Re: [PATCH 15/23] cgroup: make hierarchy iterators deal withcgroup_subsys_state instead of cgroup
From: Michal Hocko
Date: Fri Aug 02 2013 - 09:32:13 EST
On Thu 01-08-13 17:49:53, Tejun Heo wrote:
> cgroup is currently in the process of transitioning to using css
> (cgroup_subsys_state) as the primary handle instead of cgroup in
> subsystem API. For hierarchy iterators, this is beneficial because
>
> * In most cases, css is the only thing subsystems care about anyway.
>
> * On the planned unified hierarchy, iterations for different
> subsystems will need to skip over different subtrees of the
> hierarchy depending on which subsystems are enabled on each cgroup.
> Passing around css makes it unnecessary to explicitly specify the
> subsystem in question as css is intersection between cgroup and
> subsystem
>
> * For the planned unified hierarchy, css's would need to be created
> and destroyed dynamically independent from cgroup hierarchy. Having
> cgroup core manage css iteration makes enforcing deref rules a lot
> easier.
>
> Most subsystem conversions are straight-forward. Noteworthy changes
> are
>
> * blkio: cgroup_to_blkcg() is no longer used. Removed.
>
> * freezer: cgroup_freezer() is no longer used. Removed.
>
> * devices: cgroup_to_devcgroup() is no longer used. Removed.
>
> Signed-off-by: Tejun Heo <tj@xxxxxxxxxx>
> Cc: Li Zefan <lizefan@xxxxxxxxxx>
> Cc: Johannes Weiner <hannes@xxxxxxxxxxx>
> Cc: Michal Hocko <mhocko@xxxxxxx>
> Cc: Balbir Singh <bsingharora@xxxxxxxxx>
> Cc: Aristeu Rozanski <aris@xxxxxxxxxx>
> Cc: Matt Helsley <matthltc@xxxxxxxxxx>
> Cc: Vivek Goyal <vgoyal@xxxxxxxxxx>
> Cc: Jens Axboe <axboe@xxxxxxxxx>
For memcg part
Acked-by: Michal Hocko <mhocko@xxxxxxx>
(I hated additional css.cgroup step anyway)
> ---
> block/blk-cgroup.c | 8 +--
> block/blk-cgroup.h | 25 ++++-----
> block/blk-throttle.c | 8 +--
> include/linux/cgroup.h | 88 ++++++++++++++++---------------
> kernel/cgroup.c | 131 ++++++++++++++++++++++++++---------------------
> kernel/cgroup_freezer.c | 25 ++++-----
> kernel/cpuset.c | 58 ++++++++++-----------
> mm/memcontrol.c | 20 ++++----
> security/device_cgroup.c | 11 ++--
> 9 files changed, 187 insertions(+), 187 deletions(-)
>
> diff --git a/block/blk-cgroup.c b/block/blk-cgroup.c
> index f46f3c6..4b40640 100644
> --- a/block/blk-cgroup.c
> +++ b/block/blk-cgroup.c
> @@ -614,7 +614,7 @@ u64 blkg_stat_recursive_sum(struct blkg_policy_data *pd, int off)
> {
> struct blkcg_policy *pol = blkcg_policy[pd->plid];
> struct blkcg_gq *pos_blkg;
> - struct cgroup *pos_cgrp;
> + struct cgroup_subsys_state *pos_css;
> u64 sum;
>
> lockdep_assert_held(pd->blkg->q->queue_lock);
> @@ -622,7 +622,7 @@ u64 blkg_stat_recursive_sum(struct blkg_policy_data *pd, int off)
> sum = blkg_stat_read((void *)pd + off);
>
> rcu_read_lock();
> - blkg_for_each_descendant_pre(pos_blkg, pos_cgrp, pd_to_blkg(pd)) {
> + blkg_for_each_descendant_pre(pos_blkg, pos_css, pd_to_blkg(pd)) {
