squashfs: add the mount parameter theads=<single|multi|percpu>

Patch series 'squashfs: Add the mount parameter "threads="'.

Currently, Squashfs supports multiple decompressor parallel modes. 
However, this mode can be configured only during kernel building and does
not support flexible selection during runtime.

In the current patch set, the mount parameter "threads=" is added to allow
users to select the parallel decompressor mode and configure the number of
decompressors when mounting a file system.

"threads=<single|multi|percpu|1|2|3|...>"
The upper limit is num_online_cpus() * 2.


This patch (of 2):

Squashfs supports three decompression concurrency modes:
	Single-thread mode: concurrent reads are blocked and the memory
		overhead is small.
	Multi-thread mode/percpu mode: reduces concurrent read blocking but
		increases memory overhead.

The corresponding schema must be fixed at compile time. During mounting,
the concurrent decompression mode cannot be adjusted based on file read
blocking.

The mount parameter theads=<single|multi|percpu> is added to select
the concurrent decompression mode of a single SquashFS file system
image.

Link: https://lkml.kernel.org/r/20221019030930.130456-1-nixiaoming@huawei.com
Link: https://lkml.kernel.org/r/20221019030930.130456-2-nixiaoming@huawei.com
Signed-off-by: Xiaoming Ni <nixiaoming@huawei.com>
Reviewed-by: Phillip Lougher <phillip@squashfs.org.uk>
Cc: Jianguo Chen <chenjianguo3@huawei.com>
Cc: Jubin Zhong <zhongjubin@huawei.com>
Cc: Zhang Yi <yi.zhang@huawei.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

fs/squashfs/Kconfig

@@ -54,25 +54,54 @@ config SQUASHFS_FILE_DIRECT
 
 endchoice
 
+config SQUASHFS_DECOMP_SINGLE
+	depends on SQUASHFS
+	def_bool n
+
+config SQUASHFS_DECOMP_MULTI
+	depends on SQUASHFS
+	def_bool n
+
+config SQUASHFS_DECOMP_MULTI_PERCPU
+	depends on SQUASHFS
+	def_bool n
+
+config SQUASHFS_CHOICE_DECOMP_BY_MOUNT
+	bool "Select the parallel decompression mode during mount"
+	depends on SQUASHFS
+	default n
+	select SQUASHFS_DECOMP_SINGLE
+	select SQUASHFS_DECOMP_MULTI
+	select SQUASHFS_DECOMP_MULTI_PERCPU
+	help
+	  Compile all parallel decompression modes and specify the
+	  decompression mode by setting "threads=" during mount.
+	    threads=<single|multi|percpu>
+
+	  default Decompressor parallelisation is SQUASHFS_DECOMP_SINGLE
+
 choice
-	prompt "Decompressor parallelisation options"
+	prompt "Select decompression parallel mode at compile time"
 	depends on SQUASHFS
+	depends on !SQUASHFS_CHOICE_DECOMP_BY_MOUNT
 	help
 	  Squashfs now supports three parallelisation options for
 	  decompression.  Each one exhibits various trade-offs between
 	  decompression performance and CPU and memory usage.
 
 	  If in doubt, select "Single threaded compression"
 
-config SQUASHFS_DECOMP_SINGLE
+config SQUASHFS_COMPILE_DECOMP_SINGLE
 	bool "Single threaded compression"
+	select SQUASHFS_DECOMP_SINGLE
 	help
 	  Traditionally Squashfs has used single-threaded decompression.
 	  Only one block (data or metadata) can be decompressed at any
 	  one time.  This limits CPU and memory usage to a minimum.
 
-config SQUASHFS_DECOMP_MULTI
+config SQUASHFS_COMPILE_DECOMP_MULTI
 	bool "Use multiple decompressors for parallel I/O"
+	select SQUASHFS_DECOMP_MULTI
 	help
 	  By default Squashfs uses a single decompressor but it gives
 	  poor performance on parallel I/O workloads when using multiple CPU
@@ -85,8 +114,9 @@ config SQUASHFS_DECOMP_MULTI
 	  decompressors per core.  It dynamically allocates decompressors
 	  on a demand basis.
 
-config SQUASHFS_DECOMP_MULTI_PERCPU
+config SQUASHFS_COMPILE_DECOMP_MULTI_PERCPU
 	bool "Use percpu multiple decompressors for parallel I/O"
+	select SQUASHFS_DECOMP_MULTI_PERCPU
 	help
 	  By default Squashfs uses a single decompressor but it gives
 	  poor performance on parallel I/O workloads when using multiple CPU
@@ -95,7 +125,6 @@ config SQUASHFS_DECOMP_MULTI_PERCPU
 	  This decompressor implementation uses a maximum of one
 	  decompressor per core.  It uses percpu variables to ensure
 	  decompression is load-balanced across the cores.
-
 endchoice
 
 config SQUASHFS_XATTR

fs/squashfs/block.c

@@ -216,7 +216,7 @@ int squashfs_read_data(struct super_block *sb, u64 index, int length,
 			res = -EIO;
 			goto out_free_bio;
 		}
-		res = squashfs_decompress(msblk, bio, offset, length, output);
+		res = msblk->thread_ops->decompress(msblk, bio, offset, length, output);
 	} else {
 		res = copy_bio_to_actor(bio, output, offset, length);
 	}

fs/squashfs/decompressor.c

@@ -134,7 +134,7 @@ void *squashfs_decompressor_setup(struct super_block *sb, unsigned short flags)
 	if (IS_ERR(comp_opts))
 		return comp_opts;
 
