diff options
author | Kent Overstreet <kent.overstreet@gmail.com> | 2019-06-30 16:35:37 -0400 |
---|---|---|
committer | Kent Overstreet <kent.overstreet@gmail.com> | 2022-01-08 03:35:53 -0500 |
commit | 935a35ef1f79b5e53d7f3b1dfb5b28c739c712b2 (patch) | |
tree | abb8d3e5ffd253c17899907255e3af243c320540 | |
parent | ce53964ba1227f5a27c1d2b53785de1cb5daff6e (diff) |
bcachefs: Fixes for 4.19
32 files changed, 569 insertions, 211 deletions
diff --git a/fs/bcachefs/acl.c b/fs/bcachefs/acl.c index 5070caf8f349..2588812c5066 100644 --- a/fs/bcachefs/acl.c +++ b/fs/bcachefs/acl.c @@ -212,7 +212,7 @@ bch2_acl_to_xattr(struct btree_trans *trans, return xattr; } -struct posix_acl *bch2_get_acl(struct inode *vinode, int type, bool rcu) +struct posix_acl *bch2_get_acl(struct inode *vinode, int type) { struct bch_inode_info *inode = to_bch_ei(vinode); struct bch_fs *c = inode->v.i_sb->s_fs_info; @@ -224,9 +224,6 @@ struct posix_acl *bch2_get_acl(struct inode *vinode, int type, bool rcu) struct bkey_s_c k; int ret; - if (rcu) - return ERR_PTR(-ECHILD); - bch2_trans_init(&trans, c, 0, 0); retry: bch2_trans_begin(&trans); @@ -292,8 +289,7 @@ int bch2_set_acl_trans(struct btree_trans *trans, subvol_inum inum, return ret == -ENOENT ? 0 : ret; } -int bch2_set_acl(struct user_namespace *mnt_userns, - struct inode *vinode, struct posix_acl *_acl, int type) +int bch2_set_acl(struct inode *vinode, struct posix_acl *_acl, int type) { struct bch_inode_info *inode = to_bch_ei(vinode); struct bch_fs *c = inode->v.i_sb->s_fs_info; @@ -318,7 +314,7 @@ retry: mode = inode_u.bi_mode; if (type == ACL_TYPE_ACCESS) { - ret = posix_acl_update_mode(mnt_userns, &inode->v, &mode, &acl); + ret = posix_acl_update_mode(&inode->v, &mode, &acl); if (ret) goto btree_err; } diff --git a/fs/bcachefs/acl.h b/fs/bcachefs/acl.h index 2d76a4897ba8..14cabbc91808 100644 --- a/fs/bcachefs/acl.h +++ b/fs/bcachefs/acl.h @@ -26,12 +26,12 @@ typedef struct { __le32 a_version; } bch_acl_header; -struct posix_acl *bch2_get_acl(struct inode *, int, bool); +struct posix_acl *bch2_get_acl(struct inode *, int); int bch2_set_acl_trans(struct btree_trans *, subvol_inum, struct bch_inode_unpacked *, struct posix_acl *, int); -int bch2_set_acl(struct user_namespace *, struct inode *, struct posix_acl *, int); +int bch2_set_acl(struct inode *, struct posix_acl *, int); int bch2_acl_chmod(struct btree_trans *, subvol_inum, struct bch_inode_unpacked *, umode_t, struct posix_acl **); diff --git a/fs/bcachefs/alloc_background.c b/fs/bcachefs/alloc_background.c index fac2176a4f3b..b3641d196517 100644 --- a/fs/bcachefs/alloc_background.c +++ b/fs/bcachefs/alloc_background.c @@ -985,7 +985,7 @@ void bch2_recalc_capacity(struct bch_fs *c) lockdep_assert_held(&c->state_lock); for_each_online_member(ca, c, i) { - struct backing_dev_info *bdi = ca->disk_sb.bdev->bd_disk->bdi; + struct backing_dev_info *bdi = ca->disk_sb.bdev->bd_bdi; ra_pages += bdi->ra_pages; } diff --git a/fs/bcachefs/bcachefs.h b/fs/bcachefs/bcachefs.h index a28ddcd5d7b7..f1e4871a74c3 100644 --- a/fs/bcachefs/bcachefs.h +++ b/fs/bcachefs/bcachefs.h @@ -822,7 +822,7 @@ struct bch_fs { ZSTD_parameters zstd_params; struct crypto_shash *sha256; - struct crypto_sync_skcipher *chacha20; + struct crypto_skcipher *chacha20; struct crypto_shash *poly1305; atomic64_t key_version; diff --git a/fs/bcachefs/btree_cache.c b/fs/bcachefs/btree_cache.c index fc6c4d4cd02f..c964a24bb425 100644 --- a/fs/bcachefs/btree_cache.c +++ b/fs/bcachefs/btree_cache.c @@ -78,7 +78,8 @@ static int btree_node_data_alloc(struct bch_fs *c, struct btree *b, gfp_t gfp) if (!b->data) return -ENOMEM; #ifdef __KERNEL__ - b->aux_data = vmalloc_exec(btree_aux_data_bytes(b), gfp); + b->aux_data = __vmalloc(btree_aux_data_bytes(b), gfp, + PAGE_KERNEL_EXEC); #else b->aux_data = mmap(NULL, btree_aux_data_bytes(b), PROT_READ|PROT_WRITE|PROT_EXEC, diff --git a/fs/bcachefs/btree_io.c b/fs/bcachefs/btree_io.c index a3651325a022..4ff38c6395f3 100644 --- a/fs/bcachefs/btree_io.c +++ b/fs/bcachefs/btree_io.c @@ -33,8 +33,6 @@ void bch2_btree_node_io_unlock(struct btree *b) void bch2_btree_node_io_lock(struct btree *b) { - BUG_ON(lock_class_is_held(&bch2_btree_node_lock_key)); - wait_on_bit_lock_io(&b->flags, BTREE_NODE_write_in_flight, TASK_UNINTERRUPTIBLE); } @@ -53,16 +51,12 @@ void __bch2_btree_node_wait_on_write(struct btree *b) void bch2_btree_node_wait_on_read(struct btree *b) { - BUG_ON(lock_class_is_held(&bch2_btree_node_lock_key)); - wait_on_bit_io(&b->flags, BTREE_NODE_read_in_flight, TASK_UNINTERRUPTIBLE); } void bch2_btree_node_wait_on_write(struct btree *b) { - BUG_ON(lock_class_is_held(&bch2_btree_node_lock_key)); - wait_on_bit_io(&b->flags, BTREE_NODE_write_in_flight, TASK_UNINTERRUPTIBLE); } diff --git a/fs/bcachefs/btree_io.h b/fs/bcachefs/btree_io.h index 0f20224e2a77..f11a2e96227b 100644 --- a/fs/bcachefs/btree_io.h +++ b/fs/bcachefs/btree_io.h @@ -122,7 +122,7 @@ static inline void bset_encrypt(struct bch_fs *c, struct bset *i, unsigned offse bch2_encrypt(c, BSET_CSUM_TYPE(i), nonce, &bn->flags, bytes); - nonce = nonce_add(nonce, round_up(bytes, CHACHA_BLOCK_SIZE)); + nonce = nonce_add(nonce, round_up(bytes, CHACHA20_BLOCK_SIZE)); } bch2_encrypt(c, BSET_CSUM_TYPE(i), nonce, i->_data, diff --git a/fs/bcachefs/btree_iter.c b/fs/bcachefs/btree_iter.c index 2ae4e523ff3b..f4551e67bd95 100644 --- a/fs/bcachefs/btree_iter.c +++ b/fs/bcachefs/btree_iter.c @@ -562,8 +562,6 @@ void bch2_trans_unlock(struct btree_trans *trans) trans_for_each_path(trans, path) __bch2_btree_path_unlock(path); - - BUG_ON(lock_class_is_held(&bch2_btree_node_lock_key)); } /* Btree iterator: */ diff --git a/fs/bcachefs/btree_iter.h b/fs/bcachefs/btree_iter.h index eceec5d55f9b..d4189efdafae 100644 --- a/fs/bcachefs/btree_iter.h +++ b/fs/bcachefs/btree_iter.h @@ -136,7 +136,7 @@ int __must_check bch2_btree_path_traverse(struct btree_trans *, struct btree_path *, unsigned); struct btree_path *bch2_path_get(struct btree_trans *, enum btree_id, struct bpos, unsigned, unsigned, unsigned, unsigned long); -inline struct bkey_s_c bch2_btree_path_peek_slot(struct btree_path *, struct bkey *); +struct bkey_s_c bch2_btree_path_peek_slot(struct btree_path *, struct bkey *); #ifdef CONFIG_BCACHEFS_DEBUG void bch2_trans_verify_paths(struct btree_trans *); diff --git a/fs/bcachefs/btree_key_cache.c b/fs/bcachefs/btree_key_cache.c index 1d7b101224f1..17ebc92f86b1 100644 --- a/fs/bcachefs/btree_key_cache.c +++ b/fs/bcachefs/btree_key_cache.c @@ -603,7 +603,7 @@ static unsigned long bch2_btree_key_cache_scan(struct shrinker *shrink, do { struct rhash_head *pos, *next; - pos = rht_ptr_rcu(rht_bucket(tbl, bc->shrink_iter)); + pos = *rht_bucket(tbl, bc->shrink_iter); while (!rht_is_a_nulls(pos)) { next = rht_dereference_bucket_rcu(pos->next, tbl, bc->shrink_iter); diff --git a/fs/bcachefs/checksum.c b/fs/bcachefs/checksum.c index a1d89923d361..95e80dbfed4c 100644 --- a/fs/bcachefs/checksum.c +++ b/fs/bcachefs/checksum.c @@ -11,7 +11,7 @@ #include <linux/random.h> #include <linux/scatterlist.h> #include <crypto/algapi.h> -#include <crypto/chacha.h> +#include <crypto/chacha20.h> #include <crypto/hash.h> #include <crypto/poly1305.h> #include <crypto/skcipher.h> @@ -93,21 +93,21 @@ static void bch2_checksum_update(struct bch2_checksum_state *state, const void * } } -static inline void do_encrypt_sg(struct crypto_sync_skcipher *tfm, +static inline void do_encrypt_sg(struct crypto_skcipher *tfm, struct nonce nonce, struct scatterlist *sg, size_t len) { - SYNC_SKCIPHER_REQUEST_ON_STACK(req, tfm); + SKCIPHER_REQUEST_ON_STACK(req, tfm); int ret; - skcipher_request_set_sync_tfm(req, tfm); + skcipher_request_set_tfm(req, tfm); skcipher_request_set_crypt(req, sg, sg, len, nonce.d); ret = crypto_skcipher_encrypt(req); BUG_ON(ret); } -static inline void do_encrypt(struct crypto_sync_skcipher *tfm, +static inline void do_encrypt(struct crypto_skcipher *tfm, struct nonce nonce, void *buf, size_t len) { @@ -120,8 +120,8 @@ static inline void do_encrypt(struct crypto_sync_skcipher *tfm, int bch2_chacha_encrypt_key(struct bch_key *key, struct nonce nonce, void *buf, size_t len) { - struct crypto_sync_skcipher *chacha20 = - crypto_alloc_sync_skcipher("chacha20", 0, 0); + struct crypto_skcipher *chacha20 = + crypto_alloc_skcipher("chacha20", 0, 0); int ret; if (!chacha20) { @@ -129,8 +129,7 @@ int bch2_chacha_encrypt_key(struct bch_key *key, struct nonce nonce, return PTR_ERR(chacha20); } - ret = crypto_skcipher_setkey(&chacha20->base, - (void *) key, sizeof(*key)); + ret = crypto_skcipher_setkey(chacha20, (void *) key, sizeof(*key)); if (ret) { pr_err("crypto_skcipher_setkey() error: %i", ret); goto err; @@ -138,7 +137,7 @@ int bch2_chacha_encrypt_key(struct bch_key *key, struct nonce nonce, do_encrypt(chacha20, nonce, buf, len); err: - crypto_free_sync_skcipher(chacha20); + crypto_free_skcipher(chacha20); return ret; } @@ -231,7 +230,7 @@ static struct bch_csum __bch2_checksum_bio(struct bch_fs *c, unsigned type, kunmap_atomic(p); } #else - __bio_for_each_bvec(bv, bio, *iter, *iter) + __bio_for_each_contig_segment(bv, bio, *iter, *iter) bch2_checksum_update(&state, page_address(bv.bv_page) + bv.bv_offset, bv.bv_len); #endif @@ -254,7 +253,7 @@ static struct bch_csum __bch2_checksum_bio(struct bch_fs *c, unsigned type, kunmap_atomic(p); } #else - __bio_for_each_bvec(bv, bio, *iter, *iter) + __bio_for_each_contig_segment(bv, bio, *iter, *iter) crypto_shash_update(desc, page_address(bv.bv_page) + bv.bv_offset, bv.bv_len); @@ -500,7 +499,7 @@ err: static int bch2_alloc_ciphers(struct bch_fs *c) { if (!c->chacha20) - c->chacha20 = crypto_alloc_sync_skcipher("chacha20", 0, 0); + c->chacha20 = crypto_alloc_skcipher("chacha20", 0, 0); if (IS_ERR(c->chacha20)) { bch_err(c, "error requesting chacha20 module: %li", PTR_ERR(c->chacha20)); @@ -583,7 +582,7 @@ int bch2_enable_encryption(struct bch_fs *c, bool keyed) goto err; } - ret = crypto_skcipher_setkey(&c->chacha20->base, + ret = crypto_skcipher_setkey(c->chacha20, (void *) &key.key, sizeof(key.key)); if (ret) goto err; @@ -611,7 +610,7 @@ void bch2_fs_encryption_exit(struct bch_fs *c) if (!IS_ERR_OR_NULL(c->poly1305)) crypto_free_shash(c->poly1305); if (!IS_ERR_OR_NULL(c->chacha20)) - crypto_free_sync_skcipher(c->chacha20); + crypto_free_skcipher(c->chacha20); if (!IS_ERR_OR_NULL(c->sha256)) crypto_free_shash(c->sha256); } @@ -643,7 +642,7 @@ int bch2_fs_encryption_init(struct bch_fs *c) if (ret) goto out; - ret = crypto_skcipher_setkey(&c->chacha20->base, + ret = crypto_skcipher_setkey(c->chacha20, (void *) &key.key, sizeof(key.key)); if (ret) goto out; diff --git a/fs/bcachefs/checksum.h b/fs/bcachefs/checksum.h index f5c1a609c5c4..5e0e77ca71a9 100644 --- a/fs/bcachefs/checksum.h +++ b/fs/bcachefs/checksum.h @@ -7,7 +7,7 @@ #include "super-io.h" #include <linux/crc64.h> -#include <crypto/chacha.h> +#include <crypto/chacha20.h> static inline bool bch2_checksum_mergeable(unsigned type) { @@ -140,9 +140,9 @@ static inline bool bch2_crc_cmp(struct bch_csum l, struct bch_csum r) /* for skipping ahead and encrypting/decrypting at an offset: */ static inline struct nonce nonce_add(struct nonce nonce, unsigned offset) { - EBUG_ON(offset & (CHACHA_BLOCK_SIZE - 1)); + EBUG_ON(offset & (CHACHA20_BLOCK_SIZE - 1)); - le32_add_cpu(&nonce.d[0], offset / CHACHA_BLOCK_SIZE); + le32_add_cpu(&nonce.d[0], offset / CHACHA20_BLOCK_SIZE); return nonce; } diff --git a/fs/bcachefs/compress.c b/fs/bcachefs/compress.c index 8e4179d8dc27..691009fc2431 100644 --- a/fs/bcachefs/compress.c +++ b/fs/bcachefs/compress.c @@ -45,7 +45,7 @@ static bool bio_phys_contig(struct bio *bio, struct bvec_iter start) struct bvec_iter iter; void *expected_start = NULL; - __bio_for_each_bvec(bv, bio, iter, start) { + __bio_for_each_segment(bv, bio, iter, start) { if (expected_start && expected_start != page_address(bv.bv_page) + bv.bv_offset) return false; diff --git a/fs/bcachefs/ec.c b/fs/bcachefs/ec.c index 9b45640e75dc..220ced2f9a17 100644 --- a/fs/bcachefs/ec.c +++ b/fs/bcachefs/ec.c @@ -395,7 +395,7 @@ static void ec_block_io(struct bch_fs *c, struct ec_stripe_buf *buf, this_cpu_add(ca->io_done->sectors[rw][data_type], buf->size); while (offset < bytes) { - unsigned nr_iovecs = min_t(size_t, BIO_MAX_VECS, + unsigned nr_iovecs = min_t(size_t, BIO_MAX_PAGES, DIV_ROUND_UP(bytes, PAGE_SIZE)); unsigned b = min_t(size_t, bytes - offset, nr_iovecs << PAGE_SHIFT); diff --git a/fs/bcachefs/fs-io.c b/fs/bcachefs/fs-io.c index 3b9b96e5a0a2..8f0b2a745064 100644 --- a/fs/bcachefs/fs-io.c +++ b/fs/bcachefs/fs-io.c @@ -285,13 +285,28 @@ static inline struct bch_page_state *bch2_page_state(struct page *page) /* for newly allocated pages: */ static void __bch2_page_state_release(struct page *page) { - kfree(detach_page_private(page)); + struct bch_page_state *s = __bch2_page_state(page); + + if (!s) + return; + + ClearPagePrivate(page); + set_page_private(page, 0); + put_page(page); + kfree(s); } static void bch2_page_state_release(struct page *page) { - EBUG_ON(!PageLocked(page)); - __bch2_page_state_release(page); + struct bch_page_state *s = bch2_page_state(page); + + if (!s) + return; + + ClearPagePrivate(page); + set_page_private(page, 0); + put_page(page); + kfree(s); } /* for newly allocated pages: */ @@ -305,7 +320,13 @@ static struct bch_page_state *__bch2_page_state_create(struct page *page, return NULL; spin_lock_init(&s->lock); - attach_page_private(page, s); + /* + * migrate_page_move_mapping() assumes that pages with private data + * have their count elevated by 1. + */ + get_page(page); + set_page_private(page, (unsigned