[Cluster-devel] [PATCH] [DLM] Clear othercon pointers when a connection is closed

[Cluster-devel] [PATCH] [DLM] Clear othercon pointers when a connection is closed

[swhiteho]

From: Patrick Caulfield <pcaulfie@redhat.com>

This patch clears the othercon pointer and frees the memory when a connnection
is closed. This could cause a small memory leak when nodes leave the cluster.

Signed-Off-By: Patrick Caulfield <pcaulfie@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>

diff --git a/fs/dlm/lowcomms.c b/fs/dlm/lowcomms.c
index dd36273..d15fd5f 100644
--- a/fs/dlm/lowcomms.c
+++ b/fs/dlm/lowcomms.c
@@ -327,6 +327,8 @@ static void close_connection(struct connection *con, bool and_other)
 	if (con->othercon && and_other) {
 		/* Will only re-enter once. */
 		close_connection(con->othercon, false);
+		kmem_cache_free(con_cache, con->othercon);
+		con->othercon = NULL;
 	}
 	if (con->rx_page) {
 		__free_page(con->rx_page);
-- 
1.5.1.2

[Cluster-devel] [PATCH] [DLM] fix NULL ls usage

[swhiteho]

From: David Teigland <teigland@redhat.com>

Fix regression in recent patch "[DLM] variable allocation" which
attempts to dereference an "ls" struct when it's NULL.

Signed-off-by: David Teigland <teigland@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>

diff --git a/fs/dlm/rcom.c b/fs/dlm/rcom.c
index e3a1527..188b91c 100644
--- a/fs/dlm/rcom.c
+++ b/fs/dlm/rcom.c
@@ -386,8 +386,7 @@ static void receive_rcom_lock_reply(struct dlm_ls *ls, struct dlm_rcom *rc_in)
 	dlm_recover_process_copy(ls, rc_in);
 }
 
-static int send_ls_not_ready(struct dlm_ls *ls, int nodeid,
-			     struct dlm_rcom *rc_in)
+static int send_ls_not_ready(int nodeid, struct dlm_rcom *rc_in)
 {
 	struct dlm_rcom *rc;
 	struct rcom_config *rf;
@@ -395,7 +394,7 @@ static int send_ls_not_ready(struct dlm_ls *ls, int nodeid,
 	char *mb;
 	int mb_len = sizeof(struct dlm_rcom) + sizeof(struct rcom_config);
 
-	mh = dlm_lowcomms_get_buffer(nodeid, mb_len, ls->ls_allocation, &mb);
+	mh = dlm_lowcomms_get_buffer(nodeid, mb_len, GFP_NOFS, &mb);
 	if (!mh)
 		return -ENOBUFS;
 	memset(mb, 0, mb_len);
@@ -465,7 +464,7 @@ void dlm_receive_rcom(struct dlm_header *hd, int nodeid)
 		log_print("lockspace %x from %d type %x not found",
 			  hd->h_lockspace, nodeid, rc->rc_type);
 		if (rc->rc_type == DLM_RCOM_STATUS)
-			send_ls_not_ready(ls, nodeid, rc);
+			send_ls_not_ready(nodeid, rc);
 		return;
 	}
 
-- 
1.5.1.2

[Cluster-devel] [PATCH] [DLM] zero unused parts of sockaddr_storage

[swhiteho]

From: Patrick Caulfield <pcaulfie@redhat.com>

When we build a sockaddr_storage for an IP address, clear the unused parts as
they could be used for node comparisons.

I have seen this occasionally make sctp connections fail.

Signed-Off-By: Patrick Caulfield <pcaulfie@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>

diff --git a/fs/dlm/lowcomms.c b/fs/dlm/lowcomms.c
index d15fd5f..631bc43 100644
--- a/fs/dlm/lowcomms.c
+++ b/fs/dlm/lowcomms.c
@@ -313,6 +313,7 @@ static void make_sockaddr(struct sockaddr_storage *saddr, uint16_t port,
 		in6_addr->sin6_port = cpu_to_be16(port);
 		*addr_len = sizeof(struct sockaddr_in6);
 	}
+	memset((char *)saddr + *addr_len, 0, sizeof(struct sockaddr_storage) - *addr_len);
 }
 
 /* Close a remote connection and tidy up */
-- 
1.5.1.2

[Cluster-devel] [PATCH] [DLM] Fix memory leak in dlm_add_member() when dlm_node_weight() returns less than zero

[swhiteho]

From: Jesper Juhl <jesper.juhl@gmail.com>

There's a memory leak in fs/dlm/member.c::dlm_add_member().

