diff options
author | Linus Torvalds <torvalds@linux-foundation.org> | 2021-11-09 19:46:32 +0100 |
---|---|---|
committer | Linus Torvalds <torvalds@linux-foundation.org> | 2021-11-09 19:46:32 +0100 |
commit | cdd39b0539c4271d4242bc780fb1f04e130c4377 (patch) | |
tree | f2b6209244fe559bd3beebd03f8e51b9a414cf9e | |
parent | Merge tag 'for-linus-5.16-ofs1' of git://git.kernel.org/pub/scm/linux/kernel/... (diff) | |
parent | fuse: fix page stealing (diff) | |
download | linux-cdd39b0539c4271d4242bc780fb1f04e130c4377.tar.xz linux-cdd39b0539c4271d4242bc780fb1f04e130c4377.zip |
Merge tag 'fuse-update-5.16' of git://git.kernel.org/pub/scm/linux/kernel/git/mszeredi/fuse
Pull fuse updates from Miklos Szeredi:
- Fix a possible of deadlock in case inode writeback is in progress
during dentry reclaim
- Fix a crash in case of page stealing
- Selectively invalidate cached attributes, possibly improving
performance
- Allow filesystems to disable data flushing from ->flush()
- Misc fixes and cleanups
* tag 'fuse-update-5.16' of git://git.kernel.org/pub/scm/linux/kernel/git/mszeredi/fuse: (23 commits)
fuse: fix page stealing
virtiofs: use strscpy for copying the queue name
fuse: add FOPEN_NOFLUSH
fuse: only update necessary attributes
fuse: take cache_mask into account in getattr
fuse: add cache_mask
fuse: move reverting attributes to fuse_change_attributes()
fuse: simplify local variables holding writeback cache state
fuse: cleanup code conditional on fc->writeback_cache
fuse: fix attr version comparison in fuse_read_update_size()
fuse: always invalidate attributes after writes
fuse: rename fuse_write_update_size()
fuse: don't bump attr_version in cached write
fuse: selective attribute invalidation
fuse: don't increment nlink in link()
fuse: decrement nlink on overwriting rename
fuse: simplify __fuse_write_file_get()
fuse: move fuse_invalidate_attr() into fuse_update_ctime()
fuse: delete redundant code
fuse: use kmap_local_page()
...
-rw-r--r-- | fs/fuse/dax.c | 5 | ||||
-rw-r--r-- | fs/fuse/dev.c | 24 | ||||
-rw-r--r-- | fs/fuse/dir.c | 128 | ||||
-rw-r--r-- | fs/fuse/file.c | 106 | ||||
-rw-r--r-- | fs/fuse/fuse_i.h | 17 | ||||
-rw-r--r-- | fs/fuse/inode.c | 45 | ||||
-rw-r--r-- | fs/fuse/ioctl.c | 4 | ||||
-rw-r--r-- | fs/fuse/readdir.c | 6 | ||||
-rw-r--r-- | fs/fuse/virtio_fs.c | 2 | ||||
-rw-r--r-- | fs/fuse/xattr.c | 10 | ||||
-rw-r--r-- | include/uapi/linux/fuse.h | 7 |
11 files changed, 203 insertions, 151 deletions
diff --git a/fs/fuse/dax.c b/fs/fuse/dax.c index 281d79f8b3d3..713818d74de6 100644 --- a/fs/fuse/dax.c +++ b/fs/fuse/dax.c @@ -732,11 +732,8 @@ static ssize_t fuse_dax_direct_write(struct kiocb *iocb, struct iov_iter *from) ssize_t ret; ret = fuse_direct_io(&io, from, &iocb->ki_pos, FUSE_DIO_WRITE); - if (ret < 0) - return ret; - fuse_invalidate_attr(inode); - fuse_write_update_size(inode, iocb->ki_pos); + fuse_write_update_attr(inode, iocb->ki_pos, ret); return ret; } diff --git a/fs/fuse/dev.c b/fs/fuse/dev.c index dde341a6388a..79f7eda49e06 100644 --- a/fs/fuse/dev.c +++ b/fs/fuse/dev.c @@ -756,7 +756,7 @@ static int fuse_copy_do(struct fuse_copy_state *cs, void **val, unsigned *size) { unsigned ncpy = min(*size, cs->len); if (val) { - void *pgaddr = kmap_atomic(cs->pg); + void *pgaddr = kmap_local_page(cs->pg); void *buf = pgaddr + cs->offset; if (cs->write) @@ -764,7 +764,7 @@ static int fuse_copy_do(struct fuse_copy_state *cs, void **val, unsigned *size) else memcpy(*val, buf, ncpy); - kunmap_atomic(pgaddr); + kunmap_local(pgaddr); *val += ncpy; } *size -= ncpy; @@ -847,6 +847,12 @@ static int fuse_try_move_page(struct fuse_copy_state *cs, struct page **pagep) replace_page_cache_page(oldpage, newpage); + /* + * Release while we have extra ref on stolen page. Otherwise + * anon_pipe_buf_release() might think the page can be reused. + */ + pipe_buf_release(cs->pipe, buf); + get_page(newpage); if (!