diff options
Diffstat (limited to 'fs/gfs2/super.c')
| -rw-r--r-- | fs/gfs2/super.c | 934 |
1 files changed, 467 insertions, 467 deletions
diff --git a/fs/gfs2/super.c b/fs/gfs2/super.c index 32d8d26126a1..f6cd907b3ec6 100644 --- a/fs/gfs2/super.c +++ b/fs/gfs2/super.c @@ -44,6 +44,12 @@ #include "xattr.h" #include "lops.h" +enum evict_behavior { + EVICT_SHOULD_DELETE, + EVICT_SHOULD_SKIP_DELETE, + EVICT_SHOULD_DEFER_DELETE, +}; + /** * gfs2_jindex_free - Clear all the journal index information * @sdp: The GFS2 superblock @@ -61,9 +67,13 @@ void gfs2_jindex_free(struct gfs2_sbd *sdp) sdp->sd_journals = 0; spin_unlock(&sdp->sd_jindex_spin); + down_write(&sdp->sd_log_flush_lock); sdp->sd_jdesc = NULL; + up_write(&sdp->sd_log_flush_lock); + while (!list_empty(&list)) { jd = list_first_entry(&list, struct gfs2_jdesc, jd_list); + BUG_ON(jd->jd_log_bio); gfs2_free_journal_extents(jd); list_del(&jd->jd_list); iput(jd->jd_inode); @@ -75,19 +85,12 @@ void gfs2_jindex_free(struct gfs2_sbd *sdp) static struct gfs2_jdesc *jdesc_find_i(struct list_head *head, unsigned int jid) { struct gfs2_jdesc *jd; - int found = 0; list_for_each_entry(jd, head, jd_list) { - if (jd->jd_jid == jid) { - found = 1; - break; - } + if (jd->jd_jid == jid) + return jd; } - - if (!found) - jd = NULL; - - return jd; + return NULL; } struct gfs2_jdesc *gfs2_jdesc_find(struct gfs2_sbd *sdp, unsigned int jid) @@ -120,34 +123,6 @@ int gfs2_jdesc_check(struct gfs2_jdesc *jd) return 0; } -static int init_threads(struct gfs2_sbd *sdp) -{ - struct task_struct *p; - int error = 0; - - p = kthread_run(gfs2_logd, sdp, "gfs2_logd"); - if (IS_ERR(p)) { - error = PTR_ERR(p); - fs_err(sdp, "can't start logd thread: %d\n", error); - return error; - } - sdp->sd_logd_process = p; - - p = kthread_run(gfs2_quotad, sdp, "gfs2_quotad"); - if (IS_ERR(p)) { - error = PTR_ERR(p); - fs_err(sdp, "can't start quotad thread: %d\n", error); - goto fail; - } - sdp->sd_quotad_process = p; - return 0; - -fail: - kthread_stop(sdp->sd_logd_process); - sdp->sd_logd_process = NULL; - return error; -} - /** * gfs2_make_fs_rw - Turn a Read-Only FS into a Read-Write one * @sdp: the filesystem @@ -159,59 +134,23 @@ int gfs2_make_fs_rw(struct gfs2_sbd *sdp) { struct gfs2_inode *ip = GFS2_I(sdp->sd_jdesc->jd_inode); struct gfs2_glock *j_gl = ip->i_gl; - struct gfs2_holder freeze_gh; - struct gfs2_log_header_host head; int error; - error = init_threads(sdp); - if (error) - return error; - - error = gfs2_glock_nq_init(sdp->sd_freeze_gl, LM_ST_SHARED, 0, - &freeze_gh); - if (error) - goto fail_threads; - j_gl->gl_ops->go_inval(j_gl, DIO_METADATA); - if (gfs2_withdrawn(sdp)) { - error = -EIO; - goto fail; - } - - error = gfs2_find_jhead(sdp->sd_jdesc, &head, false); - if (error || gfs2_withdrawn(sdp)) - goto fail; + if (gfs2_withdrawn(sdp)) + return -EIO; - if (!(head.lh_flags & GFS2_LOG_HEAD_UNMOUNT)) { - gfs2_consist(sdp); - error = -EIO; - goto fail; + if (sdp->sd_log_sequence == 0) { + fs_err(sdp, "unknown status of our own journal jid %d", + sdp->sd_lockstruct.ls_jid); + return -EIO; } - /* Initialize some head of the log stuff */ - sdp->sd_log_sequence = head.lh_sequence + 1; - gfs2_log_pointers_init(sdp, head.lh_blkno); - error = gfs2_quota_init(sdp); - if (error || gfs2_withdrawn(sdp)) - goto fail; - - set_bit(SDF_JOURNAL_LIVE, &sdp->sd_flags); - - gfs2_glock_dq_uninit(&freeze_gh); - - return 0; - -fail: - freeze_gh.gh_flags |= GL_NOCACHE; - gfs2_glock_dq_uninit(&freeze_gh); -fail_threads: - if (sdp->sd_quotad_process) - kthread_stop(sdp->sd_quotad_process); - sdp->sd_quotad_process = NULL; - if (sdp->sd_logd_process) - kthread_stop(sdp->sd_logd_process); - sdp->sd_logd_process = NULL; + if (!error && gfs2_withdrawn(sdp)) + error = -EIO; + if (!error) + set_bit(SDF_JOURNAL_LIVE, &sdp->sd_flags); return error; } @@ -224,7 +163,7 @@ void gfs2_statfs_change_in(struct gfs2_statfs_change_host *sc, const void *buf) sc->sc_dinodes = be64_to_cpu(str->sc_dinodes); } -static void gfs2_statfs_change_out(const struct gfs2_statfs_change_host *sc, void *buf) +void gfs2_statfs_change_out(const struct gfs2_statfs_change_host *sc, void *buf) { struct gfs2_statfs_change *str = buf; @@ -237,9 +176,8 @@ int gfs2_statfs_init(struct gfs2_sbd *sdp) { struct gfs2_inode *m_ip = GFS2_I(sdp->sd_statfs_inode); struct gfs2_statfs_change_host *m_sc = &sdp->sd_statfs_master; - struct gfs2_inode *l_ip = GFS2_I(sdp->sd_sc_inode); struct gfs2_statfs_change_host *l_sc = &sdp->sd_statfs_local; - struct buffer_head *m_bh, *l_bh; + struct buffer_head *m_bh; struct gfs2_holder gh; int error; @@ -258,21 +196,15 @@ int gfs2_statfs_init(struct gfs2_sbd *sdp) sizeof(struct gfs2_dinode)); spin_unlock(&sdp->sd_statfs_spin); } else { - error = gfs2_meta_inode_buffer(l_ip, &l_bh); - if (error) - goto out_m_bh; - spin_lock(&sdp->sd_statfs_spin); gfs2_statfs_change_in(m_sc, m_bh->b_data + sizeof(struct gfs2_dinode)); - gfs2_statfs_change_in(l_sc, l_bh->b_data + + gfs2_statfs_change_in(l_sc, sdp->sd_sc_bh->b_data + sizeof(struct gfs2_dinode)); spin_unlock(&sdp->sd_statfs_spin); - brelse(l_bh); } -out_m_bh: brelse(m_bh); out: gfs2_glock_dq_uninit(&gh); @@ -285,22 +217,17 @@ void gfs2_statfs_change(struct gfs2_sbd *sdp, s64 total, s64 free, struct gfs2_inode *l_ip = GFS2_I(sdp->sd_sc_inode); struct gfs2_statfs_change_host *l_sc = &sdp->sd_statfs_local; struct gfs2_statfs_change_host *m_sc = &sdp->sd_statfs_master; - struct buffer_head *l_bh; s64 x, y; int need_sync = 0; - int error; - error = gfs2_meta_inode_buffer(l_ip, &l_bh); - if (error) - return; - - gfs2_trans_add_meta(l_ip->i_gl, l_bh); + gfs2_trans_add_meta(l_ip->i_gl, sdp->sd_sc_bh); spin_lock(&sdp->sd_statfs_spin); l_sc->sc_total += total; l_sc->sc_free += free; l_sc->sc_dinodes += dinodes; - gfs2_statfs_change_out(l_sc, l_bh->b_data + sizeof(struct gfs2_dinode)); + gfs2_statfs_change_out(l_sc, sdp->sd_sc_bh->b_data + + sizeof(struct gfs2_dinode)); if (sdp->sd_args.ar_statfs_percent) { x = 100 * l_sc->sc_free; y = m_sc->sc_free * sdp->sd_args.ar_statfs_percent; @@ -309,20 +236,18 @@ void gfs2_statfs_change(struct gfs2_sbd *sdp, s64 total, s64 free, } spin_unlock(&sdp->sd_statfs_spin); - brelse(l_bh); if (need_sync) gfs2_wake_up_statfs(sdp); } -void update_statfs(struct gfs2_sbd *sdp, struct buffer_head *m_bh, - struct buffer_head *l_bh) +void update_statfs(struct gfs2_sbd *sdp, struct buffer_head *m_bh) { struct gfs2_inode *m_ip = GFS2_I(sdp->sd_statfs_inode); struct gfs2_inode *l_ip = GFS2_I(sdp->sd_sc_inode); struct gfs2_statfs_change_host *m_sc = &sdp->sd_statfs_master; struct gfs2_statfs_change_host *l_sc = &sdp->sd_statfs_local; - gfs2_trans_add_meta(l_ip->i_gl, l_bh); + gfs2_trans_add_meta(l_ip->i_gl, sdp->sd_sc_bh); gfs2_trans_add_meta(m_ip->i_gl, m_bh); spin_lock(&sdp->sd_statfs_spin); @@ -330,7 +255,7 @@ void update_statfs(struct gfs2_sbd *sdp, struct buffer_head *m_bh, m_sc->sc_free += l_sc->sc_free; m_sc->sc_dinodes += l_sc->sc_dinodes; memset(l_sc, 0, sizeof(struct gfs2_statfs_change)); - memset(l_bh->b_data + sizeof(struct gfs2_dinode), + memset(sdp->sd_sc_bh->b_data + sizeof(struct gfs2_dinode), 0, sizeof(struct gfs2_statfs_change)); gfs2_statfs_change_out(m_sc, m_bh->b_data + sizeof(struct gfs2_dinode)); spin_unlock(&sdp->sd_statfs_spin); @@ -340,14 +265,12 @@ int gfs2_statfs_sync(struct super_block *sb, int type) { struct gfs2_sbd *sdp = sb->s_fs_info; struct gfs2_inode *m_ip = GFS2_I(sdp->sd_statfs_inode); - struct gfs2_inode *l_ip = GFS2_I(sdp->sd_sc_inode); struct gfs2_statfs_change_host *m_sc = &sdp->sd_statfs_master; struct gfs2_statfs_change_host *l_sc = &sdp->sd_statfs_local; struct gfs2_holder gh; - struct buffer_head *m_bh, *l_bh; + struct buffer_head *m_bh; int error; - sb_start_write(sb); error = gfs2_glock_nq_init(m_ip->i_gl, LM_ST_EXCLUSIVE, GL_NOCACHE, &gh); if (error) @@ -366,27 +289,20 @@ int gfs2_statfs_sync(struct super_block *sb, int type) } spin_unlock(&sdp->sd_statfs_spin); - error = gfs2_meta_inode_buffer(l_ip, &l_bh); - if (error) - goto out_bh; - error = gfs2_trans_begin(sdp, 2 * RES_DINODE, 0); if (error) - goto out_bh2; + goto out_bh; - update_statfs(sdp, m_bh, l_bh); + update_statfs(sdp, m_bh); sdp->sd_statfs_force_sync = 0; gfs2_trans_end(sdp); -out_bh2: - brelse(l_bh); out_bh: brelse(m_bh); out_unlock: gfs2_glock_dq_uninit(&gh); out: - sb_end_write(sb); return error; } @@ -399,8 +315,6 @@ struct lfcc { * gfs2_lock_fs_check_clean - Stop all writes to the FS and check that all * journals are clean * @sdp: the file system - * @state: the state to put the transaction lock into - * @t_gh: the hold on the transaction lock * * Returns: errno */ @@ -412,7 +326,12 @@ static int gfs2_lock_fs_check_clean(struct gfs2_sbd *sdp) struct lfcc *lfcc; LIST_HEAD(list); struct gfs2_log_header_host lh; - int error; + int error, error2; + + /* + * Grab all the journal glocks in SH mode. We are *probably* doing + * that to prevent recovery. + */ list_for_each_entry(jd, &sdp->sd_jindex_list, jd_list) { lfcc = kmalloc(sizeof(struct lfcc), GFP_KERNEL); @@ -429,16 +348,19 @@ static int gfs2_lock_fs_check_clean(struct gfs2_sbd *sdp) list_add(&lfcc->list, &list); } + gfs2_freeze_unlock(sdp); + error = gfs2_glock_nq_init(sdp->sd_freeze_gl, LM_ST_EXCLUSIVE, - GL_NOCACHE, &sdp->sd_freeze_gh); + LM_FLAG_RECOVER | GL_NOPID, + &sdp->sd_freeze_gh); if (error) - goto out; + goto relock_shared; list_for_each_entry(jd, &sdp->sd_jindex_list, jd_list) { error = gfs2_jdesc_check(jd); if (error) break; - error = gfs2_find_jhead(jd, &lh, false); + error = gfs2_find_jhead(jd, &lh); if (error) break; if (!(lh.lh_flags & GFS2_LOG_HEAD_UNMOUNT)) { @@ -447,8 +369,14 @@ static int gfs2_lock_fs_check_clean(struct gfs2_sbd *sdp) } } - if (error) - gfs2_glock_dq_uninit(&sdp->sd_freeze_gh); + if (!error) + goto out; /* success */ + + gfs2_freeze_unlock(sdp); + +relock_shared: + error2 = gfs2_freeze_lock_shared(sdp); + gfs2_assert_withdraw(sdp, !error2); out: while (!list_empty(&list)) { @@ -462,6 +390,7 @@ out: void gfs2_dinode_out(const struct gfs2_inode *ip, void *buf) { + const struct inode *inode = &ip->i_inode; struct gfs2_dinode *str = buf; str->di_header.mh_magic = cpu_to_be32(GFS2_MAGIC); @@ -469,15 +398,15 @@ void gfs2_dinode_out(const struct gfs2_inode *ip, void *buf) str->di_header.mh_format = cpu_to_be32(GFS2_FORMAT_DI); str->di_num.no_addr = cpu_to_be64(ip->i_no_addr); str->di_num.no_formal_ino = cpu_to_be64(ip->i_no_formal_ino); - str->di_mode = cpu_to_be32(ip->i_inode.i_mode); - str->di_uid = cpu_to_be32(i_uid_read(&ip->i_inode)); - str->di_gid = cpu_to_be32(i_gid_read(&ip->i_inode)); - str->di_nlink = cpu_to_be32(ip->i_inode.i_nlink); - str->di_size = cpu_to_be64(i_size_read(&ip->i_inode)); - str->di_blocks = cpu_to_be64(gfs2_get_inode_blocks(&ip->i_inode)); - str->di_atime = cpu_to_be64(ip->i_inode.i_atime.tv_sec); - str->di_mtime = cpu_to_be64(ip->i_inode.i_mtime.tv_sec); - str->di_ctime = cpu_to_be64(ip->i_inode.i_ctime.tv_sec); + str->di_mode = cpu_to_be32(inode->i_mode); + str->di_uid = cpu_to_be32(i_uid_read(inode)); + str->di_gid = cpu_to_be32(i_gid_read(inode)); + str->di_nlink = cpu_to_be32(inode->i_nlink); + str->di_size = cpu_to_be64(i_size_read(inode)); + str->di_blocks = cpu_to_be64(gfs2_get_inode_blocks(inode)); + str->di_atime = cpu_to_be64(inode_get_atime_sec(inode)); + str->di_mtime = cpu_to_be64(inode_get_mtime_sec(inode)); + str->di_ctime = cpu_to_be64(inode_get_ctime_sec(inode)); str->di_goal_meta = cpu_to_be64(ip->i_goal); str->di_goal_data = cpu_to_be64(ip->i_goal); @@ -485,16 +414,16 @@ void gfs2_dinode_out(const struct gfs2_inode *ip, void *buf) str->di_flags = cpu_to_be32(ip->i_diskflags); str->di_height = cpu_to_be16(ip->i_height); - str->di_payload_format = cpu_to_be32(S_ISDIR(ip->i_inode.i_mode) && + str->di_payload_format = cpu_to_be32(S_ISDIR(inode->i_mode) && !(ip->i_diskflags & GFS2_DIF_EXHASH) ? GFS2_FORMAT_DE : 0); str->di_depth = cpu_to_be16(ip->i_depth); str->di_entries = cpu_to_be32(ip->i_entries); str->di_eattr = cpu_to_be64(ip->i_eattr); - str->di_atime_nsec = cpu_to_be32(ip->i_inode.i_atime.tv_nsec); - str->di_mtime_nsec = cpu_to_be32(ip->i_inode.i_mtime.tv_nsec); - str->di_ctime_nsec = cpu_to_be32(ip->i_inode.i_ctime.tv_nsec); + str->di_atime_nsec = cpu_to_be32(inode_get_atime_nsec(inode)); + str->di_mtime_nsec = cpu_to_be32(inode_get_mtime_nsec(inode)); + str->di_ctime_nsec = cpu_to_be32(inode_get_ctime_nsec(inode)); } /** @@ -541,7 +470,7 @@ static int gfs2_write_inode(struct inode *inode, struct writeback_control *wbc) * @flags: The type of dirty * * Unfortunately it can be called under any combination of inode - * glock and transaction lock, so we have to check carefully. + * glock and freeze glock, so we have to check carefully. * * At the moment this deals only with atime - it should be possible * to expand that role in future, once a review of the locking has @@ -558,14 +487,17 @@ static void gfs2_dirty_inode(struct inode *inode, int flags) int need_endtrans = 0; int ret; - if (!(flags & I_DIRTY_INODE)) + /* This can only happen during incomplete inode creation. */ + if (unlikely(!ip->i_gl)) return; - if (unlikely(gfs2_withdrawn(sdp))) + + if (gfs2_withdrawn(sdp)) return; if (!gfs2_glock_is_locked_by_me(ip->i_gl)) { ret = gfs2_glock_nq_init(ip->i_gl, LM_ST_EXCLUSIVE, 0, &gh); if (ret) { fs_err(sdp, "dirty_inode: glock %d\n", ret); + gfs2_dump_glock(NULL, ip->i_gl, true); return; } need_unlock = 1; @@ -602,67 +534,36 @@ out: * Returns: errno */ -int gfs2_make_fs_ro(struct gfs2_sbd *sdp) +void gfs2_make_fs_ro(struct gfs2_sbd *sdp) { - struct gfs2_holder freeze_gh; - int error = 0; int log_write_allowed = test_bit(SDF_JOURNAL_LIVE, &sdp->sd_flags); - gfs2_holder_mark_uninitialized(&freeze_gh); - if (sdp->sd_freeze_gl && - !gfs2_glock_is_locked_by_me(sdp->sd_freeze_gl)) { - if (!log_write_allowed) { - error = gfs2_glock_nq_init(sdp->sd_freeze_gl, - LM_ST_SHARED, GL_NOCACHE | - LM_FLAG_TRY, &freeze_gh); - if (error == GLR_TRYFAILED) - error = 0; - } else { - error = gfs2_glock_nq_init(sdp->sd_freeze_gl, - LM_ST_SHARED, GL_NOCACHE, - &freeze_gh); - if (error && !gfs2_withdrawn(sdp)) - return error; - } - } - - gfs2_flush_delete_work(sdp); - if (!log_write_allowed && current == sdp->sd_quotad_process) - fs_warn(sdp, "The quotad daemon is withdrawing.\n"); - else if (sdp->sd_quotad_process) - kthread_stop(sdp->sd_quotad_process); - sdp->sd_quotad_process = NULL; + if (!test_bit(SDF_KILL, &sdp->sd_flags)) + gfs2_flush_delete_work(sdp); - if (!log_write_allowed && current == sdp->sd_logd_process) - fs_warn(sdp, "The logd daemon is withdrawing.\n"); - else if (sdp->sd_logd_process) - kthread_stop(sdp->sd_logd_process); - sdp->sd_logd_process = NULL; + gfs2_destroy_threads(sdp); if (log_write_allowed) { gfs2_quota_sync(sdp->sd_vfs, 0); gfs2_statfs_sync(sdp->sd_vfs, 0); + /* We do two log flushes here. The first one commits dirty inodes + * and rgrps to the journal, but queues up revokes to the ail list. + * The second flush writes out and removes the revokes. + * + * The first must be done before the FLUSH_SHUTDOWN code + * clears the LIVE flag, otherwise it will not be able to start + * a transaction to write its revokes, and the error will cause + * a withdraw of the file system. */ + gfs2_log_flush(sdp, NULL, GFS2_LFC_MAKE_FS_RO); gfs2_log_flush(sdp, NULL, GFS2_LOG_HEAD_FLUSH_SHUTDOWN | GFS2_LFC_MAKE_FS_RO); - wait_event(sdp->sd_reserving_log_wait, - atomic_read(&sdp->sd_reserving_log) == 0); - gfs2_assert_warn(sdp, atomic_read(&sdp->sd_log_blks_free) == - sdp->sd_jdesc->jd_blocks); - } else { - wait_event_timeout(sdp->sd_reserving_log_wait, - atomic_read(&sdp->sd_reserving_log) == 0, + wait_event_timeout(sdp->sd_log_waitq, + gfs2_log_is_empty(sdp), HZ * 5); + gfs2_assert_warn(sdp, gfs2_log_is_empty(sdp)); } - if (gfs2_holder_initialized(&freeze_gh)) - gfs2_glock_dq_uninit(&freeze_gh); - gfs2_quota_cleanup(sdp); - - if (!log_write_allowed) - sdp->sd_vfs->s_flags |= SB_RDONLY; - - return error; } /** @@ -674,7 +575,6 @@ int gfs2_make_fs_ro(struct gfs2_sbd *sdp) static void gfs2_put_super(struct super_block *sb) { struct gfs2_sbd *sdp = sb->s_fs_info; - int error; struct gfs2_jdesc *jd; /* No more recovery requests */ @@ -694,15 +594,23 @@ restart: } spin_unlock(&sdp->sd_jindex_spin); - if (!sb_rdonly(sb)) { - error = gfs2_make_fs_ro(sdp); - if (error) - gfs2_io_error(sdp); + if (!sb_rdonly(sb)) + gfs2_make_fs_ro(sdp); + else { + if (gfs2_withdrawn(sdp)) + gfs2_destroy_threads(sdp); + + gfs2_quota_cleanup(sdp); } + + flush_work(&sdp->sd_withdraw_work); + /* At this point, we're through modifying the disk */ /* Release stuff */ + gfs2_freeze_unlock(sdp); + iput(sdp->sd_jindex); iput(sdp->sd_statfs_inode); iput(sdp->sd_rindex); @@ -716,9 +624,10 @@ restart: gfs2_glock_dq_uninit(&sdp->sd_journal_gh); if (gfs2_holder_initialized(&sdp->sd_jinode_gh)) gfs2_glock_dq_uninit(&sdp->sd_jinode_gh); + brelse(sdp->sd_sc_bh); gfs2_glock_dq_uninit(&sdp->sd_sc_gh); gfs2_glock_dq_uninit(&sdp->sd_qc_gh); - iput(sdp->sd_sc_inode); + free_local_statfs_inodes(sdp); iput(sdp->sd_qc_inode); } @@ -727,17 +636,17 @@ restart: gfs2_jindex_free(sdp); /* Take apart glock structures and buffer lists */ gfs2_gl_hash_clear(sdp); + iput(sdp->sd_inode); gfs2_delete_debugfs_file(sdp); - /* Unmount the locking protocol */ - gfs2_lm_unmount(sdp); - /* At this point, we're through participating in the lockspace */ gfs2_sys_fs_del(sdp); + free_sbd(sdp); } /** * gfs2_sync_fs - sync the filesystem * @sb: the superblock + * @wait: true to wait for completion * * Flushes the log to disk. */ @@ -753,61 +662,94 @@ static int gfs2_sync_fs(struct super_block *sb, int wait) return sdp->sd_log_error; } -void gfs2_freeze_func(struct work_struct *work) +static int gfs2_do_thaw(struct gfs2_sbd *sdp, enum freeze_holder who, const void *freeze_owner) { + struct super_block *sb = sdp->sd_vfs; int error; - struct gfs2_holder freeze_gh; + + error = gfs2_freeze_lock_shared(sdp); + if (error) + goto fail; + error = thaw_super(sb, who, freeze_owner); + if (!error) + return 0; + +fail: + fs_info(sdp, "GFS2: couldn't thaw filesystem: %d\n", error); + gfs2_assert_withdraw(sdp, 0); + return error; +} + +void gfs2_freeze_func(struct work_struct *work) +{ struct gfs2_sbd *sdp = container_of(work, struct gfs2_sbd, sd_freeze_work); struct super_block *sb = sdp->sd_vfs; + int error; - atomic_inc(&sb->s_active); - error = gfs2_glock_nq_init(sdp->sd_freeze_gl, LM_ST_SHARED, 0, - &freeze_gh); - if (error) { - fs_info(sdp, "GFS2: couldn't get freeze lock : %d\n", error); - gfs2_assert_withdraw(sdp, 0); - } else { - atomic_set(&sdp->sd_freeze_state, SFS_UNFROZEN); - error = thaw_super(sb); - if (error) { - fs_info(sdp, "GFS2: couldn't thaw filesystem: %d\n", - error); - gfs2_assert_withdraw(sdp, 0); - } - if (!test_bit(SDF_JOURNAL_LIVE, &sdp->sd_flags)) - freeze_gh.gh_flags |= GL_NOCACHE; - gfs2_glock_dq_uninit(&freeze_gh); - } + mutex_lock(&sdp->sd_freeze_mutex); + error = -EBUSY; + if (test_bit(SDF_FROZEN, &sdp->sd_flags)) + goto freeze_failed; + + error = freeze_super(sb, FREEZE_HOLDER_USERSPACE, NULL); + if (error) + goto freeze_failed; + + gfs2_freeze_unlock(sdp); + set_bit(SDF_FROZEN, &sdp->sd_flags); + + error = gfs2_do_thaw(sdp, FREEZE_HOLDER_USERSPACE, NULL); + if (error) + goto out; + + clear_bit(SDF_FROZEN, &sdp->sd_flags); + goto out; + +freeze_failed: + fs_info(sdp, "GFS2: couldn't freeze filesystem: %d\n", error); + +out: + mutex_unlock(&sdp->sd_freeze_mutex); deactivate_super(sb); - clear_bit_unlock(SDF_FS_FROZEN, &sdp->sd_flags); - wake_up_bit(&sdp->sd_flags, SDF_FS_FROZEN); - return; } /** - * gfs2_freeze - prevent further writes to the filesystem + * gfs2_freeze_super - prevent further writes to the filesystem * @sb: the VFS structure for the filesystem + * @who: freeze flags + * @freeze_owner: owner of the freeze * */ -static int gfs2_freeze(struct super_block *sb) +static int gfs2_freeze_super(struct super_block *sb, enum freeze_holder who, + const void *freeze_owner) { struct gfs2_sbd *sdp = sb->s_fs_info; - int error = 0; + int error; - mutex_lock(&sdp->sd_freeze_mutex); - if (atomic_read(&sdp->sd_freeze_state) != SFS_UNFROZEN) - goto out; + if (!mutex_trylock(&sdp->sd_freeze_mutex)) + return -EBUSY; + if (test_bit(SDF_FROZEN, &sdp->sd_flags)) { + mutex_unlock(&sdp->sd_freeze_mutex); + return -EBUSY; + } for (;;) { - if (gfs2_withdrawn(sdp)) { - error = -EINVAL; + error = freeze_super(sb, who, freeze_owner); + if (error) { + fs_info(sdp, "GFS2: couldn't freeze filesystem: %d\n", + error); goto out; } error = gfs2_lock_fs_check_clean(sdp); - if (!error) + if (!error) { + set_bit(SDF_FREEZE_INITIATOR, &sdp->sd_flags); + set_bit(SDF_FROZEN, &sdp->sd_flags); break; + } + + (void)gfs2_do_thaw(sdp, who, freeze_owner); if (error == -EBUSY) fs_err(sdp, "waiting for recovery before freeze\n"); @@ -821,36 +763,62 @@ static int gfs2_freeze(struct super_block *sb) fs_err(sdp, "retrying...