If "dlm_node_weight(ls->ls_name, nodeid)" returns < 0, then
we'll return without freeing the memory allocated to the (at
that point yet unused) 'memb'.
This patch frees the allocated memory in that case and thus
avoids the leak.

Signed-off-by: Jesper Juhl <jesper.juhl@gmail.com>
Signed-off-by: David Teigland <teigland@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>

diff --git a/fs/dlm/member.c b/fs/dlm/member.c
index 073599d..d099775 100644
--- a/fs/dlm/member.c
+++ b/fs/dlm/member.c
@@ -56,8 +56,10 @@ static int dlm_add_member(struct dlm_ls *ls, int nodeid)
 		return -ENOMEM;
 
 	w = dlm_node_weight(ls->ls_name, nodeid);
-	if (w < 0)
+	if (w < 0) {
+		kfree(memb);
 		return w;
+	}
 
 	memb->nodeid = nodeid;
 	memb->weight = w;
-- 
1.5.1.2

[Cluster-devel] [PATCH] [DLM] More othercon fixes

[swhiteho]

From: Patrick Caulfield <pcaulfie@redhat.com>

The last patch to clean out 'othercon' structures only fixed half the problem.
The attached addresses the other situations too, and fixes bz#238490

Signed-Off-By: Patrick Caulfield <pcaulfie@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>

diff --git a/fs/dlm/lowcomms.c b/fs/dlm/lowcomms.c
index 631bc43..9e9d2e8 100644
--- a/fs/dlm/lowcomms.c
+++ b/fs/dlm/lowcomms.c
@@ -328,15 +328,24 @@ static void close_connection(struct connection *con, bool and_other)
 	if (con->othercon && and_other) {
 		/* Will only re-enter once. */
 		close_connection(con->othercon, false);
-		kmem_cache_free(con_cache, con->othercon);
-		con->othercon = NULL;
 	}
 	if (con->rx_page) {
 		__free_page(con->rx_page);
 		con->rx_page = NULL;
 	}
-	con->retries = 0;
-	mutex_unlock(&con->sock_mutex);
+
+	/* If we are an 'othercon' then NULL the pointer to us
+	   from the parent and tidy ourself up */
+	if (test_bit(CF_IS_OTHERCON, &con->flags)) {
+		struct connection *parent = __nodeid2con(con->nodeid, 0);
+		parent->othercon = NULL;
+		kmem_cache_free(con_cache, con);
+	}
+	else {
+		/* Parent connections get reused */
+		con->retries = 0;
+		mutex_unlock(&con->sock_mutex);
+	}
 }
 
 /* We only send shutdown messages to nodes that are not part of the cluster */
@@ -634,7 +643,7 @@ out_resched:
 
 out_close:
 	mutex_unlock(&con->sock_mutex);
-	if (ret != -EAGAIN && !test_bit(CF_IS_OTHERCON, &con->flags)) {
+	if (ret != -EAGAIN) {
 		close_connection(con, false);
 		/* Reconnect when there is something to send */
 	}
@@ -1125,8 +1134,6 @@ static int tcp_listen_for_all(void)
 
 	log_print("Using TCP for communications");
 
-	set_bit(CF_IS_OTHERCON, &con->flags);
-
 	sock = tcp_create_listen_sock(con, dlm_local_addr[0]);
 	if (sock) {
 		add_sock(sock, con);
@@ -1410,7 +1417,7 @@ void dlm_lowcomms_stop(void)
 	for (i = 0; i <= max_nodeid; i++) {
 		con = __nodeid2con(i, 0);
 		if (con) {
-			con->flags |= 0xFF;
+			con->flags |= 0x0F;
 			if (con->sock)
 				con->sock->sk->sk_user_data = NULL;
 		}
@@ -1426,8 +1433,6 @@ void dlm_lowcomms_stop(void)
 		con = __nodeid2con(i, 0);
 		if (con) {
 			close_connection(con, true);
-			if (con->othercon)
-				kmem_cache_free(con_cache, con->othercon);
 			kmem_cache_free(con_cache, con);
 		}
 	}
-- 
1.5.1.2

[Cluster-devel] [PATCH] [DLM] fix basts for granted PR waiting CW

[swhiteho]

From: David Teigland <teigland@redhat.com>

Fix a long standing bug where a blocking callback would be missed
when there's a granted lock in PR mode and waiting locks in both
PR and CW modes (and the PR lock was added to the waiting queue
before the CW lock).  The logic simply compared the numerical values
of the modes to determine if a blocking callback was required, but in
the one case of PR and CW, the lower valued CW mode blocks the higher
valued PR mode.  We just need to add a special check for this PR/CW
case in the tests that decide when a blocking callback is needed.

