gfs2: Wipe jdata and ail1 in gfs2_journal_wipe, formerly gfs2_meta_wipe

Before this patch, when blocks were freed, it called gfs2_meta_wipe to
take the metadata out of the pending journal blocks. It did this mostly
by calling another function called gfs2_remove_from_journal. This is
shortsighted because it does not do anything with jdata blocks which
may also be in the journal.

This patch expands the function so that it wipes out jdata blocks from
the journal as well, and it wipes it from the ail1 list if it hasn't
been written back yet. Since it now processes jdata blocks as well,
the function has been renamed from gfs2_meta_wipe to gfs2_journal_wipe.

New function gfs2_ail1_wipe wants a static view of the ail list, so it
locks the sd_ail_lock when removing items. To accomplish this, function
gfs2_remove_from_journal no longer locks the sd_ail_lock, and it's now
the caller's responsibility to do so.

I was going to make sd_ail_lock locking conditional, but the practice is
generally frowned upon. For details, see: https://lwn.net/Articles/109066/

Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>

6 files changed, 86 insertions, 12 deletions

diff --git a/fs/gfs2/aops.c b/fs/gfs2/aops.c
index 287d160ccf7a..801f244690bf 100644
--- a/fs/gfs2/aops.c
+++ b/fs/gfs2/aops.c
@@ -665,8 +665,11 @@ static void gfs2_discard(struct gfs2_sbd *sdp, struct buffer_head *bh)
 	if (bd) {
 		if (!list_empty(&bd->bd_list) && !buffer_pinned(bh))
 			list_del_init(&bd->bd_list);
-		else
+		else {
+			spin_lock(&sdp->sd_ail_lock);
 			gfs2_remove_from_journal(bh, REMOVE_JDATA);
+			spin_unlock(&sdp->sd_ail_lock);
+		}
 	}
 	bh->b_bdev = NULL;
 	clear_buffer_mapped(bh);
diff --git a/fs/gfs2/log.c b/fs/gfs2/log.c
index 3712059f19f9..cd225ccce8e0 100644
--- a/fs/gfs2/log.c
+++ b/fs/gfs2/log.c
@@ -70,7 +70,7 @@ unsigned int gfs2_struct2blk(struct gfs2_sbd *sdp, unsigned int nstruct)
  *
  */
 
-static void gfs2_remove_from_ail(struct gfs2_bufdata *bd)
+void gfs2_remove_from_ail(struct gfs2_bufdata *bd)
 {
 	bd->bd_tr = NULL;
 	list_del_init(&bd->bd_ail_st_list);
diff --git a/fs/gfs2/log.h b/fs/gfs2/log.h
index 8965c751a303..79f97290146e 100644
--- a/fs/gfs2/log.h
+++ b/fs/gfs2/log.h
@@ -63,7 +63,7 @@ static inline void gfs2_ordered_add_inode(struct gfs2_inode *ip)
 
 extern void gfs2_ordered_del_inode(struct gfs2_inode *ip);
 extern unsigned int gfs2_struct2blk(struct gfs2_sbd *sdp, unsigned int nstruct);
-
+extern void gfs2_remove_from_ail(struct gfs2_bufdata *bd);
 extern void gfs2_log_release(struct gfs2_sbd *sdp, unsigned int blks);
 extern int gfs2_log_reserve(struct gfs2_sbd *sdp, unsigned int blks);
 extern void gfs2_write_log_header(struct gfs2_sbd *sdp, struct gfs2_jdesc *jd,
diff --git a/fs/gfs2/meta_io.c b/fs/gfs2/meta_io.c
index 9856cc2e0795..2db573e31f78 100644
--- a/fs/gfs2/meta_io.c
+++ b/fs/gfs2/meta_io.c
@@ -348,38 +348,109 @@ void gfs2_remove_from_journal(struct buffer_head *bh, int meta)
 		brelse(bh);
 	}
 	if (bd) {
-		spin_lock(&sdp->sd_ail_lock);
 		if (bd->bd_tr) {
 			gfs2_trans_add_revoke(sdp, bd);
 		} else if (was_pinned) {
 			bh->b_private = NULL;
 			kmem_cache_free(gfs2_bufdata_cachep, bd);
+		} else if (!list_empty(&bd->bd_ail_st_list) &&
+					!list_empty(&bd->bd_ail_gl_list)) {
+			gfs2_remove_from_ail(bd);
 		}
-		spin_unlock(&sdp->sd_ail_lock);
 	}
 	clear_buffer_dirty(bh);
 	clear_buffer_uptodate(bh);
 }
 
