Merge tag 'gfs2-v6.4-rc5-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/gfs2/linux-gfs2

Pull gfs2 updates from Andreas Gruenbacher:

 - Move the freeze/thaw logic from glock callback context to process /
   worker thread context to prevent deadlocks

 - Fix a quota reference couting bug in do_qc()

 - Carry on deallocating inodes even when gfs2_rindex_update() fails

 - Retry filesystem-internal reads when they are interruped by a signal

 - Eliminate kmap_atomic() in favor of kmap_local_page() /
   memcpy_{from,to}_page()

 - Get rid of noop_direct_IO

 - And a few more minor fixes and cleanups

* tag 'gfs2-v6.4-rc5-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/gfs2/linux-gfs2: (23 commits)
  gfs2: Add quota_change type
  gfs2: Use memcpy_{from,to}_page where appropriate
  gfs2: Convert remaining kmap_atomic calls to kmap_local_page
  gfs2: Replace deprecated kmap_atomic with kmap_local_page
  gfs: Get rid of unnucessary locking in inode_go_dump
  gfs2: gfs2_freeze_lock_shared cleanup
  gfs2: Replace sd_freeze_state with SDF_FROZEN flag
  gfs2: Rework freeze / thaw logic
  gfs2: Rename SDF_{FS_FROZEN => FREEZE_INITIATOR}
  gfs2: Reconfiguring frozen filesystem already rejected
  gfs2: Rename gfs2_freeze_lock{ => _shared }
  gfs2: Rename the {freeze,thaw}_super callbacks
  gfs2: Rename remaining "transaction" glock references
  gfs2: retry interrupted internal reads
  gfs2: Fix possible data races in gfs2_show_options()
  gfs2: Fix duplicate should_fault_in_pages() call
  gfs2: set FMODE_CAN_ODIRECT instead of a dummy direct_IO method
  gfs2: Don't remember delete unless it's successful
  gfs2: Update rl_unlinked before releasing rgrp lock
  gfs2: Fix gfs2_qa_get imbalance in gfs2_quota_hold
  ...

Author: torvalds

fs/gfs2/aops.c

@@ -432,10 +432,10 @@ static int stuffed_readpage(struct gfs2_inode *ip, struct page *page)
 	if (error)
 		return error;
 
-	kaddr = kmap_atomic(page);
+	kaddr = kmap_local_page(page);
 	memcpy(kaddr, dibh->b_data + sizeof(struct gfs2_dinode), dsize);
 	memset(kaddr + dsize, 0, PAGE_SIZE - dsize);
-	kunmap_atomic(kaddr);
+	kunmap_local(kaddr);
 	flush_dcache_page(page);
 	brelse(dibh);
 	SetPageUptodate(page);
@@ -489,18 +489,18 @@ int gfs2_internal_read(struct gfs2_inode *ip, char *buf, loff_t *pos,
 	unsigned copied = 0;
 	unsigned amt;
 	struct page *page;
-	void *p;
 
 	do {
+		page = read_cache_page(mapping, index, gfs2_read_folio, NULL);
+		if (IS_ERR(page)) {
+			if (PTR_ERR(page) == -EINTR)
+				continue;
+			return PTR_ERR(page);
+		}
 		amt = size - copied;
 		if (offset + size > PAGE_SIZE)
 			amt = PAGE_SIZE - offset;
-		page = read_cache_page(mapping, index, gfs2_read_folio, NULL);
-		if (IS_ERR(page))
-			return PTR_ERR(page);
-		p = kmap_atomic(page);
-		memcpy(buf + copied, p + offset, amt);
-		kunmap_atomic(p);
+		memcpy_from_page(buf + copied, page, offset, amt);
 		put_page(page);
 		copied += amt;
 		index++;
@@ -751,7 +751,6 @@ static const struct address_space_operations gfs2_aops = {
 	.release_folio = iomap_release_folio,
 	.invalidate_folio = iomap_invalidate_folio,
 	.bmap = gfs2_bmap,
-	.direct_IO = noop_direct_IO,
 	.migrate_folio = filemap_migrate_folio,
 	.is_partially_uptodate = iomap_is_partially_uptodate,
 	.error_remove_page = generic_error_remove_page,

fs/gfs2/bmap.c

@@ -1729,8 +1729,8 @@ static int punch_hole(struct gfs2_inode *ip, u64 offset, u64 length)
 
