[PATCH 02/12] cgroup: introduce effective cgroup_subsys_state

From: Tejun Heo
Date: Thu Mar 27 2014 - 22:41:28 EST


In the planned default unified hierarchy, controllers may get
dynamically attached to and detached from a cgroup and a cgroup may
not have csses for all the controllers associated with the hierarchy.

When a cgroup doesn't have its own css for a given controller, the css
of the nearest ancestor with the controller enabled will be used,
which is called the effective css. This patch introduces
cgroup_e_css() and for_each_e_css() to access the effective csses and
convert compare_css_sets(), find_existing_css_set() and
cgroup_migrate() to use the effective csses so that they can handle
cgroups with partial csses correctly.

This means that for two css_sets to be considered identical, they
should have both matching csses and cgroups. compare_css_sets()
already compares both, not for correctness but for optimization. As
this now becomes a matter of correctness, update the comments
accordingly.

For all !default hierarchies, cgroup_e_css() always equals
cgroup_css(), so this patch doesn't change behavior.

While at it, fix incorrect locking comment for for_each_css().

Signed-off-by: Tejun Heo <tj@xxxxxxxxxx>
---
kernel/cgroup.c | 83 ++++++++++++++++++++++++++++++++++++++++++++-------------
1 file changed, 64 insertions(+), 19 deletions(-)

diff --git a/kernel/cgroup.c b/kernel/cgroup.c
index 3c9c623..ae1fe5a 100644
--- a/kernel/cgroup.c
+++ b/kernel/cgroup.c
@@ -208,6 +208,34 @@ static struct cgroup_subsys_state *cgroup_css(struct cgroup *cgrp,
return &cgrp->dummy_css;
}

+/**
+ * cgroup_e_css - obtain a cgroup's effective css for the specified subsystem
+ * @cgrp: the cgroup of interest
+ * @ss: the subsystem of interest (%NULL returns the dummy_css)
+ *
+ * Similar to cgroup_css() but returns the effctive css, which is defined
+ * as the matching css of the nearest ancestor including self which has @ss
+ * enabled. If @ss is associated with the hierarchy @cgrp is on, this
+ * function is guaranteed to return non-NULL css.
+ */
+static struct cgroup_subsys_state *cgroup_e_css(struct cgroup *cgrp,
+ struct cgroup_subsys *ss)
+{
+ lockdep_assert_held(&cgroup_mutex);
+
+ if (!ss)
+ return &cgrp->dummy_css;
+
+ if (!(cgrp->root->subsys_mask & (1 << ss->id)))
+ return NULL;
+
+ while (cgrp->parent &&
+ !(cgrp->parent->child_subsys_mask & (1 << ss->id)))
+ cgrp = cgrp->parent;
+
+ return cgroup_css(cgrp, ss);
+}
+
/* convenient tests for these bits */
static inline bool cgroup_is_dead(const struct cgroup *cgrp)
{
@@ -273,7 +301,7 @@ static int notify_on_release(const struct cgroup *cgrp)
* @ssid: the index of the subsystem, CGROUP_SUBSYS_COUNT after reaching the end
* @cgrp: the target cgroup to iterate css's of
*
- * Should be called under cgroup_mutex.
+ * Should be called under cgroup_[tree_]mutex.
*/
#define for_each_css(css, ssid, cgrp) \
for ((ssid) = 0; (ssid) < CGROUP_SUBSYS_COUNT; (ssid)++) \
@@ -284,6 +312,20 @@ static int notify_on_release(const struct cgroup *cgrp)
else

/**
+ * for_each_e_css - iterate all effective css's of a cgroup
+ * @css: the iteration cursor
+ * @ssid: the index of the subsystem, CGROUP_SUBSYS_COUNT after reaching the end
+ * @cgrp: the target cgroup to iterate css's of
+ *
+ * Should be called under cgroup_[tree_]mutex.
+ */
+#define for_each_e_css(css, ssid, cgrp) \
+ for ((ssid) = 0; (ssid) < CGROUP_SUBSYS_COUNT; (ssid)++) \
+ if (!((css) = cgroup_e_css(cgrp, cgroup_subsys[(ssid)]))) \
+ ; \
+ else
+
+/**
* for_each_subsys - iterate all enabled cgroup subsystems
* @ss: the iteration cursor
* @ssid: the index of @ss, CGROUP_SUBSYS_COUNT after reaching the end
@@ -452,20 +494,20 @@ static bool compare_css_sets(struct css_set *cset,
{
struct list_head *l1, *l2;

- if (memcmp(template, cset->subsys, sizeof(cset->subsys))) {
- /* Not all subsystems matched */
+ /*
+ * On the default hierarchy, there can be csets which are
+ * associated with the same set of cgroups but different csses.
+ * Let's first ensure that csses match.
+ */
+ if (memcmp(template, cset->subsys, sizeof(cset->subsys)))
return false;
- }

/*
* Compare cgroup pointers in order to distinguish between
- * different cgroups in heirarchies with no subsystems. We
- * could get by with just this check alone (and skip the
- * memcmp above) but on most setups the memcmp check will
- * avoid the need for this more expensive check on almost all
- * candidates.
+ * different cgroups in hierarchies. As different cgroups may
+ * share the same effective css, this comparison is always
+ * necessary.
*/
-
l1 = &cset->cgrp_links;
l2 = &old_cset->cgrp_links;
while (1) {
@@ -530,13 +572,16 @@ static struct css_set *find_existing_css_set(struct css_set *old_cset,
*/
for_each_subsys(ss, i) {
if (root->subsys_mask & (1UL << i)) {
- /* Subsystem is in this hierarchy. So we want
- * the subsystem state from the new
- * cgroup */
- template[i] = cgroup_css(cgrp, ss);
+ /*
+ * @ss is in this hierarchy, so we want the
+ * effective css from @cgrp.
+ */
+ template[i] = cgroup_e_css(cgrp, ss);
} else {
- /* Subsystem is not in this hierarchy, so we
- * don't want to change the subsystem state */
+ /*
+ * @ss is not in this hierarchy, so we don't want
+ * to change the css.
+ */
template[i] = old_cset->subsys[i];
}
}
@@ -1968,7 +2013,7 @@ static int cgroup_migrate(struct cgroup *cgrp, struct task_struct *leader,
return 0;

/* check that we can legitimately attach to the cgroup */
- for_each_css(css, i, cgrp) {
+ for_each_e_css(css, i, cgrp) {
if (css->ss->can_attach) {
ret = css->ss->can_attach(css, &tset);
if (ret) {
@@ -1998,7 +2043,7 @@ static int cgroup_migrate(struct cgroup *cgrp, struct task_struct *leader,
*/
tset.csets = &tset.dst_csets;

- for_each_css(css, i, cgrp)
+ for_each_e_css(css, i, cgrp)
if (css->ss->attach)
css->ss->attach(css, &tset);

@@ -2006,7 +2051,7 @@ static int cgroup_migrate(struct cgroup *cgrp, struct task_struct *leader,
goto out_release_tset;

out_cancel_attach:
- for_each_css(css, i, cgrp) {
+ for_each_e_css(css, i, cgrp) {
if (css == failed_css)
break;
if (css->ss->cancel_attach)
--
1.8.5.3

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/