\n"); msleep(1000); } - set_bit(SDF_FS_FROZEN, &sdp->sd_flags); + out: mutex_unlock(&sdp->sd_freeze_mutex); return error; } +static int gfs2_freeze_fs(struct super_block *sb) +{ + struct gfs2_sbd *sdp = sb->s_fs_info; + + if (test_bit(SDF_JOURNAL_LIVE, &sdp->sd_flags)) { + gfs2_log_flush(sdp, NULL, GFS2_LOG_HEAD_FLUSH_FREEZE | + GFS2_LFC_FREEZE_GO_SYNC); + if (gfs2_withdrawn(sdp)) + return -EIO; + } + return 0; +} + /** - * gfs2_unfreeze - reallow writes to the filesystem + * gfs2_thaw_super - reallow writes to the filesystem * @sb: the VFS structure for the filesystem + * @who: freeze flags + * @freeze_owner: owner of the freeze * */ -static int gfs2_unfreeze(struct super_block *sb) +static int gfs2_thaw_super(struct super_block *sb, enum freeze_holder who, + const void *freeze_owner) { struct gfs2_sbd *sdp = sb->s_fs_info; + int error; - mutex_lock(&sdp->sd_freeze_mutex); - if (atomic_read(&sdp->sd_freeze_state) != SFS_FROZEN || - !gfs2_holder_initialized(&sdp->sd_freeze_gh)) { + if (!mutex_trylock(&sdp->sd_freeze_mutex)) + return -EBUSY; + if (!test_bit(SDF_FREEZE_INITIATOR, &sdp->sd_flags)) { mutex_unlock(&sdp->sd_freeze_mutex); - return 0; + return -EINVAL; } - gfs2_glock_dq_uninit(&sdp->sd_freeze_gh); + atomic_inc(&sb->s_active); + gfs2_freeze_unlock(sdp); + + error = gfs2_do_thaw(sdp, who, freeze_owner); + + if (!error) { + clear_bit(SDF_FREEZE_INITIATOR, &sdp->sd_flags); + clear_bit(SDF_FROZEN, &sdp->sd_flags); + } mutex_unlock(&sdp->sd_freeze_mutex); - return wait_on_bit(&sdp->sd_flags, SDF_FS_FROZEN, TASK_INTERRUPTIBLE); + deactivate_super(sb); + return error; } /** - * statfs_fill - fill in the sg for a given RG + * statfs_slow_fill - fill in the sg for a given RG * @rgd: the RG * @sc: the sc structure * @@ -948,7 +916,7 @@ static int gfs2_statfs_slow(struct gfs2_sbd *sdp, struct gfs2_statfs_change_host /** * gfs2_statfs_i - Do a statfs * @sdp: the filesystem - * @sg: the sg structure + * @sc: the sc structure * * Returns: errno */ @@ -979,8 +947,8 @@ static int gfs2_statfs_i(struct gfs2_sbd *sdp, struct gfs2_statfs_change_host *s /** * gfs2_statfs - Gather and return stats about the filesystem - * @sb: The superblock - * @statfsbuf: The buffer + * @dentry: The name of the link + * @buf: The buffer * * Returns: 0 on success or error code */ @@ -1012,6 +980,7 @@ static int gfs2_statfs(struct dentry *dentry, struct kstatfs *buf) buf->f_files = sc.sc_dinodes + sc.sc_free; buf->f_ffree = sc.sc_free; buf->f_namelen = GFS2_FNAMESIZE; + buf->f_fsid = uuid_to_fsid(sb->s_uuid.b); return 0; } @@ -1034,12 +1003,12 @@ static int gfs2_statfs(struct dentry *dentry, struct kstatfs *buf) static int gfs2_drop_inode(struct inode *inode) { struct gfs2_inode *ip = GFS2_I(inode); + struct gfs2_sbd *sdp = GFS2_SB(inode); - if (!test_bit(GIF_FREE_VFS_INODE, &ip->i_flags) && - inode->i_nlink && + if (inode->i_nlink && gfs2_holder_initialized(&ip->i_iopen_gh)) { struct gfs2_glock *gl = ip->i_iopen_gh.gh_gl; - if (test_bit(GLF_DEMOTE, &gl->gl_flags)) + if (glock_needs_demote(gl)) clear_nlink(inode); } @@ -1054,22 +1023,18 @@ static int gfs2_drop_inode(struct inode *inode) struct gfs2_glock *gl = ip->i_iopen_gh.gh_gl; gfs2_glock_hold(gl); - if (!gfs2_queue_delete_work(gl, 0)) - gfs2_glock_queue_put(gl); - return false; + if (!gfs2_queue_verify_delete(gl, true)) + gfs2_glock_put_async(gl); + return 0; } - return generic_drop_inode(inode); -} + /* + * No longer cache inodes when trying to evict them all. + */ + if (test_bit(SDF_EVICTING, &sdp->sd_flags)) + return 1; -static int is_ancestor(const struct dentry *d1, const struct dentry *d2) -{ - do { - if (d1 == d2) - return 1; - d1 = d1->d_parent; - } while (!IS_ROOT(d1)); - return 0; + return inode_generic_drop(inode); } /** @@ -1084,9 +1049,16 @@ static int gfs2_show_options(struct seq_file *s, struct dentry *root) { struct gfs2_sbd *sdp = root->d_sb->s_fs_info; struct gfs2_args *args = &sdp->sd_args; - int val; + unsigned int logd_secs, statfs_slow, statfs_quantum, quota_quantum; - if (is_ancestor(root, sdp->sd_master_dir)) + spin_lock(&sdp->sd_tune.gt_spin); + logd_secs = sdp->sd_tune.gt_logd_secs; + quota_quantum = sdp->sd_tune.gt_quota_quantum; + statfs_quantum = sdp->sd_tune.gt_statfs_quantum; + statfs_slow = sdp->sd_tune.gt_statfs_slow; + spin_unlock(&sdp->sd_tune.gt_spin); + + if (is_subdir(root, sdp->sd_master_dir)) seq_puts(s, ",meta"); if (args->ar_lockproto[0]) seq_show_option(s, "lockproto", args->ar_lockproto); @@ -1114,6 +1086,9 @@ static int gfs2_show_options(struct seq_file *s, struct dentry *root) case GFS2_QUOTA_ON: state = "on"; break; + case GFS2_QUOTA_QUIET: + state = "quiet"; + break; default: state = "unknown"; break; @@ -1139,17 +1114,14 @@ static int gfs2_show_options(struct seq_file *s, struct dentry *root) } if (args->ar_discard) seq_puts(s, ",discard"); - val = sdp->sd_tune.gt_logd_secs; - if (val != 30) - seq_printf(s, ",commit=%d", val); - val = sdp->sd_tune.gt_statfs_quantum; - if (val != 30) - seq_printf(s, ",statfs_quantum=%d", val); - else if (sdp->sd_tune.gt_statfs_slow) + if (logd_secs != 30) + seq_printf(s, ",commit=%d", logd_secs); + if (statfs_quantum != 30) + seq_printf(s, ",statfs_quantum=%d", statfs_quantum); + else if (statfs_slow) seq_puts(s, ",statfs_quantum=0"); - val = sdp->sd_tune.gt_quota_quantum; - if (val != 60) - seq_printf(s, ",quota_quantum=%d", val); + if (quota_quantum != 60) + seq_printf(s, ",quota_quantum=%d", quota_quantum); if (args->ar_statfs_percent) seq_printf(s, ",statfs_percent=%d", args->ar_statfs_percent); if (args->ar_errors != GFS2_ERRORS_DEFAULT) { @@ -1159,6 +1131,9 @@ static int gfs2_show_options(struct seq_file *s, struct dentry *root) case GFS2_ERRORS_WITHDRAW: state = "withdraw"; break; + case GFS2_ERRORS_DEACTIVATE: + state = "deactivate"; + break; case GFS2_ERRORS_PANIC: state = "panic"; break; @@ -1179,69 +1154,6 @@ static int gfs2_show_options(struct seq_file *s, struct dentry *root) return 0; } -static void gfs2_final_release_pages(struct gfs2_inode *ip) -{ - struct inode *inode = &ip->i_inode; - struct gfs2_glock *gl = ip->i_gl; - - truncate_inode_pages(gfs2_glock2aspace(ip->i_gl), 