squashfs: extend "page actor" to handle missing pages

Patch series "Squashfs: handle missing pages decompressing into page
cache".

This patchset enables Squashfs to handle missing pages when directly
decompressing datablocks into the page cache.

Previously if the full set of pages needed was not available, Squashfs
would have to fall back to using an intermediate buffer (the older
method), which is slower, involving a memcopy, and it introduces
contention on a shared buffer.

The first patch extends the "page actor" code to handle missing pages.

The second patch updates Squashfs_readpage_block() to use the new
functionality, and removes the code that falls back to using an
intermediate buffer.

This patchset is independent of the readahead work, and it is standalone.
It can be merged on its own.

But the readahead patch for efficiency also needs this patch-set.


This patch (of 2):

This patch extends the "page actor" code to handle missing pages.

Previously if the full set of pages needed to decompress a Squashfs
datablock was unavailable, this would cause decompression to fail on the
missing pages.

In this case direct decompression into the page cache could not be
achieved and the code would fall back to using the older intermediate
buffer method.

With this patch, direct decompression into the page cache can be achieved
with missing pages.

For "multi-shot" decompressors (zlib, xz, zstd), the page actor will
allocate a temporary buffer which is passed to the decompressor, and then
freed by the page actor.

For "single shot" decompressors (lz4, lzo) which decompress into a
contiguous "bounce buffer", and which is then copied into the page cache,
it would be pointless to allocate a temporary buffer, memcpy into it, and
then free it. For these decompressors -ENOMEM is returned, which
signifies that the memcpy for that page should be skipped.

This also happens if the data block is uncompressed.

Link: https://lkml.kernel.org/r/[e-mail]
Link: https://lkml.kernel.org/r/[e-mail]
Signed-off-by: Phillip Lougher <[e-mail]>
Cc: Matthew Wilcox (Oracle) <[e-mail]>
Cc: Hsin-Yi Wang <[e-mail]>
Cc: Xiongwei Song <[e-mail]>
Signed-off-by: Andrew Morton <[e-mail]>

fs/squashfs/block.c

···
 			     struct squashfs_page_actor *actor,
 			     int offset, int req_length)
 {
-	void *actor_addr = squashfs_first_page(actor);
+	void *actor_addr;
 	struct bvec_iter_all iter_all = {};
 	struct bio_vec *bvec = bvec_init_iter_all(&iter_all);
 	int copied_bytes = 0;
 	int actor_offset = 0;
+
+	squashfs_actor_nobuff(actor);
+	actor_addr = squashfs_first_page(actor);
 
 	if (WARN_ON_ONCE(!bio_next_segment(bio, &iter_all)))
 		return 0;
···
 
 		bytes_to_copy = min_t(int, bytes_to_copy,
 				      req_length - copied_bytes);
-		memcpy(actor_addr + actor_offset, bvec_virt(bvec) + offset,
-		       bytes_to_copy);
+		if (!IS_ERR(actor_addr))
+			memcpy(actor_addr + actor_offset, bvec_virt(bvec) +
+					offset, bytes_to_copy);
 
 		actor_offset += bytes_to_copy;
 		copied_bytes += bytes_to_copy;

fs/squashfs/decompressor.h

···
 		struct bio *, int, int, struct squashfs_page_actor *);
 	int	id;
 	char	*name;
+	int	alloc_buffer;
 	int	supported;
 };
 

fs/squashfs/file_direct.c

···
 	if (page == NULL)
 		return res;
 
-	/*
-	 * Create a "page actor" which will kmap and kunmap the
-	 * page cache pages appropriately within the decompressor
-	 */
-	actor = squashfs_page_actor_init_special(page, pages, 0);
-	if (actor == NULL)
-		goto out;
-
 	/* Try to grab all the pages covered by the Squashfs block */
 	for (missing_pages = 0, i = 0, n = start_index; i < pages; i++, n++) {
 		page[i] = (n == target_page->index) ? target_page :
···
 		goto out;
 	}
 
+	/*
+	 * Create a "page actor" which will kmap and kunmap the
+	 * page cache pages appropriately within the decompressor
+	 */
+	actor = squashfs_page_actor_init_special(msblk, page, pages, 0);
+	if (actor == NULL)
+		goto out;
+
 	/* Decompress directly into the page cache buffers */
 	res = squashfs_read_data(inode->i_sb, block, bsize, NULL, actor);
+
+	kfree(actor);
+
 	if (res < 0)
 		goto mark_errored;
 