 /**
- * gfs2_meta_wipe - make inode's buffers so they aren't dirty/pinned anymore
+ * gfs2_ail1_wipe - remove deleted/freed buffers from the ail1 list
+ * @sdp: superblock
+ * @bstart: starting block address of buffers to remove
+ * @blen: length of buffers to be removed
+ *
+ * This function is called from gfs2_journal wipe, whose job is to remove
+ * buffers, corresponding to deleted blocks, from the journal. If we find any
+ * bufdata elements on the system ail1 list, they haven't been written to
+ * the journal yet. So we remove them.
+ */
+static void gfs2_ail1_wipe(struct gfs2_sbd *sdp, u64 bstart, u32 blen)
+{
+	struct gfs2_trans *tr, *s;
+	struct gfs2_bufdata *bd, *bs;
+	struct buffer_head *bh;
+	u64 end = bstart + blen;
+
+	gfs2_log_lock(sdp);
+	spin_lock(&sdp->sd_ail_lock);
+	list_for_each_entry_safe(tr, s, &sdp->sd_ail1_list, tr_list) {
+		list_for_each_entry_safe(bd, bs, &tr->tr_ail1_list,
+					 bd_ail_st_list) {
+			bh = bd->bd_bh;
+			if (bh->b_blocknr < bstart || bh->b_blocknr >= end)
+				continue;
+
+			gfs2_remove_from_journal(bh, REMOVE_JDATA);
+		}
+	}
+	spin_unlock(&sdp->sd_ail_lock);
+	gfs2_log_unlock(sdp);
+}
+
+static struct buffer_head *gfs2_getjdatabuf(struct gfs2_inode *ip, u64 blkno)
+{
+	struct address_space *mapping = ip->i_inode.i_mapping;
+	struct gfs2_sbd *sdp = GFS2_SB(&ip->i_inode);
+	struct page *page;
+	struct buffer_head *bh;
+	unsigned int shift = PAGE_SHIFT - sdp->sd_sb.sb_bsize_shift;
+	unsigned long index = blkno >> shift; /* convert block to page */
+	unsigned int bufnum = blkno - (index << shift);
+
+	page = find_get_page_flags(mapping, index, FGP_LOCK|FGP_ACCESSED);
+	if (!page)
+		return NULL;
+	if (!page_has_buffers(page)) {
+		unlock_page(page);
+		put_page(page);
+		return NULL;
+	}
+	/* Locate header for our buffer within our page */
+	for (bh = page_buffers(page); bufnum--; bh = bh->b_this_page)
+		/* Do nothing */;
+	get_bh(bh);
+	unlock_page(page);
+	put_page(page);
+	return bh;
+}
+
+/**
+ * gfs2_journal_wipe - make inode's buffers so they aren't dirty/pinned anymore
  * @ip: the inode who owns the buffers
  * @bstart: the first buffer in the run
  * @blen: the number of buffers in the run
  *
  */
 
-void gfs2_meta_wipe(struct gfs2_inode *ip, u64 bstart, u32 blen)
+void gfs2_journal_wipe(struct gfs2_inode *ip, u64 bstart, u32 blen)
 {
 	struct gfs2_sbd *sdp = GFS2_SB(&ip->i_inode);
 	struct buffer_head *bh;
+	int ty;
 
+	gfs2_ail1_wipe(sdp, bstart, blen);
 	while (blen) {
+		ty = REMOVE_META;
 		bh = gfs2_getbuf(ip->i_gl, bstart, NO_CREATE);
+		if (!bh && gfs2_is_jdata(ip)) {
+			bh = gfs2_getjdatabuf(ip, bstart);
+			ty = REMOVE_JDATA;
+		}
 		if (bh) {
 			lock_buffer(bh);
 			gfs2_log_lock(sdp);
-			gfs2_remove_from_journal(bh, REMOVE_META);
+			spin_lock(&sdp->sd_ail_lock);
+			gfs2_remove_from_journal(bh, ty);
+			spin_unlock(&sdp->sd_ail_lock);
 			gfs2_log_unlock(sdp);
 			unlock_buffer(bh);
 			brelse(bh);
diff --git a/fs/gfs2/meta_io.h b/fs/gfs2/meta_io.h
index eafb74e861c6..4a8c01929b79 100644
--- a/fs/gfs2/meta_io.h
+++ b/fs/gfs2/meta_io.h
@@ -60,7 +60,7 @@ enum {
 };
 
 extern void gfs2_remove_from_journal(struct buffer_head *bh, int meta);
-extern void gfs2_meta_wipe(struct gfs2_inode *ip, u64 bstart, u32 blen);
+extern void gfs2_journal_wipe(struct gfs2_inode *ip, u64 bstart, u32 blen);
 extern int gfs2_meta_indirect_buffer(struct gfs2_inode *ip, int height, u64 num,
 				     struct buffer_head **bhp);
 
diff --git a/fs/gfs2/rgrp.c b/fs/gfs2/rgrp.c
index 1bba5a9d45fa..9de676cfd3d7 100644
--- a/fs/gfs2/rgrp.c
+++ b/fs/gfs2/rgrp.c
@@ -2442,8 +2442,8 @@ void __gfs2_free_blocks(struct gfs2_inode *ip, struct gfs2_rgrpd *rgd,
 	gfs2_rgrp_out(rgd, rgd->rd_bits[0].bi_bh->b_data);
 
 	/* Directories keep their data in the metadata address space */
-	if (meta || ip->i_depth)
-		gfs2_meta_wipe(ip, bstart, blen);
+	if (meta || ip->i_depth || gfs2_is_jdata(ip))
+		gfs2_journal_wipe(ip, bstart, blen);
 }
 
 /**
@@ -2499,7 +2499,7 @@ void gfs2_free_di(struct gfs2_rgrpd *rgd, struct gfs2_inode *ip)
 	gfs2_statfs_change(sdp, 0, +1, -1);
 	trace_gfs2_block_alloc(ip, rgd, ip->i_no_addr, 1, GFS2_BLKST_FREE);
 	gfs2_quota_change(ip, -1, ip->i_inode.i_uid, ip->i_inode.i_gid);
-	gfs2_meta_wipe(ip, ip->i_no_addr, 1);
+	gfs2_journal_wipe(ip, ip->i_no_addr, 1);
 }
 
 /**