> struct blkg_policy_data *pos_pd = blkg_to_pd(pos_blkg, pol);
> struct blkg_stat *stat = (void *)pos_pd + off;
>
> @@ -649,7 +649,7 @@ struct blkg_rwstat blkg_rwstat_recursive_sum(struct blkg_policy_data *pd,
> {
> struct blkcg_policy *pol = blkcg_policy[pd->plid];
> struct blkcg_gq *pos_blkg;
> - struct cgroup *pos_cgrp;
> + struct cgroup_subsys_state *pos_css;
> struct blkg_rwstat sum;
> int i;
>
> @@ -658,7 +658,7 @@ struct blkg_rwstat blkg_rwstat_recursive_sum(struct blkg_policy_data *pd,
> sum = blkg_rwstat_read((void *)pd + off);
>
> rcu_read_lock();
> - blkg_for_each_descendant_pre(pos_blkg, pos_cgrp, pd_to_blkg(pd)) {
> + blkg_for_each_descendant_pre(pos_blkg, pos_css, pd_to_blkg(pd)) {
> struct blkg_policy_data *pos_pd = blkg_to_pd(pos_blkg, pol);
> struct blkg_rwstat *rwstat = (void *)pos_pd + off;
> struct blkg_rwstat tmp;
> diff --git a/block/blk-cgroup.h b/block/blk-cgroup.h
> index b6802c4..8555386 100644
> --- a/block/blk-cgroup.h
> +++ b/block/blk-cgroup.h
> @@ -184,11 +184,6 @@ static inline struct blkcg *css_to_blkcg(struct cgroup_subsys_state *css)
> return css ? container_of(css, struct blkcg, css) : NULL;
> }
>
> -static inline struct blkcg *cgroup_to_blkcg(struct cgroup *cgroup)
> -{
> - return css_to_blkcg(cgroup_css(cgroup, blkio_subsys_id));
> -}
> -
> static inline struct blkcg *task_blkcg(struct task_struct *tsk)
> {
> return css_to_blkcg(task_css(tsk, blkio_subsys_id));
> @@ -289,32 +284,31 @@ struct blkcg_gq *__blkg_lookup(struct blkcg *blkcg, struct request_queue *q,
> /**
> * blkg_for_each_descendant_pre - pre-order walk of a blkg's descendants
> * @d_blkg: loop cursor pointing to the current descendant
> - * @pos_cgrp: used for iteration
> + * @pos_css: used for iteration
> * @p_blkg: target blkg to walk descendants of
> *
> * Walk @c_blkg through the descendants of @p_blkg. Must be used with RCU
> * read locked. If called under either blkcg or queue lock, the iteration
> * is guaranteed to include all and only online blkgs. The caller may
> - * update @pos_cgrp by calling cgroup_rightmost_descendant() to skip
> - * subtree.
> + * update @pos_css by calling css_rightmost_descendant() to skip subtree.
> */
> -#define blkg_for_each_descendant_pre(d_blkg, pos_cgrp, p_blkg) \
> - cgroup_for_each_descendant_pre((pos_cgrp), (p_blkg)->blkcg->css.cgroup) \
> - if (((d_blkg) = __blkg_lookup(cgroup_to_blkcg(pos_cgrp), \
> +#define blkg_for_each_descendant_pre(d_blkg, pos_css, p_blkg) \
> + css_for_each_descendant_pre((pos_css), &(p_blkg)->blkcg->css) \
> + if (((d_blkg) = __blkg_lookup(css_to_blkcg(pos_css), \
> (p_blkg)->q, false)))
>
> /**
> * blkg_for_each_descendant_post - post-order walk of a blkg's descendants
> * @d_blkg: loop cursor pointing to the current descendant
> - * @pos_cgrp: used for iteration
> + * @pos_css: used for iteration
> * @p_blkg: target blkg to walk descendants of
> *
> * Similar to blkg_for_each_descendant_pre() but performs post-order
> * traversal instead. Synchronization rules are the same.
