squashfs: migrate from ll_rw_block usage to BIO

ll_rw_block() function has been deprecated in favor of BIO which appears
to come with large performance improvements.

This patch decreases boot time by close to 40% when using squashfs for
the root file-system.  This is observed at least in the context of
starting an Android VM on Chrome OS using crosvm.  The patch was tested
on 4.19 as well as master.

This patch is largely based on Adrien Schildknecht's patch that was
originally sent as https://lkml.org/lkml/2017/9/22/814 though with some
significant changes and simplifications while also taking Phillip
Lougher's feedback into account, around preserving support for
FILE_CACHE in particular.

[[email protected]: fix build error reported by Randy]
  Link: http://lkml.kernel.org/r/[email protected]
Signed-off-by: Philippe Liard <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Reviewed-by: Christoph Hellwig <[email protected]>
Cc: Adrien Schildknecht <[email protected]>
Cc: Phillip Lougher <[email protected]>
Cc: Guenter Roeck <[email protected]>
Cc: Daniel Rosenberg <[email protected]>
Link: https://chromium.googlesource.com/chromiumos/platform/crosvm
Link: http://lkml.kernel.org/r/[email protected]
Signed-off-by: Linus Torvalds <[email protected]>

Author: Philippe Liard

fs/squashfs/block.c

@@ -13,6 +13,7 @@
  * datablocks and metadata blocks.
  */
 
+#include <linux/blkdev.h>
 #include <linux/fs.h>
 #include <linux/vfs.h>
 #include <linux/slab.h>
@@ -27,44 +28,103 @@
 #include "page_actor.h"
 
 /*
- * Read the metadata block length, this is stored in the first two
- * bytes of the metadata block.
+ * Returns the amount of bytes copied to the page actor.
  */
-static struct buffer_head *get_block_length(struct super_block *sb,
-			u64 *cur_index, int *offset, int *length)
+static int copy_bio_to_actor(struct bio *bio,
+			     struct squashfs_page_actor *actor,
+			     int offset, int req_length)
+{
+	void *actor_addr = squashfs_first_page(actor);
+	struct bvec_iter_all iter_all = {};
+	struct bio_vec *bvec = bvec_init_iter_all(&iter_all);
+	int copied_bytes = 0;
+	int actor_offset = 0;
+
+	if (WARN_ON_ONCE(!bio_next_segment(bio, &iter_all)))
+		return 0;
+
+	while (copied_bytes < req_length) {
+		int bytes_to_copy = min_t(int, bvec->bv_len - offset,
+					  PAGE_SIZE - actor_offset);
+
+		bytes_to_copy = min_t(int, bytes_to_copy,
+				      req_length - copied_bytes);
+		memcpy(actor_addr + actor_offset,
+		       page_address(bvec->bv_page) + bvec->bv_offset + offset,
+		       bytes_to_copy);
+
+		actor_offset += bytes_to_copy;
+		copied_bytes += bytes_to_copy;
+		offset += bytes_to_copy;
+
+		if (actor_offset >= PAGE_SIZE) {
+			actor_addr = squashfs_next_page(actor);
+			if (!actor_addr)
+				break;
+			actor_offset = 0;
+		}
+		if (offset >= bvec->bv_len) {
+			if (!bio_next_segment(bio, &iter_all))
+				break;
+			offset = 0;
+		}
+	}
+	squashfs_finish_page(actor);
+	return copied_bytes;
+}
+
+static int squashfs_bio_read(struct super_block *sb, u64 index, int length,
+			     struct bio **biop, int *block_offset)
 {
 	struct squashfs_sb_info *msblk = sb->s_fs_info;
-	struct buffer_head *bh;
-
-	bh = sb_bread(sb, *cur_index);
-	if (bh == NULL)
-		return NULL;
-
-	if (msblk->devblksize - *offset == 1) {
-		*length = (unsigned char) bh->b_data[*offset];
-		put_bh(bh);
-		bh = sb_bread(sb, ++(*cur_index));
-		if (bh == NULL)
-			return NULL;
-		*length |= (unsigned char) bh->b_data[0] << 8;
-		*offset = 1;
-	} else {
-		*length = (unsigned char) bh->b_data[*offset] |
-			(unsigned char) bh->b_data[*offset + 1] << 8;
-		*offset += 2;
-
-		if (*offset == msblk->devblksize) {
-			put_bh(bh);
-			bh = sb_bread(sb, ++(*cur_index));
-			if (bh == NULL)
-				return NULL;
-			*offset = 0;
+	const u64 read_start = round_down(index, msblk->devblksize);
+	const sector_t block = read_start >> msblk->devblksize_log2;
+	const u64 read_end = round_up(index + length, msblk->devblksize);
+	const sector_t block_end = read_end >> msblk->devblksize_log2;
+	int offset = read_start - round_down(index, PAGE_SIZE);
+	int total_len = (block_end - block) << msblk->devblksize_log2;
+	const int page_count = DIV_ROUND_UP(total_len + offset, PAGE_SIZE);
+	int error, i;
+	struct bio *bio;
+
+	bio = bio_alloc(GFP_NOIO, page_count);
+	if (!bio)
+		return -ENOMEM;
+
+	bio_set_dev(bio, sb->s_bdev);
+	bio->bi_opf = READ;
+	bio->bi_iter.bi_sector = block * (msblk->devblksize >> SECTOR_SHIFT);
+
+	for (i = 0; i < page_count; ++i) {
+		unsigned int len =
+			min_t(unsigned int, PAGE_SIZE - offset, total_len);
+		struct page *page = alloc_page(GFP_NOIO);
+
+		if (!page) {
+			error = -ENOMEM;
+			goto out_free_bio;
+		}
+		if (!bio_add_page(bio, page, len, offset)) {
+			error = -EIO;
+			goto out_free_bio;
 		}
+		offset = 0;
+		total_len -= len;
 	}
 