0); - truncate_inode_pages(&inode->i_data, 0); - - if (atomic_read(&gl->gl_revokes) == 0) { - clear_bit(GLF_LFLUSH, &gl->gl_flags); - clear_bit(GLF_DIRTY, &gl->gl_flags); - } -} - -static int gfs2_dinode_dealloc(struct gfs2_inode *ip) -{ - struct gfs2_sbd *sdp = GFS2_SB(&ip->i_inode); - struct gfs2_rgrpd *rgd; - struct gfs2_holder gh; - int error; - - if (gfs2_get_inode_blocks(&ip->i_inode) != 1) { - gfs2_consist_inode(ip); - return -EIO; - } - - error = gfs2_rindex_update(sdp); - if (error) - return error; - - error = gfs2_quota_hold(ip, NO_UID_QUOTA_CHANGE, NO_GID_QUOTA_CHANGE); - if (error) - return error; - - rgd = gfs2_blk2rgrpd(sdp, ip->i_no_addr, 1); - if (!rgd) { - gfs2_consist_inode(ip); - error = -EIO; - goto out_qs; - } - - error = gfs2_glock_nq_init(rgd->rd_gl, LM_ST_EXCLUSIVE, 0, &gh); - if (error) - goto out_qs; - - error = gfs2_trans_begin(sdp, RES_RG_BIT + RES_STATFS + RES_QUOTA, - sdp->sd_jdesc->jd_blocks); - if (error) - goto out_rg_gunlock; - - gfs2_free_di(rgd, ip); - - gfs2_final_release_pages(ip); - - gfs2_trans_end(sdp); - -out_rg_gunlock: - gfs2_glock_dq_uninit(&gh); -out_qs: - gfs2_quota_unhold(ip); - return error; -} - /** * gfs2_glock_put_eventually * @gl: The glock to put @@ -1253,176 +1165,175 @@ out_qs: static void gfs2_glock_put_eventually(struct gfs2_glock *gl) { if (current->flags & PF_MEMALLOC) - gfs2_glock_queue_put(gl); + gfs2_glock_put_async(gl); else gfs2_glock_put(gl); } -static bool gfs2_upgrade_iopen_glock(struct inode *inode) +static enum evict_behavior gfs2_upgrade_iopen_glock(struct inode *inode) { struct gfs2_inode *ip = GFS2_I(inode); struct gfs2_sbd *sdp = GFS2_SB(inode); struct gfs2_holder *gh = &ip->i_iopen_gh; - long timeout = 5 * HZ; int error; gh->gh_flags |= GL_NOCACHE; gfs2_glock_dq_wait(gh); /* - * If there are no other lock holders, we'll get the lock immediately. + * If there are no other lock holders, we will immediately get + * exclusive access to the iopen glock here. + * * Otherwise, the other nodes holding the lock will be notified about - * our locking request. If they don't have the inode open, they'll - * evict the cached inode and release the lock. Otherwise, if they - * poke the inode glock, we'll take this as an indication that they - * still need the iopen glock and that they'll take care of deleting - * the inode when they're done. As a last resort, if another node - * keeps holding the iopen glock without showing any activity on the - * inode glock, we'll eventually time out. + * our locking request (see iopen_go_callback()). If they do not have + * the inode open, they are expected to evict the cached inode and + * release the lock, allowing us to proceed. + * + * Otherwise, if they cannot evict the inode, they are expected to poke + * the inode glock (note: not the iopen glock). We will notice that + * and stop waiting for the iopen glock immediately. The other node(s) + * are then expected to take care of deleting the inode when they no + * longer use it. * - * Note that we're passing the LM_FLAG_TRY_1CB flag to the first - * locking request as an optimization to notify lock holders as soon as - * possible. Without that flag, they'd be notified implicitly by the - * second locking request. + * As a last resort, if another node keeps holding the iopen glock + * without showing any activity on the inode glock, we will eventually + * time out and fail the iopen glock upgrade. */ - gfs2_holder_reinit(LM_ST_EXCLUSIVE, LM_FLAG_TRY_1CB | GL_NOCACHE, gh); - error = gfs2_glock_nq(gh); - if (error != GLR_TRYFAILED) - return !error; - gfs2_holder_reinit(LM_ST_EXCLUSIVE, GL_ASYNC | GL_NOCACHE, gh); error = gfs2_glock_nq(gh); if (error) - return false; + return EVICT_SHOULD_SKIP_DELETE; - timeout = wait_event_interruptible_timeout(sdp->sd_async_glock_wait, + wait_event_interruptible_timeout(sdp->sd_async_glock_wait, !test_bit(HIF_WAIT, &gh->gh_iflags) || - test_bit(GLF_DEMOTE, &ip->i_gl->gl_flags), - timeout); + glock_needs_demote(ip->i_gl), + 5 * HZ); if (!test_bit(HIF_HOLDER, &gh->gh_iflags)) { gfs2_glock_dq(gh); - return false; + if (glock_needs_demote(ip->i_gl)) + return EVICT_SHOULD_SKIP_DELETE; + return EVICT_SHOULD_DEFER_DELETE; } - return true; + error = gfs2_glock_holder_ready(gh); + if (error) + return EVICT_SHOULD_SKIP_DELETE; + return EVICT_SHOULD_DELETE; } /** - * gfs2_evict_inode - Remove an inode from cache + * evict_should_delete - determine whether the inode is eligible for deletion * @inode: The inode to evict + * @gh: The glock holder structure * - * There are three cases to consider: - * 1. i_nlink == 0, we are final opener (and must deallocate) - * 2. i_nlink == 0, we are not the final opener (and cannot deallocate) - * 3. i_nlink > 0 + * This function determines whether the evicted inode is eligible to be deleted + * and locks the inode glock. * - * If the fs is read only, then we have to treat all cases as per #3 - * since we are unable to do any deallocation. The inode will be - * deallocated by the next read/write node to attempt an allocation - * in the same resource group - * - * We have to (at the moment) hold the inodes main lock to cover - * the gap between unlocking the shared lock on the iopen lock and - * taking the exclusive lock. I'd rather do a shared -> exclusive - * conversion on the iopen lock, but we can change that later. This - * is safe, just less efficient. + * Returns: the fate of the dinode */ - -static void gfs2_evict_inode(struct inode *inode) +static enum evict_behavior evict_should_delete(struct inode *inode, + struct gfs2_holder *gh) { + struct gfs2_inode *ip = GFS2_I(inode); struct super_block *sb = inode->i_sb; struct gfs2_sbd *sdp = sb->s_fs_info; - struct gfs2_inode *ip = GFS2_I(inode); - struct gfs2_holder gh; - struct address_space *metamapping; - int error; - - if (test_bit(GIF_FREE_VFS_INODE, &ip->i_flags)) { - clear_inode(inode); - return; - } - - if (inode->i_nlink || sb_rdonly(sb)) - goto out; - - if (test_bit(GIF_ALLOC_FAILED, &ip->i_flags)) { - BUG_ON(!gfs2_glock_is_locked_by_me(ip->i_gl)); - gfs2_holder_mark_uninitialized(&gh); - goto out_delete; - } + int ret; - if (test_bit(GIF_DEFERRED_DELETE, &ip->i_flags)) - goto out; + if (gfs2_holder_initialized(&ip->i_iopen_gh) && + test_bit(GLF_DEFER_DELETE, &ip->i_iopen_gh.gh_gl->gl_flags)) + return EVICT_SHOULD_DEFER_DELETE; /* Deletes should never happen under memory pressure anymore. */ if (WARN_ON_ONCE(current->flags & PF_MEMALLOC)) - goto out; + return EVICT_SHOULD_DEFER_DELETE; /* Must not read inode block until block type has been verified */ - error = gfs2_glock_nq_init(ip->i_gl, LM_ST_EXCLUSIVE, GL_SKIP, &gh); - if (unlikely(error)) { - glock_clear_object(ip->i_iopen_gh.gh_gl, ip); - ip->i_iopen_gh.gh_flags |= GL_NOCACHE; - gfs2_glock_dq_uninit(&ip->i_iopen_gh); - goto out; - } + ret = gfs2_glock_nq_init(ip->i_gl, LM_ST_EXCLUSIVE, GL_SKIP, gh); + if (unlikely(ret)) + return EVICT_SHOULD_SKIP_DELETE; if (gfs2_inode_already_deleted(ip->i_gl, ip->i_no_formal_ino)) - goto out_truncate; - error = gfs2_check_blk_type(sdp, ip->i_no_addr, GFS2_BLKST_UNLINKED); - if (error) - goto out_truncate; + return EVICT_SHOULD_SKIP_DELETE; + ret = gfs2_check_blk_type(sdp, ip->i_no_addr, GFS2_BLKST_UNLINKED); + if (ret) + return EVICT_SHOULD_SKIP_DELETE; - if (test_bit(GIF_INVALID, &ip->i_flags)) { - error = gfs2_inode_refresh(ip); - if (error) - goto out_truncate; - } + ret = gfs2_instantiate(gh); + if (ret) + return EVICT_SHOULD_SKIP_DELETE; /* * The inode may have been recreated in the meantime. */ if (inode->i_nlink) - goto out_truncate; + return EVICT_SHOULD_SKIP_DELETE; -out_delete: if (gfs2_holder_initialized(&ip->i_iopen_gh) && - test_bit(HIF_HOLDER, &ip->i_iopen_gh.gh_iflags)) { - if (!gfs2_upgrade_iopen_glock(inode)) { - gfs2_holder_uninit(&ip->i_iopen_gh); - goto out_truncate; - } - } + test_bit(HIF_HOLDER, &ip->i_iopen_gh.gh_iflags)) + return gfs2_upgrade_iopen_glock(inode); + return EVICT_SHOULD_DELETE; +} + +/** + * evict_unlinked_inode - delete the pieces of an unlinked evicted inode + * @inode: The inode to evict + */ +static int evict_unlinked_inode(struct inode *inode) +{ + struct gfs2_inode *ip = GFS2_I(inode); + int ret; if (S_ISDIR(inode->i_mode) && (ip->i_diskflags & GFS2_DIF_EXHASH)) { - error = gfs2_dir_exhash_dealloc(ip); - if (error) - goto out_unlock; + ret = gfs2_dir_exhash_dealloc(ip); + if (ret) + goto out; } if (ip->i_eattr) { - error = gfs2_ea_dealloc(ip); - if (error) - goto out_unlock; + ret = gfs2_ea_dealloc(ip, true); + if (ret) + goto out; } if (!gfs2_is_stuffed(ip)) { - error = gfs2_file_dealloc(ip); - if (error) - goto out_unlock; + ret = gfs2_file_dealloc(ip); + if (ret) + goto out; } - /* We're about to clear the bitmap for the dinode, but as soon as we - do, gfs2_create_inode can create another inode at the same block - location and try to set gl_object again. We clear gl_object here so - that subsequent inode creates don't see an old gl_object. */ - glock_clear_object(ip->i_gl, ip); - error = gfs2_dinode_dealloc(ip); - gfs2_inode_remember_delete(ip->i_gl, ip->i_no_formal_ino); - goto out_unlock; + /* + * As soon as we clear the bitmap for the dinode, gfs2_create_inode() + * can get called to recreate it, or even gfs2_inode_lookup() if the + * inode was recreated on another node in the meantime. + * + * However, inserting the new inode into the inode hash table will not + * succeed until the old inode is removed, and that only happens after + * ->evict_inode() returns. The new inode is attached to its inode and + * iopen glocks after inserting it into the inode hash table, so at + * that point we can be sure that both glocks are unused. + */ + + ret = gfs2_dinode_dealloc(ip); + if (!ret && ip->i_gl) + gfs2_inode_remember_delete(ip->i_gl, ip->i_no_formal_ino); + +out: + return ret; +} + +/* + * evict_linked_inode - evict an inode whose dinode has not been unlinked + * @inode: The inode to evict + */ +static int evict_linked_inode(struct inode *inode) +{ + struct super_block *sb = inode->i_sb; + struct gfs2_sbd *sdp = sb->s_fs_info; + struct gfs2_inode *ip = GFS2_I(inode); + struct address_space *metamapping; + int ret; -out_truncate: gfs2_log_flush(sdp, ip->i_gl, GFS2_LOG_HEAD_FLUSH_NORMAL | GFS2_LFC_EVICT_INODE); metamapping = gfs2_glock2aspace(ip->i_gl); @@ -1433,65 +1344,124 @@ out_truncate: write_inode_now(inode, 1); gfs2_ail_flush(ip->i_gl, 0); - error = gfs2_trans_begin(sdp, 0, sdp->sd_jdesc->jd_blocks); - if (error) - goto out_unlock; + ret = gfs2_trans_begin(sdp, 0, sdp->sd_jdesc->jd_blocks); + if (ret) + return ret; + /* Needs to be done before glock release & also in a transaction */ truncate_inode_pages(&inode->i_data, 0); truncate_inode_pages(metamapping, 0); gfs2_trans_end(sdp); + return 0; +} + +/** + * gfs2_evict_inode - Remove an inode