Merge tag 'erofs-for-6.13-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/xiang/erofs

Pull erofs updates from Gao Xiang:
 "There is no outstanding feature for this cycle. The most useful
  changes are SEEK_{DATA,HOLE} support and some decompression
  micro-optimization. Other than those, there are some bugfixes and
  cleanups as usual:

   - Add SEEK_{DATA,HOLE} support

   - Free redundant pclusters if no cached compressed data is valid

   - Add sysfs entry to drop internal caches

   - Several bugfixes & cleanups"

* tag 'erofs-for-6.13-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/xiang/erofs:
  erofs: handle NONHEAD !delta[1] lclusters gracefully
  erofs: clarify direct I/O support
  erofs: fix blksize < PAGE_SIZE for file-backed mounts
  erofs: get rid of `buf->kmap_type`
  erofs: fix file-backed mounts over FUSE
  erofs: simplify definition of the log functions
  erofs: add sysfs node to drop internal caches
  erofs: free pclusters if no cached folio is attached
  erofs: sunset `struct erofs_workgroup`
  erofs: move erofs_workgroup operations into zdata.c
  erofs: get rid of erofs_{find,insert}_workgroup
  erofs: add SEEK_{DATA,HOLE} support

Author: torvalds

Documentation/ABI/testing/sysfs-fs-erofs

@@ -16,3 +16,14 @@ Description:	Control strategy of sync decompression:
 		  readahead on atomic contexts only.
 		- 1 (force on): enable for readpage and readahead.
 		- 2 (force off): disable for all situations.
+
+What:		/sys/fs/erofs/<disk>/drop_caches
+Date:		November 2024
+Contact:	"Guo Chunhai" <guochunhai@vivo.com>
+Description:	Writing to this will drop compression-related caches,
+		currently used to drop in-memory pclusters and cached
+		compressed folios:
+
+		- 1 : invalidate cached compressed folios
+		- 2 : drop in-memory pclusters
+		- 3 : drop in-memory pclusters and cached compressed folios

fs/erofs/data.c

@@ -10,10 +10,10 @@
 
 void erofs_unmap_metabuf(struct erofs_buf *buf)
 {
-	if (buf->kmap_type == EROFS_KMAP)
-		kunmap_local(buf->base);
+	if (!buf->base)
+		return;
+	kunmap_local(buf->base);
 	buf->base = NULL;
-	buf->kmap_type = EROFS_NO_KMAP;
 }
 
 void erofs_put_metabuf(struct erofs_buf *buf)
@@ -38,20 +38,13 @@ void *erofs_bread(struct erofs_buf *buf, erofs_off_t offset,
 	}
 	if (!folio || !folio_contains(folio, index)) {
 		erofs_put_metabuf(buf);
-		folio = read_mapping_folio(buf->mapping, index, NULL);
+		folio = read_mapping_folio(buf->mapping, index, buf->file);
 		if (IS_ERR(folio))
 			return folio;
 	}
 	buf->page = folio_file_page(folio, index);
-
-	if (buf->kmap_type == EROFS_NO_KMAP) {
-		if (type == EROFS_KMAP)
-			buf->base = kmap_local_page(buf->page);
-		buf->kmap_type = type;
-	} else if (buf->kmap_type != type) {
-		DBG_BUGON(1);
-		return ERR_PTR(-EFAULT);
-	}
+	if (!buf->base && type == EROFS_KMAP)
+		buf->base = kmap_local_page(buf->page);
 	if (type == EROFS_NO_KMAP)
 		return NULL;
 	return buf->base + (offset & ~PAGE_MASK);
@@ -61,9 +54,11 @@ void erofs_init_metabuf(struct erofs_buf *buf, struct super_block *sb)
 {
 	struct erofs_sb_info *sbi = EROFS_SB(sb);
 