-	return bh;
-}
+	error = submit_bio_wait(bio);
+	if (error)
+		goto out_free_bio;
 
+	*biop = bio;
+	*block_offset = index & ((1 << msblk->devblksize_log2) - 1);
+	return 0;
+
+out_free_bio:
+	bio_free_pages(bio);
+	bio_put(bio);
+	return error;
+}
 
 /*
  * Read and decompress a metadata block or datablock.  Length is non-zero
@@ -76,129 +136,88 @@ static struct buffer_head *get_block_length(struct super_block *sb,
  * algorithms).
  */
 int squashfs_read_data(struct super_block *sb, u64 index, int length,
-		u64 *next_index, struct squashfs_page_actor *output)
+		       u64 *next_index, struct squashfs_page_actor *output)
 {
 	struct squashfs_sb_info *msblk = sb->s_fs_info;
-	struct buffer_head **bh;
-	int offset = index & ((1 << msblk->devblksize_log2) - 1);
-	u64 cur_index = index >> msblk->devblksize_log2;
-	int bytes, compressed, b = 0, k = 0, avail, i;
-
-	bh = kcalloc(((output->length + msblk->devblksize - 1)
-		>> msblk->devblksize_log2) + 1, sizeof(*bh), GFP_KERNEL);
-	if (bh == NULL)
-		return -ENOMEM;
+	struct bio *bio = NULL;
+	int compressed;
+	int res;
+	int offset;
 
 	if (length) {
 		/*
 		 * Datablock.
 		 */
-		bytes = -offset;
 		compressed = SQUASHFS_COMPRESSED_BLOCK(length);
 		length = SQUASHFS_COMPRESSED_SIZE_BLOCK(length);
-		if (next_index)
-			*next_index = index + length;
-
 		TRACE("Block @ 0x%llx, %scompressed size %d, src size %d\n",
 			index, compressed ? "" : "un", length, output->length);
-
-		if (length < 0 || length > output->length ||
-				(index + length) > msblk->bytes_used)
-			goto read_failure;
-
-		for (b = 0; bytes < length; b++, cur_index++) {
-			bh[b] = sb_getblk(sb, cur_index);
-			if (bh[b] == NULL)
-				goto block_release;
-			bytes += msblk->devblksize;
-		}
-		ll_rw_block(REQ_OP_READ, 0, b, bh);
 	} else {
 		/*
 		 * Metadata block.
 		 */
-		if ((index + 2) > msblk->bytes_used)
-			goto read_failure;
+		const u8 *data;
+		struct bvec_iter_all iter_all = {};
+		struct bio_vec *bvec = bvec_init_iter_all(&iter_all);
 