Signed-off-by: David Teigland <teigland@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>

diff --git a/fs/dlm/lock.c b/fs/dlm/lock.c
index b455919..2082daf 100644
--- a/fs/dlm/lock.c
+++ b/fs/dlm/lock.c
@@ -1670,9 +1670,10 @@ static int can_be_granted(struct dlm_rsb *r, struct dlm_lkb *lkb, int now,
    with a deadlk here, we'd have to generate something like grant_lock with
    the deadlk error.) */
 
-/* returns the highest requested mode of all blocked conversions */
+/* Returns the highest requested mode of all blocked conversions; sets
+   cw if there's a blocked conversion to DLM_LOCK_CW. */
 
-static int grant_pending_convert(struct dlm_rsb *r, int high)
+static int grant_pending_convert(struct dlm_rsb *r, int high, int *cw)
 {
 	struct dlm_lkb *lkb, *s;
 	int hi, demoted, quit, grant_restart, demote_restart;
@@ -1709,6 +1710,9 @@ static int grant_pending_convert(struct dlm_rsb *r, int high)
 		}
 
 		hi = max_t(int, lkb->lkb_rqmode, hi);
+
+		if (cw && lkb->lkb_rqmode == DLM_LOCK_CW)
+			*cw = 1;
 	}
 
 	if (grant_restart)
@@ -1721,29 +1725,52 @@ static int grant_pending_convert(struct dlm_rsb *r, int high)
 	return max_t(int, high, hi);
 }
 
-static int grant_pending_wait(struct dlm_rsb *r, int high)
+static int grant_pending_wait(struct dlm_rsb *r, int high, int *cw)
 {
 	struct dlm_lkb *lkb, *s;
 
 	list_for_each_entry_safe(lkb, s, &r->res_waitqueue, lkb_statequeue) {
 		if (can_be_granted(r, lkb, 0, NULL))
 			grant_lock_pending(r, lkb);
-                else
+                else {
 			high = max_t(int, lkb->lkb_rqmode, high);
+			if (lkb->lkb_rqmode == DLM_LOCK_CW)
+				*cw = 1;
+		}
 	}
 
 	return high;
 }
 
+/* cw of 1 means there's a lock with a rqmode of DLM_LOCK_CW that's blocked
+   on either the convert or waiting queue.
+   high is the largest rqmode of all locks blocked on the convert or
+   waiting queue. */
+
+static int lock_requires_bast(struct dlm_lkb *gr, int high, int cw)
+{
+	if (gr->lkb_grmode == DLM_LOCK_PR && cw) {
+		if (gr->lkb_highbast < DLM_LOCK_EX)
+			return 1;
+		return 0;
+	}
+
+	if (gr->lkb_highbast < high &&
+	    !__dlm_compat_matrix[gr->lkb_grmode+1][high+1])
+		return 1;
+	return 0;
+}
+
 static void grant_pending_locks(struct dlm_rsb *r)
 {
 	struct dlm_lkb *lkb, *s;
 	int high = DLM_LOCK_IV;
+	int cw = 0;
 
 	DLM_ASSERT(is_master(r), dlm_dump_rsb(r););
 
