zram_entry_free(struct zram *zram, struct zram_entry *entry)
if (!zram_dedup_put_entry(zram, entry))
return;

- zs_free(zram->mem_pool, entry->handle);
+ zs_free(zram->mem_pool, zram_entry_handle(zram, entry));
+
+ if (!zram_dedup_enabled(zram))
+ return;
+
kfree(entry);

atomic64_sub(sizeof(*entry), &zram->stats.meta_data_size);
@@ -598,7 +649,8 @@ static int zram_decompress_page(struct zram *zram, struct page *page, u32 index)
entry = zram_get_entry(zram, index);
size = zram_get_obj_size(zram, index);

- src = zs_map_object(zram->mem_pool, entry->handle, ZS_MM_RO);
+ src = zs_map_object(zram->mem_pool,
+ zram_entry_handle(zram, entry), ZS_MM_RO);
if (size == PAGE_SIZE) {
dst = kmap_atomic(page);
memcpy(dst, src, PAGE_SIZE);
@@ -612,7 +664,7 @@ static int zram_decompress_page(struct zram *zram, struct page *page, u32 index)
kunmap_atomic(dst);
zcomp_stream_put(zram->comp);
}
- zs_unmap_object(zram->mem_pool, entry->handle);
+ zs_unmap_object(zram->mem_pool, zram_entry_handle(zram, entry));
zram_slot_unlock(zram, index);

/* Should NEVER happen. Return bio error if it does. */
@@ -750,7 +802,8 @@ static int __zram_bvec_write(struct zram *zram, struct bio_vec *bvec, u32 index)
return ret;
}

- dst = zs_map_object(zram->mem_pool, entry->handle, ZS_MM_WO);
+ dst = zs_map_object(zram->mem_pool,
+ zram_entry_handle(zram, entry), ZS_MM_WO);

src = zstrm->buffer;
if (comp_len == PAGE_SIZE)
@@ -760,7 +813,7 @@ static int __zram_bvec_write(struct zram *zram, struct bio_vec *bvec, u32 index)
kunmap_atomic(src);

zcomp_stream_put(zram->comp);
- zs_unmap_object(zram->mem_pool, entry->handle);
+ zs_unmap_object(zram->mem_pool, zram_entry_handle(zram, entry));
zram_dedup_insert(zram, entry, checksum);

found_dup:
@@ -1159,6 +1212,11 @@ static DEVICE_ATTR_WO(mem_limit);
static DEVICE_ATTR_WO(mem_used_max);
static DEVICE_ATTR_RW(max_comp_streams);
static DEVICE_ATTR_RW(comp_algorithm);
+#ifdef CONFIG_ZRAM_DEDUP
+static DEVICE_ATTR_RW(use_dedup);
+#else
+static DEVICE_ATTR_RO(use_dedup);
+#endif

static struct attribute *zram_disk_attrs[] = {
&dev_attr_disksize.attr,
@@ -1169,6 +1227,7 @@ static struct attribute *zram_disk_attrs[] = {
&dev_attr_mem_used_max.attr,
&dev_attr_max_comp_streams.attr,
&dev_attr_comp_algorithm.attr,
+ &dev_attr_use_dedup.attr,
&dev_attr_io_stat.attr,
&dev_attr_mm_stat.attr,
&dev_attr_debug_stat.attr,
diff --git a/drivers/block/zram/zram_drv.h b/drivers/block/zram/zram_drv.h
index 4b86921..102fb7a 100644
--- a/drivers/block/zram/zram_drv.h
+++ b/drivers/block/zram/zram_drv.h
@@ -134,7 +134,16 @@ struct zram {
* zram is claimed so open request will be failed
*/
bool claim; /* Protected by bdev->bd_mutex */
+ bool use_dedup;
};

+static inline bool zram_dedup_enabled(struct zram *zram)
+{
+#ifdef CONFIG_ZRAM_DEDUP
+ return zram->use_dedup;
+#else
+ return 0;
+#endif
+}
void zram_entry_free(struct zram *zram, struct zram_entry *entry);
#endif
--
2.7.4