···
 			put_page(page[i]);
 	}
 
-	kfree(actor);
 	kfree(page);
 
 	return 0;
···
 	}
 
 out:
-	kfree(actor);
 	kfree(page);
 	return res;
 }

fs/squashfs/lz4_wrapper.c

···
 	buff = stream->output;
 	while (data) {
 		if (bytes <= PAGE_SIZE) {
-			memcpy(data, buff, bytes);
+			if (!IS_ERR(data))
+				memcpy(data, buff, bytes);
 			break;
 		}
-		memcpy(data, buff, PAGE_SIZE);
+		if (!IS_ERR(data))
+			memcpy(data, buff, PAGE_SIZE);
 		buff += PAGE_SIZE;
 		bytes -= PAGE_SIZE;
 		data = squashfs_next_page(output);
···
 	.decompress = lz4_uncompress,
 	.id = LZ4_COMPRESSION,
 	.name = "lz4",
+	.alloc_buffer = 0,
 	.supported = 1
 };

fs/squashfs/lzo_wrapper.c

···
 	buff = stream->output;
 	while (data) {
 		if (bytes <= PAGE_SIZE) {
-			memcpy(data, buff, bytes);
+			if (!IS_ERR(data))
+				memcpy(data, buff, bytes);
 			break;
 		} else {
-			memcpy(data, buff, PAGE_SIZE);
+			if (!IS_ERR(data))
+				memcpy(data, buff, PAGE_SIZE);
 			buff += PAGE_SIZE;
 			bytes -= PAGE_SIZE;
 			data = squashfs_next_page(output);
···
 	.decompress = lzo_uncompress,
 	.id = LZO_COMPRESSION,
 	.name = "lzo",
+	.alloc_buffer = 0,
 	.supported = 1
 };

fs/squashfs/page_actor.c

···
 #include <linux/kernel.h>
 #include <linux/slab.h>
 #include <linux/pagemap.h>
+#include "squashfs_fs_sb.h"
+#include "decompressor.h"
 #include "page_actor.h"
 
 /*
···
 }
 
 /* Implementation of page_actor for decompressing directly into page cache. */
+static void *handle_next_page(struct squashfs_page_actor *actor)
+{
+	int max_pages = (actor->length + PAGE_SIZE - 1) >> PAGE_SHIFT;
+
+	if (actor->returned_pages == max_pages)
+		return NULL;
+
+	if ((actor->next_page == actor->pages) ||
+			(actor->next_index != actor->page[actor->next_page]->index)) {
+		if (actor->alloc_buffer) {
+			void *tmp_buffer = kmalloc(PAGE_SIZE, GFP_KERNEL);
+
+			if (tmp_buffer) {
+				actor->tmp_buffer = tmp_buffer;
+				actor->next_index++;
+				actor->returned_pages++;
+				return tmp_buffer;
+			}
+		}
+
+		actor->next_index++;
+		actor->returned_pages++;
+		return ERR_PTR(-ENOMEM);
+	}
+
+	actor->next_index++;
+	actor->returned_pages++;
+	return actor->pageaddr = kmap_local_page(actor->page[actor->next_page++]);
+}
+
 static void *direct_first_page(struct squashfs_page_actor *actor)
 {
-	actor->next_page = 1;
-	return actor->pageaddr = kmap_atomic(actor->page[0]);
+	return handle_next_page(actor);
 }
 
 static void *direct_next_page(struct squashfs_page_actor *actor)
 {
 	if (actor->pageaddr)
-		kunmap_atomic(actor->pageaddr);
+		kunmap_local(actor->pageaddr);
 
-	return actor->pageaddr = actor->next_page == actor->pages ? NULL :
-		kmap_atomic(actor->page[actor->next_page++]);
+	kfree(actor->tmp_buffer);
+	actor->pageaddr = actor->tmp_buffer = NULL;
+
+	return handle_next_page(actor);
 }
 
 static void direct_finish_page(struct squashfs_page_actor *actor)
 {
 	if (actor->pageaddr)
-		kunmap_atomic(actor->pageaddr);
+		kunmap_local(actor->pageaddr);
+
+	kfree(actor->tmp_buffer);
 }
 
-struct squashfs_page_actor *squashfs_page_actor_init_special(struct page **page,
-	int pages, int length)
+struct squashfs_page_actor *squashfs_page_actor_init_special(struct squashfs_sb_info *msblk,
+	struct page **page, int pages, int length)
 {
 	struct squashfs_page_actor *actor = kmalloc(sizeof(*actor), GFP_KERNEL);
 
···
 	actor->page = page;
 	actor->pages = pages;
 	actor->next_page = 0;
+	actor->returned_pages = 0;
+	actor->next_index = page[0]->index & ~((1 << (msblk->block_log - PAGE_SHIFT)) - 1);
 	actor->pageaddr = NULL;
+	actor->tmp_buffer = NULL;
+	actor->alloc_buffer = msblk->decompressor->alloc_buffer;
 	actor->squashfs_first_page = direct_first_page;
 	actor->squashfs_next_page = direct_next_page;
 	actor->squashfs_finish_page = direct_finish_page;

fs/squashfs/page_actor.h

···
 {
 	/* empty */
 }
+
+static inline void squashfs_actor_nobuff(struct squashfs_page_actor *actor)
+{
+	/* empty */
+}
 #else
 struct squashfs_page_actor {
 	union {
···
 		struct page	**page;
 	};
 	void	*pageaddr;
+	void	*tmp_buffer;
 	void    *(*squashfs_first_page)(struct squashfs_page_actor *);
 	void    *(*squashfs_next_page)(struct squashfs_page_actor *);
 	void    (*squashfs_finish_page)(struct squashfs_page_actor *);
 	int	pages;
 	int	length;
 	int	next_page;
+	int	alloc_buffer;
+	int	returned_pages;
+	pgoff_t	next_index;
 };
 