-	if (erofs_is_fileio_mode(sbi))
-		buf->mapping = file_inode(sbi->fdev)->i_mapping;
-	else if (erofs_is_fscache_mode(sb))
+	buf->file = NULL;
+	if (erofs_is_fileio_mode(sbi)) {
+		buf->file = sbi->fdev;		/* some fs like FUSE needs it */
+		buf->mapping = buf->file->f_mapping;
+	} else if (erofs_is_fscache_mode(sb))
 		buf->mapping = sbi->s_fscache->inode->i_mapping;
 	else
 		buf->mapping = sb->s_bdev->bd_mapping;
@@ -350,7 +345,6 @@ static int erofs_iomap_end(struct inode *inode, loff_t pos, loff_t length,
 		struct erofs_buf buf = {
 			.page = kmap_to_page(ptr),
 			.base = ptr,
-			.kmap_type = EROFS_KMAP,
 		};
 
 		DBG_BUGON(iomap->type != IOMAP_INLINE);
@@ -411,22 +405,9 @@ static ssize_t erofs_file_read_iter(struct kiocb *iocb, struct iov_iter *to)
 	if (IS_DAX(inode))
 		return dax_iomap_rw(iocb, to, &erofs_iomap_ops);
 #endif
-	if (iocb->ki_flags & IOCB_DIRECT) {
-		struct block_device *bdev = inode->i_sb->s_bdev;
-		unsigned int blksize_mask;
-
-		if (bdev)
-			blksize_mask = bdev_logical_block_size(bdev) - 1;
-		else
-			blksize_mask = i_blocksize(inode) - 1;
-
-		if ((iocb->ki_pos | iov_iter_count(to) |
-		     iov_iter_alignment(to)) & blksize_mask)
-			return -EINVAL;
-
+	if ((iocb->ki_flags & IOCB_DIRECT) && inode->i_sb->s_bdev)
 		return iomap_dio_rw(iocb, to, &erofs_iomap_ops,
 				    NULL, 0, NULL, 0);
-	}
 	return filemap_read(iocb, to, 0);
 }
 
@@ -473,8 +454,32 @@ static int erofs_file_mmap(struct file *file, struct vm_area_struct *vma)
 #define erofs_file_mmap	generic_file_readonly_mmap
 #endif
 
+static loff_t erofs_file_llseek(struct file *file, loff_t offset, int whence)
+{
+	struct inode *inode = file->f_mapping->host;
+	const struct iomap_ops *ops = &erofs_iomap_ops;
+
+	if (erofs_inode_is_data_compressed(EROFS_I(inode)->datalayout))
+#ifdef CONFIG_EROFS_FS_ZIP
+		ops = &z_erofs_iomap_report_ops;
+#else
+		return generic_file_llseek(file, offset, whence);
+#endif
+
+	if (whence == SEEK_HOLE)
+		offset = iomap_seek_hole(inode, offset, ops);
+	else if (whence == SEEK_DATA)
+		offset = iomap_seek_data(inode, offset, ops);
+	else
+		return generic_file_llseek(file, offset, whence);
+
+	if (offset < 0)
+		return offset;
+	return vfs_setpos(file, offset, inode->i_sb->s_maxbytes);
+}
+
 const struct file_operations erofs_file_fops = {
-	.llseek		= generic_file_llseek,
+	.llseek		= erofs_file_llseek,
 	.read_iter	= erofs_file_read_iter,
 	.mmap		= erofs_file_mmap,
 	.get_unmapped_area = thp_get_unmapped_area,

fs/erofs/inode.c

@@ -318,6 +318,7 @@ int erofs_getattr(struct mnt_idmap *idmap, const struct path *path,
 		  unsigned int query_flags)
 {
 	struct inode *const inode = d_inode(path->dentry);
+	struct block_device *bdev = inode->i_sb->s_bdev;
 	bool compressed =
 		erofs_inode_is_data_compressed(EROFS_I(inode)->datalayout);
 