 	if (offset >= maxsize) {
 		/*
-		 * The starting point lies beyond the allocated meta-data;
-		 * there are no blocks do deallocate.
+		 * The starting point lies beyond the allocated metadata;
+		 * there are no blocks to deallocate.
 		 */
 		return 0;
 	}

fs/gfs2/file.c

@@ -630,6 +630,9 @@ int gfs2_open_common(struct inode *inode, struct file *file)
 		ret = generic_file_open(inode, file);
 		if (ret)
 			return ret;
+
+		if (!gfs2_is_jdata(GFS2_I(inode)))
+			file->f_mode |= FMODE_CAN_ODIRECT;
 	}
 
 	fp = kzalloc(sizeof(struct gfs2_file), GFP_NOFS);
@@ -1030,8 +1033,8 @@ static ssize_t gfs2_file_buffered_write(struct kiocb *iocb,
 	}
 
 	gfs2_holder_init(ip->i_gl, LM_ST_EXCLUSIVE, 0, gh);
-retry:
 	if (should_fault_in_pages(from, iocb, &prev_count, &window_size)) {
+retry:
 		window_size -= fault_in_iov_iter_readable(from, window_size);
 		if (!window_size) {
 			ret = -EFAULT;

fs/gfs2/glock.c

@@ -145,8 +145,8 @@ static void gfs2_glock_dealloc(struct rcu_head *rcu)
  *
  * We need to allow some glocks to be enqueued, dequeued, promoted, and demoted
  * when we're withdrawn. For example, to maintain metadata integrity, we should
- * disallow the use of inode and rgrp glocks when withdrawn. Other glocks, like
- * iopen or the transaction glocks may be safely used because none of their
+ * disallow the use of inode and rgrp glocks when withdrawn. Other glocks like
+ * the iopen or freeze glock may be safely used because none of their
  * metadata goes through the journal. So in general, we should disallow all
  * glocks that are journaled, and allow all the others. One exception is:
  * we need to allow our active journal to be promoted and demoted so others

fs/gfs2/glops.c

@@ -236,7 +236,7 @@ static void rgrp_go_inval(struct gfs2_glock *gl, int flags)
 	truncate_inode_pages_range(mapping, start, end);
 }
 
-static void gfs2_rgrp_go_dump(struct seq_file *seq, struct gfs2_glock *gl,
+static void gfs2_rgrp_go_dump(struct seq_file *seq, const struct gfs2_glock *gl,
 			      const char *fs_id_buf)
 {
 	struct gfs2_rgrpd *rgd = gl->gl_object;
@@ -536,72 +536,53 @@ static int inode_go_held(struct gfs2_holder *gh)
  *
  */
 
-static void inode_go_dump(struct seq_file *seq, struct gfs2_glock *gl,
+static void inode_go_dump(struct seq_file *seq, const struct gfs2_glock *gl,
 			  const char *fs_id_buf)
 {
 	struct gfs2_inode *ip = gl->gl_object;
-	struct inode *inode;
-	unsigned long nrpages;
+	const struct inode *inode = &ip->i_inode;
 
 	if (ip == NULL)
 		return;
 
-	inode = &ip->i_inode;
-	xa_lock_irq(&inode->i_data.i_pages);
-	nrpages = inode->i_data.nrpages;
-	xa_unlock_irq(&inode->i_data.i_pages);
-
 	gfs2_print_dbg(seq, "%s I: n:%llu/%llu t:%u f:0x%02lx d:0x%08x s:%llu "
 		       "p:%lu\n", fs_id_buf,
 		  (unsigned long long)ip->i_no_formal_ino,
 		  (unsigned long long)ip->i_no_addr,
-		  IF2DT(ip->i_inode.i_mode), ip->i_flags,
+		  IF2DT(inode->i_mode), ip->i_flags,
 		  (unsigned int)ip->i_diskflags,
-		  (unsigned long long)i_size_read(inode), nrpages);
+		  (unsigned long long)i_size_read(inode),
+		  inode->i_data.nrpages);
 }
 
 /**
- * freeze_go_sync - promote/demote the freeze glock
+ * freeze_go_callback - A cluster node is requesting a freeze
  * @gl: the glock
+ * @remote: true if this came from a different cluster node
  */
 