> */
> -#define blkg_for_each_descendant_post(d_blkg, pos_cgrp, p_blkg) \
> - cgroup_for_each_descendant_post((pos_cgrp), (p_blkg)->blkcg->css.cgroup) \
> - if (((d_blkg) = __blkg_lookup(cgroup_to_blkcg(pos_cgrp), \
> +#define blkg_for_each_descendant_post(d_blkg, pos_css, p_blkg) \
> + css_for_each_descendant_post((pos_css), &(p_blkg)->blkcg->css) \
> + if (((d_blkg) = __blkg_lookup(css_to_blkcg(pos_css), \
> (p_blkg)->q, false)))
>
> /**
> @@ -577,7 +571,6 @@ static inline int blkcg_activate_policy(struct request_queue *q,
> static inline void blkcg_deactivate_policy(struct request_queue *q,
> const struct blkcg_policy *pol) { }
>
> -static inline struct blkcg *cgroup_to_blkcg(struct cgroup *cgroup) { return NULL; }
> static inline struct blkcg *bio_blkcg(struct bio *bio) { return NULL; }
>
> static inline struct blkg_policy_data *blkg_to_pd(struct blkcg_gq *blkg,
> diff --git a/block/blk-throttle.c b/block/blk-throttle.c
> index 88bcfb6..8cefa7f 100644
> --- a/block/blk-throttle.c
> +++ b/block/blk-throttle.c
> @@ -1349,7 +1349,7 @@ static int tg_set_conf(struct cgroup_subsys_state *css, struct cftype *cft,
> struct throtl_grp *tg;
> struct throtl_service_queue *sq;
> struct blkcg_gq *blkg;
> - struct cgroup *pos_cgrp;
> + struct cgroup_subsys_state *pos_css;
> int ret;
>
> ret = blkg_conf_prep(blkcg, &blkcg_policy_throtl, buf, &ctx);
> @@ -1380,7 +1380,7 @@ static int tg_set_conf(struct cgroup_subsys_state *css, struct cftype *cft,
> * blk-throttle.
> */
> tg_update_has_rules(tg);
> - blkg_for_each_descendant_pre(blkg, pos_cgrp, ctx.blkg)
> + blkg_for_each_descendant_pre(blkg, pos_css, ctx.blkg)
> tg_update_has_rules(blkg_to_tg(blkg));
>
> /*
> @@ -1623,7 +1623,7 @@ void blk_throtl_drain(struct request_queue *q)
> {
> struct throtl_data *td = q->td;
> struct blkcg_gq *blkg;
> - struct cgroup *pos_cgrp;
> + struct cgroup_subsys_state *pos_css;
> struct bio *bio;
> int rw;
>
> @@ -1636,7 +1636,7 @@ void blk_throtl_drain(struct request_queue *q)
> * better to walk service_queue tree directly but blkg walk is
> * easier.
> */
> - blkg_for_each_descendant_post(blkg, pos_cgrp, td->queue->root_blkg)
> + blkg_for_each_descendant_post(blkg, pos_css, td->queue->root_blkg)
> tg_drain_bios(&blkg_to_tg(blkg)->service_queue);
>
> tg_drain_bios(&td_root_tg(td)->service_queue);
> diff --git a/include/linux/cgroup.h b/include/linux/cgroup.h
> index df6ab19..7fba0d0 100644
> --- a/include/linux/cgroup.h
> +++ b/include/linux/cgroup.h
> @@ -780,68 +780,72 @@ static inline struct cgroup *cgroup_from_id(struct cgroup_subsys *ss, int id)
> return idr_find(&ss->root->cgroup_idr, id);
> }
>
> -struct cgroup *cgroup_next_child(struct cgroup *pos, struct cgroup *cgrp);
> +struct cgroup_subsys_state *css_next_child(struct cgroup_subsys_state *pos,
> + struct cgroup_subsys_state *parent);
>
> /**
> - * cgroup_for_each_child - iterate through children of a cgroup
> - * @pos: the cgroup * to use as the loop cursor
> - * @cgrp: cgroup whose children to walk
> + * css_for_each_child - iterate through children of a css
> + * @pos: the css * to use as the loop cursor
> + * @parent: css whose children to walk
> *
> - * Walk @cgrp's children. Must be called under rcu_read_lock(). A child
> - * cgroup which hasn't finished ->css_online() or already has finished
> + * Walk @parent's children. Must be called under rcu_read_lock(). A child
> + * css which hasn't finished ->css_online() or already has finished
> * ->css_offline() may show up during traversal and it's each subsystem's
> * responsibility to verify that each @pos is alive.
> *
> * If a subsystem synchronizes against the parent in its ->css_online() and
> - * before starting iterating, a cgroup which finished ->css_online() is
> + * before starting iterating, a css which finished ->css_online() is
> * guaranteed to be visible in the future iterations.
> *
> * It is allowed to temporarily drop RCU read lock during iteration. The
> * caller is responsible for ensuring that @pos remains accessible until
> * the start of the next iteration by, for example, bumping the css refcnt.