@@ -330,15 +331,14 @@ int erofs_getattr(struct mnt_idmap *idmap, const struct path *path,
 	/*
 	 * Return the DIO alignment restrictions if requested.
 	 *
-	 * In EROFS, STATX_DIOALIGN is not supported in ondemand mode and
-	 * compressed files, so in these cases we report no DIO support.
+	 * In EROFS, STATX_DIOALIGN is only supported in bdev-based mode
+	 * and uncompressed inodes, otherwise we report no DIO support.
 	 */
 	if ((request_mask & STATX_DIOALIGN) && S_ISREG(inode->i_mode)) {
 		stat->result_mask |= STATX_DIOALIGN;
-		if (!erofs_is_fscache_mode(inode->i_sb) && !compressed) {
-			stat->dio_mem_align =
-				bdev_logical_block_size(inode->i_sb->s_bdev);
-			stat->dio_offset_align = stat->dio_mem_align;
+		if (bdev && !compressed) {
+			stat->dio_mem_align = bdev_dma_alignment(bdev) + 1;
+			stat->dio_offset_align = bdev_logical_block_size(bdev);
 		}
 	}
 	generic_fillattr(idmap, request_mask, inode, stat);

fs/erofs/internal.h

@@ -20,18 +20,12 @@
 #include <linux/iomap.h>
 #include "erofs_fs.h"
 
-/* redefine pr_fmt "erofs: " */
-#undef pr_fmt
-#define pr_fmt(fmt) "erofs: " fmt
-
-__printf(3, 4) void _erofs_err(struct super_block *sb,
-			       const char *function, const char *fmt, ...);
+__printf(2, 3) void _erofs_printk(struct super_block *sb, const char *fmt, ...);
 #define erofs_err(sb, fmt, ...)	\
-	_erofs_err(sb, __func__, fmt "\n", ##__VA_ARGS__)
-__printf(3, 4) void _erofs_info(struct super_block *sb,
-			       const char *function, const char *fmt, ...);
+	_erofs_printk(sb, KERN_ERR fmt "\n", ##__VA_ARGS__)
 #define erofs_info(sb, fmt, ...) \
-	_erofs_info(sb, __func__, fmt "\n", ##__VA_ARGS__)
+	_erofs_printk(sb, KERN_INFO fmt "\n", ##__VA_ARGS__)
+
 #ifdef CONFIG_EROFS_FS_DEBUG
 #define DBG_BUGON               BUG_ON
 #else
@@ -208,22 +202,16 @@ enum {
 	EROFS_ZIP_CACHE_READAROUND
 };
 
-/* basic unit of the workstation of a super_block */
-struct erofs_workgroup {
-	pgoff_t index;
-	struct lockref lockref;
-};
-
 enum erofs_kmap_type {
 	EROFS_NO_KMAP,		/* don't map the buffer */
 	EROFS_KMAP,		/* use kmap_local_page() to map the buffer */
 };
 
 struct erofs_buf {
 	struct address_space *mapping;
+	struct file *file;
 	struct page *page;
 	void *base;
-	enum erofs_kmap_type kmap_type;
 };
 #define __EROFS_BUF_INITIALIZER	((struct erofs_buf){ .page = NULL })
 
@@ -456,20 +444,17 @@ static inline void erofs_pagepool_add(struct page **pagepool, struct page *page)
 void erofs_release_pages(struct page **pagepool);
 
 #ifdef CONFIG_EROFS_FS_ZIP
-void erofs_workgroup_put(struct erofs_workgroup *grp);
-struct erofs_workgroup *erofs_find_workgroup(struct super_block *sb,
-					     pgoff_t index);
-struct erofs_workgroup *erofs_insert_workgroup(struct super_block *sb,
-					       struct erofs_workgroup *grp);
-void erofs_workgroup_free_rcu(struct erofs_workgroup *grp);
+#define MNGD_MAPPING(sbi)	((sbi)->managed_cache->i_mapping)
+
+extern atomic_long_t erofs_global_shrink_cnt;
 void erofs_shrinker_register(struct super_block *sb);
 void erofs_shrinker_unregister(struct super_block *sb);
 int __init erofs_init_shrinker(void);
 void erofs_exit_shrinker(void);
 int __init z_erofs_init_subsystem(void);
 void z_erofs_exit_subsystem(void);
-int erofs_try_to_free_all_cached_folios(struct erofs_sb_info *sbi,
-					struct erofs_workgroup *egrp);
+unsigned long z_erofs_shrink_scan(struct erofs_sb_info *sbi,
+				  unsigned long nr_shrink);
 int z_erofs_map_blocks_iter(struct inode *inode, struct erofs_map_blocks *map,
 			    int flags);
 void *z_erofs_get_gbuf(unsigned int requiredpages);

fs/erofs/super.c

@@ -18,37 +18,22 @@
 
 static struct kmem_cache *erofs_inode_cachep __read_mostly;
 