(buf->flags & PIPE_BUF_FLAG_LRU)) @@ -949,10 +955,10 @@ static int fuse_copy_page(struct fuse_copy_state *cs, struct page **pagep, } } if (page) { - void *mapaddr = kmap_atomic(page); + void *mapaddr = kmap_local_page(page); void *buf = mapaddr + offset; offset += fuse_copy_do(cs, &buf, &count); - kunmap_atomic(mapaddr); + kunmap_local(mapaddr); } else offset += fuse_copy_do(cs, NULL, &count); } @@ -1591,7 +1597,7 @@ static int fuse_notify_store(struct fuse_conn *fc, unsigned int size, end = outarg.offset + outarg.size; if (end > file_size) { file_size = end; - fuse_write_update_size(inode, file_size); + fuse_write_update_attr(inode, file_size, outarg.size); } num = outarg.size; @@ -2031,8 +2037,12 @@ static ssize_t fuse_dev_splice_write(struct pipe_inode_info *pipe, pipe_lock(pipe); out_free: - for (idx = 0; idx < nbuf; idx++) - pipe_buf_release(pipe, &bufs[idx]); + for (idx = 0; idx < nbuf; idx++) { + struct pipe_buffer *buf = &bufs[idx]; + + if (buf->ops) + pipe_buf_release(pipe, buf); + } pipe_unlock(pipe); kvfree(bufs); diff --git a/fs/fuse/dir.c b/fs/fuse/dir.c index d9b977c0f38d..0654bfedcbb0 100644 --- a/fs/fuse/dir.c +++ b/fs/fuse/dir.c @@ -116,7 +116,7 @@ u64 entry_attr_timeout(struct fuse_entry_out *o) return time_to_jiffies(o->attr_valid, o->attr_valid_nsec); } -static void fuse_invalidate_attr_mask(struct inode *inode, u32 mask) +void fuse_invalidate_attr_mask(struct inode *inode, u32 mask) { set_mask_bits(&get_fuse_inode(inode)->inval_mask, 0, mask); } @@ -738,14 +738,51 @@ static int fuse_symlink(struct user_namespace *mnt_userns, struct inode *dir, return create_new_entry(fm, &args, dir, entry, S_IFLNK); } -void fuse_update_ctime(struct inode *inode) +void fuse_flush_time_update(struct inode *inode) +{ + int err = sync_inode_metadata(inode, 1); + + mapping_set_error(inode->i_mapping, err); +} + +static void fuse_update_ctime_in_cache(struct inode *inode) { if (!IS_NOCMTIME(inode)) { inode->i_ctime = current_time(inode); mark_inode_dirty_sync(inode); + fuse_flush_time_update(inode); } } +void fuse_update_ctime(struct inode *inode) +{ + fuse_invalidate_attr_mask(inode, STATX_CTIME); + fuse_update_ctime_in_cache(inode); +} + +static void fuse_entry_unlinked(struct dentry *entry) +{ + struct inode *inode = d_inode(entry); + struct fuse_conn *fc = get_fuse_conn(inode); + struct fuse_inode *fi = get_fuse_inode(inode); + + spin_lock(&fi->lock); + fi->attr_version = atomic64_inc_return(&fc->attr_version); + /* + * If i_nlink == 0 then unlink doesn't make sense, yet this can + * happen if userspace filesystem is careless. It would be + * difficult to enforce correct nlink usage so just ignore this + * condition here + */ + if (S_ISDIR(inode->i_mode)) + clear_nlink(inode); + else if (inode->i_nlink > 0) + drop_nlink(inode); + spin_unlock(&fi->lock); + fuse_invalidate_entry_cache(entry); + fuse_update_ctime(inode); +} + static int fuse_unlink(struct inode *dir, struct dentry *entry) { int err; @@ -762,24 +799,8 @@ static int fuse_unlink(struct inode *dir, struct dentry *entry) args.in_args[0].value = entry->d_name.name; err = fuse_simple_request(fm, &args); if (!err) { - struct inode *inode = d_inode(entry); - struct fuse_inode *fi = get_fuse_inode(inode); - - spin_lock(&fi->lock); - fi->attr_version = atomic64_inc_return(&fm->fc->attr_version); - /* - * If i_nlink == 0 then unlink doesn't make sense, yet this can - * happen if userspace filesystem is careless. It would be - * difficult to enforce correct nlink usage so just ignore this - * condition here - */ - if (inode->i_nlink > 0) - drop_nlink(inode); - spin_unlock(&fi->lock); - fuse_invalidate_attr(inode); fuse_dir_changed(dir); - fuse_invalidate_entry_cache(entry); - fuse_update_ctime(inode); + fuse_entry_unlinked(entry); } else if (err == -EINTR) fuse_invalidate_entry(entry); return err; @@ -801,9 +822,8 @@ static int fuse_rmdir(struct inode *dir, struct dentry *entry) args.in_args[0].value = entry->d_name.name; err = fuse_simple_request(fm, &args); if (!err) { - clear_nlink(d_inode(entry)); fuse_dir_changed(dir); - fuse_invalidate_entry_cache(entry); + fuse_entry_unlinked(entry); } else if (err == -EINTR) fuse_invalidate_entry(entry); return err; @@ -833,24 +853,18 @@ static int fuse_rename_common(struct inode *olddir, struct dentry *oldent, err = fuse_simple_request(fm, &args); if (!err) { /* ctime changes */ - fuse_invalidate_attr(d_inode(oldent)); fuse_update_ctime(d_inode(oldent)); - if (flags & RENAME_EXCHANGE) { - fuse_invalidate_attr(d_inode(newent)); + if (flags & RENAME_EXCHANGE) fuse_update_ctime(d_inode(newent)); - } fuse_dir_changed(olddir); if (olddir != newdir) fuse_dir_changed(newdir); /* newent will end up negative */ - if (!(flags & RENAME_EXCHANGE) && d_really_is_positive(newent)) { - fuse_invalidate_attr(d_inode(newent)); - fuse_invalidate_entry_cache(newent); - fuse_update_ctime(d_inode(newent)); - } + if (!(flags & RENAME_EXCHANGE) && d_really_is_positive(newent)) + fuse_entry_unlinked(newent); } else if (err == -EINTR) { /* If request was interrupted, DEITY only knows if the rename actually took place. If the invalidation @@ -916,25 +930,11 @@ static int fuse_link(struct dentry *entry, struct inode *newdir, args.in_args[1].size = newent->d_name.len + 1; args.in_args[1].value = newent->d_name.name; err = create_new_entry(fm, &args, newdir, newent, inode->i_mode); - /* Contrary to "normal" filesystems it can happen that link - makes two "logical" inodes point to the same "physical" - inode. We invalidate the attributes of the old one, so it - will reflect changes in the backing inode (link count, - etc.) - */ - if (!err) { - struct fuse_inode *fi = get_fuse_inode(inode); - - spin_lock(&fi->lock); - fi->attr_version = atomic64_inc_return(&fm->fc->attr_version); - if (likely(inode->i_nlink < UINT_MAX)) - inc_nlink(inode); - spin_unlock(&fi->lock); - fuse_invalidate_attr(inode); - fuse_update_ctime(inode); - } else if (err == -EINTR) { + if (!err) + fuse_update_ctime_in_cache(inode); + else if (err == -EINTR) fuse_invalidate_attr(inode); - } + return err; } @@ -944,15 +944,6 @@ static void fuse_fillattr(struct inode *inode, struct fuse_attr *attr, unsigned int blkbits; struct fuse_conn *fc = get_fuse_conn(inode); - /* see the comment in fuse_change_attributes() */ - if (fc->writeback_cache && S_ISREG(inode->i_mode)) { - attr->size = i_size_read(inode); - attr->mtime = inode->i_mtime.tv_sec; - attr->mtimensec = inode->i_mtime.tv_nsec; - attr->ctime = inode->i_ctime.tv_sec; - attr->ctimensec = inode->i_ctime.tv_nsec; - } - stat->dev = inode->i_sb->s_dev; stat->ino = attr->ino; stat->mode = (inode->i_mode & S_IFMT) | (attr->mode & 07777); @@ -1030,12 +1021,14 @@ static int fuse_update_get_attr(struct inode *inode, struct file *file, struct fuse_inode *fi = get_fuse_inode(inode); int err = 0; bool sync; + u32 inval_mask = READ_ONCE(fi->inval_mask); + u32 cache_mask = fuse_get_cache_mask(inode); if (flags & AT_STATX_FORCE_SYNC) sync = true; else if (flags & AT_STATX_DONT_SYNC) sync = false; - else if (request_mask & READ_ONCE(fi->inval_mask)) + else if (request_mask & inval_mask & ~cache_mask) sync = true; else sync = time_before64(fi->i_time, get_jiffies_64()); @@ -1052,11 +1045,9 @@ static int fuse_update_get_attr(struct inode *inode, struct file *file, return err; } -int fuse_update_attributes(struct inode *inode, struct file *file) +int fuse_update_attributes(struct inode *inode, struct file *file, u32 mask) { - /* Do *not* need to get atime for internal purposes */ - return fuse_update_get_attr(inode, file, NULL, - STATX_BASIC_STATS & ~STATX_ATIME, 0); + return fuse_update_get_attr(inode, file, NULL, mask, 0); } int fuse_reverse_inval_entry(struct fuse_conn *fc, u64 parent_nodeid, @@ -1071,7 +1062,7 @@ int fuse_reverse_inval_entry(struct fuse_conn *fc, u64 parent_nodeid, if (!parent) return -ENOENT; - inode_lock(parent); + inode_lock_nested(parent, I_MUTEX_PARENT); if (!S_ISDIR(parent->i_mode)) goto unlock; @@ -1561,10 +1552,10 @@ int fuse_do_setattr(struct dentry *dentry, struct iattr *attr, struct fuse_setattr_in inarg; struct fuse_attr_out outarg; bool is_truncate = false; - bool is_wb = fc->writeback_cache; + bool is_wb = fc->writeback_cache && S_ISREG(inode->i_mode); loff_t oldsize; int err; - bool trust_local_cmtime = is_wb && S_ISREG(inode->i_mode); + bool trust_local_cmtime = is_wb; bool fault_blocked = false; if (!fc->default_permissions) @@ -1608,7 +1599,7 @@ int fuse_do_setattr(struct dentry *dentry, struct iattr *attr, } /* Flush dirty data/metadata before non-truncate SETATTR */ - if (is_wb && S_ISREG(inode->i_mode) && + if (is_wb && attr->ia_valid & (ATTR_MODE | ATTR_UID | ATTR_GID | ATTR_MTIME_SET | ATTR_TIMES_SET)) { @@ -1676,10 +1667,11 @@ int fuse_do_setattr(struct dentry *dentry, struct iattr *attr, } fuse_change_attributes_common(inode, &outarg.attr, - attr_timeout(&outarg)); + attr_timeout(&outarg), + fuse_get_cache_mask(inode)); oldsize = inode->i_size; /* see the comment in fuse_change_attributes() */ - if (!is_wb || is_truncate || !S_ISREG(inode->i_mode)) + if (!is_wb || is_truncate) i_size_write(inode, outarg.attr.size); if (is_truncate) { diff --git a/fs/fuse/file.c b/fs/fuse/file.c index 34b6d0650e66..9d6c5f6361f7 100644 --- a/fs/fuse/file.c +++ b/fs/fuse/file.c @@ -211,9 +211,8 @@ void fuse_finish_open(struct inode *inode, struct file *file) i_size_write(inode, 0); spin_unlock(&fi->lock); truncate_pagecache(inode, 0); - fuse_invalidate_attr(inode); - if (fc->writeback_cache) - file_update_time(file); + file_update_time(file); + fuse_invalidate_attr_mask(inode, FUSE_STATX_MODSIZE); } else if (!(ff->open_flags & FOPEN_KEEP_CACHE)) { invalidate_inode_pages2(inode->i_mapping); } @@ -339,12 +338,6 @@ static int fuse_open(struct inode *inode, struct file *file) static int fuse_release(struct inode *inode, struct file *file) { - struct fuse_conn *fc = get_fuse_conn(inode); - - /* see fuse_vma_close() for !writeback_cache case */ - if (fc->writeback_cache) - write_inode_now(inode, 1); - fuse_release_common(file, false); /* return value is ignored by VFS */ @@ -483,6 +476,9 @@ static int fuse_flush(struct file *file, fl_owner_t id) if (fuse_is_bad(inode)) return -EIO; + if (ff->open_flags & FOPEN_NOFLUSH && !fm->fc->writeback_cache) + return 0; + err = write_inode_now(inode, 1); if (err) return err; @@ -521,7 +517,7 @@ inval_attr_out: * enabled, i_blocks from cached attr may not be accurate. */ if (!err && fm->fc->writeback_cache) - fuse_invalidate_attr(inode); + fuse_invalidate_attr_mask(inode, STATX_BLOCKS); return err; } @@ -793,7 +789,7 @@ static void fuse_read_update_size(struct inode *inode, loff_t size, struct fuse_inode *fi = get_fuse_inode(inode); spin_lock(&fi->lock); - if (attr_ver == fi->attr_version && size < inode->i_size && + if (attr_ver >= fi->attr_version && size < inode->i_size && !test_bit(FUSE_I_SIZE_UNSTABLE, &fi->state)) { fi->attr_version = atomic64_inc_return(&fc->attr_version); i_size_write(inode, size); @@ -1003,7 +999,7 @@ static ssize_t fuse_cache_read_iter(struct kiocb *iocb, struct iov_iter *to) if (fc->auto_inval_data || (iocb->ki_pos + iov_iter_count(to) > i_size_read(inode))) { int err; - err = fuse_update_attributes(inode, iocb->ki_filp); + err = fuse_update_attributes(inode, iocb->ki_filp, STATX_SIZE); if (err) return err; } @@ -1072,7 +1068,7 @@ static ssize_t fuse_send_write(struct fuse_io_args *ia, loff_t pos, return err ?: ia->write.out.size; } -bool fuse_write_update_size(struct inode *inode, loff_t pos) +bool fuse_write_update_attr(struct inode *inode, loff_t pos, ssize_t written) { struct fuse_conn *fc = get_fuse_conn(inode); struct fuse_inode *fi = get_fuse_inode(inode); @@ -1080,12 +1076,14 @@ bool fuse_write_update_size(struct inode *inode, loff_t