> */
> -#define cgroup_for_each_child(pos, cgrp) \
> - for ((pos) = cgroup_next_child(NULL, (cgrp)); (pos); \
> - (pos) = cgroup_next_child((pos), (cgrp)))
> +#define css_for_each_child(pos, parent) \
> + for ((pos) = css_next_child(NULL, (parent)); (pos); \
> + (pos) = css_next_child((pos), (parent)))
>
> -struct cgroup *cgroup_next_descendant_pre(struct cgroup *pos,
> - struct cgroup *cgroup);
> -struct cgroup *cgroup_rightmost_descendant(struct cgroup *pos);
> +struct cgroup_subsys_state *
> +css_next_descendant_pre(struct cgroup_subsys_state *pos,
> + struct cgroup_subsys_state *css);
> +
> +struct cgroup_subsys_state *
> +css_rightmost_descendant(struct cgroup_subsys_state *pos);
>
> /**
> - * cgroup_for_each_descendant_pre - pre-order walk of a cgroup's descendants
> - * @pos: the cgroup * to use as the loop cursor
> - * @cgroup: cgroup whose descendants to walk
> + * css_for_each_descendant_pre - pre-order walk of a css's descendants
> + * @pos: the css * to use as the loop cursor
> + * @root: css whose descendants to walk
> *
> - * Walk @cgroup's descendants. Must be called under rcu_read_lock(). A
> - * descendant cgroup which hasn't finished ->css_online() or already has
> + * Walk @root's descendants. Must be called under rcu_read_lock(). A
> + * descendant css which hasn't finished ->css_online() or already has
> * finished ->css_offline() may show up during traversal and it's each
> * subsystem's responsibility to verify that each @pos is alive.
> *
> * If a subsystem synchronizes against the parent in its ->css_online() and
> * before starting iterating, and synchronizes against @pos on each
> - * iteration, any descendant cgroup which finished ->css_online() is
> + * iteration, any descendant css which finished ->css_online() is
> * guaranteed to be visible in the future iterations.
> *
> * In other words, the following guarantees that a descendant can't escape
> * state updates of its ancestors.
> *
> - * my_online(@cgrp)
> + * my_online(@css)
> * {
> - * Lock @cgrp->parent and @cgrp;
> - * Inherit state from @cgrp->parent;
> + * Lock @css's parent and @css;
> + * Inherit state from the parent;
> * Unlock both.
> * }
> *
> - * my_update_state(@cgrp)
> + * my_update_state(@css)
> * {
> - * Lock @cgrp;
> - * Update @cgrp's state;
> - * Unlock @cgrp;
> + * Lock @css;
> + * Update @css's state;
> + * Unlock @css;
> *
> - * cgroup_for_each_descendant_pre(@pos, @cgrp) {
> + * css_for_each_descendant_pre(@pos, @css) {
> * Lock @pos;
> - * Verify @pos is alive and inherit state from @pos->parent;
> + * Verify @pos is alive and inherit state from @pos's parent;
> * Unlock @pos;
> * }
> * }
> @@ -852,8 +856,7 @@ struct cgroup *cgroup_rightmost_descendant(struct cgroup *pos);
> * visible by walking order and, as long as inheriting operations to the
> * same @pos are atomic to each other, multiple updates racing each other
> * still result in the correct state. It's guaranateed that at least one
> - * inheritance happens for any cgroup after the latest update to its
> - * parent.
> + * inheritance happens for any css after the latest update to its parent.
> *
> * If checking parent's state requires locking the parent, each inheriting
> * iteration should lock and unlock both @pos->parent and @pos.
> @@ -866,25 +869,26 @@ struct cgroup *cgroup_rightmost_descendant(struct cgroup *pos);
> * caller is responsible for ensuring that @pos remains accessible until
> * the start of the next iteration by, for example, bumping the css refcnt.
> */
> -#define cgroup_for_each_descendant_pre(pos, cgroup) \
> - for (pos = cgroup_next_descendant_pre(NULL, (cgroup)); (pos); \
> - pos = cgroup_next_descendant_pre((pos), (cgroup)))
> +#define css_for_each_descendant_pre(pos, css) \
> + for ((pos) = css_next_descendant_pre(NULL, (css)); (pos); \
> + (pos) = css_next_descendant_pre((pos), (css)))
>
> -struct cgroup *cgroup_next_descendant_post(struct cgroup *pos,
> - struct cgroup *cgroup);
> +struct cgroup_subsys_state *
> +css_next_descendant_post(struct cgroup_subsys_state *pos,
> + struct cgroup_subsys_state *css);
>
> /**
> - * cgroup_for_each_descendant_post - post-order walk of a cgroup's descendants
> - * @pos: the cgroup * to use as the loop cursor
> - * @cgroup: cgroup whose descendants to walk
> + * css_for_each_descendant_post - post-order walk of a css's descendants
> + * @pos: the css * to use as the loop cursor
> + * @css: css whose descendants to walk
> *
> - * Similar to cgroup_for_each_descendant_pre() but performs post-order
> + * Similar to css_for_each_descendant_pre() but performs post-order
> * traversal instead. Note that the walk visibility guarantee described in
> * pre-order walk doesn't apply the same to post-order walks.