long) s); + SetPagePrivate(page); return s; } @@ -857,12 +878,18 @@ int bch2_migrate_page(struct address_space *mapping, struct page *newpage, EBUG_ON(!PageLocked(page)); EBUG_ON(!PageLocked(newpage)); - ret = migrate_page_move_mapping(mapping, newpage, page, 0); + ret = migrate_page_move_mapping(mapping, newpage, page, NULL, mode, 0); if (ret != MIGRATEPAGE_SUCCESS) return ret; - if (PagePrivate(page)) - attach_page_private(newpage, detach_page_private(page)); + if (PagePrivate(page)) { + ClearPagePrivate(page); + get_page(newpage); + set_page_private(newpage, page_private(page)); + set_page_private(page, 0); + put_page(page); + SetPagePrivate(newpage); + } if (mode != MIGRATE_SYNC_NO_COPY) migrate_page_copy(newpage, page); @@ -876,10 +903,10 @@ int bch2_migrate_page(struct address_space *mapping, struct page *newpage, static void bch2_readpages_end_io(struct bio *bio) { - struct bvec_iter_all iter; struct bio_vec *bv; + unsigned i; - bio_for_each_segment_all(bv, bio, iter) { + bio_for_each_segment_all(bv, bio, i) { struct page *page = bv->bv_page; if (!bio->bi_status) { @@ -898,29 +925,31 @@ struct readpages_iter { struct address_space *mapping; struct page **pages; unsigned nr_pages; + unsigned nr_added; unsigned idx; pgoff_t offset; }; static int readpages_iter_init(struct readpages_iter *iter, - struct readahead_control *ractl) + struct address_space *mapping, + struct list_head *pages, unsigned nr_pages) { - unsigned i, nr_pages = readahead_count(ractl); - memset(iter, 0, sizeof(*iter)); - iter->mapping = ractl->mapping; - iter->offset = readahead_index(ractl); - iter->nr_pages = nr_pages; + iter->mapping = mapping; + iter->offset = list_last_entry(pages, struct page, lru)->index; iter->pages = kmalloc_array(nr_pages, sizeof(struct page *), GFP_NOFS); if (!iter->pages) return -ENOMEM; - nr_pages = __readahead_batch(ractl, iter->pages, nr_pages); - for (i = 0; i < nr_pages; i++) { - __bch2_page_state_create(iter->pages[i], __GFP_NOFAIL); - put_page(iter->pages[i]); + while (!list_empty(pages)) { + struct page *page = list_last_entry(pages, struct page, lru); + + __bch2_page_state_create(page, __GFP_NOFAIL); + + iter->pages[iter->nr_pages++] = page; + list_del(&page->lru); } return 0; @@ -928,9 +957,41 @@ static int readpages_iter_init(struct readpages_iter *iter, static inline struct page *readpage_iter_next(struct readpages_iter *iter) { - if (iter->idx >= iter->nr_pages) - return NULL; + struct page *page; + unsigned i; + int ret; + + BUG_ON(iter->idx > iter->nr_added); + BUG_ON(iter->nr_added > iter->nr_pages); + + if (iter->idx < iter->nr_added) + goto out; + + while (1) { + if (iter->idx == iter->nr_pages) + return NULL; + + ret = add_to_page_cache_lru_vec(iter->mapping, + iter->pages + iter->nr_added, + iter->nr_pages - iter->nr_added, + iter->offset + iter->nr_added, + GFP_NOFS); + if (ret > 0) + break; + + page = iter->pages[iter->nr_added]; + iter->idx++; + iter->nr_added++; + + __bch2_page_state_release(page); + put_page(page); + } + + iter->nr_added += ret; + for (i = iter->idx; i < iter->nr_added; i++) + put_page(iter->pages[i]); +out: EBUG_ON(iter->pages[iter->idx]->index != iter->offset + iter->idx); return iter->pages[iter->idx]; @@ -968,8 +1029,11 @@ static void readpage_bio_extend(struct readpages_iter *iter, if (!get_more) break; - page = xa_load(&iter->mapping->i_pages, page_offset); - if (page && !xa_is_value(page)) + rcu_read_lock(); + page = radix_tree_lookup(&iter->mapping->i_pages, page_offset); + rcu_read_unlock(); + + if (page && !radix_tree_exceptional_entry(page)) break; page = __page_cache_alloc(readahead_gfp_mask(iter->mapping)); @@ -1105,9 +1169,10 @@ err: bch2_bkey_buf_exit(&sk, c); } -void bch2_readahead(struct readahead_control *ractl) +int bch2_readpages(struct file *file, struct address_space *mapping, + struct list_head *pages, unsigned nr_pages) { - struct bch_inode_info *inode = to_bch_ei(ractl->mapping->host); + struct bch_inode_info *inode = to_bch_ei(mapping->host); struct bch_fs *c = inode->v.i_sb->s_fs_info; struct bch_io_opts opts = io_opts(c, &inode->ei_inode); struct btree_trans trans; @@ -1115,7 +1180,7 @@ void bch2_readahead(struct readahead_control *ractl) struct readpages_iter readpages_iter; int ret; - ret = readpages_iter_init(&readpages_iter, ractl); + ret = readpages_iter_init(&readpages_iter, mapping, pages, nr_pages); BUG_ON(ret); bch2_trans_init(&trans, c, 0, 0); @@ -1127,7 +1192,7 @@ void bch2_readahead(struct readahead_control *ractl) unsigned n = min_t(unsigned, readpages_iter.nr_pages - readpages_iter.idx, - BIO_MAX_VECS); + BIO_MAX_PAGES); struct bch_read_bio *rbio = rbio_init(bio_alloc_bioset(GFP_NOFS, n, &c->bio_read), opts); @@ -1147,6 +1212,8 @@ void bch2_readahead(struct readahead_control *ractl) bch2_trans_exit(&trans); kfree(readpages_iter.pages); + + return 0; } static void __bchfs_readpage(struct bch_fs *c, struct bch_read_bio *rbio, @@ -1241,37 +1308,36 @@ static void bch2_writepage_io_done(struct closure *cl) struct bch_writepage_io, cl); struct bch_fs *c = io->op.c; struct bio *bio = &io->op.wbio.bio; - struct bvec_iter_all iter; struct bio_vec *bvec; - unsigned i; + unsigned i, j; up(&io->op.c->io_in_flight); if (io->op.error) { set_bit(EI_INODE_ERROR, &io->inode->ei_flags); - bio_for_each_segment_all(bvec, bio, iter) { + bio_for_each_segment_all(bvec, bio, i) { struct bch_page_state *s; SetPageError(bvec->bv_page); - mapping_set_error(bvec->bv_page->mapping, -EIO); + mapping_set_error(io->inode->v.i_mapping, -EIO); s = __bch2_page_state(bvec->bv_page); spin_lock(&s->lock); - for (i = 0; i < PAGE_SECTORS; i++) - s->s[i].nr_replicas = 0; + for (j = 0; j < PAGE_SECTORS; j++) + s->s[j].nr_replicas = 0; spin_unlock(&s->lock); } } if (io->op.flags & BCH_WRITE_WROTE_DATA_INLINE) { - bio_for_each_segment_all(bvec, bio, iter) { + bio_for_each_segment_all(bvec, bio, i) { struct bch_page_state *s; s = __bch2_page_state(bvec->bv_page); spin_lock(&s->lock); - for (i = 0; i < PAGE_SECTORS; i++) - s->s[i].nr_replicas = 0; + for (j = 0; j < PAGE_SECTORS; j++) + s->s[j].nr_replicas = 0; spin_unlock(&s->lock); } } @@ -1295,7 +1361,7 @@ static void bch2_writepage_io_done(struct closure *cl) */ i_sectors_acct(c, io->inode, NULL, io->op.i_sectors_delta); - bio_for_each_segment_all(bvec, bio, iter) { + bio_for_each_segment_all(bvec, bio, i) { struct bch_page_state *s = __bch2_page_state(bvec->bv_page); if (atomic_dec_and_test(&s->write_count)) @@ -1329,7 +1395,7 @@ static void bch2_writepage_io_alloc(struct bch_fs *c, { struct bch_write_op *op; - w->io = container_of(bio_alloc_bioset(GFP_NOFS, BIO_MAX_VECS, + w->io = container_of(bio_alloc_bioset(GFP_NOFS, BIO_MAX_PAGES, &c->writepage_bioset), struct bch_writepage_io, op.wbio.bio); @@ -1449,9 +1515,9 @@ do_io: if (w->io && (w->io->op.res.nr_replicas != nr_replicas_this_write || - bio_full(&w->io->op.wbio.bio, PAGE_SIZE) || + bio_full(&w->io->op.wbio.bio) || w->io->op.wbio.bio.bi_iter.bi_size + (sectors << 9) >= - (BIO_MAX_VECS * PAGE_SIZE) || + (BIO_MAX_PAGES * PAGE_SIZE) || bio_end_sector(&w->io->op.wbio.bio) != sector)) bch2_writepage_do_io(w); @@ -1727,8 +1793,8 @@ static int __bch2_buffered_write(struct bch_inode_info *inode, unsigned pg_offset = (offset + copied) & (PAGE_SIZE - 1); unsigned pg_len = min_t(unsigned, len - copied, PAGE_SIZE - pg_offset); - unsigned pg_copied = copy_page_from_iter_atomic(page, - pg_offset, pg_len,iter); + unsigned pg_copied = iov_iter_copy_from_user_atomic(page, + iter, pg_offset, pg_len); if (!pg_copied) break; @@ -1741,6 +1807,7 @@ static int __bch2_buffered_write(struct bch_inode_info *inode, } flush_dcache_page(page); + iov_iter_advance(iter, pg_copied); copied += pg_copied; if (pg_copied != pg_len) @@ -1858,6 +1925,18 @@ again: /* O_DIRECT reads */ +static void bio_release_pages(struct bio *bio, bool mark_dirty) +{ + struct bio_vec *bvec; + unsigned i; + + bio_for_each_segment_all(bvec, bio, i) { + if (mark_dirty && !PageCompound(bvec->bv_page)) + set_page_dirty_lock(bvec->bv_page); + put_page(bvec->bv_page); + } +} + static void bio_check_or_release(struct bio *bio, bool check_dirty) { if (check_dirty) { @@ -1921,7 +2000,7 @@ static int bch2_direct_IO_read(struct kiocb *req, struct iov_iter *iter) iter->count -= shorten; bio = bio_alloc_bioset(GFP_KERNEL, - iov_iter_npages(iter, BIO_MAX_VECS), + iov_iter_npages(iter, BIO_MAX_PAGES), &c->dio_read_bioset); bio->bi_end_io = bch2_direct_IO_read_endio; @@ -1956,7 +2035,7 @@ static int bch2_direct_IO_read(struct kiocb *req, struct iov_iter *iter) goto start; while (iter->count) { bio = bio_alloc_bioset(GFP_KERNEL, - iov_iter_npages(iter, BIO_MAX_VECS), + iov_iter_npages(iter, BIO_MAX_PAGES), &c->bio_read); bio->bi_end_io = bch2_direct_IO_read_split_endio; start: @@ -2089,9 +2168,8 @@ static long bch2_dio_write_loop(struct dio_write *dio) struct bch_inode_info *inode = file_bch_inode(req->ki_filp); struct bch_fs *c = inode->v.i_sb->s_fs_info; struct bio *bio = &dio->op.wbio.bio; - struct bvec_iter_all iter; struct bio_vec *bv; - unsigned unaligned, iter_count; + unsigned i, unaligned, iter_count; bool sync = dio->sync, dropped_locks; long ret; @@ -2104,7 +2182,7 @@ static long bch2_dio_write_loop(struct dio_write *dio) iter_count = dio->iter.count; if (kthread) - kthread_use_mm(dio->mm); + use_mm(dio->mm); BUG_ON(current->faults_disabled_mapping); current->faults_disabled_mapping = mapping; @@ -2114,7 +2192,7 @@ static long bch2_dio_write_loop(struct dio_write *dio) current->faults_disabled_mapping = NULL; if (kthread) - kthread_unuse_mm(dio->mm); + unuse_mm(dio->mm); /* * If the fault handler returned an error but also signalled @@ -2211,9 +2289,8 @@ loop: i_size_write(&inode->v, req->ki_pos); spin_unlock(&inode->v.i_lock); - if (likely(!bio_flagged(bio, BIO_NO_PAGE_REF))) - bio_for_each_segment_all(bv, bio, iter) - put_page(bv->bv_page); + bio_for_each_segment_all(bv, bio, i) + put_page(bv->bv_page); bio->bi_vcnt = 0; if (dio->op.error) { @@ -2237,9 +2314,8 @@ err: if (dio->free_iov) kfree(dio->iter.iov); - if (likely(!bio_flagged(bio, BIO_NO_PAGE_REF))) - bio_for_each_segment_all(bv, bio, iter) - put_page(bv->bv_page); + bio_for_each_segment_all(bv, bio, i) + put_page(bv->bv_page); bio_put(bio); /* inode->i_dio_count is our ref on inode and thus bch_fs */ @@ -2306,9 +2382,7 @@ ssize_t bch2_direct_write(struct kiocb *req, struct iov_iter *iter) } bio = bio_alloc_bioset(GFP_KERNEL, - iov_iter_is_bvec(iter) - ? 0 - : iov_iter_npages(iter, BIO_MAX_VECS), + iov_iter_npages(iter, BIO_MAX_PAGES), &c->dio_write_bioset); dio = container_of(bio, struct dio_write, op.wbio.bio); init_completion(&dio->done); @@ -2595,7 +2669,7 @@ static int bch2_extend(struct user_namespace *mnt_userns, truncate_setsize(&inode->v, iattr->ia_size); - return bch2_setattr_nonsize(mnt_userns, inode, iattr); + return bch2_setattr_nonsize(inode, iattr); } static int bch2_truncate_finish_fn(struct bch_inode_info *inode, @@ -2715,7 +2789,7 @@ int bch2_truncate(struct user_namespace *mnt_userns, ret = bch2_write_inode(c, inode, bch2_truncate_finish_fn, NULL, 0); mutex_unlock(&inode->ei_update_lock); - ret = bch2_setattr_nonsize(mnt_userns, inode, iattr); + ret = bch2_setattr_nonsize(inode, iattr); err: bch2_pagecache_block_put(&inode->ei_pagecache_lock); return ret; @@ -3156,6 +3230,235 @@ long bch2_fallocate_dispatch(struct file *file, int mode, return ret; } +static int generic_access_check_limits(struct file *file, loff_t pos, + loff_t *count) +{ + struct inode *inode = file->f_mapping->host; + loff_t max_size = inode->i_sb->s_maxbytes; + + if (!(file->f_flags & O_LARGEFILE)) + max_size = MAX_NON_LFS; + + if (unlikely(pos >= max_size)) + return -EFBIG; + *count = min(*count, max_size - pos); + return 0; +} + +static int generic_write_check_limits(struct file *file, loff_t pos, + loff_t *count) +{ + loff_t limit = rlimit(RLIMIT_FSIZE); + + if (limit != RLIM_INFINITY) { + if (pos >= limit) { + send_sig(SIGXFSZ, current, 0); + return -EFBIG; + } + *count = min(*count, limit - pos); + } + + return generic_access_check_limits(file, pos, count); +} + +static int generic_remap_checks(struct file *file_in, loff_t pos_in, + struct file *file_out, loff_t pos_out, + loff_t *req_count, unsigned int remap_flags) +{ + struct inode *inode_in = file_in->f_mapping->host; + struct inode *inode_out = file_out->f_mapping->host; + uint64_t count = *req_count; + uint64_t bcount; + loff_t size_in, size_out; + loff_t bs = inode_out->i_sb->s_blocksize; + int ret; + + /* The start of both ranges must be aligned to an fs block. */ + if (!IS_ALIGNED(pos_in, bs) || !IS_ALIGNED(pos_out, bs)) + return -EINVAL; + + /* Ensure offsets don't wrap. */ + if (pos_in + count < pos_in || pos_out + count < pos_out) + return -EINVAL; + + size_in = i_size_read(inode_in); + size_out = i_size_read(inode_out); + + /* Dedupe requires both ranges to be within EOF. */ + if ((remap_flags & REMAP_FILE_DEDUP) && + (pos_in >= size_in || pos_in + count > size_in || + pos_out >= size_out || pos_out + count > size_out)) + return -EINVAL; + + /* Ensure the infile range is within the infile. */ + if (pos_in >= size_in) + return -EINVAL; + count = min(count, size_in - (uint64_t)pos_in); + + ret = generic_access_check_limits(file_in, pos_in, &count); + if (ret) + return ret; + + ret = generic_write_check_limits(file_out, pos_out, &count); + if (ret) + return ret; + + /* + * If the user wanted us to link to the infile's EOF, round up to the + * next block boundary for this check. + * + * Otherwise, make sure the count is also block-aligned, having + * already confirmed the starting offsets' block alignment. + */ + if (pos_in + count == size_in) { + bcount = ALIGN(size_in, bs) - pos_in; + } else { + if (!IS_ALIGNED(count, bs)) + count = ALIGN_DOWN(count, bs); + bcount = count; + } + + /* Don't allow overlapped cloning within the same file. */ + if (inode_in == inode_out && + pos_out + bcount > pos_in && + pos_out < pos_in + bcount) + return -EINVAL; + + /* + * We shortened the request but the caller can't deal with that, so + * bounce the request back to userspace. + */ + if (*req_count != count && !