-		bh[0] = get_block_length(sb, &cur_index, &offset, &length);
-		if (bh[0] == NULL)
-			goto read_failure;
-		b = 1;
+		if (index + 2 > msblk->bytes_used) {
+			res = -EIO;
+			goto out;
+		}
+		res = squashfs_bio_read(sb, index, 2, &bio, &offset);
+		if (res)
+			goto out;
+
+		if (WARN_ON_ONCE(!bio_next_segment(bio, &iter_all))) {
+			res = -EIO;
+			goto out_free_bio;
+		}
+		/* Extract the length of the metadata block */
+		data = page_address(bvec->bv_page) + bvec->bv_offset;
+		length = data[offset];
+		if (offset <= bvec->bv_len - 1) {
+			length |= data[offset + 1] << 8;
+		} else {
+			if (WARN_ON_ONCE(!bio_next_segment(bio, &iter_all))) {
+				res = -EIO;
+				goto out_free_bio;
+			}
+			data = page_address(bvec->bv_page) + bvec->bv_offset;
+			length |= data[0] << 8;
+		}
+		bio_free_pages(bio);
+		bio_put(bio);
 
-		bytes = msblk->devblksize - offset;
 		compressed = SQUASHFS_COMPRESSED(length);
 		length = SQUASHFS_COMPRESSED_SIZE(length);
-		if (next_index)
-			*next_index = index + length + 2;
+		index += 2;
 
 		TRACE("Block @ 0x%llx, %scompressed size %d\n", index,
-				compressed ? "" : "un", length);
-
-		if (length < 0 || length > output->length ||
-					(index + length) > msblk->bytes_used)
-			goto block_release;
-
-		for (; bytes < length; b++) {
-			bh[b] = sb_getblk(sb, ++cur_index);
-			if (bh[b] == NULL)
-				goto block_release;
-			bytes += msblk->devblksize;
-		}
-		ll_rw_block(REQ_OP_READ, 0, b - 1, bh + 1);
+		      compressed ? "" : "un", length);
 	}
+	if (next_index)
+		*next_index = index + length;
 
-	for (i = 0; i < b; i++) {
-		wait_on_buffer(bh[i]);
-		if (!buffer_uptodate(bh[i]))
-			goto block_release;
-	}
+	res = squashfs_bio_read(sb, index, length, &bio, &offset);
+	if (res)
+		goto out;
 
 	if (compressed) {
-		if (!msblk->stream)
-			goto read_failure;
-		length = squashfs_decompress(msblk, bh, b, offset, length,
-			output);
-		if (length < 0)
-			goto read_failure;
-	} else {
-		/*
-		 * Block is uncompressed.
-		 */
-		int in, pg_offset = 0;
-		void *data = squashfs_first_page(output);
-
-		for (bytes = length; k < b; k++) {
-			in = min(bytes, msblk->devblksize - offset);
-			bytes -= in;
-			while (in) {
-				if (pg_offset == PAGE_SIZE) {
-					data = squashfs_next_page(output);
-					pg_offset = 0;
-				}
-				avail = min_t(int, in, PAGE_SIZE -
-						pg_offset);
-				memcpy(data + pg_offset, bh[k]->b_data + offset,
-						avail);
-				in -= avail;
-				pg_offset += avail;
-				offset += avail;
-			}
-			offset = 0;
-			put_bh(bh[k]);
+		if (!msblk->stream) {
+			res = -EIO;
+			goto out_free_bio;
 		}
-		squashfs_finish_page(output);
+		res = squashfs_decompress(msblk, bio, offset, length, output);
+	} else {
+		res = copy_bio_to_actor(bio, output, offset, length);
 	}
 