-	high = grant_pending_convert(r, high);
-	high = grant_pending_wait(r, high);
+	high = grant_pending_convert(r, high, &cw);
+	high = grant_pending_wait(r, high, &cw);
 
 	if (high == DLM_LOCK_IV)
 		return;
@@ -1751,27 +1778,41 @@ static void grant_pending_locks(struct dlm_rsb *r)
 	/*
 	 * If there are locks left on the wait/convert queue then send blocking
 	 * ASTs to granted locks based on the largest requested mode (high)
-	 * found above. FIXME: highbast < high comparison not valid for PR/CW.
+	 * found above.
 	 */
 
 	list_for_each_entry_safe(lkb, s, &r->res_grantqueue, lkb_statequeue) {
-		if (lkb->lkb_bastaddr && (lkb->lkb_highbast < high) &&
-		    !__dlm_compat_matrix[lkb->lkb_grmode+1][high+1]) {
-			queue_bast(r, lkb, high);
+		if (lkb->lkb_bastaddr && lock_requires_bast(lkb, high, cw)) {
+			if (cw && high == DLM_LOCK_PR)
+				queue_bast(r, lkb, DLM_LOCK_CW);
+			else
+				queue_bast(r, lkb, high);
 			lkb->lkb_highbast = high;
 		}
 	}
 }
 
+static int modes_require_bast(struct dlm_lkb *gr, struct dlm_lkb *rq)
+{
+	if ((gr->lkb_grmode == DLM_LOCK_PR && rq->lkb_rqmode == DLM_LOCK_CW) ||
+	    (gr->lkb_grmode == DLM_LOCK_CW && rq->lkb_rqmode == DLM_LOCK_PR)) {
+		if (gr->lkb_highbast < DLM_LOCK_EX)
+			return 1;
+		return 0;
+	}
+
+	if (gr->lkb_highbast < rq->lkb_rqmode && !modes_compat(gr, rq))
+		return 1;
+	return 0;
+}
+
 static void send_bast_queue(struct dlm_rsb *r, struct list_head *head,
 			    struct dlm_lkb *lkb)
 {
 	struct dlm_lkb *gr;
 
 	list_for_each_entry(gr, head, lkb_statequeue) {
-		if (gr->lkb_bastaddr &&
-		    gr->lkb_highbast < lkb->lkb_rqmode &&
-		    !modes_compat(gr, lkb)) {
+		if (gr->lkb_bastaddr && modes_require_bast(gr, lkb)) {
 			queue_bast(r, gr, lkb->lkb_rqmode);
 			gr->lkb_highbast = lkb->lkb_rqmode;
 		}
@@ -2235,7 +2276,7 @@ static int do_convert(struct dlm_rsb *r, struct dlm_lkb *lkb)
 	   before we try again to grant this one. */
 
 	if (is_demoted(lkb)) {
-		grant_pending_convert(r, DLM_LOCK_IV);
+		grant_pending_convert(r, DLM_LOCK_IV, NULL);
 		if (_can_be_granted(r, lkb, 1)) {
 			grant_lock(r, lkb);
 			queue_cast(r, lkb, 0);
-- 
1.5.1.2

[Cluster-devel] [PATCH] [GFS2] soft lockup detected in databuf_lo_before_commit

[swhiteho]

From: Bob Peterson <rpeterso@redhat.com>

This is part 2 of the patch for bug #245832, part 1 of which is already
in the git tree.

The problem was that sdp->sd_log_num_databuf was not always being
protected by the gfs2_log_lock spinlock, but the sd_log_le_databuf
(which it is supposed to reflect) was protected.  That meant there
was a timing window during which gfs2_log_flush called
databuf_lo_before_commit and the count didn't match what was
really on the linked list in that window.  So when it ran out of
items on the linked list, it decremented total_dbuf from 0 to -1 and
thus never left the "while(total_dbuf)" loop.

The solution is to protect the variable sdp->sd_log_num_databuf so
that the value will always match the contents of the linked list,
and therefore the number will never go negative, and therefore, the
loop will be exited properly.

Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>

diff --git a/fs/gfs2/lops.c b/fs/gfs2/lops.c
index aff70f0..3b395c4 100644
--- a/fs/gfs2/lops.c
+++ b/fs/gfs2/lops.c
@@ -486,8 +486,8 @@ static void databuf_lo_add(struct gfs2_sbd *sdp, struct gfs2_log_element *le)
 		gfs2_pin(sdp, bd->bd_bh);
 		tr->tr_num_databuf_new++;
 	}
-	sdp->sd_log_num_databuf++;
 	gfs2_log_lock(sdp);
+	sdp->sd_log_num_databuf++;
 	list_add(&le->le_list, &sdp->sd_log_le_databuf);
 	gfs2_log_unlock(sdp);
 }
@@ -523,7 +523,7 @@ static void databuf_lo_before_commit(struct gfs2_sbd *sdp)
 	struct buffer_head *bh = NULL,*bh1 = NULL;
 	struct gfs2_log_descriptor *ld;
 	unsigned int limit;
-	unsigned int total_dbuf = sdp->sd_log_num_databuf;
+	unsigned int total_dbuf;
 	unsigned int total_jdata = sdp->sd_log_num_jdata;
 	unsigned int num, n;
 	__be64 *ptr = NULL;
@@ -535,6 +535,7 @@ static void databuf_lo_before_commit(struct gfs2_sbd *sdp)
 	 * into the log along with a header
 	 */
 	gfs2_log_lock(sdp);
+	total_dbuf = sdp->sd_log_num_databuf;
 	bd2 = bd1 = list_prepare_entry(bd1, &sdp->sd_log_le_databuf,
 				       bd_le.le_list);
 	while(total_dbuf) {
@@ -653,6 +654,7 @@ static void databuf_lo_before_commit(struct gfs2_sbd *sdp)
 				break;
 		}
 		bh = NULL;
+		BUG_ON(total_dbuf < num);
 		total_dbuf -= num;
 		total_jdata -= num;
 	}
-- 
1.5.1.2

[Cluster-devel] [PATCH] [GFS2] soft lockup in rgblk_search

[swhiteho]

From: Bob Peterson <rpeterso@redhat.com>

This patch seems to fix the problem described in bugzilla bug 246114.
It was written by Steve Whitehouse with some tweaking by me.

The code was looping in the relatively new section of code designed to
search for and reuse unlinked inodes.  In cases where it was finding an
appropriate inode to reuse, it was looping around and finding the same
block over and over because a "<=" check should have been a "<" when
comparing the goal block to the last unlinked block found.

Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>

diff --git a/fs/gfs2/rgrp.c b/fs/gfs2/rgrp.c
index e4e0406..bb58e69 100644
--- a/fs/gfs2/rgrp.c
+++ b/fs/gfs2/rgrp.c
@@ -863,16 +863,19 @@ static struct inode *try_rgrp_unlink(struct gfs2_rgrpd *rgd, u64 *last_unlinked)
 	u64 no_addr;
 
 	for(;;) {
+		if (goal >= rgd->rd_data)
+			break;
 		goal = rgblk_search(rgd, goal, GFS2_BLKST_UNLINKED,
 				    GFS2_BLKST_UNLINKED);
 		if (goal == 0)
-			return 0;
+			break;
 		no_addr = goal + rgd->rd_data0;
-		if (no_addr <= *last_unlinked)
+		goal++;
+		if (no_addr < *last_unlinked)
 			continue;
 		*last_unlinked = no_addr;
 		inode = gfs2_inode_lookup(rgd->rd_sbd->sd_vfs, DT_UNKNOWN,
-					no_addr, -1);
+					  no_addr, -1);
 		if (!IS_ERR(inode))
 			return inode;
 	}
-- 
1.5.1.2

[Cluster-devel] [PATCH] [GFS2] Fix incorrect return code in rgrp.c

[swhiteho]

From: Steven Whitehouse <swhiteho@redhat.com>

The following patch fixes a bug where 0 was being used as a return code
to indicate "nothing to do" when in fact 0 was a valid block location
which might be returned by the function.

Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>

diff --git a/fs/gfs2/rgrp.c b/fs/gfs2/rgrp.c
index bb58e69..ce48c45 100644
--- a/fs/gfs2/rgrp.c
+++ b/fs/gfs2/rgrp.c
@@ -867,7 +867,7 @@ static struct inode *try_rgrp_unlink(struct gfs2_rgrpd *rgd, u64 *last_unlinked)
 			break;
 		goal = rgblk_search(rgd, goal, GFS2_BLKST_UNLINKED,
 				    GFS2_BLKST_UNLINKED);
-		if (goal == 0)
+		if (goal == BFITNOENT)
 			break;
 		no_addr = goal + rgd->rd_data0;
 		goal++;
@@ -1316,7 +1316,7 @@ static u32 rgblk_search(struct gfs2_rgrpd *rgd, u32 goal,
 				    bi->bi_len, blk, new_state);
 	}
 
-	return (blk == BFITNOENT) ? 0 : (bi->bi_start * GFS2_NBBY) + blk;
+	return (blk == BFITNOENT) ? blk : (bi->bi_start * GFS2_NBBY) + blk;
 }
 