(remap_flags & REMAP_FILE_CAN_SHORTEN)) + return -EINVAL; + + *req_count = count; + return 0; +} + +static int generic_remap_check_len(struct inode *inode_in, + struct inode *inode_out, + loff_t pos_out, + loff_t *len, + unsigned int remap_flags) +{ + u64 blkmask = i_blocksize(inode_in) - 1; + loff_t new_len = *len; + + if ((*len & blkmask) == 0) + return 0; + + if ((remap_flags & REMAP_FILE_DEDUP) || + pos_out + *len < i_size_read(inode_out)) + new_len &= ~blkmask; + + if (new_len == *len) + return 0; + + if (remap_flags & REMAP_FILE_CAN_SHORTEN) { + *len = new_len; + return 0; + } + + return (remap_flags & REMAP_FILE_DEDUP) ? -EBADE : -EINVAL; +} + +static int generic_remap_file_range_prep(struct file *file_in, loff_t pos_in, + struct file *file_out, loff_t pos_out, + loff_t *len, unsigned int remap_flags) +{ + struct inode *inode_in = file_inode(file_in); + struct inode *inode_out = file_inode(file_out); + bool same_inode = (inode_in == inode_out); + int ret; + + /* Don't touch certain kinds of inodes */ + if (IS_IMMUTABLE(inode_out)) + return -EPERM; + + if (IS_SWAPFILE(inode_in) || IS_SWAPFILE(inode_out)) + return -ETXTBSY; + + /* Don't reflink dirs, pipes, sockets... */ + if (S_ISDIR(inode_in->i_mode) || S_ISDIR(inode_out->i_mode)) + return -EISDIR; + if (!S_ISREG(inode_in->i_mode) || !S_ISREG(inode_out->i_mode)) + return -EINVAL; + + /* Zero length dedupe exits immediately; reflink goes to EOF. */ + if (*len == 0) { + loff_t isize = i_size_read(inode_in); + + if ((remap_flags & REMAP_FILE_DEDUP) || pos_in == isize) + return 0; + if (pos_in > isize) + return -EINVAL; + *len = isize - pos_in; + if (*len == 0) + return 0; + } + + /* Check that we don't violate system file offset limits. */ + ret = generic_remap_checks(file_in, pos_in, file_out, pos_out, len, + remap_flags); + if (ret) + return ret; + + /* Wait for the completion of any pending IOs on both files */ + inode_dio_wait(inode_in); + if (!same_inode) + inode_dio_wait(inode_out); + + ret = filemap_write_and_wait_range(inode_in->i_mapping, + pos_in, pos_in + *len - 1); + if (ret) + return ret; + + ret = filemap_write_and_wait_range(inode_out->i_mapping, + pos_out, pos_out + *len - 1); + if (ret) + return ret; + + /* + * Check that the extents are the same. + */ + if (remap_flags & REMAP_FILE_DEDUP) { + bool is_same = false; + + ret = vfs_dedupe_file_range_compare(inode_in, pos_in, + inode_out, pos_out, *len, &is_same); + if (ret) + return ret; + if (!is_same) + return -EBADE; + } + + ret = generic_remap_check_len(inode_in, inode_out, pos_out, len, + remap_flags); + if (ret) + return ret; + + /* If can't alter the file contents, we're done. */ + if (!(remap_flags & REMAP_FILE_DEDUP)) { + /* Update the timestamps, since we can alter file contents. */ + if (!(file_out->f_mode & FMODE_NOCMTIME)) { + ret = file_update_time(file_out); + if (ret) + return ret; + } + + /* + * Clear the security bits if the process is not being run by + * root. This keeps people from modifying setuid and setgid + * binaries. + */ + ret = file_remove_privs(file_out); + if (ret) + return ret; + } + + return 0; +} + loff_t bch2_remap_file_range(struct file *file_src, loff_t pos_src, struct file *file_dst, loff_t pos_dst, loff_t len, unsigned remap_flags) diff --git a/fs/bcachefs/fs-io.h b/fs/bcachefs/fs-io.h index b24efeaf343e..f9e7f49b13c7 100644 --- a/fs/bcachefs/fs-io.h +++ b/fs/bcachefs/fs-io.h @@ -19,7 +19,8 @@ int bch2_writepage(struct page *, struct writeback_control *); int bch2_readpage(struct file *, struct page *); int bch2_writepages(struct address_space *, struct writeback_control *); -void bch2_readahead(struct readahead_control *); +int bch2_readpages(struct file *, struct address_space *, + struct list_head *, unsigned); int bch2_write_begin(struct file *, struct address_space *, loff_t, unsigned, unsigned, struct page **, void **); @@ -35,6 +36,10 @@ int bch2_truncate(struct user_namespace *, struct bch_inode_info *, struct iattr *); long bch2_fallocate_dispatch(struct file *, int, loff_t, loff_t); +#define REMAP_FILE_ADVISORY (0) +#define REMAP_FILE_DEDUP (1 << 0) +#define REMAP_FILE_CAN_SHORTEN (1 << 1) + loff_t bch2_remap_file_range(struct file *, loff_t, struct file *, loff_t, loff_t, unsigned); diff --git a/fs/bcachefs/fs-ioctl.c b/fs/bcachefs/fs-ioctl.c index 9f329a624c12..a76017386593 100644 --- a/fs/bcachefs/fs-ioctl.c +++ b/fs/bcachefs/fs-ioctl.c @@ -85,7 +85,7 @@ static int bch2_ioc_setflags(struct bch_fs *c, return ret; inode_lock(&inode->v); - if (!inode_owner_or_capable(file_mnt_user_ns(file), &inode->v)) { + if (!inode_owner_or_capable(&inode->v)) { ret = -EACCES; goto setflags_out; } @@ -156,7 +156,7 @@ static int bch2_ioc_fssetxattr(struct bch_fs *c, return ret; inode_lock(&inode->v); - if (!inode_owner_or_capable(file_mnt_user_ns(file), &inode->v)) { + if (!inode_owner_or_capable(&inode->v)) { ret = -EACCES; goto err; } @@ -268,20 +268,22 @@ static int bch2_ioc_goingdown(struct bch_fs *c, u32 __user *arg) down_write(&c->vfs_sb->s_umount); switch (flags) { - case FSOP_GOING_FLAGS_DEFAULT: - ret = freeze_bdev(c->vfs_sb->s_bdev); + case FSOP_GOING_FLAGS_DEFAULT: { + struct super_block *sb = freeze_bdev(c->vfs_sb->s_bdev); if (ret) goto err; - bch2_journal_flush(&c->journal); - c->vfs_sb->s_flags |= SB_RDONLY; - bch2_fs_emergency_read_only(c); - thaw_bdev(c->vfs_sb->s_bdev); + if (sb && !IS_ERR(sb)) { + bch2_journal_flush(&c->journal); + c->vfs_sb->s_flags |= SB_RDONLY; + bch2_fs_emergency_read_only(c); + thaw_bdev(c->vfs_sb->s_bdev, sb); + } break; + } case FSOP_GOING_FLAGS_LOGFLUSH: bch2_journal_flush(&c->journal); - fallthrough; case FSOP_GOING_FLAGS_NOLOGFLUSH: c->vfs_sb->s_flags |= SB_RDONLY; @@ -377,8 +379,7 @@ retry: goto err3; } - error = inode_permission(file_mnt_user_ns(filp), - dir, MAY_WRITE | MAY_EXEC); + error = inode_permission(dir, MAY_WRITE | MAY_EXEC); if (error) goto err3; @@ -393,7 +394,7 @@ retry: !arg.src_ptr) snapshot_src.subvol = to_bch_ei(dir)->ei_inode.bi_subvol; - inode = __bch2_create(file_mnt_user_ns(filp), to_bch_ei(dir), + inode = __bch2_create(NULL, to_bch_ei(dir), dst_dentry, arg.mode|S_IFDIR, 0, snapshot_src, create_flags); error = PTR_ERR_OR_ZERO(inode); @@ -442,10 +443,8 @@ static long bch2_ioctl_subvolume_destroy(struct bch_fs *c, struct file *filp, dir = path.dentry->d_parent->d_inode; ret = __bch2_unlink(dir, path.dentry, true); - if (!ret) { - fsnotify_rmdir(dir, path.dentry); + if (!ret) d_delete(path.dentry); - } path_put(&path); return ret; diff --git a/fs/bcachefs/fs.c b/fs/bcachefs/fs.c index 472c03d2adb5..60e6ba4918c4 100644 --- a/fs/bcachefs/fs.c +++ b/fs/bcachefs/fs.c @@ -263,6 +263,7 @@ __bch2_create(struct user_namespace *mnt_userns, unsigned flags) { struct bch_fs *c = dir->v.i_sb->s_fs_info; + struct user_namespace *ns = dir->v.i_sb->s_user_ns; struct btree_trans trans; struct bch_inode_unpacked dir_u; struct bch_inode_info *inode, *old; @@ -303,8 +304,8 @@ retry: inode_inum(dir), &dir_u, &inode_u, !