-	kfree(bh);
-	return length;
-
-block_release:
-	for (; k < b; k++)
-		put_bh(bh[k]);
+out_free_bio:
+	bio_free_pages(bio);
+	bio_put(bio);
+out:
+	if (res < 0)
+		ERROR("Failed to read block 0x%llx: %d\n", index, res);
 
-read_failure:
-	ERROR("squashfs_read_data failed to read block 0x%llx\n",
-					(unsigned long long) index);
-	kfree(bh);
-	return -EIO;
+	return res;
 }

fs/squashfs/decompressor.h

@@ -10,13 +10,14 @@
  * decompressor.h
  */
 
+#include <linux/bio.h>
+
 struct squashfs_decompressor {
 	void	*(*init)(struct squashfs_sb_info *, void *);
 	void	*(*comp_opts)(struct squashfs_sb_info *, void *, int);
 	void	(*free)(void *);
 	int	(*decompress)(struct squashfs_sb_info *, void *,
-		struct buffer_head **, int, int, int,
-		struct squashfs_page_actor *);
+		struct bio *, int, int, struct squashfs_page_actor *);
 	int	id;
 	char	*name;
 	int	supported;

fs/squashfs/decompressor_multi.c

@@ -6,7 +6,7 @@
 #include <linux/types.h>
 #include <linux/mutex.h>
 #include <linux/slab.h>
-#include <linux/buffer_head.h>
+#include <linux/bio.h>
 #include <linux/sched.h>
 #include <linux/wait.h>
 #include <linux/cpumask.h>
@@ -180,14 +180,15 @@ static struct decomp_stream *get_decomp_stream(struct squashfs_sb_info *msblk,
 }
 
 
-int squashfs_decompress(struct squashfs_sb_info *msblk, struct buffer_head **bh,
-	int b, int offset, int length, struct squashfs_page_actor *output)
+int squashfs_decompress(struct squashfs_sb_info *msblk, struct bio *bio,
+			int offset, int length,
+			struct squashfs_page_actor *output)
 {
 	int res;
 	struct squashfs_stream *stream = msblk->stream;
 	struct decomp_stream *decomp_stream = get_decomp_stream(msblk, stream);
 	res = msblk->decompressor->decompress(msblk, decomp_stream->stream,
-		bh, b, offset, length, output);
+		bio, offset, length, output);
 	put_decomp_stream(decomp_stream, stream);
 	if (res < 0)
 		ERROR("%s decompression failed, data probably corrupt\n",

fs/squashfs/decompressor_multi_percpu.c

@@ -72,14 +72,17 @@ void squashfs_decompressor_destroy(struct squashfs_sb_info *msblk)
 	}
 }
 
-int squashfs_decompress(struct squashfs_sb_info *msblk, struct buffer_head **bh,
-	int b, int offset, int length, struct squashfs_page_actor *output)
+int squashfs_decompress(struct squashfs_sb_info *msblk, struct bio *bio,
+	int offset, int length, struct squashfs_page_actor *output)
 {
-	struct squashfs_stream __percpu *percpu =
-			(struct squashfs_stream __percpu *) msblk->stream;
-	struct squashfs_stream *stream = get_cpu_ptr(percpu);
-	int res = msblk->decompressor->decompress(msblk, stream->stream, bh, b,
-		offset, length, output);
+	struct squashfs_stream __percpu *percpu;
+	struct squashfs_stream *stream;
+	int res;
+
+	percpu = (struct squashfs_stream __percpu *)msblk->stream;
+	stream = get_cpu_ptr(percpu);
+	res = msblk->decompressor->decompress(msblk, stream->stream, bio,
+					      offset, length, output);
 	put_cpu_ptr(stream);
 
 	if (res < 0)