pos) spin_lock(&fi->lock); fi->attr_version = atomic64_inc_return(&fc->attr_version); - if (pos > inode->i_size) { + if (written > 0 && pos > inode->i_size) { i_size_write(inode, pos); ret = true; } spin_unlock(&fi->lock); + fuse_invalidate_attr_mask(inode, FUSE_STATX_MODSIZE); + return ret; } @@ -1268,11 +1266,8 @@ static ssize_t fuse_perform_write(struct kiocb *iocb, kfree(ap->pages); } while (!err && iov_iter_count(ii)); - if (res > 0) - fuse_write_update_size(inode, pos); - + fuse_write_update_attr(inode, pos, res); clear_bit(FUSE_I_SIZE_UNSTABLE, &fi->state); - fuse_invalidate_attr(inode); return res > 0 ? res : err; } @@ -1290,7 +1285,8 @@ static ssize_t fuse_cache_write_iter(struct kiocb *iocb, struct iov_iter *from) if (fc->writeback_cache) { /* Update size (EOF optimization) and mode (SUID clearing) */ - err = fuse_update_attributes(mapping->host, file); + err = fuse_update_attributes(mapping->host, file, + STATX_SIZE | STATX_MODE); if (err) return err; @@ -1451,7 +1447,6 @@ ssize_t fuse_direct_io(struct fuse_io_priv *io, struct iov_iter *iter, if (!ia) return -ENOMEM; - ia->io = io; if (!cuse && fuse_range_is_writeback(inode, idx_from, idx_to)) { if (!write) inode_lock(inode); @@ -1561,11 +1556,9 @@ static ssize_t fuse_direct_write_iter(struct kiocb *iocb, struct iov_iter *from) } else { res = fuse_direct_io(&io, from, &iocb->ki_pos, FUSE_DIO_WRITE); + fuse_write_update_attr(inode, iocb->ki_pos, res); } } - fuse_invalidate_attr(inode); - if (res > 0) - fuse_write_update_size(inode, iocb->ki_pos); inode_unlock(inode); return res; @@ -1776,7 +1769,7 @@ static void fuse_writepage_end(struct fuse_mount *fm, struct fuse_args *args, * is enabled, we trust local ctime/mtime. */ if (!fc->writeback_cache) - fuse_invalidate_attr(inode); + fuse_invalidate_attr_mask(inode, FUSE_STATX_MODIFY); spin_lock(&fi->lock); rb_erase(&wpa->writepages_entry, &fi->writepages); while (wpa->next) { @@ -1822,14 +1815,13 @@ static void fuse_writepage_end(struct fuse_mount *fm, struct fuse_args *args, static struct fuse_file *__fuse_write_file_get(struct fuse_inode *fi) { - struct fuse_file *ff = NULL; + struct fuse_file *ff; spin_lock(&fi->lock); - if (!list_empty(&fi->write_files)) { - ff = list_entry(fi->write_files.next, struct fuse_file, - write_entry); + ff = list_first_entry_or_null(&fi->write_files, struct fuse_file, + write_entry); + if (ff) fuse_file_get(ff); - } spin_unlock(&fi->lock); return ff; @@ -1848,6 +1840,17 @@ int fuse_write_inode(struct inode *inode, struct writeback_control *wbc) struct fuse_file *ff; int err; + /* + * Inode is always written before the last reference is dropped and + * hence this should not be reached from reclaim. + * + * Writing back the inode from reclaim can deadlock if the request + * processing itself needs an allocation. Allocations triggering + * reclaim while serving a request can't be prevented, because it can + * involve any number of unrelated userspace processes. + */ + WARN_ON(wbc->for_reclaim); + ff = __fuse_write_file_get(fi); err = fuse_flush_times(inode, ff); if (ff) @@ -2306,15 +2309,18 @@ static int fuse_write_end(struct file *file, struct address_space *mapping, if (!copied) goto unlock; + pos += copied; if (!PageUptodate(page)) { /* Zero any unwritten bytes at the end of the page */ - size_t endoff = (pos + copied) & ~PAGE_MASK; + size_t endoff = pos & ~PAGE_MASK; if (endoff) zero_user_segment(page, endoff, PAGE_SIZE); SetPageUptodate(page); } - fuse_write_update_size(inode, pos + copied); + if (pos > inode->i_size) + i_size_write(inode, pos); + set_page_dirty(page); unlock: @@ -2340,12 +2346,15 @@ static int fuse_launder_page(struct page *page) } /* - * Write back dirty pages now, because there may not be any suitable - * open files later + * Write back dirty data/metadata now (there may not be any suitable + * open files later for data) */ static void fuse_vma_close(struct vm_area_struct *vma) { - filemap_write_and_wait(vma->vm_file->f_mapping); + int err; + + err = write_inode_now(vma->vm_file->f_mapping->host, 1); + mapping_set_error(vma->vm_file->f_mapping, err); } /* @@ -2628,7 +2637,7 @@ static loff_t fuse_lseek(struct file *file, loff_t offset, int whence) return vfs_setpos(file, outarg.offset, inode->i_sb->s_maxbytes); fallback: - err = fuse_update_attributes(inode, file); + err = fuse_update_attributes(inode, file, STATX_SIZE); if (!err) return generic_file_llseek(file, offset, whence); else @@ -2648,7 +2657,7 @@ static loff_t fuse_file_llseek(struct file *file, loff_t offset, int whence) break; case SEEK_END: inode_lock(inode); - retval = fuse_update_attributes(inode, file); + retval = fuse_update_attributes(inode, file, STATX_SIZE); if (!retval) retval = generic_file_llseek(file, offset, whence); inode_unlock(inode); @@ -2869,7 +2878,7 @@ fuse_direct_IO(struct kiocb *iocb, struct iov_iter *iter) if (iov_iter_rw(iter) == WRITE) { ret = fuse_direct_io(io, iter, &pos, FUSE_DIO_WRITE); - fuse_invalidate_attr(inode); + fuse_invalidate_attr_mask(inode, FUSE_STATX_MODSIZE); } else { ret = __fuse_direct_read(io, iter, &pos); } @@ -2891,9 +2900,8 @@ fuse_direct_IO(struct kiocb *iocb, struct iov_iter *iter) kref_put(&io->refcnt, fuse_io_release); if (iov_iter_rw(iter) == WRITE) { - if (ret > 0) - fuse_write_update_size(inode, pos); - else if (ret < 0 && offset + count > i_size) + fuse_write_update_attr(inode, pos, ret); + if (ret < 0 && offset + count > i_size) fuse_do_truncate(file); } @@ -2981,16 +2989,14 @@ static long fuse_file_fallocate(struct file *file, int mode, loff_t offset, /* we could have extended the file */ if (!(mode & FALLOC_FL_KEEP_SIZE)) { - bool changed = fuse_write_update_size(inode, offset + length); - - if (changed && fm->fc->writeback_cache) + if (fuse_write_update_attr(inode, offset + length, length)) file_update_time(file); } if (mode & (FALLOC_FL_PUNCH_HOLE | FALLOC_FL_ZERO_RANGE)) truncate_pagecache_range(inode, offset, offset + length - 1); - fuse_invalidate_attr(inode); + fuse_invalidate_attr_mask(inode, FUSE_STATX_MODSIZE); out: if (!(mode & FALLOC_FL_KEEP_SIZE)) @@ -3002,6 +3008,8 @@ out: if (lock_inode) inode_unlock(inode); + fuse_flush_time_update(inode); + return err; } @@ -3096,12 +3104,8 @@ static ssize_t __fuse_copy_file_range(struct file *file_in, loff_t pos_in, ALIGN_DOWN(pos_out, PAGE_SIZE), ALIGN(pos_out + outarg.size, PAGE_SIZE) - 1); - if (fc->writeback_cache) { - fuse_write_update_size(inode_out, pos_out + outarg.size); - file_update_time(file_out); - } - - fuse_invalidate_attr(inode_out); + file_update_time(file_out); + fuse_write_update_attr(inode_out, pos_out + outarg.size, outarg.size); err = outarg.size; out: @@ -3111,6 +3115,8 @@ out: inode_unlock(inode_out); file_accessed(file_in); + fuse_flush_time_update(inode_out); + return err; } diff --git a/fs/fuse/fuse_i.h b/fs/fuse/fuse_i.h index f55f9f94b1a4..198637b41e19 100644 --- a/fs/fuse/fuse_i.h +++ b/fs/fuse/fuse_i.h @@ -1031,7 +1031,9 @@ void fuse_change_attributes(struct inode *inode, struct fuse_attr *attr, u64 attr_valid, u64 attr_version); void fuse_change_attributes_common(struct inode *inode, struct fuse_attr *attr, - u64 attr_valid); + u64 attr_valid, u32 cache_mask); + +u32 fuse_get_cache_mask(struct inode *inode); /** * Initialize the client device @@ -1065,7 +1067,15 @@ void fuse_wait_aborted(struct fuse_conn *fc); /** * Invalidate inode attributes */ + +/* Attributes possibly changed on data modification */ +#define FUSE_STATX_MODIFY (STATX_MTIME | STATX_CTIME | STATX_BLOCKS) + +/* Attributes possibly changed on data and/or size modification */ +#define FUSE_STATX_MODSIZE (FUSE_STATX_MODIFY | STATX_SIZE) + void fuse_invalidate_attr(struct inode *inode); +void fuse_invalidate_attr_mask(struct inode *inode, u32 mask); void fuse_invalidate_entry_cache(struct dentry *entry); @@ -1148,9 +1158,10 @@ int