(flags & BCH_CREATE_TMPFILE) ? &dentry->d_name : NULL, - from_kuid(mnt_userns, current_fsuid()), - from_kgid(mnt_userns, current_fsgid()), + from_kuid(ns, current_fsuid()), + from_kgid(ns, current_fsgid()), mode, rdev, default_acl, acl, snapshot_src, flags) ?: bch2_quota_acct(c, bch_qid(&inode_u), Q_INO, 1, @@ -408,12 +409,11 @@ static struct dentry *bch2_lookup(struct inode *vdir, struct dentry *dentry, return d_splice_alias(vinode, dentry); } -static int bch2_mknod(struct user_namespace *mnt_userns, - struct inode *vdir, struct dentry *dentry, +static int bch2_mknod(struct inode *vdir, struct dentry *dentry, umode_t mode, dev_t rdev) { struct bch_inode_info *inode = - __bch2_create(mnt_userns, to_bch_ei(vdir), dentry, mode, rdev, + __bch2_create(NULL, to_bch_ei(vdir), dentry, mode, rdev, (subvol_inum) { 0 }, 0); if (IS_ERR(inode)) @@ -423,11 +423,10 @@ static int bch2_mknod(struct user_namespace *mnt_userns, return 0; } -static int bch2_create(struct user_namespace *mnt_userns, - struct inode *vdir, struct dentry *dentry, +static int bch2_create(struct inode *vdir, struct dentry *dentry, umode_t mode, bool excl) { - return bch2_mknod(mnt_userns, vdir, dentry, mode|S_IFREG, 0); + return bch2_mknod(vdir, dentry, mode|S_IFREG, 0); } static int __bch2_link(struct bch_fs *c, @@ -516,15 +515,14 @@ static int bch2_unlink(struct inode *vdir, struct dentry *dentry) return __bch2_unlink(vdir, dentry, false); } -static int bch2_symlink(struct user_namespace *mnt_userns, - struct inode *vdir, struct dentry *dentry, +static int bch2_symlink(struct inode *vdir, struct dentry *dentry, const char *symname) { struct bch_fs *c = vdir->i_sb->s_fs_info; struct bch_inode_info *dir = to_bch_ei(vdir), *inode; int ret; - inode = __bch2_create(mnt_userns, dir, dentry, S_IFLNK|S_IRWXUGO, 0, + inode = __bch2_create(NULL, dir, dentry, S_IFLNK|S_IRWXUGO, 0, (subvol_inum) { 0 }, BCH_CREATE_TMPFILE); if (unlikely(IS_ERR(inode))) return PTR_ERR(inode); @@ -551,14 +549,12 @@ err: return ret; } -static int bch2_mkdir(struct user_namespace *mnt_userns, - struct inode *vdir, struct dentry *dentry, umode_t mode) +static int bch2_mkdir(struct inode *vdir, struct dentry *dentry, umode_t mode) { - return bch2_mknod(mnt_userns, vdir, dentry, mode|S_IFDIR, 0); + return bch2_mknod(vdir, dentry, mode|S_IFDIR, 0); } -static int bch2_rename2(struct user_namespace *mnt_userns, - struct inode *src_vdir, struct dentry *src_dentry, +static int bch2_rename2(struct inode *src_vdir, struct dentry *src_dentry, struct inode *dst_vdir, struct dentry *dst_dentry, unsigned flags) { @@ -664,8 +660,7 @@ err: return ret; } -static void bch2_setattr_copy(struct user_namespace *mnt_userns, - struct bch_inode_info *inode, +static void bch2_setattr_copy(struct bch_inode_info *inode, struct bch_inode_unpacked *bi, struct iattr *attr) { @@ -673,9 +668,9 @@ static void bch2_setattr_copy(struct user_namespace *mnt_userns, unsigned int ia_valid = attr->ia_valid; if (ia_valid & ATTR_UID) - bi->bi_uid = from_kuid(mnt_userns, attr->ia_uid); + bi->bi_uid = from_kuid(c->vfs_sb->s_user_ns, attr->ia_uid); if (ia_valid & ATTR_GID) - bi->bi_gid = from_kgid(mnt_userns, attr->ia_gid); + bi->bi_gid = from_kgid(c->vfs_sb->s_user_ns, attr->ia_gid); if (ia_valid & ATTR_SIZE) bi->bi_size = attr->ia_size; @@ -694,14 +689,13 @@ static void bch2_setattr_copy(struct user_namespace *mnt_userns, : inode->v.i_gid; if (!in_group_p(gid) && - !capable_wrt_inode_uidgid(mnt_userns, &inode->v, CAP_FSETID)) + !capable_wrt_inode_uidgid(&inode->v, CAP_FSETID)) mode &= ~S_ISGID; bi->bi_mode = mode; } } -int bch2_setattr_nonsize(struct user_namespace *mnt_userns, - struct bch_inode_info *inode, +int bch2_setattr_nonsize(struct bch_inode_info *inode, struct iattr *attr) { struct bch_fs *c = inode->v.i_sb->s_fs_info; @@ -738,7 +732,7 @@ retry: if (ret) goto btree_err; - bch2_setattr_copy(mnt_userns, inode, &inode_u, attr); + bch2_setattr_copy(inode, &inode_u, attr); if (attr->ia_valid & ATTR_MODE) { ret = bch2_acl_chmod(&trans, inode_inum(inode), &inode_u, @@ -770,8 +764,7 @@ err: return ret; } -static int bch2_getattr(struct user_namespace *mnt_userns, - const struct path *path, struct kstat *stat, +static int bch2_getattr(const struct path *path, struct kstat *stat, u32 request_mask, unsigned query_flags) { struct bch_inode_info *inode = to_bch_ei(d_inode(path->dentry)); @@ -811,28 +804,26 @@ static int bch2_getattr(struct user_namespace *mnt_userns, return 0; } -static int bch2_setattr(struct user_namespace *mnt_userns, - struct dentry *dentry, struct iattr *iattr) +static int bch2_setattr(struct dentry *dentry, struct iattr *iattr) { struct bch_inode_info *inode = to_bch_ei(dentry->d_inode); int ret; lockdep_assert_held(&inode->v.i_rwsem); - ret = setattr_prepare(mnt_userns, dentry, iattr); + ret = setattr_prepare(dentry, iattr); if (ret) return ret; return iattr->ia_valid & ATTR_SIZE - ? bch2_truncate(mnt_userns, inode, iattr) - : bch2_setattr_nonsize(mnt_userns, inode, iattr); + ? bch2_truncate(NULL, inode, iattr) + : bch2_setattr_nonsize(inode, iattr); } -static int bch2_tmpfile(struct user_namespace *mnt_userns, - struct inode *vdir, struct dentry *dentry, umode_t mode) +static int bch2_tmpfile(struct inode *vdir, struct dentry *dentry, umode_t mode) { struct bch_inode_info *inode = - __bch2_create(mnt_userns, to_bch_ei(vdir), dentry, mode, 0, + __bch2_create(NULL, to_bch_ei(vdir), dentry, mode, 0, (subvol_inum) { 0 }, BCH_CREATE_TMPFILE); if (IS_ERR(inode)) @@ -911,10 +902,6 @@ static int bch2_fiemap(struct inode *vinode, struct fiemap_extent_info *info, u32 snapshot; int ret = 0; - ret = fiemap_prep(&ei->v, info, start, &len, FIEMAP_FLAG_SYNC); - if (ret) - return ret; - if (start + len < start) return -EINVAL; @@ -1030,6 +1017,15 @@ static int bch2_vfs_readdir(struct file *file, struct dir_context *ctx) return bch2_readdir(c, inode_inum(inode), ctx); } +static int bch2_clone_file_range(struct file *file_src, loff_t pos_src, + struct file *file_dst, loff_t pos_dst, + u64 len) +{ + return bch2_remap_file_range(file_src, pos_src, + file_dst, pos_dst, + len, 0); +} + static