cgroup: make cgroup_path() and friends behave in the style of strlcpy()

cgroup_path() and friends used to format the path from the end and
thus the resulting path usually didn't start at the start of the
passed in buffer.  Also, when the buffer was too small, the partial
result was truncated from the head rather than tail and there was no
way to tell how long the full path would be.  These make the functions
less robust and more awkward to use.

With recent updates to kernfs_path(), cgroup_path() and friends can be
made to behave in strlcpy() style.

* cgroup_path(), cgroup_path_ns[_locked]() and task_cgroup_path() now
  always return the length of the full path.  If buffer is too small,
  it contains nul terminated truncated output.

* All users updated accordingly.

v2: cgroup_path() usage in kernel/sched/debug.c converted.

Signed-off-by: Tejun Heo <tj@kernel.org>
Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Serge Hallyn <serge.hallyn@ubuntu.com>
Cc: Peter Zijlstra <peterz@infradead.org>

1 files changed, 1 insertions, 10 deletions

diff --git a/include/linux/blk-cgroup.h b/include/linux/blk-cgroup.h
index 10648e300c93..4e8c215e185c 100644
--- a/include/linux/blk-cgroup.h
+++ b/include/linux/blk-cgroup.h
@@ -343,16 +343,7 @@ static inline struct blkcg *cpd_to_blkcg(struct blkcg_policy_data *cpd)
  */
 static inline int blkg_path(struct blkcg_gq *blkg, char *buf, int buflen)
 {
-	char *p;
-
-	p = cgroup_path(blkg->blkcg->css.cgroup, buf, buflen);
-	if (!p) {
-		strncpy(buf, "<unavailable>", buflen);
-		return -ENAMETOOLONG;
-	}
-
-	memmove(buf, p, buf + buflen - p);
-	return 0;
+	return cgroup_path(blkg->blkcg->css.cgroup, buf, buflen);
 }
 
 /**