from cache + * @inode: The inode to evict + * + * There are three cases to consider: + * 1. i_nlink == 0, we are final opener (and must deallocate) + * 2. i_nlink == 0, we are not the final opener (and cannot deallocate) + * 3. i_nlink > 0 + * + * If the fs is read only, then we have to treat all cases as per #3 + * since we are unable to do any deallocation. The inode will be + * deallocated by the next read/write node to attempt an allocation + * in the same resource group + * + * We have to (at the moment) hold the inodes main lock to cover + * the gap between unlocking the shared lock on the iopen lock and + * taking the exclusive lock. I'd rather do a shared -> exclusive + * conversion on the iopen lock, but we can change that later. This + * is safe, just less efficient. + */ + +static void gfs2_evict_inode(struct inode *inode) +{ + struct super_block *sb = inode->i_sb; + struct gfs2_sbd *sdp = sb->s_fs_info; + struct gfs2_inode *ip = GFS2_I(inode); + struct gfs2_holder gh; + enum evict_behavior behavior; + int ret; + + gfs2_holder_mark_uninitialized(&gh); + if (inode->i_nlink || sb_rdonly(sb) || !ip->i_no_addr) + goto out; + + /* + * In case of an incomplete mount, gfs2_evict_inode() may be called for + * system files without having an active journal to write to. In that + * case, skip the filesystem evict. + */ + if (!sdp->sd_jdesc) + goto out; + + behavior = evict_should_delete(inode, &gh); + if (behavior == EVICT_SHOULD_DEFER_DELETE && + !test_bit(SDF_KILL, &sdp->sd_flags)) { + struct gfs2_glock *io_gl = ip->i_iopen_gh.gh_gl; + + if (io_gl) { + gfs2_glock_hold(io_gl); + if (!gfs2_queue_verify_delete(io_gl, true)) + gfs2_glock_put(io_gl); + goto out; + } + behavior = EVICT_SHOULD_SKIP_DELETE; + } + if (behavior == EVICT_SHOULD_DELETE) + ret = evict_unlinked_inode(inode); + else + ret = evict_linked_inode(inode); -out_unlock: if (gfs2_rs_active(&ip->i_res)) gfs2_rs_deltree(&ip->i_res); - if (gfs2_holder_initialized(&gh)) { - glock_clear_object(ip->i_gl, ip); - gfs2_glock_dq_uninit(&gh); - } - if (error && error != GLR_TRYFAILED && error != -EROFS) - fs_warn(sdp, "gfs2_evict_inode: %d\n", error); + if (ret && ret != GLR_TRYFAILED && ret != -EROFS) + fs_warn(sdp, "gfs2_evict_inode: %d\n", ret); out: + if (gfs2_holder_initialized(&gh)) + gfs2_glock_dq_uninit(&gh); truncate_inode_pages_final(&inode->i_data); if (ip->i_qadata) gfs2_assert_warn(sdp, ip->i_qadata->qa_ref == 0); - gfs2_rs_delete(ip, NULL); + gfs2_rs_deltree(&ip->i_res); gfs2_ordered_del_inode(ip); clear_inode(inode); gfs2_dir_hash_inval(ip); - if (ip->i_gl) { - glock_clear_object(ip->i_gl, ip); - wait_on_bit_io(&ip->i_flags, GIF_GLOP_PENDING, TASK_UNINTERRUPTIBLE); - gfs2_glock_add_to_lru(ip->i_gl); - gfs2_glock_put_eventually(ip->i_gl); - ip->i_gl = NULL; - } if (gfs2_holder_initialized(&ip->i_iopen_gh)) { struct gfs2_glock *gl = ip->i_iopen_gh.gh_gl; glock_clear_object(gl, ip); - if (test_bit(HIF_HOLDER, &ip->i_iopen_gh.gh_iflags)) { - ip->i_iopen_gh.gh_flags |= GL_NOCACHE; - gfs2_glock_dq(&ip->i_iopen_gh); - } gfs2_glock_hold(gl); - gfs2_holder_uninit(&ip->i_iopen_gh); + ip->i_iopen_gh.gh_flags |= GL_NOCACHE; + gfs2_glock_dq_uninit(&ip->i_iopen_gh); gfs2_glock_put_eventually(gl); } + if (ip->i_gl) { + glock_clear_object(ip->i_gl, ip); + wait_on_bit_io(&ip->i_flags, GIF_GLOP_PENDING, TASK_UNINTERRUPTIBLE); + gfs2_glock_put_eventually(ip->i_gl); + rcu_assign_pointer(ip->i_gl, NULL); + } } static struct inode *gfs2_alloc_inode(struct super_block *sb) { struct gfs2_inode *ip; - ip = kmem_cache_alloc(gfs2_inode_cachep, GFP_KERNEL); + ip = alloc_inode_sb(sb, gfs2_inode_cachep, GFP_KERNEL); if (!ip) return NULL; + ip->i_no_addr = 0; + ip->i_no_formal_ino = 0; ip->i_flags = 0; ip->i_gl = NULL; gfs2_holder_mark_uninitialized(&ip->i_iopen_gh); memset(&ip->i_res, 0, sizeof(ip->i_res)); RB_CLEAR_NODE(&ip->i_res.rs_node); + ip->i_diskflags = 0; ip->i_rahead = 0; return &ip->i_inode; } @@ -1501,6 +1471,35 @@ static void gfs2_free_inode(struct inode *inode) kmem_cache_free(gfs2_inode_cachep, GFS2_I(inode)); } +void free_local_statfs_inodes(struct gfs2_sbd *sdp) +{ + struct local_statfs_inode *lsi, *safe; + + /* Run through the statfs inodes list to iput and free memory */ + list_for_each_entry_safe(lsi, safe, &sdp->sd_sc_inodes_list, si_list) { + if (lsi->si_jid == sdp->sd_jdesc->jd_jid) + sdp->sd_sc_inode = NULL; /* belongs to this node */ + if (lsi->si_sc_inode) + iput(lsi->si_sc_inode); + list_del(&lsi->si_list); + kfree(lsi); + } +} + +struct inode *find_local_statfs_inode(struct gfs2_sbd *sdp, + unsigned int index) +{ + struct local_statfs_inode *lsi; + + /* Return the local (per node) statfs inode in the + * sdp->sd_sc_inodes_list corresponding to the 'index'. */ + list_for_each_entry(lsi, &sdp->sd_sc_inodes_list, si_list) { + if (lsi->si_jid == index) + return lsi->si_sc_inode; + } + return NULL; +} + const struct super_operations gfs2_super_ops = { .alloc_inode = gfs2_alloc_inode, .free_inode = gfs2_free_inode, @@ -1509,8 +1508,9 @@ const struct super_operations gfs2_super_ops = { .evict_inode = gfs2_evict_inode, .put_super = gfs2_put_super, .sync_fs = gfs2_sync_fs, - .freeze_super = gfs2_freeze, - .thaw_super = gfs2_unfreeze, + .freeze_super = gfs2_freeze_super, + .freeze_fs = gfs2_freeze_fs, + .thaw_super = gfs2_thaw_super, .statfs = gfs2_statfs, .drop_inode = gfs2_drop_inode, .show_options = gfs2_show_options, |