-static int freeze_go_sync(struct gfs2_glock *gl)
+static void freeze_go_callback(struct gfs2_glock *gl, bool remote)
 {
-	int error = 0;
 	struct gfs2_sbd *sdp = gl->gl_name.ln_sbd;
+	struct super_block *sb = sdp->sd_vfs;
+
+	if (!remote ||
+	    gl->gl_state != LM_ST_SHARED ||
+	    gl->gl_demote_state != LM_ST_UNLOCKED)
+		return;
 
 	/*
-	 * We need to check gl_state == LM_ST_SHARED here and not gl_req ==
-	 * LM_ST_EXCLUSIVE. That's because when any node does a freeze,
-	 * all the nodes should have the freeze glock in SH mode and they all
-	 * call do_xmote: One for EX and the others for UN. They ALL must
-	 * freeze locally, and they ALL must queue freeze work. The freeze_work
-	 * calls freeze_func, which tries to reacquire the freeze glock in SH,
-	 * effectively waiting for the thaw on the node who holds it in EX.
-	 * Once thawed, the work func acquires the freeze glock in
-	 * SH and everybody goes back to thawed.
+	 * Try to get an active super block reference to prevent racing with
+	 * unmount (see trylock_super()).  But note that unmount isn't the only
+	 * place where a write lock on s_umount is taken, and we can fail here
+	 * because of things like remount as well.
 	 */
-	if (gl->gl_state == LM_ST_SHARED && !gfs2_withdrawn(sdp) &&
-	    !test_bit(SDF_NORECOVERY, &sdp->sd_flags)) {
-		atomic_set(&sdp->sd_freeze_state, SFS_STARTING_FREEZE);
-		error = freeze_super(sdp->sd_vfs);
-		if (error) {
-			fs_info(sdp, "GFS2: couldn't freeze filesystem: %d\n",
-				error);
-			if (gfs2_withdrawn(sdp)) {
-				atomic_set(&sdp->sd_freeze_state, SFS_UNFROZEN);
-				return 0;
-			}
-			gfs2_assert_withdraw(sdp, 0);
-		}
-		queue_work(gfs2_freeze_wq, &sdp->sd_freeze_work);
-		if (test_bit(SDF_JOURNAL_LIVE, &sdp->sd_flags))
-			gfs2_log_flush(sdp, NULL, GFS2_LOG_HEAD_FLUSH_FREEZE |
-				       GFS2_LFC_FREEZE_GO_SYNC);
-		else /* read-only mounts */
-			atomic_set(&sdp->sd_freeze_state, SFS_FROZEN);
+	if (down_read_trylock(&sb->s_umount)) {
+		atomic_inc(&sb->s_active);
+		up_read(&sb->s_umount);
+		if (!queue_work(gfs2_freeze_wq, &sdp->sd_freeze_work))
+			deactivate_super(sb);
 	}
-	return 0;
 }
 
 /**
@@ -761,9 +742,9 @@ const struct gfs2_glock_operations gfs2_rgrp_glops = {
 };
 
 const struct gfs2_glock_operations gfs2_freeze_glops = {
-	.go_sync = freeze_go_sync,
 	.go_xmote_bh = freeze_go_xmote_bh,
 	.go_demote_ok = freeze_go_demote_ok,
+	.go_callback = freeze_go_callback,
 	.go_type = LM_TYPE_NONDISK,
 	.go_flags = GLOF_NONDISK,
 };

fs/gfs2/incore.h

@@ -221,7 +221,7 @@ struct gfs2_glock_operations {
 	int (*go_demote_ok) (const struct gfs2_glock *gl);
 	int (*go_instantiate) (struct gfs2_glock *gl);
 	int (*go_held)(struct gfs2_holder *gh);
-	void (*go_dump)(struct seq_file *seq, struct gfs2_glock *gl,
+	void (*go_dump)(struct seq_file *seq, const struct gfs2_glock *gl,
 			const char *fs_id_buf);
 	void (*go_callback)(struct gfs2_glock *gl, bool remote);
 	void (*go_free)(struct gfs2_glock *gl);
@@ -600,20 +600,15 @@ enum {
 	SDF_RORECOVERY		= 7, /* read only recovery */
 	SDF_SKIP_DLM_UNLOCK	= 8,
 	SDF_FORCE_AIL_FLUSH     = 9,
-	SDF_FS_FROZEN           = 10,
+	SDF_FREEZE_INITIATOR	= 10,
 	SDF_WITHDRAWING		= 11, /* Will withdraw eventually */
 	SDF_WITHDRAW_IN_PROG	= 12, /* Withdraw is in progress */
 	SDF_REMOTE_WITHDRAW	= 13, /* Performing remote recovery */
 	SDF_WITHDRAW_RECOVERY	= 14, /* Wait for journal recovery when we are
 					 withdrawing */
 	SDF_DEACTIVATING	= 15,
 	SDF_EVICTING		= 16,
-};
-
-enum gfs2_freeze_state {
-	SFS_UNFROZEN		= 0,
-	SFS_STARTING_FREEZE	= 1,
-	SFS_FROZEN		= 2,
+	SDF_FROZEN		= 17,
 };
 