const struct file_operations bch_file_operations = { .llseek = bch2_llseek, .read_iter = bch2_read_iter, @@ -1044,7 +1040,7 @@ static const struct file_operations bch_file_operations = { #ifdef CONFIG_COMPAT .compat_ioctl = bch2_compat_fs_ioctl, #endif - .remap_file_range = bch2_remap_file_range, + .clone_file_range = bch2_clone_file_range, }; static const struct inode_operations bch_file_inode_operations = { @@ -1114,7 +1110,7 @@ static const struct address_space_operations bch_address_space_operations = { .writepage = bch2_writepage, .readpage = bch2_readpage, .writepages = bch2_writepages, - .readahead = bch2_readahead, + .readpages = bch2_readpages, .set_page_dirty = __set_page_dirty_nobuffers, .write_begin = bch2_write_begin, .write_end = bch2_write_end, @@ -1571,14 +1567,13 @@ static int bch2_sync_fs(struct super_block *sb, int wait) static struct bch_fs *bch2_path_to_fs(const char *path) { struct bch_fs *c; - dev_t dev; - int ret; + struct block_device *bdev = lookup_bdev(path); - ret = lookup_bdev(path, &dev); - if (ret) - return ERR_PTR(ret); + if (IS_ERR(bdev)) + return ERR_CAST(bdev); - c = bch2_dev_to_fs(dev); + c = bch2_dev_to_fs(bdev->bd_dev); + bdput(bdev); if (c) closure_put(&c->cl); return c ?: ERR_PTR(-ENOENT); @@ -1834,8 +1829,6 @@ got_sb: sb->s_xattr = bch2_xattr_handlers; sb->s_magic = BCACHEFS_STATFS_MAGIC; sb->s_time_gran = c->sb.nsec_per_time_unit; - sb->s_time_min = div_s64(S64_MIN, c->sb.time_units_per_sec) + 1; - sb->s_time_max = div_s64(S64_MAX, c->sb.time_units_per_sec); c->vfs_sb = sb; strlcpy(sb->s_id, c->name, sizeof(sb->s_id)); @@ -1843,7 +1836,9 @@ got_sb: if (ret) goto err_put_super; - sb->s_bdi->ra_pages = VM_READAHEAD_PAGES; + sb->s_bdi->congested_fn = bch2_congested; + sb->s_bdi->congested_data = c; + sb->s_bdi->ra_pages = VM_MAX_READAHEAD * 1024 / PAGE_SIZE; for_each_online_member(ca, c, i) { struct block_device *bdev = ca->disk_sb.bdev; @@ -1862,7 +1857,7 @@ got_sb: sb->s_flags |= SB_POSIXACL; #endif - sb->s_shrink.seeks = 0; + sb->s_shrink.seeks = 1; vinode = bch2_vfs_inode_get(c, BCACHEFS_ROOT_SUBVOL_INUM); if (IS_ERR(vinode)) { diff --git a/fs/bcachefs/fs.h b/fs/bcachefs/fs.h index b2211ec7f302..b5bc70afb100 100644 --- a/fs/bcachefs/fs.h +++ b/fs/bcachefs/fs.h @@ -186,8 +186,7 @@ void bch2_inode_update_after_write(struct btree_trans *, int __must_check bch2_write_inode(struct bch_fs *, struct bch_inode_info *, inode_set_fn, void *, unsigned); -int bch2_setattr_nonsize(struct user_namespace *, - struct bch_inode_info *, +int bch2_setattr_nonsize(struct bch_inode_info *, struct iattr *); int __bch2_unlink(struct inode *, struct dentry *, bool); diff --git a/fs/bcachefs/fsck.c b/fs/bcachefs/fsck.c index 43b6159be01b..69b4136364c3 100644 --- a/fs/bcachefs/fsck.c +++ b/fs/bcachefs/fsck.c @@ -2052,8 +2052,8 @@ static void inc_link(struct bch_fs *c, struct snapshots_seen *s, if (inum < range_start || inum >= range_end) return; - link = __inline_bsearch(&key, links->d, links->nr, - sizeof(links->d[0]), nlink_cmp); + link = bsearch(&key, links->d, links->nr, + sizeof(links->d[0]), nlink_cmp); if (!link) return; diff --git a/fs/bcachefs/io.c b/fs/bcachefs/io.c index 73558cd00350..21e1e3956f10 100644 --- a/fs/bcachefs/io.c +++ b/fs/bcachefs/io.c @@ -136,10 +136,10 @@ void bch2_latency_acct(struct bch_dev *ca, u64 submit_time, int rw) void bch2_bio_free_pages_pool(struct bch_fs *c, struct bio *bio) { - struct bvec_iter_all iter; struct bio_vec *bv; + unsigned i; - bio_for_each_segment_all(bv, bio, iter) + bio_for_each_segment_all(bv, bio, i) if (bv->bv_page != ZERO_PAGE(0)) mempool_free(bv->bv_page, &c->bio_bounce_pages); bio->bi_vcnt = 0; @@ -699,7 +699,7 @@ static struct bio *bch2_write_bio_alloc(struct bch_fs *c, ? ((unsigned long) buf & (PAGE_SIZE - 1)) : 0), PAGE_SIZE); - pages = min(pages, BIO_MAX_VECS); + pages = min_t(unsigned, pages, BIO_MAX_PAGES); bio = bio_alloc_bioset(GFP_NOIO, pages, &c->bio_write); wbio = wbio_init(bio); diff --git a/fs/bcachefs/journal.h b/fs/bcachefs/journal.h index b298873212d2..62f9aec4a427 100644 --- a/fs/bcachefs/journal.h +++ b/fs/bcachefs/journal.h @@ -278,7 +278,7 @@ static inline void bch2_journal_res_put(struct journal *j, if (!res->ref) return; - lock_release(&j->res_map, _THIS_IP_); + lock_release(&j->res_map, 0, _THIS_IP_); while (res->u64s) bch2_journal_add_entry(j, res, diff --git a/fs/bcachefs/journal_reclaim.c b/fs/bcachefs/journal_reclaim.c index 52a3935cff53..b9bf381ef15e 100644 --- a/fs/bcachefs/journal_reclaim.c +++ b/fs/bcachefs/journal_reclaim.c @@ -11,6 +11,7 @@ #include <linux/kthread.h> #include <linux/sched/mm.h> +#include <linux/sched/task.h> #include <trace/events/bcachefs.h> /* Free space calculations: */ diff --git a/fs/bcachefs/move.c b/fs/bcachefs/move.c index f73be9cb7ac3..9b6e402e19f0 100644 --- a/fs/bcachefs/move.c +++ b/fs/bcachefs/move.c @@ -427,12 +427,12 @@ static void move_free(struct closure *cl) { struct moving_io *io = container_of(cl, struct moving_io, cl); struct moving_context *ctxt = io->write.ctxt; - struct bvec_iter_all iter; struct bio_vec *bv; + unsigned i; bch2_disk_reservation_put(io->write.op.c, &io->write.op.res); - bio_for_each_segment_all(bv, &io->write.op.wbio.bio, iter) + bio_for_each_segment_all(bv, &io->write.op.wbio.bio, i) if (bv->bv_page) __free_page(bv->bv_page); @@ -784,14 +784,6 @@ out: return ret; } -inline void bch_move_stats_init(struct bch_move_stats *stats, char *name) -{ - memset(stats, 0, sizeof(*stats)); - - scnprintf(stats->name, sizeof(stats->name), - "%s", name); -} - static inline void progress_list_add(struct bch_fs *c, struct bch_move_stats *stats) { diff --git a/fs/bcachefs/move.h b/fs/bcachefs/move.h index 2a789a1158ca..98323ad93e7c 100644 --- a/fs/bcachefs/move.h +++ b/fs/bcachefs/move.h @@ -66,8 +66,13 @@ int bch2_data_job(struct bch_fs *, struct bch_move_stats *, struct bch_ioctl_data); -inline void bch_move_stats_init(struct bch_move_stats *stats, - char *name); +static inline void bch_move_stats_init(struct bch_move_stats *stats, char *name) +{ + memset(stats, 0, sizeof(*stats)); + + scnprintf(stats->name, sizeof(stats->name), + "%s", name); +} #endif /* _BCACHEFS_MOVE_H */ diff --git a/fs/bcachefs/str_hash.h b/fs/bcachefs/str_hash.h index 57d636740d2f..5de733b95aa4 100644 --- a/fs/bcachefs/str_hash.h +++ b/fs/bcachefs/str_hash.h @@ -13,7 +13,7 @@ #include <linux/crc32c.h> #include <crypto/hash.h> -#include <crypto/sha2.h> +#include <crypto/sha.h> static inline enum bch_str_hash_type bch2_str_hash_opt_to_type(struct bch_fs *c, enum bch_str_hash_opts opt) diff --git a/fs/bcachefs/super-io.c b/fs/bcachefs/super-io.c index 49dafdad77cd..6e2e077f5f8d 100644 --- a/fs/bcachefs/super-io.c +++ b/fs/bcachefs/super-io.c @@ -288,7 +288,7 @@ static int bch2_sb_validate(struct bch_sb_handle *disk_sb, struct printbuf *out) block_size = le16_to_cpu(sb->block_size); if (block_size > PAGE_SECTORS) { - pr_buf(out, "Block size too big (got %u, max %u)", + pr_buf(out, "Block size too big (got %u, max %lu)", block_size, PAGE_SECTORS); return -EINVAL; } diff --git a/fs/bcachefs/super.c b/fs/bcachefs/super.c index 586ba60d03ea..e706b203a030 100644 --- a/fs/bcachefs/super.c +++ b/fs/bcachefs/super.c @@ -167,6 +167,44 @@ static void bch2_dev_usage_journal_reserve(struct bch_fs *c) &c->dev_usage_journal_res, u64s * nr); } +int bch2_congested(void *data, int bdi_bits) +{ + struct bch_fs *c = data; + struct backing_dev_info *bdi; + struct bch_dev *ca; + unsigned i; + int ret = 0; + + rcu_read_lock(); + if (bdi_bits & (1 << WB_sync_congested)) { + /* Reads - check all devices: */ + for_each_readable_member(ca, c, i) { + bdi = ca->disk_sb.bdev->bd_bdi; + + if (bdi_congested(bdi, bdi_bits)) { + ret = 1; + break; + } + } + } else { + const struct bch_devs_mask *devs = + bch2_target_to_mask(c, c->opts.foreground_target) ?: + &c->rw_devs[BCH_DATA_user]; + + for_each_member_device_rcu(ca, c, i, devs) { + bdi = ca->disk_sb.bdev->bd_bdi; + + if (bdi_congested(bdi, bdi_bits)) { + ret = 1; + break; + } + } + } + rcu_read_unlock(); + + return ret; +} + /* Filesystem RO/RW: */ /* @@ -539,7 +577,8 @@ void __bch2_fs_stop(struct bch_fs *c) for_each_member_device(ca, c, i) if (ca->kobj.state_in_sysfs && ca->disk_sb.bdev) - sysfs_remove_link(bdev_kobj(ca->disk_sb.bdev), "bcachefs"); + sysfs_remove_link(&part_to_dev(ca->disk_sb.bdev->bd_part)->kobj, + "bcachefs"); if (c->kobj.state_in_sysfs) kobject_del(&c->kobj); @@ -1046,7 +1085,8 @@ static void bch2_dev_free(struct bch_dev *ca) if (ca->kobj.state_in_sysfs && ca->disk_sb.bdev) - sysfs_remove_link(bdev_kobj(ca->disk_sb.bdev), "bcachefs"); + sysfs_remove_link(&part_to_dev(ca->disk_sb.bdev->bd_part)->kobj, + "bcachefs"); if (ca->kobj.state_in_sysfs) kobject_del(&ca->kobj); @@ -1082,7 +1122,10 @@ static void __bch2_dev_offline(struct bch_fs *c, struct bch_dev *ca) wait_for_completion(&ca->io_ref_completion); if (ca->kobj.state_in_sysfs) { - sysfs_remove_link(bdev_kobj(ca->disk_sb.bdev), "bcachefs"); + struct kobject *block = + &part_to_dev(ca->disk_sb.bdev->bd_part)->kobj; + + sysfs_remove_link(block, "bcachefs"); sysfs_remove_link(&ca->kobj, "block"); } @@ -1119,12 +1162,12 @@ static int bch2_dev_sysfs_online(struct bch_fs *c, struct bch_dev *ca) } if (ca->disk_sb.bdev) { - struct kobject *block = bdev_kobj(ca->disk_sb.bdev); + struct kobject *block = + &part_to_dev(ca->disk_sb.bdev->bd_part)->kobj; ret = sysfs_create_link(block, &ca->kobj, "bcachefs"); if (ret) return ret; - ret = sysfs_create_link(&ca->kobj, block, "block"); if (ret) return ret; @@ -1874,23 +1917,20 @@ err: /* return with ref on ca->ref: */ struct bch_dev *bch2_dev_lookup(struct bch_fs *c, const char *path) { + struct block_device *bdev = lookup_bdev(path); struct bch_dev *ca; - dev_t dev; unsigned i; - int ret; - ret = lookup_bdev(path, &dev); - if (ret) - return ERR_PTR(ret); + if (IS_ERR(bdev)) + return ERR_CAST(bdev); - rcu_read_lock(); - for_each_member_device_rcu(ca, c, i, NULL) - if (ca->dev == dev) + for_each_member_device(ca, c, i) + if (ca->disk_sb.bdev == bdev) goto found; + ca = ERR_PTR(-ENOENT); found: - rcu_read_unlock(); - + bdput(bdev); return ca; } diff --git a/fs/bcachefs/super.h b/fs/bcachefs/super.h index 3f24ca5a853d..7e118244abe6 100644 --- a/fs/bcachefs/super.h +++ b/fs/bcachefs/super.h @@ -217,6 +217,7 @@ static inline bool is_superblock_bucket(struct bch_dev *ca, u64 b) struct bch_fs *bch2_dev_to_fs(dev_t); struct bch_fs *bch2_uuid_to_fs(uuid_le); +int bch2_congested(void *, int); bool bch2_dev_state_allowed(struct bch_fs *, struct bch_dev *, enum bch_member_state, int); diff --git a/fs/bcachefs/util.h b/fs/bcachefs/util.h index 3196bc303182..df8c5d52f98a 100644 --- a/fs/bcachefs/util.h +++ b/fs/bcachefs/util.h @@ -18,6 +18,9 @@ #include <linux/vmalloc.h> #include <linux/workqueue.h> +#define PAGE_SECTORS_SHIFT (PAGE_SHIFT - 9) +#define PAGE_SECTORS (1UL << PAGE_SECTORS_SHIFT) + struct closure; #ifdef CONFIG_BCACHEFS_DEBUG @@ -85,7 +88,7 @@ static inline void *vpmalloc(size_t size, gfp_t gfp_mask) { return (void *) __get_free_pages(gfp_mask|__GFP_NOWARN, get_order(size)) ?: - __vmalloc(size, gfp_mask); + __vmalloc(size, gfp_mask, PAGE_KERNEL); } static inline void kvpfree(void *p, size_t size) @@ -650,6 +653,35 @@ static inline void memset_u64s_tail(void *s, int c, unsigned bytes) memset(s + bytes, c, rem); } +static inline struct bio_vec next_contig_bvec(struct bio *bio, + struct bvec_iter *iter) +{ + struct bio_vec bv = bio_iter_iovec(bio, *iter); + + bio_advance_iter(bio, iter, bv.bv_len); +#ifndef CONFIG_HIGHMEM + while (iter->bi_size) { + struct bio_vec next = bio_iter_iovec(bio, *iter); + + if (page_address(bv.bv_page) + bv.bv_offset + bv.bv_len != + page_address(next.bv_page) + next.bv_offset) + break; + + bv.bv_len += next.bv_len; + bio_advance_iter(bio, iter, next.bv_len); + } +#endif + return bv; +} + +#define __bio_for_each_contig_segment(bv, bio, iter, start) \ + for (iter = (start); \ + (iter).bi_size && \ + ((bv = next_contig_bvec((bio), &(iter))), 1);) + +#define bio_for_each_contig_segment(bv, bio, iter) \ + __bio_for_each_contig_segment(bv, bio, iter, (bio)->bi_iter) + void sort_cmp_size(void *base, size_t num, size_t size, int (*cmp_func)(const void *, const void *, size_t), void (*swap_func)(void *, void *, size_t)); diff --git a/fs/bcachefs/xattr.c b/fs/bcachefs/xattr.c index 4d7db64e3ef3..7dec2f5e573e 100644 --- a/fs/bcachefs/xattr.c +++ b/fs/bcachefs/xattr.c @@ -359,7 +359,6 @@ static int bch2_xattr_get_handler(const struct xattr_handler *handler, } static int bch2_xattr_set_handler(const struct xattr_handler *handler, - struct user_namespace *mnt_userns, struct dentry *dentry, struct inode *vinode, const char *name, const void *value, size_t size, int flags) @@ -492,7 +491,6 @@ static int inode_opt_set_fn(struct bch_inode_info *inode, } static int bch2_xattr_bcachefs_set(const struct xattr_handler *handler, - struct user_namespace *mnt_userns, struct dentry *dentry, struct inode *vinode, const char *name, const void *value, size_t size, int flags) diff --git a/include/trace/events/bcachefs.h b/include/trace/events/bcachefs.h index 295dcd60e704..4371e80173da 100644 --- a/include/trace/events/bcachefs.h +++ b/include/trace/events/bcachefs.h @@ -71,10 +71,10 @@ DECLARE_EVENT_CLASS(bio, ), TP_fast_assign( - __entry->dev = bio->bi_bdev ? bio_dev(bio) : 0; + __entry->dev = bio->bi_disk ? bio_dev(bio) : 0; __entry->sector = bio->bi_iter.bi_sector; __entry->nr_sector = bio->bi_iter.bi_size >> 9; - blk_fill_rwbs(__entry->rwbs, bio->bi_opf); + blk_fill_rwbs(__entry->rwbs, bio->bi_opf, bio->bi_iter.bi_size); ), TP_printk("%d,%d %s %llu + %u", |