diff options
Diffstat (limited to 'fs')
35 files changed, 548 insertions, 356 deletions
diff --git a/fs/Kconfig.binfmt b/fs/Kconfig.binfmt index 06fb7a93a1bd..4d5ae61580aa 100644 --- a/fs/Kconfig.binfmt +++ b/fs/Kconfig.binfmt @@ -168,21 +168,6 @@ config OSF4_COMPAT with v4 shared libraries freely available from Compaq. If you're going to use shared libraries from Tru64 version 5.0 or later, say N. -config BINFMT_EM86 - tristate "Kernel support for Linux/Intel ELF binaries" - depends on ALPHA - help - Say Y here if you want to be able to execute Linux/Intel ELF - binaries just like native Alpha binaries on your Alpha machine. For - this to work, you need to have the emulator /usr/bin/em86 in place. - - You can get the same functionality by saying N here and saying Y to - "Kernel support for MISC binaries". - - You may answer M to compile the emulation support as a module and - later load the module when you want to use a Linux/Intel binary. The - module will be called binfmt_em86. If unsure, say Y. - config BINFMT_MISC tristate "Kernel support for MISC binaries" help diff --git a/fs/Makefile b/fs/Makefile index 9c708e1fbe8f..f98f3e691c37 100644 --- a/fs/Makefile +++ b/fs/Makefile @@ -39,7 +39,6 @@ obj-$(CONFIG_FS_ENCRYPTION) += crypto/ obj-$(CONFIG_FS_VERITY) += verity/ obj-$(CONFIG_FILE_LOCKING) += locks.o obj-$(CONFIG_BINFMT_AOUT) += binfmt_aout.o -obj-$(CONFIG_BINFMT_EM86) += binfmt_em86.o obj-$(CONFIG_BINFMT_MISC) += binfmt_misc.o obj-$(CONFIG_BINFMT_SCRIPT) += binfmt_script.o obj-$(CONFIG_BINFMT_ELF) += binfmt_elf.o diff --git a/fs/binfmt_em86.c b/fs/binfmt_em86.c deleted file mode 100644 index 06b9b9fddf70..000000000000 --- a/fs/binfmt_em86.c +++ /dev/null @@ -1,110 +0,0 @@ -// SPDX-License-Identifier: GPL-2.0-only -/* - * linux/fs/binfmt_em86.c - * - * Based on linux/fs/binfmt_script.c - * Copyright (C) 1996 Martin von Löwis - * original #!-checking implemented by tytso. - * - * em86 changes Copyright (C) 1997 Jim Paradis - */ - -#include <linux/module.h> -#include <linux/string.h> -#include <linux/stat.h> -#include <linux/binfmts.h> -#include <linux/elf.h> -#include <linux/init.h> -#include <linux/fs.h> -#include <linux/file.h> -#include <linux/errno.h> - - -#define EM86_INTERP "/usr/bin/em86" -#define EM86_I_NAME "em86" - -static int load_em86(struct linux_binprm *bprm) -{ - const char *i_name, *i_arg; - char *interp; - struct file * file; - int retval; - struct elfhdr elf_ex; - - /* Make sure this is a Linux/Intel ELF executable... */ - elf_ex = *((struct elfhdr *)bprm->buf); - - if (memcmp(elf_ex.e_ident, ELFMAG, SELFMAG) != 0) - return -ENOEXEC; - - /* First of all, some simple consistency checks */ - if ((elf_ex.e_type != ET_EXEC && elf_ex.e_type != ET_DYN) || - (!((elf_ex.e_machine == EM_386) || (elf_ex.e_machine == EM_486))) || - !bprm->file->f_op->mmap) { - return -ENOEXEC; - } - - /* Need to be able to load the file after exec */ - if (bprm->interp_flags & BINPRM_FLAGS_PATH_INACCESSIBLE) - return -ENOENT; - - /* Unlike in the script case, we don't have to do any hairy - * parsing to find our interpreter... it's hardcoded! - */ - interp = EM86_INTERP; - i_name = EM86_I_NAME; - i_arg = NULL; /* We reserve the right to add an arg later */ - - /* - * Splice in (1) the interpreter's name for argv[0] - * (2) (optional) argument to interpreter - * (3) filename of emulated file (replace argv[0]) - * - * This is done in reverse order, because of how the - * user environment and arguments are stored. - */ - remove_arg_zero(bprm); - retval = copy_string_kernel(bprm->filename, bprm); - if (retval < 0) return retval; - bprm->argc++; - if (i_arg) { - retval = copy_string_kernel(i_arg, bprm); - if (retval < 0) return retval; - bprm->argc++; - } - retval = copy_string_kernel(i_name, bprm); - if (retval < 0) return retval; - bprm->argc++; - - /* - * OK, now restart the process with the interpreter's inode. - * Note that we use open_exec() as the name is now in kernel - * space, and we don't need to copy it. - */ - file = open_exec(interp); - if (IS_ERR(file)) - return PTR_ERR(file); - - bprm->interpreter = file; - return 0; -} - -static struct linux_binfmt em86_format = { - .module = THIS_MODULE, - .load_binary = load_em86, -}; - -static int __init init_em86_binfmt(void) -{ - register_binfmt(&em86_format); - return 0; -} - -static void __exit exit_em86_binfmt(void) -{ - unregister_binfmt(&em86_format); -} - -core_initcall(init_em86_binfmt); -module_exit(exit_em86_binfmt); -MODULE_LICENSE("GPL"); diff --git a/fs/block_dev.c b/fs/block_dev.c index 0c424a0cadaa..9ef4f1fc2cb0 100644 --- a/fs/block_dev.c +++ b/fs/block_dev.c @@ -812,6 +812,8 @@ static void bdev_free_inode(struct inode *inode) free_percpu(bdev->bd_stats); kfree(bdev->bd_meta_info); + if (!bdev_is_partition(bdev)) + kfree(bdev->bd_disk); kmem_cache_free(bdev_cachep, BDEV_I(inode)); } diff --git a/fs/btrfs/compression.c b/fs/btrfs/compression.c index 9a023ae0f98b..30d82cdf128c 100644 --- a/fs/btrfs/compression.c +++ b/fs/btrfs/compression.c @@ -352,7 +352,7 @@ static void end_compressed_bio_write(struct bio *bio) btrfs_record_physical_zoned(inode, cb->start, bio); btrfs_writepage_endio_finish_ordered(BTRFS_I(inode), NULL, cb->start, cb->start + cb->len - 1, - bio->bi_status == BLK_STS_OK); + !cb->errors); end_compressed_writeback(inode, cb); /* note, our inode could be gone now */ diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index b117dd3b8172..a59ab7b9aea0 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -209,7 +209,7 @@ void btrfs_set_buffer_lockdep_class(u64 objectid, struct extent_buffer *eb, static void csum_tree_block(struct extent_buffer *buf, u8 *result) { struct btrfs_fs_info *fs_info = buf->fs_info; - const int num_pages = fs_info->nodesize >> PAGE_SHIFT; + const int num_pages = num_extent_pages(buf); const int first_page_part = min_t(u32, PAGE_SIZE, fs_info->nodesize); SHASH_DESC_ON_STACK(shash, fs_info->csum_shash); char *kaddr; diff --git a/fs/btrfs/tree-log.c b/fs/btrfs/tree-log.c index 9fd0348be7f5..e6430ac9bbe8 100644 --- a/fs/btrfs/tree-log.c +++ b/fs/btrfs/tree-log.c @@ -6503,8 +6503,8 @@ void btrfs_log_new_name(struct btrfs_trans_handle *trans, * if this inode hasn't been logged and directory we're renaming it * from hasn't been logged, we don't need to log it */ - if (inode->logged_trans < trans->transid && - (!old_dir || old_dir->logged_trans < trans->transid)) + if (!inode_logged(trans, inode) && + (!old_dir || !inode_logged(trans, old_dir))) return; /* diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c index 1e4d43ffe38b..70f94b75f25a 100644 --- a/fs/btrfs/volumes.c +++ b/fs/btrfs/volumes.c @@ -1078,6 +1078,7 @@ static void __btrfs_free_extra_devids(struct btrfs_fs_devices *fs_devices, if (test_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state)) { list_del_init(&device->dev_alloc_list); clear_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state); + fs_devices->rw_devices--; } list_del_init(&device->dev_list); fs_devices->num_devices--; diff --git a/fs/cifs/file.c b/fs/cifs/file.c index cd108607a070..0a72840a88f1 100644 --- a/fs/cifs/file.c +++ b/fs/cifs/file.c @@ -4619,7 +4619,7 @@ read_complete: static int cifs_readpage(struct file *file, struct page *page) { - loff_t offset = (loff_t)page->index << PAGE_SHIFT; + loff_t offset = page_file_offset(page); int rc = -EACCES; unsigned int xid; diff --git a/fs/cifs/fs_context.c b/fs/cifs/fs_context.c index 9a59d7ff9a11..eed59bc1d913 100644 --- a/fs/cifs/fs_context.c +++ b/fs/cifs/fs_context.c @@ -925,6 +925,13 @@ static int smb3_fs_context_parse_param(struct fs_context *fc, ctx->cred_uid = uid; ctx->cruid_specified = true; break; + case Opt_backupuid: + uid = make_kuid(current_user_ns(), result.uint_32); + if (!uid_valid(uid)) + goto cifs_parse_mount_err; + ctx->backupuid = uid; + ctx->backupuid_specified = true; + break; case Opt_backupgid: gid = make_kgid(current_user_ns(), result.uint_32); if (!gid_valid(gid)) diff --git a/fs/cifs/smb2ops.c b/fs/cifs/smb2ops.c index 23d6f4d71649..2dfd0d8297eb 100644 --- a/fs/cifs/smb2ops.c +++ b/fs/cifs/smb2ops.c @@ -3617,7 +3617,8 @@ static int smb3_simple_fallocate_write_range(unsigned int xid, char *buf) { struct cifs_io_parms io_parms = {0}; - int rc, nbytes; + int nbytes; + int rc = 0; struct kvec iov[2]; io_parms.netfid = cfile->fid.netfid; diff --git a/fs/ext2/dir.c b/fs/ext2/dir.c index 14292dba3a12..2c2f179b6977 100644 --- a/fs/ext2/dir.c +++ b/fs/ext2/dir.c @@ -106,12 +106,11 @@ static int ext2_commit_chunk(struct page *page, loff_t pos, unsigned len) return err; } -static bool ext2_check_page(struct page *page, int quiet) +static bool ext2_check_page(struct page *page, int quiet, char *kaddr) { struct inode *dir = page->mapping->host; struct super_block *sb = dir->i_sb; unsigned chunk_size = ext2_chunk_size(dir); - char *kaddr = page_address(page); u32 max_inumber = le32_to_cpu(EXT2_SB(sb)->s_es->s_inodes_count); unsigned offs, rec_len; unsigned limit = PAGE_SIZE; @@ -205,7 +204,8 @@ static struct page * ext2_get_page(struct inode *dir, unsigned long n, if (!IS_ERR(page)) { *page_addr = kmap_local_page(page); if (unlikely(!PageChecked(page))) { - if (PageError(page) || !ext2_check_page(page, quiet)) + if (PageError(page) || !ext2_check_page(page, quiet, + *page_addr)) goto fail; } } @@ -584,10 +584,10 @@ out_unlock: * ext2_delete_entry deletes a directory entry by merging it with the * previous entry. Page is up-to-date. */ -int ext2_delete_entry (struct ext2_dir_entry_2 * dir, struct page * page ) +int ext2_delete_entry (struct ext2_dir_entry_2 *dir, struct page *page, + char *kaddr) { struct inode *inode = page->mapping->host; - char *kaddr = page_address(page); unsigned from = ((char*)dir - kaddr) & ~(ext2_chunk_size(inode)-1); unsigned to = ((char *)dir - kaddr) + ext2_rec_len_from_disk(dir->rec_len); @@ -607,7 +607,7 @@ int ext2_delete_entry (struct ext2_dir_entry_2 * dir, struct page * page ) de = ext2_next_entry(de); } if (pde) - from = (char*)pde - (char*)page_address(page); + from = (char *)pde - kaddr; pos = page_offset(page) + from; lock_page(page); err = ext2_prepare_chunk(page, pos, to - from); diff --git a/fs/ext2/ext2.h b/fs/ext2/ext2.h index b0a694820cb7..e512630cb63e 100644 --- a/fs/ext2/ext2.h +++ b/fs/ext2/ext2.h @@ -740,7 +740,8 @@ extern int ext2_inode_by_name(struct inode *dir, extern int ext2_make_empty(struct inode *, struct inode *); extern struct ext2_dir_entry_2 *ext2_find_entry(struct inode *, const struct qstr *, struct page **, void **res_page_addr); -extern int ext2_delete_entry (struct ext2_dir_entry_2 *, struct page *); +extern int ext2_delete_entry(struct ext2_dir_entry_2 *dir, struct page *page, + char *kaddr); extern int ext2_empty_dir (struct inode *); extern struct ext2_dir_entry_2 *ext2_dotdot(struct inode *dir, struct page **p, void **pa); extern void ext2_set_link(struct inode *, struct ext2_dir_entry_2 *, struct page *, void *, diff --git a/fs/ext2/namei.c b/fs/ext2/namei.c index 1f69b81655b6..5f6b7560eb3f 100644 --- a/fs/ext2/namei.c +++ b/fs/ext2/namei.c @@ -293,7 +293,7 @@ static int ext2_unlink(struct inode * dir, struct dentry *dentry) goto out; } - err = ext2_delete_entry (de, page); + err = ext2_delete_entry (de, page, page_addr); ext2_put_page(page, page_addr); if (err) goto out; @@ -397,7 +397,7 @@ static int ext2_rename (struct user_namespace * mnt_userns, old_inode->i_ctime = current_time(old_inode); mark_inode_dirty(old_inode); - ext2_delete_entry(old_de, old_page); + ext2_delete_entry(old_de, old_page, old_page_addr); if (dir_de) { if (old_dir != new_dir) diff --git a/fs/ext4/ext4_jbd2.c b/fs/ext4/ext4_jbd2.c index b96ecba91899..b60f0152ea57 100644 --- a/fs/ext4/ext4_jbd2.c +++ b/fs/ext4/ext4_jbd2.c @@ -244,9 +244,6 @@ int __ext4_journal_get_write_access(const char *where, unsigned int line, * "bh" may be NULL: a metadata block may have been freed from memory * but there may still be a record of it in the journal, and that record * still needs to be revoked. - * - * If the handle isn't valid we're not journaling, but we still need to - * call into ext4_journal_revoke() to put the buffer head. */ int __ext4_forget(const char *where, unsigned int line, handle_t *handle, int is_metadata, struct inode *inode, diff --git a/fs/ext4/mmp.c b/fs/ext4/mmp.c index bc364c119af6..cebea4270817 100644 --- a/fs/ext4/mmp.c +++ b/fs/ext4/mmp.c @@ -138,7 +138,7 @@ static int kmmpd(void *data) unsigned mmp_check_interval; unsigned long last_update_time; unsigned long diff; - int retval; + int retval = 0; mmp_block = le64_to_cpu(es->s_mmp_block); mmp = (struct mmp_struct *)(bh->b_data); diff --git a/fs/ext4/namei.c b/fs/ext4/namei.c index 5fd56f616cf0..f3bbcd4efb56 100644 --- a/fs/ext4/namei.c +++ b/fs/ext4/namei.c @@ -2517,7 +2517,7 @@ again: goto journal_error; err = ext4_handle_dirty_dx_node(handle, dir, frame->bh); - if (err) + if (restart || err) goto journal_error; } else { struct dx_root *dxroot; diff --git a/fs/internal.h b/fs/internal.h index 3ce8edbaa3ca..82e8eb32ff3d 100644 --- a/fs/internal.h +++ b/fs/internal.h @@ -61,7 +61,6 @@ extern void __init chrdev_init(void); */ extern const struct fs_context_operations legacy_fs_context_ops; extern int parse_monolithic_mount_data(struct fs_context *, void *); -extern void fc_drop_locked(struct fs_context *); extern void vfs_clean_context(struct fs_context *fc); extern int finish_clean_context(struct fs_context *fc); diff --git a/fs/io-wq.c b/fs/io-wq.c index cf086b01c6c6..12fc19353bb0 100644 --- a/fs/io-wq.c +++ b/fs/io-wq.c @@ -130,6 +130,7 @@ struct io_cb_cancel_data { }; static void create_io_worker(struct io_wq *wq, struct io_wqe *wqe, int index); +static void io_wqe_dec_running(struct io_worker *worker); static bool io_worker_get(struct io_worker *worker) { @@ -168,26 +169,21 @@ static void io_worker_exit(struct io_worker *worker) { struct io_wqe *wqe = worker->wqe; struct io_wqe_acct *acct = io_wqe_get_acct(worker); - unsigned flags; if (refcount_dec_and_test(&worker->ref)) complete(&worker->ref_done); wait_for_completion(&worker->ref_done); - preempt_disable(); - current->flags &= ~PF_IO_WORKER; - flags = worker->flags; - worker->flags = 0; - if (flags & IO_WORKER_F_RUNNING) - atomic_dec(&acct->nr_running); - worker->flags = 0; - preempt_enable(); - raw_spin_lock_irq(&wqe->lock); - if (flags & IO_WORKER_F_FREE) + if (worker->flags & IO_WORKER_F_FREE) hlist_nulls_del_rcu(&worker->nulls_node); list_del_rcu(&worker->all_list); acct->nr_workers--; + preempt_disable(); + io_wqe_dec_running(worker); + worker->flags = 0; + current->flags &= ~PF_IO_WORKER; + preempt_enable(); raw_spin_unlock_irq(&wqe->lock); kfree_rcu(worker, rcu); @@ -214,15 +210,19 @@ static bool io_wqe_activate_free_worker(struct io_wqe *wqe) struct hlist_nulls_node *n; struct io_worker *worker; - n = rcu_dereference(hlist_nulls_first_rcu(&wqe->free_list)); - if (is_a_nulls(n)) - return false; - - worker = hlist_nulls_entry(n, struct io_worker, nulls_node); - if (io_worker_get(worker)) { - wake_up_process(worker->task); + /* + * Iterate free_list and see if we can find an idle worker to + * activate. If a given worker is on the free_list but in the process + * of exiting, keep trying. + */ + hlist_nulls_for_each_entry_rcu(worker, n, &wqe->free_list, nulls_node) { + if (!io_worker_get(worker)) + continue; + if (wake_up_process(worker->task)) { + io_worker_release(worker); + return true; + } io_worker_release(worker); - return true; } return false; @@ -247,10 +247,19 @@ static void io_wqe_wake_worker(struct io_wqe *wqe, struct io_wqe_acct *acct) ret = io_wqe_activate_free_worker(wqe); rcu_read_unlock(); - if (!ret && acct->nr_workers < acct->max_workers) { - atomic_inc(&acct->nr_running); - atomic_inc(&wqe->wq->worker_refs); - create_io_worker(wqe->wq, wqe, acct->index); + if (!ret) { + bool do_create = false; + + raw_spin_lock_irq(&wqe->lock); + if (acct->nr_workers < acct->max_workers) { + atomic_inc(&acct->nr_running); + atomic_inc(&wqe->wq->worker_refs); + acct->nr_workers++; + do_create = true; + } + raw_spin_unlock_irq(&wqe->lock); + if (do_create) + create_io_worker(wqe->wq, wqe, acct->index); } } @@ -271,9 +280,17 @@ static void create_worker_cb(struct callback_head *cb) { struct create_worker_data *cwd; struct io_wq *wq; + struct io_wqe *wqe; + struct io_wqe_acct *acct; cwd = container_of(cb, struct create_worker_data, work); - wq = cwd->wqe->wq; + wqe = cwd->wqe; + wq = wqe->wq; + acct = &wqe->acct[cwd->index]; + raw_spin_lock_irq(&wqe->lock); + if (acct->nr_workers < acct->max_workers) + acct->nr_workers++; + raw_spin_unlock_irq(&wqe->lock); create_io_worker(wq, cwd->wqe, cwd->index); kfree(cwd); } @@ -635,6 +652,9 @@ static void create_io_worker(struct io_wq *wq, struct io_wqe *wqe, int index) kfree(worker); fail: atomic_dec(&acct->nr_running); + raw_spin_lock_irq(&wqe->lock); + acct->nr_workers--; + raw_spin_unlock_irq(&wqe->lock); io_worker_ref_put(wq); return; } @@ -650,9 +670,8 @@ fail: worker->flags |= IO_WORKER_F_FREE; if (index == IO_WQ_ACCT_BOUND) worker->flags |= IO_WORKER_F_BOUND; - if (!acct->nr_workers && (worker->flags & IO_WORKER_F_BOUND)) + if ((acct->nr_workers == 1) && (worker->flags & IO_WORKER_F_BOUND)) worker->flags |= IO_WORKER_F_FIXED; - acct->nr_workers++; raw_spin_unlock_irq(&wqe->lock); wake_up_new_task(tsk); } diff --git a/fs/io_uring.c b/fs/io_uring.c index 5a0fd6bcd318..bf548af0426c 100644 --- a/fs/io_uring.c +++ b/fs/io_uring.c @@ -1279,8 +1279,17 @@ static void io_prep_async_link(struct io_kiocb *req) { struct io_kiocb *cur; - io_for_each_link(cur, req) - io_prep_async_work(cur); + if (req->flags & REQ_F_LINK_TIMEOUT) { + struct io_ring_ctx *ctx = req->ctx; + + spin_lock_irq(&ctx->completion_lock); + io_for_each_link(cur, req) + io_prep_async_work(cur); + spin_unlock_irq(&ctx->completion_lock); + } else { + io_for_each_link(cur, req) + io_prep_async_work(cur); + } } static void io_queue_async_work(struct io_kiocb *req) @@ -1950,9 +1959,13 @@ static void tctx_task_work(struct callback_head *cb) node = next; } if (wq_list_empty(&tctx->task_list)) { + spin_lock_irq(&tctx->task_lock); clear_bit(0, &tctx->task_state); - if (wq_list_empty(&tctx->task_list)) + if (wq_list_empty(&tctx->task_list)) { + spin_unlock_irq(&tctx->task_lock); break; + } + spin_unlock_irq(&tctx->task_lock); /* another tctx_task_work() is enqueued, yield */ if (test_and_set_bit(0, &tctx->task_state)) break; @@ -2047,6 +2060,12 @@ static void io_req_task_queue(struct io_kiocb *req) io_req_task_work_add(req); } +static void io_req_task_queue_reissue(struct io_kiocb *req) +{ + req->io_task_work.func = io_queue_async_work; + io_req_task_work_add(req); +} + static inline void io_queue_next(struct io_kiocb *req) { struct io_kiocb *nxt = io_req_find_next(req); @@ -2235,7 +2254,7 @@ static void io_iopoll_complete(struct io_ring_ctx *ctx, unsigned int *nr_events, !(req->flags & REQ_F_DONT_REISSUE)) { req->iopoll_completed = 0; req_ref_get(req); - io_queue_async_work(req); + io_req_task_queue_reissue(req); continue; } @@ -2428,6 +2447,12 @@ static bool io_rw_should_reissue(struct io_kiocb *req) */ if (percpu_ref_is_dying(&ctx->refs)) return false; + /* + * Play it safe and assume not safe to re-import and reissue if we're + * not in the original thread group (or in task context). + */ + if (!same_thread_group(req->task, current) || !in_task()) + return false; return true; } #else @@ -2758,7 +2783,7 @@ static void kiocb_done(struct kiocb *kiocb, ssize_t ret, req->flags &= ~REQ_F_REISSUE; if (io_resubmit_prep(req)) { req_ref_get(req); - io_queue_async_work(req); + io_req_task_queue_reissue(req); } else { int cflags = 0; @@ -4914,7 +4939,6 @@ static bool io_poll_complete(struct io_kiocb *req, __poll_t mask) if (req->poll.events & EPOLLONESHOT) flags = 0; if (!io_cqring_fill_event(ctx, req->user_data, error, flags)) { - io_poll_remove_waitqs(req); req->poll.done = true; flags = 0; } @@ -4937,6 +4961,7 @@ static void io_poll_task_func(struct io_kiocb *req) done = io_poll_complete(req, req->result); if (done) { + io_poll_remove_double(req); hash_del(&req->hash_node); } else { req->result = 0; @@ -5124,7 +5149,7 @@ static __poll_t __io_arm_poll_handler(struct io_kiocb *req, ipt->error = -EINVAL; spin_lock_irq(&ctx->completion_lock); - if (ipt->error) + if (ipt->error || (mask && (poll->events & EPOLLONESHOT))) io_poll_remove_double(req); if (likely(poll->head)) { spin_lock(&poll->head->lock); @@ -5196,7 +5221,6 @@ static int io_arm_poll_handler(struct io_kiocb *req) ret = __io_arm_poll_handler(req, &apoll->poll, &ipt, mask, io_async_wake); if (ret || ipt.error) { - io_poll_remove_double(req); spin_unlock_irq(&ctx->completion_lock); if (ret) return IO_APOLL_READY; diff --git a/fs/namespace.c b/fs/namespace.c index ab4174a3c802..f79d9471cb76 100644 --- a/fs/namespace.c +++ b/fs/namespace.c @@ -1938,6 +1938,20 @@ void drop_collected_mounts(struct vfsmount *mnt) namespace_unlock(); } +static bool has_locked_children(struct mount *mnt, struct dentry *dentry) +{ + struct mount *child; + + list_for_each_entry(child, &mnt->mnt_mounts, mnt_child) { + if (!is_subdir(child->mnt_mountpoint, dentry)) + continue; + + if (child->mnt.mnt_flags & MNT_LOCKED) + return true; + } + return false; +} + /** * clone_private_mount - create a private clone of a path * @path: path to clone @@ -1953,10 +1967,19 @@ struct vfsmount *clone_private_mount(const struct path *path) struct mount *old_mnt = real_mount(path->mnt); struct mount *new_mnt; + down_read(&namespace_sem); if (IS_MNT_UNBINDABLE(old_mnt)) - return ERR_PTR(-EINVAL); + goto invalid; + + if (!check_mnt(old_mnt)) + goto invalid; + + if (has_locked_children(old_mnt, path->dentry)) + goto invalid; new_mnt = clone_mnt(old_mnt, path->dentry, CL_PRIVATE); + up_read(&namespace_sem); + if (IS_ERR(new_mnt)) return ERR_CAST(new_mnt); @@ -1964,6 +1987,10 @@ struct vfsmount *clone_private_mount(const struct path *path) new_mnt->mnt_ns = MNT_NS_INTERNAL; return &new_mnt->mnt; + +invalid: + up_read(&namespace_sem); + return ERR_PTR(-EINVAL); } EXPORT_SYMBOL_GPL(clone_private_mount); @@ -2315,19 +2342,6 @@ static int do_change_type(struct path *path, int ms_flags) return err; } -static bool has_locked_children(struct mount *mnt, struct dentry *dentry) -{ - struct mount *child; - list_for_each_entry(child, &mnt->mnt_mounts, mnt_child) { - if (!is_subdir(child->mnt_mountpoint, dentry)) - continue; - - if (child->mnt.mnt_flags & MNT_LOCKED) - return true; - } - return false; -} - static struct mount *__do_loopback(struct path *old_path, int recurse) { struct mount *mnt = ERR_PTR(-EINVAL), *old = real_mount(old_path->mnt); diff --git a/fs/ocfs2/file.c b/fs/ocfs2/file.c index 775657943057..54d7843c0211 100644 --- a/fs/ocfs2/file.c +++ b/fs/ocfs2/file.c @@ -1529,6 +1529,45 @@ static void ocfs2_truncate_cluster_pages(struct inode *inode, u64 byte_start, } } +/* + * zero out partial blocks of one cluster. + * + * start: file offset where zero starts, will be made upper block aligned. + * len: it will be trimmed to the end of current cluster if "start + len" + * is bigger than it. + */ +static int ocfs2_zeroout_partial_cluster(struct inode *inode, + u64 start, u64 len) +{ + int ret; + u64 start_block, end_block, nr_blocks; + u64 p_block, offset; + u32 cluster, p_cluster, nr_clusters; + struct super_block *sb = inode->i_sb; + u64 end = ocfs2_align_bytes_to_clusters(sb, start); + + if (start + len < end) + end = start + len; + + start_block = ocfs2_blocks_for_bytes(sb, start); + end_block = ocfs2_blocks_for_bytes(sb, end); + nr_blocks = end_block - start_block; + if (!nr_blocks) + return 0; + + cluster = ocfs2_bytes_to_clusters(sb, start); + ret = ocfs2_get_clusters(inode, cluster, &p_cluster, + &nr_clusters, NULL); + if (ret) + return ret; + if (!p_cluster) + return 0; + + offset = start_block - ocfs2_clusters_to_blocks(sb, cluster); + p_block = ocfs2_clusters_to_blocks(sb, p_cluster) + offset; + return sb_issue_zeroout(sb, p_block, nr_blocks, GFP_NOFS); +} + static int ocfs2_zero_partial_clusters(struct inode *inode, u64 start, u64 len) { @@ -1538,6 +1577,7 @@ static int ocfs2_zero_partial_clusters(struct inode *inode, struct ocfs2_super *osb = OCFS2_SB(inode->i_sb); unsigned int csize = osb->s_clustersize; handle_t *handle; + loff_t isize = i_size_read(inode); /* * The "start" and "end" values are NOT necessarily part of @@ -1558,6 +1598,26 @@ static int ocfs2_zero_partial_clusters(struct inode *inode, if ((start & (csize - 1)) == 0 && (end & (csize - 1)) == 0) goto out; + /* No page cache for EOF blocks, issue zero out to disk. */ + if (end > isize) { + /* + * zeroout eof blocks in last cluster starting from + * "isize" even "start" > "isize" because it is + * complicated to zeroout just at "start" as "start" + * may be not aligned with block size, buffer write + * would be required to do that, but out of eof buffer + * write is not supported. + */ + ret = ocfs2_zeroout_partial_cluster(inode, isize, + end - isize); + if (ret) { + mlog_errno(ret); + goto out; + } + if (start >= isize) + goto out; + end = isize; + } handle = ocfs2_start_trans(osb, OCFS2_INODE_UPDATE_CREDITS); if (IS_ERR(handle)) { ret = PTR_ERR(handle); @@ -1856,45 +1916,6 @@ out: } /* - * zero out partial blocks of one cluster. - * - * start: file offset where zero starts, will be made upper block aligned. - * len: it will be trimmed to the end of current cluster if "start + len" - * is bigger than it. - */ -static int ocfs2_zeroout_partial_cluster(struct inode *inode, - u64 start, u64 len) -{ - int ret; - u64 start_block, end_block, nr_blocks; - u64 p_block, offset; - u32 cluster, p_cluster, nr_clusters; - struct super_block *sb = inode->i_sb; - u64 end = ocfs2_align_bytes_to_clusters(sb, start); - - if (start + len < end) - end = start + len; - - start_block = ocfs2_blocks_for_bytes(sb, start); - end_block = ocfs2_blocks_for_bytes(sb, end); - nr_blocks = end_block - start_block; - if (!nr_blocks) - return 0; - - cluster = ocfs2_bytes_to_clusters(sb, start); - ret = ocfs2_get_clusters(inode, cluster, &p_cluster, - &nr_clusters, NULL); - if (ret) - return ret; - if (!p_cluster) - return 0; - - offset = start_block - ocfs2_clusters_to_blocks(sb, cluster); - p_block = ocfs2_clusters_to_blocks(sb, p_cluster) + offset; - return sb_issue_zeroout(sb, p_block, nr_blocks, GFP_NOFS); -} - -/* * Parts of this function taken from xfs_change_file_space() */ static int __ocfs2_change_file_space(struct file *file, struct inode *inode, @@ -1935,7 +1956,6 @@ static int __ocfs2_change_file_space(struct file *file, struct inode *inode, goto out_inode_unlock; } - orig_isize = i_size_read(inode); switch (sr->l_whence) { case 0: /*SEEK_SET*/ break; @@ -1943,7 +1963,7 @@ static int __ocfs2_change_file_space(struct file *file, struct inode *inode, sr->l_start += f_pos; break; case 2: /*SEEK_END*/ - sr->l_start += orig_isize; + sr->l_start += i_size_read(inode); break; default: ret = -EINVAL; @@ -1998,6 +2018,7 @@ static int __ocfs2_change_file_space(struct file *file, struct inode *inode, ret = -EINVAL; } + orig_isize = i_size_read(inode); /* zeroout eof blocks in the cluster. */ if (!ret && change_size && orig_isize < size) { ret = ocfs2_zeroout_partial_cluster(inode, orig_isize, diff --git a/fs/overlayfs/export.c b/fs/overlayfs/export.c index 41ebf52f1bbc..ebde05c9cf62 100644 --- a/fs/overlayfs/export.c +++ b/fs/overlayfs/export.c @@ -392,6 +392,7 @@ static struct dentry *ovl_lookup_real_one(struct dentry *connected, */ take_dentry_name_snapshot(&name, real); this = lookup_one_len(name.name.name, connected, name.name.len); + release_dentry_name_snapshot(&name); err = PTR_ERR(this); if (IS_ERR(this)) { goto fail; @@ -406,7 +407,6 @@ static struct dentry *ovl_lookup_real_one(struct dentry *connected, } out: - release_dentry_name_snapshot(&name); dput(parent); inode_unlock(dir); return this; diff --git a/fs/overlayfs/file.c b/fs/overlayfs/file.c index 4d53d3b7e5fe..d081faa55e83 100644 --- a/fs/overlayfs/file.c +++ b/fs/overlayfs/file.c @@ -392,6 +392,51 @@ out_unlock: return ret; } +/* + * Calling iter_file_splice_write() directly from overlay's f_op may deadlock + * due to lock order inversion between pipe->mutex in iter_file_splice_write() + * and file_start_write(real.file) in ovl_write_iter(). + * + * So do everything ovl_write_iter() does and call iter_file_splice_write() on + * the real file. + */ +static ssize_t ovl_splice_write(struct pipe_inode_info *pipe, struct file *out, + loff_t *ppos, size_t len, unsigned int flags) +{ + struct fd real; + const struct cred *old_cred; + struct inode *inode = file_inode(out); + struct inode *realinode = ovl_inode_real(inode); + ssize_t ret; + + inode_lock(inode); + /* Update mode */ + ovl_copyattr(realinode, inode); + ret = file_remove_privs(out); + if (ret) + goto out_unlock; + + ret = ovl_real_fdget(out, &real); + if (ret) + goto out_unlock; + + old_cred = ovl_override_creds(inode->i_sb); + file_start_write(real.file); + + ret = iter_file_splice_write(pipe, real.file, ppos, len, flags); + + file_end_write(real.file); + /* Update size */ + ovl_copyattr(realinode, inode); + revert_creds(old_cred); + fdput(real); + +out_unlock: + inode_unlock(inode); + + return ret; +} + static int ovl_fsync(struct file *file, loff_t start, loff_t end, int datasync) { struct fd real; @@ -603,7 +648,7 @@ const struct file_operations ovl_file_operations = { .fadvise = ovl_fadvise, .flush = ovl_flush, .splice_read = generic_file_splice_read, - .splice_write = iter_file_splice_write, + .splice_write = ovl_splice_write, .copy_file_range = ovl_copy_file_range, .remap_file_range = ovl_remap_file_range, diff --git a/fs/overlayfs/readdir.c b/fs/overlayfs/readdir.c index e8ad2c2c77dd..150fdf3bc68d 100644 --- a/fs/overlayfs/readdir.c +++ b/fs/overlayfs/readdir.c @@ -481,6 +481,8 @@ static int ovl_cache_update_ino(struct path *path, struct ovl_cache_entry *p) } this = lookup_one_len(p->name, dir, p->len); if (IS_ERR_OR_NULL(this) || !this->d_inode) { + /* Mark a stale entry */ + p->is_whiteout = true; if (IS_ERR(this)) { err = PTR_ERR(this); this = NULL; @@ -776,6 +778,9 @@ static int ovl_iterate(struct file *file, struct dir_context *ctx) if (err) goto out; } + } + /* ovl_cache_update_ino() sets is_whiteout on stale entry */ + if (!p->is_whiteout) { if (!dir_emit(ctx, p->name, p->len, p->ino, p->type)) break; } diff --git a/fs/pipe.c b/fs/pipe.c index bfd946a9ad01..8e6ef62aeb1c 100644 --- a/fs/pipe.c +++ b/fs/pipe.c @@ -32,6 +32,21 @@ #include "internal.h" /* + * New pipe buffers will be restricted to this size while the user is exceeding + * their pipe buffer quota. The general pipe use case needs at least two + * buffers: one for data yet to be read, and one for new data. If this is less + * than two, then a write to a non-empty pipe may block even if the pipe is not + * full. This can occur with GNU make jobserver or similar uses of pipes as + * semaphores: multiple processes may be waiting to write tokens back to the + * pipe before reading tokens: https://lore.kernel.org/lkml/1628086770.5rn8p04n6j.none@localhost/. + * + * Users can reduce their pipe buffers with F_SETPIPE_SZ below this at their + * own risk, namely: pipe writes to non-full pipes may block until the pipe is + * emptied. + */ +#define PIPE_MIN_DEF_BUFFERS 2 + +/* * The max size that a non-root user is allowed to grow the pipe. Can * be set by root in /proc/sys/fs/pipe-max-size */ @@ -429,20 +444,20 @@ pipe_write(struct kiocb *iocb, struct iov_iter *from) #endif /* - * Only wake up if the pipe started out empty, since - * otherwise there should be no readers waiting. + * Epoll nonsensically wants a wakeup whether the pipe + * was already empty or not. * * If it wasn't empty we try to merge new data into * the last buffer. * * That naturally merges small writes, but it also - * page-aligs the rest of the writes for large writes + * page-aligns the rest of the writes for large writes * spanning multiple pages. */ head = pipe->head; - was_empty = pipe_empty(head, pipe->tail); + was_empty = true; chars = total_len & (PAGE_SIZE-1); - if (chars && !was_empty) { + if (chars && !pipe_empty(head, pipe->tail)) { unsigned int mask = pipe->ring_size - 1; struct pipe_buffer *buf = &pipe->bufs[(head - 1) & mask]; int offset = buf->offset + buf->len; @@ -781,8 +796,8 @@ struct pipe_inode_info *alloc_pipe_info(void) user_bufs = account_pipe_buffers(user, 0, pipe_bufs); if (too_many_pipe_buffers_soft(user_bufs) && pipe_is_unprivileged_user()) { - user_bufs = account_pipe_buffers(user, pipe_bufs, 1); - pipe_bufs = 1; + user_bufs = account_pipe_buffers(user, pipe_bufs, PIPE_MIN_DEF_BUFFERS); + pipe_bufs = PIPE_MIN_DEF_BUFFERS; } if (too_many_pipe_buffers_hard(user_bufs) && pipe_is_unprivileged_user()) diff --git a/fs/reiserfs/stree.c b/fs/reiserfs/stree.c index 476a7ff49482..ef42729216d1 100644 --- a/fs/reiserfs/stree.c +++ b/fs/reiserfs/stree.c @@ -387,6 +387,24 @@ void pathrelse(struct treepath *search_path) search_path->path_length = ILLEGAL_PATH_ELEMENT_OFFSET; } +static int has_valid_deh_location(struct buffer_head *bh, struct item_head *ih) +{ + struct reiserfs_de_head *deh; + int i; + + deh = B_I_DEH(bh, ih); + for (i = 0; i < ih_entry_count(ih); i++) { + if (deh_location(&deh[i]) > ih_item_len(ih)) { + reiserfs_warning(NULL, "reiserfs-5094", + "directory entry location seems wrong %h", + &deh[i]); + return 0; + } + } + + return 1; +} + static int is_leaf(char *buf, int blocksize, struct buffer_head *bh) { struct block_head *blkh; @@ -454,11 +472,14 @@ static int is_leaf(char *buf, int blocksize, struct buffer_head *bh) "(second one): %h", ih); return 0; } - if (is_direntry_le_ih(ih) && (ih_item_len(ih) < (ih_entry_count(ih) * IH_SIZE))) { - reiserfs_warning(NULL, "reiserfs-5093", - "item entry count seems wrong %h", - ih); - return 0; + if (is_direntry_le_ih(ih)) { + if (ih_item_len(ih) < (ih_entry_count(ih) * IH_SIZE)) { + reiserfs_warning(NULL, "reiserfs-5093", + "item entry count seems wrong %h", + ih); + return 0; + } + return has_valid_deh_location(bh, ih); } prev_location = ih_location(ih); } diff --git a/fs/reiserfs/super.c b/fs/reiserfs/super.c index 3ffafc73acf0..58481f8d63d5 100644 --- a/fs/reiserfs/super.c +++ b/fs/reiserfs/super.c @@ -2082,6 +2082,14 @@ static int reiserfs_fill_super(struct super_block *s, void *data, int silent) unlock_new_inode(root_inode); } + if (!S_ISDIR(root_inode->i_mode) || !inode_get_bytes(root_inode) || + !root_inode->i_size) { + SWARN(silent, s, "", "corrupt root inode, run fsck"); + iput(root_inode); + errval = -EUCLEAN; + goto error; + } + s->s_root = d_make_root(root_inode); if (!s->s_root) goto error; diff --git a/fs/xfs/libxfs/xfs_log_format.h b/fs/xfs/libxfs/xfs_log_format.h index d548ea4b6aab..2c5bcbc19264 100644 --- a/fs/xfs/libxfs/xfs_log_format.h +++ b/fs/xfs/libxfs/xfs_log_format.h @@ -411,7 +411,16 @@ struct xfs_log_dinode { /* start of the extended dinode, writable fields */ uint32_t di_crc; /* CRC of the inode */ uint64_t di_changecount; /* number of attribute changes */ - xfs_lsn_t di_lsn; /* flush sequence */ + + /* + * The LSN we write to this field during formatting is not a reflection + * of the current on-disk LSN. It should never be used for recovery + * sequencing, nor should it be recovered into the on-disk inode at all. + * See xlog_recover_inode_commit_pass2() and xfs_log_dinode_to_disk() + * for details. + */ + xfs_lsn_t di_lsn; + uint64_t di_flags2; /* more random flags */ uint32_t di_cowextsize; /* basic cow extent size for file */ uint8_t di_pad2[12]; /* more padding for future expansion */ diff --git