 /**
@@ -1396,6 +1396,7 @@ u64 gfs2_alloc_data(struct gfs2_inode *ip)
 		goal = rgd->rd_last_alloc_data;
 
 	blk = rgblk_search(rgd, goal, GFS2_BLKST_FREE, GFS2_BLKST_USED);
+	BUG_ON(blk == BFITNOENT);
 	rgd->rd_last_alloc_data = blk;
 
 	block = rgd->rd_data0 + blk;
@@ -1440,6 +1441,7 @@ u64 gfs2_alloc_meta(struct gfs2_inode *ip)
 		goal = rgd->rd_last_alloc_meta;
 
 	blk = rgblk_search(rgd, goal, GFS2_BLKST_FREE, GFS2_BLKST_USED);
+	BUG_ON(blk == BFITNOENT);
 	rgd->rd_last_alloc_meta = blk;
 
 	block = rgd->rd_data0 + blk;
@@ -1481,6 +1483,7 @@ u64 gfs2_alloc_di(struct gfs2_inode *dip, u64 *generation)
 
 	blk = rgblk_search(rgd, rgd->rd_last_alloc_meta,
 			   GFS2_BLKST_FREE, GFS2_BLKST_DINODE);
+	BUG_ON(blk == BFITNOENT);
 
 	rgd->rd_last_alloc_meta = blk;
 
-- 
1.5.1.2

[Cluster-devel] [PATCH] [GFS2] Fix incorrect error path in prepare_write()

[swhiteho]

From: Steven Whitehouse <swhiteho@redhat.com>

The error path in prepare_write() was incorrect in the (very rare) event
that the transaction fails to start. The following prevents a NULL
pointer dereference,

Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>

diff --git a/fs/gfs2/ops_address.c b/fs/gfs2/ops_address.c
index ce90032..42a5f58 100644
--- a/fs/gfs2/ops_address.c
+++ b/fs/gfs2/ops_address.c
@@ -416,7 +416,7 @@ static int gfs2_prepare_write(struct file *file, struct page *page,
 
 	error = gfs2_trans_begin(sdp, rblocks, 0);
 	if (error)
-		goto out;
+		goto out_trans_fail;
 
 	if (gfs2_is_stuffed(ip)) {
 		if (end > sdp->sd_sb.sb_bsize - sizeof(struct gfs2_dinode)) {
@@ -434,6 +434,7 @@ prepare_write:
 out:
 	if (error) {
 		gfs2_trans_end(sdp);
+out_trans_fail:
 		if (alloc_required) {
 			gfs2_inplace_release(ip);
 out_qunlock:
-- 
1.5.1.2

[Cluster-devel] [PATCH] [GFS2] Fix setting of inherit jdata attr

[swhiteho]

From: Steven Whitehouse <swhiteho@redhat.com>

Due to a mix up between the jdata attribute and inherit jdata attribute
it has not been possible to set the inherit jdata attribute on
directories. This is now fixed and the ioctl will report the inherit
jdata attribute for directories rather than the jdata attribute as it
did previously. This stems from our need to have the one bit in the
ioctl attr flags mean two different things according to whether the
underlying inode is a directory or not.

Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>

diff --git a/fs/gfs2/ops_file.c b/fs/gfs2/ops_file.c
index 7734211..94d76ac 100644
--- a/fs/gfs2/ops_file.c
+++ b/fs/gfs2/ops_file.c
@@ -177,8 +177,8 @@ static const u32 fsflags_to_gfs2[32] = {
 	[5] = GFS2_DIF_APPENDONLY,
 	[7] = GFS2_DIF_NOATIME,
 	[12] = GFS2_DIF_EXHASH,
-	[14] = GFS2_DIF_JDATA,
-	[20] = GFS2_DIF_DIRECTIO,
+	[14] = GFS2_DIF_INHERIT_JDATA,
+	[20] = GFS2_DIF_INHERIT_DIRECTIO,
 };
 
 static const u32 gfs2_to_fsflags[32] = {
@@ -187,8 +187,6 @@ static const u32 gfs2_to_fsflags[32] = {
 	[gfs2fl_AppendOnly] = FS_APPEND_FL,
 	[gfs2fl_NoAtime] = FS_NOATIME_FL,
 	[gfs2fl_ExHash] = FS_INDEX_FL,
-	[gfs2fl_Jdata] = FS_JOURNAL_DATA_FL,
-	[gfs2fl_Directio] = FS_DIRECTIO_FL,
 	[gfs2fl_InheritDirectio] = FS_DIRECTIO_FL,
 	[gfs2fl_InheritJdata] = FS_JOURNAL_DATA_FL,
 };
@@ -207,6 +205,12 @@ static int gfs2_get_flags(struct file *filp, u32 __user *ptr)
 		return error;
 
