[to-be-updated] zram-preparation-for-multi-zcomp-support.patch removed from -mm tree

[to-be-updated] zram-preparation-for-multi-zcomp-support.patch removed from -mm tree

[Andrew Morton]

The quilt patch titled
     Subject: zram: preparation for multi-zcomp support
has been removed from the -mm tree.  Its filename was
     zram-preparation-for-multi-zcomp-support.patch

This patch was dropped because an updated version will be merged

------------------------------------------------------
From: Sergey Senozhatsky <senozhatsky@chromium.org>
Subject: zram: preparation for multi-zcomp support
Date: Tue, 18 Oct 2022 13:55:25 +0900

Patch series "zram: Support multiple compression streams", v4.

This series adds support for multiple (per-CPU) compression streams (at
point only 2).  The main idea is that different compression algorithms
have different characteristics and zram may benefit when it uses a
combination of algorithms: a default algorithm that is faster but have
lower compression rate and a secondary algorithm that can use higher
compression rate at a price of slower compression/decompression.

There are several use cases for this functionality:

- huge pages recompression: zstd or deflate can successfully compress
  huge pages (~50% of huge pages on my synthetic ChromeOS tests), IOW
  pages that lzo was not able to compress.

- idle pages recompression: idle/cold pages sit in the memory and we may
  reduce zsmalloc memory usage if we recompress those idle pages.

Userspace has a number of ways to control the behavior and impact of zram
recompression: what type of pages should be recompressed, size watermarks,
etc.  Please refer to documentation patch.


This patch (of 9):

Turn compression streams and compressor algorithm name struct zram members
into arrays, so that we can have multiple compression streams support (in
the next patches).

The patch uses a rather explicit API for compressor selection:

- Get primary (default) compression stream
	zcomp_stream_get(zram->comps[ZRAM_PRIMARY_ZCOMP])
- Get secondary compression stream
	zcomp_stream_get(zram->comps[ZRAM_SECONDARY_ZCOMP])

We use similar API for compression streams put().

At this point we always have just one compression stream, since
CONFIG_ZRAM_MULTI_COMP is not yet defined.

Link: https://lkml.kernel.org/r/20221018045533.2396670-1-senozhatsky@chromium.org
Link: https://lkml.kernel.org/r/20221018045533.2396670-2-senozhatsky@chromium.org
Signed-off-by: Sergey Senozhatsky <senozhatsky@chromium.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Nitin Gupta <ngupta@vflare.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
---

 drivers/block/zram/zcomp.c    |    6 +-
 drivers/block/zram/zcomp.h    |    2 
 drivers/block/zram/zram_drv.c |   87 ++++++++++++++++++++++----------
 drivers/block/zram/zram_drv.h |   14 ++++-
 4 files changed, 77 insertions(+), 32 deletions(-)

--- a/drivers/block/zram/zcomp.c~zram-preparation-for-multi-zcomp-support
+++ a/drivers/block/zram/zcomp.c
@@ -206,7 +206,7 @@ void zcomp_destroy(struct zcomp *comp)
  * case of allocation error, or any other error potentially
  * returned by zcomp_init().
  */
-struct zcomp *zcomp_create(const char *compress)
+struct zcomp *zcomp_create(const char *alg)
 {
 	struct zcomp *comp;
 	int error;
@@ -216,14 +216,14 @@ struct zcomp *zcomp_create(const char *c
 	 * is not loaded yet. We must do it here, otherwise we are about to
 	 * call /sbin/modprobe under CPU hot-plug lock.
 	 */
-	if (!zcomp_available_algorithm(compress))
+	if (!zcomp_available_algorithm(alg))
 		return ERR_PTR(-EINVAL);
 
 	comp = kzalloc(sizeof(struct zcomp), GFP_KERNEL);
 	if (!comp)
 		return ERR_PTR(-ENOMEM);
 
