ocfs2: Fix possible deadlock when extending quota file

In OCFS2, allocator locks rank above transaction start. Thus we cannot extend quota file from inside a transaction less we could deadlock.

We solve the problem by starting transaction not already in ocfs2_acquire_dquot() but only in ocfs2_local_read_dquot() and ocfs2_global_read_dquot() and we allocate blocks to quota files before starting the transaction. In case we crash, quota files will just have a few blocks more but that's no problem since we just use them next time we extend the quota file.

---
 fs/ocfs2/journal.h      |    5 --
 fs/ocfs2/quota_global.c |  115 +++++++++++++++++++++++------------------------
 2 files changed, 57 insertions(+), 63 deletions(-)

Joel, could you please merge this fix? Thanks.

index 4a4d3b5..2c3222a 100644
@@ -363,11 +363,6 @@ static inline int ocfs2_quota_trans_credits(struct super_block *sb)
 	return credits;
 }
 
-/* Number of credits needed for removing quota structure from file */
-int ocfs2_calc_qdel_credits(struct super_block *sb, int type);
-/* Number of credits needed for initialization of new quota structure */
-int ocfs2_calc_qinit_credits(struct super_block *sb, int type);
-
 /* group extend. inode update and last group update. */
 #define OCFS2_GROUP_EXTEND_CREDITS	(OCFS2_INODE_UPDATE_CREDITS + 1)
 

1.6.0.2

This patch is now in the fixes branch of ocfs2.git.

Joel

Life's Little Instruction Book #43

	"Never give up on somebody.  Miracles happen every day."

Joel Becker
Principal Software Developer
Oracle
E-mail: joel.becker at oracle.com
Phone: (650) 506-8127