-	stream = squashfs_decompressor_create(msblk, comp_opts);
+	stream = msblk->thread_ops->create(msblk, comp_opts);
 	if (IS_ERR(stream))
 		kfree(comp_opts);
 

fs/squashfs/decompressor_multi.c

@@ -29,12 +29,11 @@
 #define MAX_DECOMPRESSOR	(num_online_cpus() * 2)
 
 
-int squashfs_max_decompressors(void)
+static int squashfs_max_decompressors(void)
 {
 	return MAX_DECOMPRESSOR;
 }
 
-
 struct squashfs_stream {
 	void			*comp_opts;
 	struct list_head	strm_list;
@@ -59,7 +58,7 @@ static void put_decomp_stream(struct decomp_stream *decomp_strm,
 	wake_up(&stream->wait);
 }
 
-void *squashfs_decompressor_create(struct squashfs_sb_info *msblk,
+static void *squashfs_decompressor_create(struct squashfs_sb_info *msblk,
 				void *comp_opts)
 {
 	struct squashfs_stream *stream;
@@ -103,7 +102,7 @@ void *squashfs_decompressor_create(struct squashfs_sb_info *msblk,
 }
 
 
-void squashfs_decompressor_destroy(struct squashfs_sb_info *msblk)
+static void squashfs_decompressor_destroy(struct squashfs_sb_info *msblk)
 {
 	struct squashfs_stream *stream = msblk->stream;
 	if (stream) {
@@ -180,7 +179,7 @@ static struct decomp_stream *get_decomp_stream(struct squashfs_sb_info *msblk,
 }
 
 
-int squashfs_decompress(struct squashfs_sb_info *msblk, struct bio *bio,
+static int squashfs_decompress(struct squashfs_sb_info *msblk, struct bio *bio,
 			int offset, int length,
 			struct squashfs_page_actor *output)
 {
@@ -195,3 +194,10 @@ int squashfs_decompress(struct squashfs_sb_info *msblk, struct bio *bio,
 			msblk->decompressor->name);
 	return res;
 }
+
+const struct squashfs_decompressor_thread_ops squashfs_decompressor_multi = {
+	.create = squashfs_decompressor_create,
+	.destroy = squashfs_decompressor_destroy,
+	.decompress = squashfs_decompress,
+	.max_decompressors = squashfs_max_decompressors,
+};

fs/squashfs/decompressor_multi_percpu.c

@@ -25,7 +25,7 @@ struct squashfs_stream {
 	local_lock_t	lock;
 };
 
-void *squashfs_decompressor_create(struct squashfs_sb_info *msblk,
+static void *squashfs_decompressor_create(struct squashfs_sb_info *msblk,
 						void *comp_opts)
 {
 	struct squashfs_stream *stream;
@@ -59,7 +59,7 @@ void *squashfs_decompressor_create(struct squashfs_sb_info *msblk,
 	return ERR_PTR(err);
 }
 
-void squashfs_decompressor_destroy(struct squashfs_sb_info *msblk)
+static void squashfs_decompressor_destroy(struct squashfs_sb_info *msblk)
 {
 	struct squashfs_stream __percpu *percpu =
 			(struct squashfs_stream __percpu *) msblk->stream;
@@ -75,19 +75,21 @@ void squashfs_decompressor_destroy(struct squashfs_sb_info *msblk)
 	}
 }
 
-int squashfs_decompress(struct squashfs_sb_info *msblk, struct bio *bio,
+static int squashfs_decompress(struct squashfs_sb_info *msblk, struct bio *bio,
 	int offset, int length, struct squashfs_page_actor *output)
 {
 	struct squashfs_stream *stream;
+	struct squashfs_stream __percpu *percpu =
+			(struct squashfs_stream __percpu *) msblk->stream;
 	int res;
 
-	local_lock(&msblk->stream->lock);
-	stream = this_cpu_ptr(msblk->stream);
+	local_lock(&percpu->lock);
+	stream = this_cpu_ptr(percpu);
 
 	res = msblk->decompressor->decompress(msblk, stream->stream, bio,
 					      offset, length, output);
 