a/fs/xfs/xfs_buf_item_recover.c b/fs/xfs/xfs_buf_item_recover.c index d44e8b4a3391..4775485b4062 100644 --- a/fs/xfs/xfs_buf_item_recover.c +++ b/fs/xfs/xfs_buf_item_recover.c @@ -698,7 +698,8 @@ xlog_recover_do_inode_buffer( static xfs_lsn_t xlog_recover_get_buf_lsn( struct xfs_mount *mp, - struct xfs_buf *bp) + struct xfs_buf *bp, + struct xfs_buf_log_format *buf_f) { uint32_t magic32; uint16_t magic16; @@ -706,11 +707,20 @@ xlog_recover_get_buf_lsn( void *blk = bp->b_addr; uuid_t *uuid; xfs_lsn_t lsn = -1; + uint16_t blft; /* v4 filesystems always recover immediately */ if (!xfs_sb_version_hascrc(&mp->m_sb)) goto recover_immediately; + /* + * realtime bitmap and summary file blocks do not have magic numbers or + * UUIDs, so we must recover them immediately. + */ + blft = xfs_blft_from_flags(buf_f); + if (blft == XFS_BLFT_RTBITMAP_BUF || blft == XFS_BLFT_RTSUMMARY_BUF) + goto recover_immediately; + magic32 = be32_to_cpu(*(__be32 *)blk); switch (magic32) { case XFS_ABTB_CRC_MAGIC: @@ -796,6 +806,7 @@ xlog_recover_get_buf_lsn( switch (magicda) { case XFS_DIR3_LEAF1_MAGIC: case XFS_DIR3_LEAFN_MAGIC: + case XFS_ATTR3_LEAF_MAGIC: case XFS_DA3_NODE_MAGIC: lsn = be64_to_cpu(((struct xfs_da3_blkinfo *)blk)->lsn); uuid = &((struct xfs_da3_blkinfo *)blk)->uuid; @@ -919,7 +930,7 @@ xlog_recover_buf_commit_pass2( * the verifier will be reset to match whatever recover turns that * buffer into. */ - lsn = xlog_recover_get_buf_lsn(mp, bp); + lsn = xlog_recover_get_buf_lsn(mp, bp, buf_f); if (lsn && lsn != -1 && XFS_LSN_CMP(lsn, current_lsn) >= 0) { trace_xfs_log_recover_buf_skip(log, buf_f); xlog_recover_validate_buf_type(mp, bp, buf_f, NULLCOMMITLSN); diff --git a/fs/xfs/xfs_inode_item_recover.c b/fs/xfs/xfs_inode_item_recover.c index 7b79518b6c20..e0072a6cd2d3 100644 --- a/fs/xfs/xfs_inode_item_recover.c +++ b/fs/xfs/xfs_inode_item_recover.c @@ -145,7 +145,8 @@ xfs_log_dinode_to_disk_ts( STATIC void xfs_log_dinode_to_disk( struct xfs_log_dinode *from, - struct xfs_dinode *to) + struct xfs_dinode *to, + xfs_lsn_t lsn) { to->di_magic = cpu_to_be16(from->di_magic); to->di_mode = cpu_to_be16(from->di_mode); @@ -182,7 +183,7 @@ xfs_log_dinode_to_disk( to->di_flags2 = cpu_to_be64(from->di_flags2); to->di_cowextsize = cpu_to_be32(from->di_cowextsize); to->di_ino = cpu_to_be64(from->di_ino); - to->di_lsn = cpu_to_be64(from->di_lsn); + to->di_lsn = cpu_to_be64(lsn); memcpy(to->di_pad2, from->di_pad2, sizeof(to->di_pad2)); uuid_copy(&to->di_uuid, &from->di_uuid); to->di_flushiter = 0; @@ -261,16 +262,25 @@ xlog_recover_inode_commit_pass2( } /* - * If the inode has an LSN in it, recover the inode only if it's less - * than the lsn of the transaction we are replaying. Note: we still - * need to replay an owner change even though the inode is more recent - * than the transaction as there is no guarantee that all the btree - * blocks are more recent than this transaction, too. + * If the inode has an LSN in it, recover the inode only if the on-disk + * inode's LSN is older than the lsn of the transaction we are + * replaying. We can have multiple checkpoints with the same start LSN, + * so the current LSN being equal to the on-disk LSN doesn't necessarily + * mean that the on-disk inode is more recent than the change being + * replayed. + * + * We must check the current_lsn against the on-disk inode + * here because the we can't trust the log dinode to contain a valid LSN + * (see comment below before replaying the log dinode for details). + * + * Note: we still need to replay an owner change even though the inode + * is more recent than the transaction as there is no guarantee that all + * the btree blocks are more recent than this transaction, too. */ if (dip->di_version >= 3) { xfs_lsn_t lsn = be64_to_cpu(dip->di_lsn); - if (lsn && lsn != -1 && XFS_LSN_CMP(lsn, current_lsn) >= 0) { + if (lsn && lsn != -1 && XFS_LSN_CMP(lsn, current_lsn) > 0) { trace_xfs_log_recover_inode_skip(log, in_f); error = 0; goto out_owner_change; @@ -368,8 +378,17 @@ xlog_recover_inode_commit_pass2( goto out_release; } - /* recover the log dinode inode into the on disk inode */ - xfs_log_dinode_to_disk(ldip, dip); + /* + * Recover the log dinode inode into the on disk inode. + * + * The LSN in the log dinode is garbage - it can be zero or reflect + * stale in-memory runtime state that isn't coherent with the changes + * logged in this transaction or the changes written to the on-disk + * inode. Hence we write the current lSN into the inode because that + * matches what xfs_iflush() would write inode the inode when flushing + * the changes in this transaction. + */ + xfs_log_dinode_to_disk(ldip, dip, current_lsn); fields = in_f->ilf_fields; if (fields & XFS_ILOG_DEV) diff --git a/fs/xfs/xfs_log.c b/fs/xfs/xfs_log.c index 36fa2650b081..60ac5fd63f1e 100644 --- a/fs/xfs/xfs_log.c +++ b/fs/xfs/xfs_log.c @@ -78,13 +78,12 @@ xlog_verify_iclog( STATIC void xlog_verify_tail_lsn( struct xlog *log, - struct xlog_in_core *iclog, - xfs_lsn_t tail_lsn); + struct xlog_in_core *iclog); #else #define xlog_verify_dest_ptr(a,b) #define xlog_verify_grant_tail(a) #define xlog_verify_iclog(a,b,c) -#define xlog_verify_tail_lsn(a,b,c) +#define xlog_verify_tail_lsn(a,b) #endif STATIC int @@ -487,51 +486,80 @@ out_error: return error; } -static bool -__xlog_state_release_iclog( - struct xlog *log, - struct xlog_in_core *iclog) -{ - lockdep_assert_held(&log->l_icloglock); - - if (iclog->ic_state == XLOG_STATE_WANT_SYNC) { - /* update tail before writing to iclog */ - xfs_lsn_t tail_lsn = xlog_assign_tail_lsn(log->l_mp); - - iclog->ic_state = XLOG_STATE_SYNCING; - iclog->ic_header.h_tail_lsn = cpu_to_be64(tail_lsn); - xlog_verify_tail_lsn(log, iclog, tail_lsn); - /* cycle incremented when incrementing curr_block */ - trace_xlog_iclog_syncing(iclog, _RET_IP_); - return true; - } - - ASSERT(iclog->ic_state == XLOG_STATE_ACTIVE); - return false; -} - /* * Flush iclog to disk if this is the last reference to the given iclog and the * it is in the WANT_SYNC state. + * + * If the caller passes in a non-zero @old_tail_lsn and the current log tail + * does not match, there may be metadata on disk that must be persisted before + * this iclog is written. To satisfy that requirement, set the + * XLOG_ICL_NEED_FLUSH flag as a condition for writing this iclog with the new + * log tail value. + * + * If XLOG_ICL_NEED_FUA is already set on the iclog, we need to ensure that the + * log tail is updated correctly. NEED_FUA indicates that the iclog will be + * written to stable storage, and implies that a commit record is contained + * within the iclog. We need to ensure that the log tail does not move beyond + * the tail that the first commit record in the iclog ordered against, otherwise + * correct recovery of that checkpoint becomes dependent on future operations + * performed on this iclog. + * + * Hence if NEED_FUA is set and the current iclog tail lsn is empty, write the + * current tail into iclog. Once the iclog tail is set, future operations must + * not modify it, otherwise they potentially violate ordering constraints for + * the checkpoint