-	comp->name = compress;
+	comp->name = alg;
 	error = zcomp_init(comp);
 	if (error) {
 		kfree(comp);
--- a/drivers/block/zram/zcomp.h~zram-preparation-for-multi-zcomp-support
+++ a/drivers/block/zram/zcomp.h
@@ -27,7 +27,7 @@ int zcomp_cpu_dead(unsigned int cpu, str
 ssize_t zcomp_available_show(const char *comp, char *buf);
 bool zcomp_available_algorithm(const char *comp);
 
-struct zcomp *zcomp_create(const char *comp);
+struct zcomp *zcomp_create(const char *alg);
 void zcomp_destroy(struct zcomp *comp);
 
 struct zcomp_strm *zcomp_stream_get(struct zcomp *comp);
--- a/drivers/block/zram/zram_drv.c~zram-preparation-for-multi-zcomp-support
+++ a/drivers/block/zram/zram_drv.c
@@ -1004,36 +1004,53 @@ static ssize_t comp_algorithm_show(struc
 	struct zram *zram = dev_to_zram(dev);
 
 	down_read(&zram->init_lock);
-	sz = zcomp_available_show(zram->compressor, buf);
+	sz = zcomp_available_show(zram->comp_algs[ZRAM_PRIMARY_ZCOMP], buf);
 	up_read(&zram->init_lock);
 
 	return sz;
 }
 
+static void comp_algorithm_set(struct zram *zram, u32 idx, const char *alg)
+{
+	/* Do not kfree() algs that we didn't allocate, IOW the default ones */
+	if (zram->comp_algs[idx] != default_compressor)
+		kfree(zram->comp_algs[idx]);
+	zram->comp_algs[idx] = alg;
+}
+
 static ssize_t comp_algorithm_store(struct device *dev,
 		struct device_attribute *attr, const char *buf, size_t len)
 {
 	struct zram *zram = dev_to_zram(dev);
-	char compressor[ARRAY_SIZE(zram->compressor)];
+	char *compressor;
 	size_t sz;
 
-	strscpy(compressor, buf, sizeof(compressor));
+	sz = strlen(buf);
+	if (sz >= CRYPTO_MAX_ALG_NAME)
+		return -E2BIG;
+
+	compressor = kstrdup(buf, GFP_KERNEL);
+	if (!compressor)
+		return -ENOMEM;
+
 	/* ignore trailing newline */
-	sz = strlen(compressor);
 	if (sz > 0 && compressor[sz - 1] == '\n')
 		compressor[sz - 1] = 0x00;
 
-	if (!zcomp_available_algorithm(compressor))
+	if (!zcomp_available_algorithm(compressor)) {
+		kfree(compressor);
 		return -EINVAL;
+	}
 
 	down_write(&zram->init_lock);
 	if (init_done(zram)) {
 		up_write(&zram->init_lock);
+		kfree(compressor);
 		pr_info("Can't change algorithm for initialized device\n");
 		return -EBUSY;
 	}
 
-	strcpy(zram->compressor, compressor);
+	comp_algorithm_set(zram, ZRAM_PRIMARY_ZCOMP, compressor);
 	up_write(&zram->init_lock);
 	return len;
 }
@@ -1281,7 +1298,7 @@ static int __zram_bvec_read(struct zram
 	size = zram_get_obj_size(zram, index);
 
 	if (size != PAGE_SIZE)
-		zstrm = zcomp_stream_get(zram->comp);
+		zstrm = zcomp_stream_get(zram->comps[ZRAM_PRIMARY_ZCOMP]);
 
 	src = zs_map_object(zram->mem_pool, handle, ZS_MM_RO);
 	if (size == PAGE_SIZE) {
@@ -1293,7 +1310,7 @@ static int __zram_bvec_read(struct zram
 		dst = kmap_atomic(page);
 		ret = zcomp_decompress(zstrm, src, size, dst);
 		kunmap_atomic(dst);
-		zcomp_stream_put(zram->comp);
+		zcomp_stream_put(zram->comps[ZRAM_PRIMARY_ZCOMP]);
 	}
 	zs_unmap_object(zram->mem_pool, handle);
 	zram_slot_unlock(zram, index);
@@ -1360,13 +1377,13 @@ static int __zram_bvec_write(struct zram
 	kunmap_atomic(mem);
 