fuse_allow_current_process(struct fuse_conn *fc); u64 fuse_lock_owner_id(struct fuse_conn *fc, fl_owner_t id); +void fuse_flush_time_update(struct inode *inode); void fuse_update_ctime(struct inode *inode); -int fuse_update_attributes(struct inode *inode, struct file *file); +int fuse_update_attributes(struct inode *inode, struct file *file, u32 mask); void fuse_flush_writepages(struct inode *inode); @@ -1208,7 +1219,7 @@ long fuse_ioctl_common(struct file *file, unsigned int cmd, __poll_t fuse_file_poll(struct file *file, poll_table *wait); int fuse_dev_release(struct inode *inode, struct file *file); -bool fuse_write_update_size(struct inode *inode, loff_t pos); +bool fuse_write_update_attr(struct inode *inode, loff_t pos, ssize_t written); int fuse_flush_times(struct inode *inode, struct fuse_file *ff); int fuse_write_inode(struct inode *inode, struct writeback_control *wbc); diff --git a/fs/fuse/inode.c b/fs/fuse/inode.c index 12d49a1914e8..8b89e3ba7df3 100644 --- a/fs/fuse/inode.c +++ b/fs/fuse/inode.c @@ -118,6 +118,9 @@ static void fuse_evict_inode(struct inode *inode) { struct fuse_inode *fi = get_fuse_inode(inode); + /* Will write inode on close/munmap and in all other dirtiers */ + WARN_ON(inode->i_state & I_DIRTY_INODE); + truncate_inode_pages_final(&inode->i_data); clear_inode(inode); if (inode->i_sb->s_flags & SB_ACTIVE) { @@ -161,7 +164,7 @@ static ino_t fuse_squash_ino(u64 ino64) } void fuse_change_attributes_common(struct inode *inode, struct fuse_attr *attr, - u64 attr_valid) + u64 attr_valid, u32 cache_mask) { struct fuse_conn *fc = get_fuse_conn(inode); struct fuse_inode *fi = get_fuse_inode(inode); @@ -181,9 +184,11 @@ void fuse_change_attributes_common(struct inode *inode, struct fuse_attr *attr, inode->i_atime.tv_sec = attr->atime; inode->i_atime.tv_nsec = attr->atimensec; /* mtime from server may be stale due to local buffered write */ - if (!fc->writeback_cache || !S_ISREG(inode->i_mode)) { + if (!(cache_mask & STATX_MTIME)) { inode->i_mtime.tv_sec = attr->mtime; inode->i_mtime.tv_nsec = attr->mtimensec; + } + if (!(cache_mask & STATX_CTIME)) { inode->i_ctime.tv_sec = attr->ctime; inode->i_ctime.tv_nsec = attr->ctimensec; } @@ -215,16 +220,44 @@ void fuse_change_attributes_common(struct inode *inode, struct fuse_attr *attr, inode->i_flags &= ~S_NOSEC; } +u32 fuse_get_cache_mask(struct inode *inode) +{ + struct fuse_conn *fc = get_fuse_conn(inode); + + if (!fc->writeback_cache || !S_ISREG(inode->i_mode)) + return 0; + + return STATX_MTIME | STATX_CTIME | STATX_SIZE; +} + void fuse_change_attributes(struct inode *inode, struct fuse_attr *attr, u64 attr_valid, u64 attr_version) { struct fuse_conn *fc = get_fuse_conn(inode); struct fuse_inode *fi = get_fuse_inode(inode); - bool is_wb = fc->writeback_cache; + u32 cache_mask; loff_t oldsize; struct timespec64 old_mtime; spin_lock(&fi->lock); + /* + * In case of writeback_cache enabled, writes update mtime, ctime and + * may update i_size. In these cases trust the cached value in the + * inode. + */ + cache_mask = fuse_get_cache_mask(inode); + if (cache_mask & STATX_SIZE) + attr->size = i_size_read(inode); + + if (cache_mask & STATX_MTIME) { + attr->mtime = inode->i_mtime.tv_sec; + attr->mtimensec = inode->i_mtime.tv_nsec; + } + if (cache_mask & STATX_CTIME) { + attr->ctime = inode->i_ctime.tv_sec; + attr->ctimensec = inode->i_ctime.tv_nsec; + } + if ((attr_version != 0 && fi->attr_version > attr_version) || test_bit(FUSE_I_SIZE_UNSTABLE, &fi->state)) { spin_unlock(&fi->lock); @@ -232,7 +265,7 @@ void fuse_change_attributes(struct inode *inode, struct fuse_attr *attr, } old_mtime = inode->i_mtime; - fuse_change_attributes_common(inode, attr, attr_valid); + fuse_change_attributes_common(inode, attr, attr_valid, cache_mask); oldsize = inode->i_size; /* @@ -240,11 +273,11 @@ void fuse_change_attributes(struct inode *inode, struct fuse_attr *attr, * extend local i_size without keeping userspace server in sync. So, * attr->size coming from server can be stale. We cannot trust it. */ - if (!is_wb || !S_ISREG(inode->i_mode)) + if (!