Skip to contentSkip to conversation

GFS2: Pre-pull patch posting (fixes)

PATCH New Steven Whitehouse 1 participant 5 messages May 24, 2013 GFS2 file system

lore
Share & tools
lore
#0 lore raw

This time there are just four fixes. There are a couple of minor updates to the quota code, a fix for KConfig to ensure that only valid combinations including GFS2 can be built, and a fix for a typo affecting end i/o processing when writing the journal.

Also, there is a temporary fix for a performance regression relating to block reservations and directories. A longer fix will be applied in due course, but this deals with the most immediate problem for now,

Steve.

#1 lore raw

PATCH 1/4 GFS2: two minor quota fixups

From: Bob Peterson <[email protected]>

This patch fixes two regression problems that Abhi found in the GFS2 quota code.

fs/gfs2/quota.c +2 -2
index c7c840e..c253b13 100644
@@ -121,7 +121,7 @@ static u64 qd2index(struct gfs2_quota_data *qd)
 {
 	struct kqid qid = qd->qd_id;
 	return (2 * (u64)from_kqid(&init_user_ns, qid)) +
-		(qid.type == USRQUOTA) ? 0 : 1;
+		((qid.type == USRQUOTA) ? 0 : 1);
 }
 
 static u64 qd2offset(struct gfs2_quota_data *qd)
@@ -721,7 +721,7 @@ get_a_page:
 			goto unlock_out;
 	}
 
-	gfs2_trans_add_meta(ip->i_gl, bh);
+	gfs2_trans_add_data(ip->i_gl, bh);
 
 	kaddr = kmap_atomic(page);
 	if (offset + sizeof(struct gfs2_quota) > PAGE_CACHE_SIZE)
Signature · 1 line
1.7.4
#2 lore raw

PATCH 2/4 GFS2: Use single-block reservations for directories

In reply to Steven Whitehouse

From: Bob Peterson <[email protected]>

This patch changes the multi-block allocation code, such that directory inodes only get a single block reserved in the bitmap. That way, the bitmaps are more tightly packed together, and there are fewer spans of free blocks for in-use block reservations. This means it takes less time to find a free span of blocks in the bitmap, which speeds things up. This increases the performance of some workloads by almost 2X. In Nate's mockup.py script (which does
(1) create dir, (2) create dir in dir, (3) create file in that dir) the test executes in 23 steps rather than 43 steps, a 47% performance improvement.

fs/gfs2/rgrp.c +7 -2
index 0c5a575..5232525 100644
@@ -1401,9 +1401,14 @@ static void rg_mblk_search(struct gfs2_rgrpd *rgd, struct gfs2_inode *ip,
 	u32 extlen;
 	u32 free_blocks = rgd->rd_free_clone - rgd->rd_reserved;
 	int ret;
+	struct inode *inode = &ip->i_inode;
 
-	extlen = max_t(u32, atomic_read(&rs->rs_sizehint), requested);
-	extlen = clamp(extlen, RGRP_RSRV_MINBLKS, free_blocks);
+	if (S_ISDIR(inode->i_mode))
+		extlen = 1;
+	else {
+		extlen = max_t(u32, atomic_read(&rs->rs_sizehint), requested);
+		extlen = clamp(extlen, RGRP_RSRV_MINBLKS, free_blocks);
+	}
 	if ((rgd->rd_free_clone < rgd->rd_reserved) || (free_blocks < extlen))
 		return;
 
Signature · 1 line
1.7.4
#3 lore raw

PATCH 3/4 GFS2: fix DLM depends to fix build errors

In reply to Steven Whitehouse

From: Randy Dunlap <[email protected]>

Fix build errors by correcting DLM dependencies in GFS2. Build errors happen when CONFIG_GFS2_FS_LOCKING_DLM=y and CONFIG_DLM=m:

fs/built-in.o: In function `gfs2_lock':
file.c:(.text+0xc7abd): undefined reference to dlm_posix_get' file.c:(.text+0xc7ad0): undefined reference to dlm_posix_unlock' file.c:(.text+0xc7ad9): undefined reference to dlm_posix_lock' fs/built-in.o: In function gdlm_unmount':
lock_dlm.c:(.text+0xd6e5b): undefined reference to dlm_release_lockspace' fs/built-in.o: In function sync_unlock':
lock_dlm.c:(.text+0xd6e9e): undefined reference to dlm_unlock' fs/built-in.o: In function sync_lock':
lock_dlm.c:(.text+0xd6fb6): undefined reference to dlm_lock' fs/built-in.o: In function gdlm_put_lock':
lock_dlm.c:(.text+0xd7238): undefined reference to dlm_unlock' fs/built-in.o: In function gdlm_mount':
lock_dlm.c:(.text+0xd753e): undefined reference to dlm_new_lockspace' lock_dlm.c:(.text+0xd79d3): undefined reference to dlm_release_lockspace' fs/built-in.o: In function `gdlm_lock':
lock_dlm.c:(.text+0xd8179): undefined reference to dlm_lock' fs/built-in.o: In function gdlm_cancel':
lock_dlm.c:(.text+0xd6b22): undefined reference to `dlm_unlock'

fs/gfs2/Kconfig +1 -1
index eb08c9e..5a376ab 100644
@@ -26,7 +26,7 @@ config GFS2_FS
 config GFS2_FS_LOCKING_DLM
 	bool "GFS2 DLM locking"
 	depends on (GFS2_FS!=n) && NET && INET && (IPV6 || IPV6=n) && \
-		HOTPLUG && DLM && CONFIGFS_FS && SYSFS
+		HOTPLUG && CONFIGFS_FS && SYSFS && (DLM=y || DLM=GFS2_FS)
 	help
 	  Multiple node locking module for GFS2
 
Signature · 1 line
1.7.4
#4 lore raw

PATCH 4/4 GFS2: Fix typo in gfs2_log_end_write loop

In reply to Steven Whitehouse

There was a missing _all in this loop iterator

fs/gfs2/lops.c +1 -1
index c5fa758..68b4c8f 100644
@@ -212,7 +212,7 @@ static void gfs2_end_log_write(struct bio *bio, int error)
 		fs_err(sdp, "Error %d writing to log\n", error);
 	}
 
-	bio_for_each_segment(bvec, bio, i) {
+	bio_for_each_segment_all(bvec, bio, i) {
 		page = bvec->bv_page;
 		if (page_has_buffers(page))
 			gfs2_end_log_write_bh(sdp, bvec, error);
Signature · 1 line
1.7.4