 	fsflags = fsflags_cvt(gfs2_to_fsflags, ip->i_di.di_flags);
+	if (!S_ISDIR(inode->i_mode)) {
+		if (ip->i_di.di_flags & GFS2_DIF_JDATA)
+			fsflags |= FS_JOURNAL_DATA_FL;
+		if (ip->i_di.di_flags & GFS2_DIF_DIRECTIO)
+			fsflags |= FS_DIRECTIO_FL;
+	}
 	if (put_user(fsflags, ptr))
 		error = -EFAULT;
 
@@ -270,13 +274,6 @@ static int do_gfs2_set_flags(struct file *filp, u32 reqflags, u32 mask)
 	if ((new_flags ^ flags) == 0)
 		goto out;
 
-	if (S_ISDIR(inode->i_mode)) {
-		if ((new_flags ^ flags) & GFS2_DIF_JDATA)
-			new_flags ^= (GFS2_DIF_JDATA|GFS2_DIF_INHERIT_JDATA);
-		if ((new_flags ^ flags) & GFS2_DIF_DIRECTIO)
-			new_flags ^= (GFS2_DIF_DIRECTIO|GFS2_DIF_INHERIT_DIRECTIO);
-	}
-
 	error = -EINVAL;
 	if ((new_flags ^ flags) & ~GFS2_FLAGS_USER_SET)
 		goto out;
@@ -315,11 +312,19 @@ out:
 
 static int gfs2_set_flags(struct file *filp, u32 __user *ptr)
 {
+	struct inode *inode = filp->f_path.dentry->d_inode;
 	u32 fsflags, gfsflags;
 	if (get_user(fsflags, ptr))
 		return -EFAULT;
 	gfsflags = fsflags_cvt(fsflags_to_gfs2, fsflags);
-	return do_gfs2_set_flags(filp, gfsflags, ~0);
+	if (!S_ISDIR(inode->i_mode)) {
+		if (gfsflags & GFS2_DIF_INHERIT_JDATA)
+			gfsflags ^= (GFS2_DIF_JDATA | GFS2_DIF_INHERIT_JDATA);
+		if (gfsflags & GFS2_DIF_INHERIT_DIRECTIO)
+			gfsflags ^= (GFS2_DIF_DIRECTIO | GFS2_DIF_INHERIT_DIRECTIO);
+		return do_gfs2_set_flags(filp, gfsflags, ~0);
+	}
+	return do_gfs2_set_flags(filp, gfsflags, ~GFS2_DIF_JDATA);
 }
 
 static long gfs2_ioctl(struct file *filp, unsigned int cmd, unsigned long arg)
-- 
1.5.1.2

[Cluster-devel] [PATCH] [GFS2] Revert remounting w/o acl option leaves acls enabled

[swhiteho]

From: Steven Whitehouse <swhiteho@redhat.com>

This reverts commit 569a7b6c2e8965ff4908003b925757703a3d649c. The
code was correct originally. The default setting for ACLs after a
remount should be to be the same as before the remount.

Signed-off-by: Abhijith Das <adas@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>

diff --git a/fs/gfs2/mount.c b/fs/gfs2/mount.c
index 6f006a8..4864659 100644
--- a/fs/gfs2/mount.c
+++ b/fs/gfs2/mount.c
@@ -82,19 +82,20 @@ int gfs2_mount_args(struct gfs2_sbd *sdp, char *data_arg, int remount)
 	char *options, *o, *v;
 	int error = 0;
 
-	/*  If someone preloaded options, use those instead  */
-	spin_lock(&gfs2_sys_margs_lock);
-	if (!remount && gfs2_sys_margs) {
-		data = gfs2_sys_margs;
-		gfs2_sys_margs = NULL;
-	}
-	spin_unlock(&gfs2_sys_margs_lock);
+	if (!remount) {
+		/*  If someone preloaded options, use those instead  */
+		spin_lock(&gfs2_sys_margs_lock);
+		if (gfs2_sys_margs) {
+			data = gfs2_sys_margs;
+			gfs2_sys_margs = NULL;
+		}
+		spin_unlock(&gfs2_sys_margs_lock);
 