 compress_again:
-	zstrm = zcomp_stream_get(zram->comp);
+	zstrm = zcomp_stream_get(zram->comps[ZRAM_PRIMARY_ZCOMP]);
 	src = kmap_atomic(page);
 	ret = zcomp_compress(zstrm, src, &comp_len);
 	kunmap_atomic(src);
 
 	if (unlikely(ret)) {
-		zcomp_stream_put(zram->comp);
+		zcomp_stream_put(zram->comps[ZRAM_PRIMARY_ZCOMP]);
 		pr_err("Compression failed! err=%d\n", ret);
 		zs_free(zram->mem_pool, handle);
 		return ret;
@@ -1394,7 +1411,7 @@ compress_again:
 				__GFP_HIGHMEM |
 				__GFP_MOVABLE);
 	if (IS_ERR((void *)handle)) {
-		zcomp_stream_put(zram->comp);
+		zcomp_stream_put(zram->comps[ZRAM_PRIMARY_ZCOMP]);
 		atomic64_inc(&zram->stats.writestall);
 		handle = zs_malloc(zram->mem_pool, comp_len,
 				GFP_NOIO | __GFP_HIGHMEM |
@@ -1411,14 +1428,14 @@ compress_again:
 		 * zstrm buffer back. It is necessary that the dereferencing
 		 * of the zstrm variable below occurs correctly.
 		 */
-		zstrm = zcomp_stream_get(zram->comp);
+		zstrm = zcomp_stream_get(zram->comps[ZRAM_PRIMARY_ZCOMP]);
 	}
 
 	alloced_pages = zs_get_total_pages(zram->mem_pool);
 	update_used_max(zram, alloced_pages);
 
 	if (zram->limit_pages && alloced_pages > zram->limit_pages) {
-		zcomp_stream_put(zram->comp);
+		zcomp_stream_put(zram->comps[ZRAM_PRIMARY_ZCOMP]);
 		zs_free(zram->mem_pool, handle);
 		return -ENOMEM;
 	}
@@ -1432,7 +1449,7 @@ compress_again:
 	if (comp_len == PAGE_SIZE)
 		kunmap_atomic(src);
 
-	zcomp_stream_put(zram->comp);
+	zcomp_stream_put(zram->comps[ZRAM_PRIMARY_ZCOMP]);
 	zs_unmap_object(zram->mem_pool, handle);
 	atomic64_add(comp_len, &zram->stats.compr_data_size);
 out:
@@ -1707,6 +1724,20 @@ out:
 	return ret;
 }
 
+static void zram_destroy_comps(struct zram *zram)
+{
+	u32 idx;
+
+	for (idx = 0; idx < ZRAM_MAX_ZCOMPS; idx++) {
+		struct zcomp *comp = zram->comps[idx];
+
+		zram->comps[idx] = NULL;
+		if (IS_ERR_OR_NULL(comp))
+			continue;
+		zcomp_destroy(comp);
+	}
+}
+
 static void zram_reset_device(struct zram *zram)
 {
 	down_write(&zram->init_lock);
@@ -1724,11 +1755,11 @@ static void zram_reset_device(struct zra
 	/* I/O operation under all of CPU are done so let's free */
 	zram_meta_free(zram, zram->disksize);
 	zram->disksize = 0;
+	zram_destroy_comps(zram);
 	memset(&zram->stats, 0, sizeof(zram->stats));
-	zcomp_destroy(zram->comp);
-	zram->comp = NULL;
 	reset_bdev(zram);
 
+	comp_algorithm_set(zram, ZRAM_PRIMARY_ZCOMP, default_compressor);
 	up_write(&zram->init_lock);
 }
 
@@ -1739,6 +1770,7 @@ static ssize_t disksize_store(struct dev
 	struct zcomp *comp;
 	struct zram *zram = dev_to_zram(dev);
 	int err;
+	u32 idx;
 
 	disksize = memparse(buf, NULL);
 	if (!disksize)
@@ -1757,22 +1789,25 @@ static ssize_t disksize_store(struct dev
 		goto out_unlock;
 	}
 