(cache_mask & STATX_SIZE)) i_size_write(inode, attr->size); spin_unlock(&fi->lock); - if (!is_wb && S_ISREG(inode->i_mode)) { + if (!cache_mask && S_ISREG(inode->i_mode)) { bool inval = false; if (oldsize != attr->size) { diff --git a/fs/fuse/ioctl.c b/fs/fuse/ioctl.c index 546ea3d58fb4..fbc09dab1f85 100644 --- a/fs/fuse/ioctl.c +++ b/fs/fuse/ioctl.c @@ -286,11 +286,11 @@ long fuse_do_ioctl(struct file *file, unsigned int cmd, unsigned long arg, in_iovs + out_iovs > FUSE_IOCTL_MAX_IOV) goto out; - vaddr = kmap_atomic(ap.pages[0]); + vaddr = kmap_local_page(ap.pages[0]); err = fuse_copy_ioctl_iovec(fm->fc, iov_page, vaddr, transferred, in_iovs + out_iovs, (flags & FUSE_IOCTL_COMPAT) != 0); - kunmap_atomic(vaddr); + kunmap_local(vaddr); if (err) goto out; diff --git a/fs/fuse/readdir.c b/fs/fuse/readdir.c index bc267832310c..b4e565711045 100644 --- a/fs/fuse/readdir.c +++ b/fs/fuse/readdir.c @@ -76,11 +76,11 @@ static void fuse_add_dirent_to_cache(struct file *file, WARN_ON(fi->rdc.pos != pos)) goto unlock; - addr = kmap_atomic(page); + addr = kmap_local_page(page); if (!offset) clear_page(addr); memcpy(addr + offset, dirent, reclen); - kunmap_atomic(addr); + kunmap_local(addr); fi->rdc.size = (index << PAGE_SHIFT) + offset + reclen; fi->rdc.pos = dirent->off; unlock: @@ -454,7 +454,7 @@ static int fuse_readdir_cached(struct file *file, struct dir_context *ctx) * cache; both cases require an up-to-date mtime value. */ if (!ctx->pos && fc->auto_inval_data) { - int err = fuse_update_attributes(inode, file); + int err = fuse_update_attributes(inode, file, STATX_MTIME); if (err) return err; diff --git a/fs/fuse/virtio_fs.c b/fs/fuse/virtio_fs.c index 94fc874f5de7..4cfa4bc1f579 100644 --- a/fs/fuse/virtio_fs.c +++ b/fs/fuse/virtio_fs.c @@ -649,7 +649,7 @@ static void virtio_fs_vq_done(struct virtqueue *vq) static void virtio_fs_init_vq(struct virtio_fs_vq *fsvq, char *name, int vq_type) { - strncpy(fsvq->name, name, VQ_NAME_LEN); + strscpy(fsvq->name, name, VQ_NAME_LEN); spin_lock_init(&fsvq->lock); INIT_LIST_HEAD(&fsvq->queued_reqs); INIT_LIST_HEAD(&fsvq->end_reqs); diff --git a/fs/fuse/xattr.c b/fs/fuse/xattr.c index 61dfaf7b7d20..0d3e7177fce0 100644 --- a/fs/fuse/xattr.c +++ b/fs/fuse/xattr.c @@ -42,10 +42,9 @@ int fuse_setxattr(struct inode *inode, const char *name, const void *value, fm->fc->no_setxattr = 1; err = -EOPNOTSUPP; } - if (!err) { - fuse_invalidate_attr(inode); + if (!err) fuse_update_ctime(inode); - } + return err; } @@ -173,10 +172,9 @@ int fuse_removexattr(struct inode *inode, const char *name) fm->fc->no_removexattr = 1; err = -EOPNOTSUPP; } - if (!err) { - fuse_invalidate_attr(inode); + if (!err) fuse_update_ctime(inode); - } + return err; } diff --git a/include/uapi/linux/fuse.h b/include/uapi/linux/fuse.h index 36ed092227fa..a1dc3ee1d17c 100644 --- a/include/uapi/linux/fuse.h +++ b/include/uapi/linux/fuse.h @@ -184,6 +184,9 @@ * * 7.34 * - add FUSE_SYNCFS + * + * 7.35 + * - add FOPEN_NOFLUSH */ #ifndef _LINUX_FUSE_H @@ -219,7 +222,7 @@ #define FUSE_KERNEL_VERSION 7 /** Minor version number of this interface */ -#define FUSE_KERNEL_MINOR_VERSION 34 +#define FUSE_KERNEL_MINOR_VERSION 35 /** The node ID of the root inode */ #define FUSE_ROOT_ID 1 @@ -290,12 +293,14 @@ struct fuse_file_lock { * FOPEN_NONSEEKABLE: the file is not seekable * FOPEN_CACHE_DIR: allow caching this directory * FOPEN_STREAM: the file is stream-like (no file position at all) + * FOPEN_NOFLUSH: don't flush data cache on close (unless FUSE_WRITEBACK_CACHE) */ #define FOPEN_DIRECT_IO (1 << 0) #define FOPEN_KEEP_CACHE (1 << 1) #define FOPEN_NONSEEKABLE (1 << 2) #define FOPEN_CACHE_DIR (1 << 3) #define FOPEN_STREAM (1 << 4) +#define FOPEN_NOFLUSH (1 << 5) /** * INIT request/reply flags |