-extern struct squashfs_page_actor *squashfs_page_actor_init(void **, int, int);
-extern struct squashfs_page_actor *squashfs_page_actor_init_special(struct page
-							 **, int, int);
+extern struct squashfs_page_actor *squashfs_page_actor_init(void **buffer,
+				int pages, int length);
+extern struct squashfs_page_actor *squashfs_page_actor_init_special(
+				struct squashfs_sb_info *msblk,
+				struct page **page, int pages, int length);
 static inline void *squashfs_first_page(struct squashfs_page_actor *actor)
 {
 	return actor->squashfs_first_page(actor);
···
 static inline void squashfs_finish_page(struct squashfs_page_actor *actor)
 {
 	actor->squashfs_finish_page(actor);
+}
+static inline void squashfs_actor_nobuff(struct squashfs_page_actor *actor)
+{
+	actor->alloc_buffer = 0;
 }
 #endif
 #endif

fs/squashfs/xz_wrapper.c

···
 	stream->buf.out_pos = 0;
 	stream->buf.out_size = PAGE_SIZE;
 	stream->buf.out = squashfs_first_page(output);
+	if (IS_ERR(stream->buf.out)) {
+		error = PTR_ERR(stream->buf.out);
+		goto finish;
+	}
 
 	for (;;) {
 		enum xz_ret xz_err;
···
 
 		if (stream->buf.out_pos == stream->buf.out_size) {
 			stream->buf.out = squashfs_next_page(output);
-			if (stream->buf.out != NULL) {
+			if (IS_ERR(stream->buf.out)) {
+				error = PTR_ERR(stream->buf.out);
+				break;
+			} else if (stream->buf.out != NULL) {
 				stream->buf.out_pos = 0;
 				total += PAGE_SIZE;
 			}
···
 		}
 	}
 
+finish:
 	squashfs_finish_page(output);
 
 	return error ? error : total + stream->buf.out_pos;
···
 	.decompress = squashfs_xz_uncompress,
 	.id = XZ_COMPRESSION,
 	.name = "xz",
+	.alloc_buffer = 1,
 	.supported = 1
 };

fs/squashfs/zlib_wrapper.c

···
 	stream->next_out = squashfs_first_page(output);
 	stream->avail_in = 0;
 
+	if (IS_ERR(stream->next_out)) {
+		error = PTR_ERR(stream->next_out);
+		goto finish;
+	}
+
 	for (;;) {
 		int zlib_err;
 
···
 
 		if (stream->avail_out == 0) {
 			stream->next_out = squashfs_next_page(output);
-			if (stream->next_out != NULL)
+			if (IS_ERR(stream->next_out)) {
+				error = PTR_ERR(stream->next_out);
+				break;
+			} else if (stream->next_out != NULL)
 				stream->avail_out = PAGE_SIZE;
 		}
 
···
 		}
 	}
 
+finish:
 	squashfs_finish_page(output);
 
 	if (!error)
···
 	.decompress = zlib_uncompress,
 	.id = ZLIB_COMPRESSION,
 	.name = "zlib",
+	.alloc_buffer = 1,
 	.supported = 1
 };
 

fs/squashfs/zstd_wrapper.c

···
 
 	out_buf.size = PAGE_SIZE;
 	out_buf.dst = squashfs_first_page(output);
+	if (IS_ERR(out_buf.dst)) {
+		error = PTR_ERR(out_buf.dst);
+		goto finish;
+	}
 
 	for (;;) {
 		size_t zstd_err;
···
 
 		if (out_buf.pos == out_buf.size) {
 			out_buf.dst = squashfs_next_page(output);
-			if (out_buf.dst == NULL) {
+			if (IS_ERR(out_buf.dst)) {
+				error = PTR_ERR(out_buf.dst);
+				break;
+			} else if (out_buf.dst == NULL) {
 				/* Shouldn't run out of pages
 				 * before stream is done.
 				 */
···
 		}
 	}
 
+finish:
+
 	squashfs_finish_page(output);
 
 	return error ? error : total_out;
···
 	.decompress = zstd_uncompress,
 	.id = ZSTD_COMPRESSION,
 	.name = "zstd",
+	.alloc_buffer = 1,
 	.supported = 1
 };