commit that wrote the initial tail lsn value. The tail lsn in + * the iclog will get zeroed on activation of the iclog after sync, so we + * always capture the tail lsn on the iclog on the first NEED_FUA release + * regardless of the number of active reference counts on this iclog. */ + int xlog_state_release_iclog( struct xlog *log, - struct xlog_in_core *iclog) + struct xlog_in_core *iclog, + xfs_lsn_t old_tail_lsn) { + xfs_lsn_t tail_lsn; lockdep_assert_held(&log->l_icloglock); trace_xlog_iclog_release(iclog, _RET_IP_); if (iclog->ic_state == XLOG_STATE_IOERROR) return -EIO; - if (atomic_dec_and_test(&iclog->ic_refcnt) && - __xlog_state_release_iclog(log, iclog)) { - spin_unlock(&log->l_icloglock); - xlog_sync(log, iclog); - spin_lock(&log->l_icloglock); + /* + * Grabbing the current log tail needs to be atomic w.r.t. the writing + * of the tail LSN into the iclog so we guarantee that the log tail does + * not move between deciding if a cache flush is required and writing + * the LSN into the iclog below. + */ + if (old_tail_lsn || iclog->ic_state == XLOG_STATE_WANT_SYNC) { + tail_lsn = xlog_assign_tail_lsn(log->l_mp); + + if (old_tail_lsn && tail_lsn != old_tail_lsn) + iclog->ic_flags |= XLOG_ICL_NEED_FLUSH; + + if ((iclog->ic_flags & XLOG_ICL_NEED_FUA) && + !iclog->ic_header.h_tail_lsn) + iclog->ic_header.h_tail_lsn = cpu_to_be64(tail_lsn); } + if (!atomic_dec_and_test(&iclog->ic_refcnt)) + return 0; + + if (iclog->ic_state != XLOG_STATE_WANT_SYNC) { + ASSERT(iclog->ic_state == XLOG_STATE_ACTIVE); + return 0; + } + + iclog->ic_state = XLOG_STATE_SYNCING; + if (!iclog->ic_header.h_tail_lsn) + iclog->ic_header.h_tail_lsn = cpu_to_be64(tail_lsn); + xlog_verify_tail_lsn(log, iclog); + trace_xlog_iclog_syncing(iclog, _RET_IP_); + + spin_unlock(&log->l_icloglock); + xlog_sync(log, iclog); + spin_lock(&log->l_icloglock); return 0; } @@ -774,6 +802,21 @@ xfs_log_mount_cancel( } /* + * Flush out the iclog to disk ensuring that device caches are flushed and + * the iclog hits stable storage before any completion waiters are woken. + */ +static inline int +xlog_force_iclog( + struct xlog_in_core *iclog) +{ + atomic_inc(&iclog->ic_refcnt); + iclog->ic_flags |= XLOG_ICL_NEED_FLUSH | XLOG_ICL_NEED_FUA; + if (iclog->ic_state == XLOG_STATE_ACTIVE) + xlog_state_switch_iclogs(iclog->ic_log, iclog, 0); + return xlog_state_release_iclog(iclog->ic_log, iclog, 0); +} + +/* * Wait for the iclog and all prior iclogs to be written disk as required by the * log force state machine. Waiting on ic_force_wait ensures iclog completions * have been ordered and callbacks run before we are woken here, hence @@ -827,13 +870,6 @@ xlog_write_unmount_record( /* account for space used by record data */ ticket->t_curr_res -= sizeof(ulf); - /* - * For external log devices, we need to flush the data device cache - * first to ensure all metadata writeback is on stable storage before we - * stamp the tail LSN into the unmount record. - */ - if (log->l_targ != log->l_mp->m_ddev_targp) - blkdev_issue_flush(log->l_targ->bt_bdev); return xlog_write(log, &vec, ticket, NULL, NULL, XLOG_UNMOUNT_TRANS); } @@ -865,18 +901,7 @@ out_err: spin_lock(&log->l_icloglock); iclog = log->l_iclog; - atomic_inc(&iclog->ic_refcnt); - if (iclog->ic_state == XLOG_STATE_ACTIVE) - xlog_state_switch_iclogs(log, iclog, 0); - else - ASSERT(iclog->ic_state == XLOG_STATE_WANT_SYNC || - iclog->ic_state == XLOG_STATE_IOERROR); - /* - * Ensure the journal is fully flushed and on stable storage once the - * iclog containing the unmount record is written. - */ - iclog->ic_flags |= (XLOG_ICL_NEED_FLUSH | XLOG_ICL_NEED_FUA); - error = xlog_state_release_iclog(log, iclog); + error = xlog_force_iclog(iclog); xlog_wait_on_iclog(iclog); if (tic) { @@ -1796,10 +1821,20 @@ xlog_write_iclog( * metadata writeback and causing priority inversions. */ iclog->ic_bio.bi_opf = REQ_OP_WRITE | REQ_META | REQ_SYNC | REQ_IDLE; - if (iclog->ic_flags & XLOG_ICL_NEED_FLUSH) + if (iclog->ic_flags & XLOG_ICL_NEED_FLUSH) { iclog->ic_bio.bi_opf |= REQ_PREFLUSH; + /* + * For external log devices, we also need to flush the data + * device cache first to ensure all metadata writeback covered + * by the LSN in this iclog is on stable storage. This is slow, + * but it *must* complete before we issue the external log IO. + */ + if (log->l_targ != log->l_mp->m_ddev_targp) + blkdev_issue_flush(log->l_mp->m_ddev_targp->bt_bdev); + } if (iclog->ic_flags & XLOG_ICL_NEED_FUA) iclog->ic_bio.bi_opf |= REQ_FUA; + iclog->ic_flags &= ~(XLOG_ICL_NEED_FLUSH | XLOG_ICL_NEED_FUA); if (xlog_map_iclog_data(&iclog->ic_bio, iclog->ic_data, count)) { @@ -2310,7 +2345,7 @@ xlog_write_copy_finish( return 0; release_iclog: - error = xlog_state_release_iclog(log, iclog); + error = xlog_state_release_iclog(log, iclog, 0); spin_unlock(&log->l_icloglock); return error; } @@ -2529,7 +2564,7 @@ next_lv: ASSERT(optype & XLOG_COMMIT_TRANS); *commit_iclog = iclog; } else { - error = xlog_state_release_iclog(log, iclog); + error = xlog_state_release_iclog(log, iclog, 0); } spin_unlock(&log->l_icloglock); @@ -2567,6 +2602,7 @@ xlog_state_activate_iclog( memset(iclog->ic_header.h_cycle_data, 0, sizeof(iclog->ic_header.h_cycle_data)); iclog->ic_header.h_lsn = 0; + iclog->ic_header.h_tail_lsn = 0; } /* @@ -2967,7 +3003,7 @@ restart: * reference to the iclog. */ if (!atomic_add_unless(&iclog->ic_refcnt, -1, 1)) - error = xlog_state_release_iclog(log, iclog); + error = xlog_state_release_iclog(log, iclog, 0); spin_unlock(&log->l_icloglock); if (error) return error; @@ -3132,6 +3168,35 @@ xlog_state_switch_iclogs( } /* + * Force the iclog to disk and check if the iclog has been completed before + * xlog_force_iclog() returns. This can happen on synchronous (e.g. + * pmem) or fast async storage because we drop the icloglock to issue the IO. + * If completion has already occurred, tell the caller so that it can avoid an + * unnecessary wait on the iclog. + */ +static int +xlog_force_and_check_iclog( + struct xlog_in_core *iclog, + bool *completed) +{ + xfs_lsn_t lsn = be64_to_cpu(iclog->ic_header.h_lsn); + int error; + + *completed = false; + error = xlog_force_iclog(iclog); + if (error) + return error; + + /* + * If the iclog has already been completed and reused the header LSN + * will have been rewritten by completion + */ + if (be64_to_cpu(iclog->ic_header.h_lsn) != lsn) + *completed = true; + return 0; +} + +/* * Write out all data in the in-core log as of this exact moment in time. * * Data may be written to the in-core log during this call. However, @@ -3165,7 +3230,6 @@ xfs_log_force( { struct xlog *log = mp->m_log; struct xlog_in_core *iclog; - xfs_lsn_t lsn; XFS_STATS_INC(mp, xs_log_force); trace_xfs_log_force(mp, 0, _RET_IP_); @@ -3193,39 +3257,33 @@ xfs_log_force( iclog = iclog->ic_prev; } else if (iclog->ic_state == XLOG_STATE_ACTIVE) { if (atomic_read(&iclog->ic_refcnt) == 0) { - /* - * We are the only one with access to this iclog. - * - * Flush it out now. There should be a roundoff of zero - * to show that someone has already taken care of the - * roundoff from the previous sync. - */ - atomic_inc(&iclog->ic_refcnt); - lsn = be64_to_cpu(iclog->ic_header.h_lsn); - xlog_state_switch_iclogs(log, iclog, 0); - if (xlog_state_release_iclog(log, iclog)) + /* We have exclusive access to this iclog. */ + bool completed; + + if (xlog_force_and_check_iclog(iclog, &completed)) goto out_error; - if (be64_to_cpu(iclog->ic_header.h_lsn) != lsn) + if (completed) goto out_unlock; } else { /* - * Someone else is writing to this iclog. - * - * Use its call to flush out the data. However, the - * other thread may not force out this LR, so we mark - * it WANT_SYNC. + * Someone else is still writing to this iclog, so we + * need to ensure that when they release the iclog it + * gets synced immediately as we may be waiting on it. */ xlog_state_switch_iclogs(log, iclog, 0); } - } else { - /* - * If the head iclog is not active nor dirty, we just attach - * ourselves to the head and go to sleep if necessary. - */ - ; } + /* + * The iclog we are about to wait on may contain the checkpoint pushed + * by the above xlog_cil_force() call, but it may not have been pushed + * to disk yet. Like the ACTIVE case above, we need to make sure caches + * are flushed when this iclog is written. + */ + if (iclog->ic_state == XLOG_STATE_WANT_SYNC) + iclog->ic_flags |= XLOG_ICL_NEED_FLUSH | XLOG_ICL_NEED_FUA; + if (flags & XFS_LOG_SYNC) return xlog_wait_on_iclog(iclog); out_unlock: @@ -3245,6 +3303,7 @@ xlog_force_lsn( bool already_slept) { struct xlog_in_core *iclog; + bool completed; spin_lock(&log->l_icloglock); iclog = log->l_iclog; @@ -3258,7 +3317,8 @@ xlog_force_lsn( goto out_unlock; } - if (iclog->ic_state == XLOG_STATE_ACTIVE) { + switch (iclog->ic_state) { + case XLOG_STATE_ACTIVE: /* * We sleep here if we haven't already slept (e.g. this is the * first time we've looked at the correct iclog buf) and the @@ -3281,12 +3341,31 @@ xlog_force_lsn( &log->l_icloglock); return -EAGAIN; } - atomic_inc(&iclog->ic_refcnt); - xlog_state_switch_iclogs(log, iclog, 0); - if (xlog_state_release_iclog(log, iclog)) + if (xlog_force_and_check_iclog(iclog, &completed)) goto out_error; if (log_flushed) *log_flushed = 1; + if (completed) + goto out_unlock; + break; + case XLOG_STATE_WANT_SYNC: + /* + * This iclog may contain the checkpoint pushed by the + * xlog_cil_force_seq() call, but there are other writers still + * accessing it so it hasn't been pushed to disk yet. Like the + * ACTIVE case above, we need to make sure caches are flushed + * when this iclog is written. + */ + iclog->ic_flags |= XLOG_ICL_NEED_FLUSH | XLOG_ICL_NEED_FUA; + break; + default: + /* + * The entire checkpoint was written by the CIL force and is on + * its way to disk already. It will be stable when it + * completes, so we don't need to manipulate caches here at all. + * We just need to wait for completion if necessary. + */ + break; } if (flags & XFS_LOG_SYNC) @@ -3559,10 +3638,10 @@ xlog_verify_grant_tail( STATIC void xlog_verify_tail_lsn( struct xlog *log, - struct xlog_in_core *iclog, - xfs_lsn_t tail_lsn) + struct xlog_in_core *iclog) { - int blocks; + xfs_lsn_t tail_lsn = be64_to_cpu(iclog->ic_header.h_tail_lsn); + int blocks; if (CYCLE_LSN(tail_lsn) == log->l_prev_cycle) { blocks = diff --git a/fs/xfs/xfs_log_cil.c b/fs/xfs/xfs_log_cil.c index b128aaa9b870..4c44bc3786c0 100644 --- a/fs/xfs/xfs_log_cil.c +++ b/fs/xfs/xfs_log_cil.c @@ -654,8 +654,9 @@ xlog_cil_push_work( struct xfs_trans_header thdr; struct xfs_log_iovec lhdr; struct xfs_log_vec lvhdr = { NULL }; + xfs_lsn_t preflush_tail_lsn; xfs_lsn_t commit_lsn; - xfs_lsn_t push_seq; + xfs_csn_t push_seq; struct bio bio; DECLARE_COMPLETION_ONSTACK(bdev_flush); @@ -730,7 +731,15 @@ xlog_cil_push_work( * because we hold the flush lock exclusively. Hence we can now issue * a cache flush to ensure all the completed metadata in the journal we * are about to overwrite is on stable storage. + * + * Because we are issuing this cache flush before we've written the + * tail lsn to the iclog, we can have metadata IO completions move the + * tail forwards between the completion of this flush and the iclog + * being written. In this case, we need to re-issue the cache flush + * before the iclog write. To detect whether the log tail moves, sample + * the tail LSN *before* we issue the flush. */ + preflush_tail_lsn = atomic64_read(&log->l_tail_lsn); xfs_flush_bdev_async(&bio, log->l_mp->m_ddev_targp->bt_bdev, &bdev_flush); @@ -941,7 +950,7 @@ restart: * storage. */ commit_iclog->ic_flags |= XLOG_ICL_NEED_FUA; - xlog_state_release_iclog(log, commit_iclog); + xlog_state_release_iclog(log, commit_iclog, preflush_tail_lsn); spin_unlock(&log->l_icloglock); return; diff --git a/fs/xfs/xfs_log_priv.h b/fs/xfs/xfs_log_priv.h index 4c41bbfa33b0..f3e79a45d60a 100644 --- a/fs/xfs/xfs_log_priv.h +++ b/fs/xfs/xfs_log_priv.h @@ -59,6 +59,16 @@ enum xlog_iclog_state { { XLOG_STATE_DIRTY, "XLOG_STATE_DIRTY" }, \ { XLOG_STATE_IOERROR, "XLOG_STATE_IOERROR" } +/* + * In core log flags + */ +#define XLOG_ICL_NEED_FLUSH (1 << 0) /* iclog needs REQ_PREFLUSH */ +#define XLOG_ICL_NEED_FUA (1 << 1) /* iclog needs REQ_FUA */ + +#define XLOG_ICL_STRINGS \ + { XLOG_ICL_NEED_FLUSH, "XLOG_ICL_NEED_FLUSH" }, \ + { XLOG_ICL_NEED_FUA, "XLOG_ICL_NEED_FUA" } + /* * Log ticket flags @@ -143,9 +153,6 @@ enum xlog_iclog_state { #define XLOG_COVER_OPS 5 -#define XLOG_ICL_NEED_FLUSH (1 << 0) /* iclog needs REQ_PREFLUSH */ -#define XLOG_ICL_NEED_FUA (1 << 1) /* iclog needs REQ_FUA */ - /* Ticket reservation region accounting */ #define XLOG_TIC_LEN_MAX 15 @@ -497,7 +504,8 @@ int xlog_commit_record(struct xlog *log, struct xlog_ticket *ticket, void xfs_log_ticket_ungrant(struct xlog *log, struct xlog_ticket *ticket); void xfs_log_ticket_regrant(struct xlog *log, struct xlog_ticket *ticket); -int xlog_state_release_iclog(struct xlog *log, struct xlog_in_core *iclog); +int xlog_state_release_iclog(struct xlog *log, struct xlog_in_core *iclog, + xfs_lsn_t log_tail_lsn); /* * When we crack an atomic LSN, we sample it first so that the value will not diff --git a/fs/xfs/xfs_trace.h b/fs/xfs/xfs_trace.h index f9d8d605f9b1..19260291ff8b 100644 --- a/fs/xfs/xfs_trace.h +++ b/fs/xfs/xfs_trace.h @@ -3944,6 +3944,7 @@ DECLARE_EVENT_CLASS(xlog_iclog_class, __field(uint32_t, state) __field(int32_t, refcount) __field(uint32_t, offset) + __field(uint32_t, flags) __field(unsigned long long, lsn) __field(unsigned long, caller_ip) ), @@ -3952,15 +3953,17 @@ DECLARE_EVENT_CLASS(xlog_iclog_class, __entry->state = iclog->ic_state; __entry->refcount = atomic_read(&iclog->ic_refcnt); __entry->offset = iclog->ic_offset; + __entry->flags = iclog->ic_flags; __entry->lsn = be64_to_cpu(iclog->ic_header.h_lsn); __entry->caller_ip = caller_ip; ), - TP_printk("dev %d:%d state %s refcnt %d offset %u lsn 0x%llx caller %pS", + TP_printk("dev %d:%d state %s refcnt %d offset %u lsn 0x%llx flags %s caller %pS", MAJOR(__entry->dev), MINOR(__entry->dev), __print_symbolic(__entry->state, XLOG_STATE_STRINGS), __entry->refcount, __entry->offset, __entry->lsn, + __print_flags(__entry->flags, "|", XLOG_ICL_STRINGS), (char *)__entry->caller_ip) ); |