-	local_unlock(&msblk->stream->lock);
+	local_unlock(&percpu->lock);
 
 	if (res < 0)
 		ERROR("%s decompression failed, data probably corrupt\n",
@@ -96,7 +98,14 @@ int squashfs_decompress(struct squashfs_sb_info *msblk, struct bio *bio,
 	return res;
 }
 
-int squashfs_max_decompressors(void)
+static int squashfs_max_decompressors(void)
 {
 	return num_possible_cpus();
 }
+
+const struct squashfs_decompressor_thread_ops squashfs_decompressor_percpu = {
+	.create = squashfs_decompressor_create,
+	.destroy = squashfs_decompressor_destroy,
+	.decompress = squashfs_decompress,
+	.max_decompressors = squashfs_max_decompressors,
+};

fs/squashfs/decompressor_single.c

@@ -24,7 +24,7 @@ struct squashfs_stream {
 	struct mutex	mutex;
 };
 
-void *squashfs_decompressor_create(struct squashfs_sb_info *msblk,
+static void *squashfs_decompressor_create(struct squashfs_sb_info *msblk,
 						void *comp_opts)
 {
 	struct squashfs_stream *stream;
@@ -49,7 +49,7 @@ void *squashfs_decompressor_create(struct squashfs_sb_info *msblk,
 	return ERR_PTR(err);
 }
 
-void squashfs_decompressor_destroy(struct squashfs_sb_info *msblk)
+static void squashfs_decompressor_destroy(struct squashfs_sb_info *msblk)
 {
 	struct squashfs_stream *stream = msblk->stream;
 
@@ -59,7 +59,7 @@ void squashfs_decompressor_destroy(struct squashfs_sb_info *msblk)
 	}
 }
 
-int squashfs_decompress(struct squashfs_sb_info *msblk, struct bio *bio,
+static int squashfs_decompress(struct squashfs_sb_info *msblk, struct bio *bio,
 			int offset, int length,
 			struct squashfs_page_actor *output)
 {
@@ -78,7 +78,14 @@ int squashfs_decompress(struct squashfs_sb_info *msblk, struct bio *bio,
 	return res;
 }
 
-int squashfs_max_decompressors(void)
+static int squashfs_max_decompressors(void)
 {
 	return 1;
 }
+
+const struct squashfs_decompressor_thread_ops squashfs_decompressor_single = {
+	.create = squashfs_decompressor_create,
+	.destroy = squashfs_decompressor_destroy,
+	.decompress = squashfs_decompress,
+	.max_decompressors = squashfs_max_decompressors,
+};

fs/squashfs/squashfs.h

@@ -38,11 +38,24 @@ extern const struct squashfs_decompressor *squashfs_lookup_decompressor(int);
 extern void *squashfs_decompressor_setup(struct super_block *, unsigned short);
 
 /* decompressor_xxx.c */
-extern void *squashfs_decompressor_create(struct squashfs_sb_info *, void *);
-extern void squashfs_decompressor_destroy(struct squashfs_sb_info *);
-extern int squashfs_decompress(struct squashfs_sb_info *, struct bio *,
-				int, int, struct squashfs_page_actor *);
-extern int squashfs_max_decompressors(void);
+
+struct squashfs_decompressor_thread_ops {
+	void * (*create)(struct squashfs_sb_info *msblk, void *comp_opts);
+	void (*destroy)(struct squashfs_sb_info *msblk);
+	int (*decompress)(struct squashfs_sb_info *msblk, struct bio *bio,
+			  int offset, int length, struct squashfs_page_actor *output);
+	int (*max_decompressors)(void);
+};
+
+#ifdef CONFIG_SQUASHFS_DECOMP_SINGLE
+extern const struct squashfs_decompressor_thread_ops squashfs_decompressor_single;
+#endif
+#ifdef CONFIG_SQUASHFS_DECOMP_MULTI
+extern const struct squashfs_decompressor_thread_ops squashfs_decompressor_multi;
+#endif
+#ifdef CONFIG_SQUASHFS_DECOMP_MULTI_PERCPU
+extern const struct squashfs_decompressor_thread_ops squashfs_decompressor_percpu;
+#endif
 
 /* export.c */
 extern __le64 *squashfs_read_inode_lookup_table(struct super_block *, u64, u64,

fs/squashfs/squashfs_fs_sb.h

@@ -53,7 +53,7 @@ struct squashfs_sb_info {
 	__le64					*xattr_id_table;
 	struct mutex				meta_index_mutex;
 	struct meta_index			*meta_index;
-	struct squashfs_stream			*stream;
+	void					*stream;
 	__le64					*inode_lookup_table;
 	u64					inode_table;
 	u64					directory_table;
@@ -66,5 +66,6 @@ struct squashfs_sb_info {
 	int					xattr_ids;
 	unsigned int				ids;
 	bool					panic_on_errors;
+	const struct squashfs_decompressor_thread_ops *thread_ops;
 };
 #endif