-	/*  Set some defaults  */
-	memset(args, 0, sizeof(struct gfs2_args));
-	args->ar_num_glockd = GFS2_GLOCKD_DEFAULT;
-	args->ar_quota = GFS2_QUOTA_DEFAULT;
-	args->ar_data = GFS2_DATA_DEFAULT;
+		/*  Set some defaults  */
+		args->ar_num_glockd = GFS2_GLOCKD_DEFAULT;
+		args->ar_quota = GFS2_QUOTA_DEFAULT;
+		args->ar_data = GFS2_DATA_DEFAULT;
+	}
 
 	/* Split the options into tokens with the "," character and
 	   process them */
-- 
1.5.1.2

[Cluster-devel] [GFS2/DLM] [0/12] Pre pull request patch posting

[Steven Whitehouse]

Hi,

This set of patches are from the GFS2 -fixes tree. There are further
fixes currently in the -nmw git tree, but these are the ones which have
been there some time and are known to fix important issues, so these are
the ones which I'll send a pull request for shortly,

Steve.

[Cluster-devel] [GFS2/DLM] Pull request

[Steven Whitehouse]

Hi,

Please consider pulling the following bug fixes from the GFS2/DLM -fixes git tree,

Steve.

The following changes since commit 39d3520c92cf7a28c07229ca00cc35a1e8026c77:
  Linus Torvalds (1):
        Linux 2.6.23-rc3

are found in the git repository at:

  git://git.kernel.org/pub/scm/linux/kernel/git/steve/gfs2-2.6-fixes.git

Bob Peterson (2):
      [GFS2] soft lockup detected in databuf_lo_before_commit
      [GFS2] soft lockup in rgblk_search

David Teigland (2):
      [DLM] fix NULL ls usage
      [DLM] fix basts for granted PR waiting CW

Jesper Juhl (1):
      [DLM] Fix memory leak in dlm_add_member() when dlm_node_weight() returns less than zero

Patrick Caulfield (3):
      [DLM] Clear othercon pointers when a connection is closed
      [DLM] zero unused parts of sockaddr_storage
      [DLM] More othercon fixes

Steven Whitehouse (4):
      [GFS2] Fix incorrect return code in rgrp.c
      [GFS2] Fix incorrect error path in prepare_write()
      [GFS2] Fix setting of inherit jdata attr
      [GFS2] Revert remounting w/o acl option leaves acls enabled

 fs/dlm/lock.c         |   69 +++++++++++++++++++++++++++++++++++++++----------
 fs/dlm/lowcomms.c     |   24 +++++++++++-----
 fs/dlm/member.c       |    4 ++-
 fs/dlm/rcom.c         |    7 ++---
 fs/gfs2/lops.c        |    6 +++-
 fs/gfs2/mount.c       |   25 +++++++++--------
 fs/gfs2/ops_address.c |    3 +-
 fs/gfs2/ops_file.c    |   29 ++++++++++++--------
 fs/gfs2/rgrp.c        |   16 ++++++++---
 9 files changed, 124 insertions(+), 59 deletions(-)

[Cluster-devel] [PATCH] dlm: More othercon fixes

[Patrick Caulfield]

The last patch to clean out 'othercon' structures only fixed half the problem.
The attached addresses the other situations too, and fixes bz#238490

Signed-Off-By: Patrick Caulfield <pcaulfie@redhat.com>

Patrick


-------------- next part --------------
A non-text attachment was scrubbed...
Name: othercon2.patch
Type: text/x-patch
Size: 2006 bytes
Desc: not available
URL: <http://listman.redhat.com/archives/cluster-devel/attachments/20070802/963eacf9/attachment.bin>

[Cluster-devel] [PATCH] dlm: More othercon fixes

[Steven Whitehouse]

Hi,

On Thu, 2007-08-02 at 14:58 +0100, Patrick Caulfield wrote:
> The last patch to clean out 'othercon' structures only fixed half the problem.
> The attached addresses the other situations too, and fixes bz#238490
> 
> Signed-Off-By: Patrick Caulfield <pcaulfie@redhat.com>
> 
> Patrick
> 
> 

Now in the -nmw git tree. Thanks,

Steve.