 #define GFS2_FSNAME_LEN		256
@@ -841,7 +836,6 @@ struct gfs2_sbd {
 
 	/* For quiescing the filesystem */
 	struct gfs2_holder sd_freeze_gh;
-	atomic_t sd_freeze_state;
 	struct mutex sd_freeze_mutex;
 
 	char sd_fsname[GFS2_FSNAME_LEN + 3 * sizeof(int) + 2];

fs/gfs2/lock_dlm.c

@@ -296,28 +296,22 @@ static void gdlm_put_lock(struct gfs2_glock *gl)
 	struct lm_lockstruct *ls = &sdp->sd_lockstruct;
 	int error;
 
-	if (gl->gl_lksb.sb_lkid == 0) {
-		gfs2_glock_free(gl);
-		return;
-	}
+	if (gl->gl_lksb.sb_lkid == 0)
+		goto out_free;
 
 	clear_bit(GLF_BLOCKING, &gl->gl_flags);
 	gfs2_glstats_inc(gl, GFS2_LKS_DCOUNT);
 	gfs2_sbstats_inc(gl, GFS2_LKS_DCOUNT);
 	gfs2_update_request_times(gl);
 
 	/* don't want to call dlm if we've unmounted the lock protocol */
-	if (test_bit(DFL_UNMOUNT, &ls->ls_recover_flags)) {
-		gfs2_glock_free(gl);
-		return;
-	}
+	if (test_bit(DFL_UNMOUNT, &ls->ls_recover_flags))
+		goto out_free;
 	/* don't want to skip dlm_unlock writing the lvb when lock has one */
 
 	if (test_bit(SDF_SKIP_DLM_UNLOCK, &sdp->sd_flags) &&
-	    !gl->gl_lksb.sb_lvbptr) {
-		gfs2_glock_free(gl);
-		return;
-	}
+	    !gl->gl_lksb.sb_lvbptr)
+		goto out_free;
 
 again:
 	error = dlm_unlock(ls->ls_dlm, gl->gl_lksb.sb_lkid, DLM_LKF_VALBLK,
@@ -331,8 +325,11 @@ static void gdlm_put_lock(struct gfs2_glock *gl)
 		fs_err(sdp, "gdlm_unlock %x,%llx err=%d\n",
 		       gl->gl_name.ln_type,
 		       (unsigned long long)gl->gl_name.ln_number, error);
-		return;
 	}
+	return;
+
+out_free:
+	gfs2_glock_free(gl);
 }
 
 static void gdlm_cancel(struct gfs2_glock *gl)

fs/gfs2/log.c

@@ -914,9 +914,8 @@ void gfs2_write_log_header(struct gfs2_sbd *sdp, struct gfs2_jdesc *jd,
 static void log_write_header(struct gfs2_sbd *sdp, u32 flags)
 {
 	blk_opf_t op_flags = REQ_PREFLUSH | REQ_FUA | REQ_META | REQ_SYNC;
-	enum gfs2_freeze_state state = atomic_read(&sdp->sd_freeze_state);
 
-	gfs2_assert_withdraw(sdp, (state != SFS_FROZEN));
+	gfs2_assert_withdraw(sdp, !test_bit(SDF_FROZEN, &sdp->sd_flags));
 
 	if (test_bit(SDF_NOBARRIERS, &sdp->sd_flags)) {
 		gfs2_ordered_wait(sdp);
@@ -1036,7 +1035,7 @@ void gfs2_log_flush(struct gfs2_sbd *sdp, struct gfs2_glock *gl, u32 flags)
 {
 	struct gfs2_trans *tr = NULL;
 	unsigned int reserved_blocks = 0, used_blocks = 0;
-	enum gfs2_freeze_state state = atomic_read(&sdp->sd_freeze_state);
+	bool frozen = test_bit(SDF_FROZEN, &sdp->sd_flags);
 	unsigned int first_log_head;
 	unsigned int reserved_revokes = 0;
 