-void _erofs_err(struct super_block *sb, const char *func, const char *fmt, ...)
+void _erofs_printk(struct super_block *sb, const char *fmt, ...)
 {
 	struct va_format vaf;
 	va_list args;
+	int level;
 
 	va_start(args, fmt);
 
-	vaf.fmt = fmt;
+	level = printk_get_level(fmt);
+	vaf.fmt = printk_skip_level(fmt);
 	vaf.va = &args;
-
 	if (sb)
-		pr_err("(device %s): %s: %pV", sb->s_id, func, &vaf);
+		printk("%c%cerofs (device %s): %pV",
+				KERN_SOH_ASCII, level, sb->s_id, &vaf);
 	else
-		pr_err("%s: %pV", func, &vaf);
-	va_end(args);
-}
-
-void _erofs_info(struct super_block *sb, const char *func, const char *fmt, ...)
-{
-	struct va_format vaf;
-	va_list args;
-
-	va_start(args, fmt);
-
-	vaf.fmt = fmt;
-	vaf.va = &args;
-
-	if (sb)
-		pr_info("(device %s): %pV", sb->s_id, &vaf);
-	else
-		pr_info("%pV", &vaf);
+		printk("%c%cerofs: %pV", KERN_SOH_ASCII, level, &vaf);
 	va_end(args);
 }
 
@@ -631,7 +616,11 @@ static int erofs_fc_fill_super(struct super_block *sb, struct fs_context *fc)
 			errorfc(fc, "unsupported blksize for fscache mode");
 			return -EINVAL;
 		}
-		if (!sb_set_blocksize(sb, 1 << sbi->blkszbits)) {
+
+		if (erofs_is_fileio_mode(sbi)) {
+			sb->s_blocksize = 1 << sbi->blkszbits;
+			sb->s_blocksize_bits = sbi->blkszbits;
+		} else if (!sb_set_blocksize(sb, 1 << sbi->blkszbits)) {
 			errorfc(fc, "failed to set erofs blksize");
 			return -EINVAL;
 		}

fs/erofs/sysfs.c

@@ -10,6 +10,7 @@
 
 enum {
 	attr_feature,
+	attr_drop_caches,
 	attr_pointer_ui,
 	attr_pointer_bool,
 };
@@ -57,11 +58,13 @@ static struct erofs_attr erofs_attr_##_name = {			\
 
 #ifdef CONFIG_EROFS_FS_ZIP
 EROFS_ATTR_RW_UI(sync_decompress, erofs_mount_opts);
+EROFS_ATTR_FUNC(drop_caches, 0200);
 #endif
 
 static struct attribute *erofs_attrs[] = {
 #ifdef CONFIG_EROFS_FS_ZIP
 	ATTR_LIST(sync_decompress),
+	ATTR_LIST(drop_caches),
 #endif
 	NULL,
 };
@@ -163,6 +166,20 @@ static ssize_t erofs_attr_store(struct kobject *kobj, struct attribute *attr,
 			return -EINVAL;
 		*(bool *)ptr = !!t;
 		return len;
+#ifdef CONFIG_EROFS_FS_ZIP
+	case attr_drop_caches:
+		ret = kstrtoul(skip_spaces(buf), 0, &t);
+		if (ret)
+			return ret;
+		if (t < 1 || t > 3)
+			return -EINVAL;
+
+		if (t & 2)
+			z_erofs_shrink_scan(sbi, ~0UL);
+		if (t & 1)
+			invalidate_mapping_pages(MNGD_MAPPING(sbi), 0, -1);
+		return len;
+#endif
 	}
 	return 0;
 }