xfs_quota -x -c "project -d DEPTH" stops in the middle of its operation
if a directory whose level is more than DEPTH was found.
To set/check/clear a project quota, xfs_quota performs pre-order tree traversal
by using nftw(3). However, if the directory whose level is more than
the recursion level specified by -d option is found, the function called
by nftw() returns -1 and nftw() stops the tree walk.
The following steps show that only one directory is set a project quota,
even though another directory whose level is the same as the recursion level
isn't set a project quota.
# mkfs.xfs -f /dev/sdb1
# mount -o pquota /dev/sdb1 /mnt
# mkdir -p /mnt/pquota/dirA/dirAA /mnt/pquota/dirB/dirBB
# xfs_quota -x -c "project -d 1 -s -p /mnt/pquota 100"
# xfs_io -c stat /mnt/pquota/dirA | grep projid
fsxattr.projid = 100
# xfs_io -c stat /mnt/pquota/dirB | grep projid
fsxattr.projid = 0
To fix above problem, if the directory whose level is more than the recursion
level is found, the function called by nftw() returns 0 instead of -1.
Note that the time of tree traversal is the same as no limitation of
recursion level. To reduce the time, we have to use FTW_SKIP_SUBTREE
that is glibc-specified flag. Is it better to use this flag?
Signed-off-by: Kazuya Mio <k-mio@xxxxxxxxxxxxx>
---
quota/project.c | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
diff --git a/quota/project.c b/quota/project.c
index e4e7a01..20bc01b 100644
--- a/quota/project.c
+++ b/quota/project.c
@@ -101,7 +101,7 @@ check_project(
int fd;
if (recurse_depth >= 0 && data->level > recurse_depth)
- return -1;
+ return 0;
if (flag == FTW_NS ){
exitcode = 1;
@@ -146,7 +146,7 @@ clear_project(
int fd;
if (recurse_depth >= 0 && data->level > recurse_depth)
- return -1;
+ return 0;
if (flag == FTW_NS ){
exitcode = 1;
@@ -193,7 +193,7 @@ setup_project(
int fd;
if (recurse_depth >= 0 && data->level > recurse_depth)
- return -1;
+ return 0;
if (flag == FTW_NS ){
exitcode = 1;
--
To unsubscribe from this list: send the line "unsubscribe linux-xfs" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