@@ -1067,7 +1066,7 @@ void gfs2_log_flush(struct gfs2_sbd *sdp, struct gfs2_glock *gl, u32 flags)
 		if (tr) {
 			sdp->sd_log_tr = NULL;
 			tr->tr_first = first_log_head;
-			if (unlikely (state == SFS_FROZEN)) {
+			if (unlikely(frozen)) {
 				if (gfs2_assert_withdraw_delayed(sdp,
 				       !tr->tr_num_buf_new && !tr->tr_num_databuf_new))
 					goto out_withdraw;
@@ -1092,7 +1091,7 @@ void gfs2_log_flush(struct gfs2_sbd *sdp, struct gfs2_glock *gl, u32 flags)
 	if (flags & GFS2_LOG_HEAD_FLUSH_SHUTDOWN)
 		clear_bit(SDF_JOURNAL_LIVE, &sdp->sd_flags);
 
-	if (unlikely(state == SFS_FROZEN))
+	if (unlikely(frozen))
 		if (gfs2_assert_withdraw_delayed(sdp, !reserved_revokes))
 			goto out_withdraw;
 
@@ -1136,8 +1135,6 @@ void gfs2_log_flush(struct gfs2_sbd *sdp, struct gfs2_glock *gl, u32 flags)
 		if (flags & (GFS2_LOG_HEAD_FLUSH_SHUTDOWN |
 			     GFS2_LOG_HEAD_FLUSH_FREEZE))
 			gfs2_log_shutdown(sdp);
-		if (flags & GFS2_LOG_HEAD_FLUSH_FREEZE)
-			atomic_set(&sdp->sd_freeze_state, SFS_FROZEN);
 	}
 
 out_end:

fs/gfs2/lops.c

@@ -427,10 +427,11 @@ static bool gfs2_jhead_pg_srch(struct gfs2_jdesc *jd,
 {
 	struct gfs2_sbd *sdp = GFS2_SB(jd->jd_inode);
 	struct gfs2_log_header_host lh;
-	void *kaddr = kmap_atomic(page);
+	void *kaddr;
 	unsigned int offset;
 	bool ret = false;
 
+	kaddr = kmap_local_page(page);
 	for (offset = 0; offset < PAGE_SIZE; offset += sdp->sd_sb.sb_bsize) {
 		if (!__get_log_header(sdp, kaddr + offset, 0, &lh)) {
 			if (lh.lh_sequence >= head->lh_sequence)
@@ -441,7 +442,7 @@ static bool gfs2_jhead_pg_srch(struct gfs2_jdesc *jd,
 			}
 		}
 	}
-	kunmap_atomic(kaddr);
+	kunmap_local(kaddr);
 	return ret;
 }
 
@@ -626,11 +627,11 @@ static void gfs2_check_magic(struct buffer_head *bh)
 	__be32 *ptr;
 
 	clear_buffer_escaped(bh);
-	kaddr = kmap_atomic(bh->b_page);
+	kaddr = kmap_local_page(bh->b_page);
 	ptr = kaddr + bh_offset(bh);
 	if (*ptr == cpu_to_be32(GFS2_MAGIC))
 		set_buffer_escaped(bh);
-	kunmap_atomic(kaddr);
+	kunmap_local(kaddr);
 }
 
 static int blocknr_cmp(void *priv, const struct list_head *a,
@@ -696,14 +697,12 @@ static void gfs2_before_commit(struct gfs2_sbd *sdp, unsigned int limit,
 			lock_buffer(bd2->bd_bh);
 
 			if (buffer_escaped(bd2->bd_bh)) {
-				void *kaddr;
+				void *p;
+
 				page = mempool_alloc(gfs2_page_pool, GFP_NOIO);
-				ptr = page_address(page);
-				kaddr = kmap_atomic(bd2->bd_bh->b_page);
-				memcpy(ptr, kaddr + bh_offset(bd2->bd_bh),
-				       bd2->bd_bh->b_size);
-				kunmap_atomic(kaddr);
-				*(__be32 *)ptr = 0;
+				p = page_address(page);
+				memcpy_from_page(p, page, bh_offset(bd2->bd_bh), bd2->bd_bh->b_size);
+				*(__be32 *)p = 0;
 				clear_buffer_escaped(bd2->bd_bh);
 				unlock_buffer(bd2->bd_bh);
 				brelse(bd2->bd_bh);