-	comp = zcomp_create(zram->compressor);
-	if (IS_ERR(comp)) {
-		pr_err("Cannot initialise %s compressing backend\n",
-				zram->compressor);
-		err = PTR_ERR(comp);
-		goto out_free_meta;
-	}
+	for (idx = 0; idx < ZRAM_MAX_ZCOMPS; idx++) {
+		comp = zcomp_create(zram->comp_algs[idx]);
+		if (IS_ERR(comp)) {
+			pr_err("Cannot initialise %s compressing backend\n",
+			       zram->comp_algs[idx]);
+			err = PTR_ERR(comp);
+			goto out_free_comps;
+		}
 
-	zram->comp = comp;
+		zram->comps[idx] = comp;
+	}
 	zram->disksize = disksize;
 	set_capacity_and_notify(zram->disk, zram->disksize >> SECTOR_SHIFT);
 	up_write(&zram->init_lock);
 
 	return len;
 
-out_free_meta:
+out_free_comps:
+	zram_destroy_comps(zram);
 	zram_meta_free(zram, disksize);
 out_unlock:
 	up_write(&zram->init_lock);
@@ -1959,7 +1994,7 @@ static int zram_add(void)
 	if (ret)
 		goto out_cleanup_disk;
 
-	strscpy(zram->compressor, default_compressor, sizeof(zram->compressor));
+	zram->comp_algs[ZRAM_PRIMARY_ZCOMP] = default_compressor;
 
 	zram_debugfs_register(zram);
 	pr_info("Added device: %s\n", zram->disk->disk_name);
--- a/drivers/block/zram/zram_drv.h~zram-preparation-for-multi-zcomp-support
+++ a/drivers/block/zram/zram_drv.h
@@ -89,10 +89,20 @@ struct zram_stats {
 #endif
 };
 
+#ifdef CONFIG_ZRAM_MULTI_COMP
+#define ZRAM_PRIMARY_ZCOMP	0
+#define ZRAM_SECONDARY_ZCOMP	1
+#define ZRAM_MAX_ZCOMPS	2
+#else
+#define ZRAM_PRIMARY_ZCOMP	0
+#define ZRAM_SECONDARY_ZCOMP	0
+#define ZRAM_MAX_ZCOMPS	1
+#endif
+
 struct zram {
 	struct zram_table_entry *table;
 	struct zs_pool *mem_pool;
-	struct zcomp *comp;
+	struct zcomp *comps[ZRAM_MAX_ZCOMPS];
 	struct gendisk *disk;
 	/* Prevent concurrent execution of device init */
 	struct rw_semaphore init_lock;
@@ -107,7 +117,7 @@ struct zram {
 	 * we can store in a disk.
 	 */
 	u64 disksize;	/* bytes */
-	char compressor[CRYPTO_MAX_ALG_NAME];
+	const char *comp_algs[ZRAM_MAX_ZCOMPS];
 	/*
 	 * zram is claimed so open request will be failed
 	 */
_

Patches currently in -mm which might be from senozhatsky@chromium.org are

zram-factor-out-wb-and-non-wb-zram-read-functions.patch
zram-introduce-recompress-sysfs-knob.patch
documentation-add-recompression-documentation.patch
zram-add-recompression-algorithm-choice-to-kconfig.patch
zram-add-recompress-flag-to-read_block_state.patch
zram-clarify-writeback_store-comment.patch
zram-use-is_err_value-to-check-for-zs_malloc-errors.patch
zsmalloc-turn-zspage-order-into-runtime-variable.patch
zsmalloc-move-away-from-page-order-defines.patch
zsmalloc-make-huge-class-watermark-zs_pool-member.patch
zram-huge-size-watermark-cannot-be-global.patch
zsmalloc-pass-limit-on-pages-per-zspage-to-zs_create_pool.patch
zram-add-pages_per_pool_page-device-attribute.patch
documentation-document-zram-pages_per_pool_page-attribute.patch
zsmalloc-break-out-of-loop-when-found-perfect-zspage-order.patch