> */
> -#define cgroup_for_each_descendant_post(pos, cgroup) \
> - for (pos = cgroup_next_descendant_post(NULL, (cgroup)); (pos); \
> - pos = cgroup_next_descendant_post((pos), (cgroup)))
> +#define css_for_each_descendant_post(pos, css) \
> + for ((pos) = css_next_descendant_post(NULL, (css)); (pos); \
> + (pos) = css_next_descendant_post((pos), (css)))
>
> /* A cgroup_iter should be treated as an opaque object */
> struct cgroup_iter {
> diff --git a/kernel/cgroup.c b/kernel/cgroup.c
> index 7b53b58..850ad87 100644
> --- a/kernel/cgroup.c
> +++ b/kernel/cgroup.c
> @@ -2807,8 +2807,8 @@ static void cgroup_cfts_prepare(void)
> /*
> * Thanks to the entanglement with vfs inode locking, we can't walk
> * the existing cgroups under cgroup_mutex and create files.
> - * Instead, we use cgroup_for_each_descendant_pre() and drop RCU
> - * read lock before calling cgroup_addrm_files().
> + * Instead, we use css_for_each_descendant_pre() and drop RCU read
> + * lock before calling cgroup_addrm_files().
> */
> mutex_lock(&cgroup_mutex);
> }
> @@ -2818,10 +2818,11 @@ static int cgroup_cfts_commit(struct cftype *cfts, bool is_add)
> {
> LIST_HEAD(pending);
> struct cgroup_subsys *ss = cfts[0].ss;
> - struct cgroup *cgrp, *root = &ss->root->top_cgroup;
> + struct cgroup *root = &ss->root->top_cgroup;
> struct super_block *sb = ss->root->sb;
> struct dentry *prev = NULL;
> struct inode *inode;
> + struct cgroup_subsys_state *css;
> u64 update_before;
> int ret = 0;
>
> @@ -2854,7 +2855,9 @@ static int cgroup_cfts_commit(struct cftype *cfts, bool is_add)
>
> /* add/rm files for all cgroups created before */
> rcu_read_lock();
> - cgroup_for_each_descendant_pre(cgrp, root) {
> + css_for_each_descendant_pre(css, cgroup_css(root, ss->subsys_id)) {
> + struct cgroup *cgrp = css->cgroup;
> +
> if (cgroup_is_dead(cgrp))
> continue;
>
> @@ -3030,17 +3033,21 @@ static void cgroup_enable_task_cg_lists(void)
> }
>
> /**
> - * cgroup_next_child - find the next child of a given cgroup
> - * @pos: the current position (%NULL to initiate traversal)
> - * @cgrp: cgroup whose descendants to walk
> + * css_next_child - find the next child of a given css
> + * @pos_css: the current position (%NULL to initiate traversal)
> + * @parent_css: css whose children to walk
> *
> - * This function returns the next child of @cgrp and should be called under
> - * RCU read lock. The only requirement is that @cgrp and @pos are
> - * accessible. The next sibling is guaranteed to be returned regardless of
> - * their states.
> + * This function returns the next child of @parent_css and should be called
> + * under RCU read lock. The only requirement is that @parent_css and
> + * @pos_css are accessible. The next sibling is guaranteed to be returned
> + * regardless of their states.
> */
> -struct cgroup *cgroup_next_child(struct cgroup *pos, struct cgroup *cgrp)
> +struct cgroup_subsys_state *
> +css_next_child(struct cgroup_subsys_state *pos_css,
> + struct cgroup_subsys_state *parent_css)
> {
> + struct cgroup *pos = pos_css ? pos_css->cgroup : NULL;
> + struct cgroup *cgrp = parent_css->cgroup;
> struct cgroup *next;
>
> WARN_ON_ONCE(!rcu_read_lock_held());
> @@ -3074,59 +3081,64 @@ struct cgroup *cgroup_next_child(struct cgroup *pos, struct cgroup *cgrp)
> break;
> }
>
> - if (&next->sibling != &cgrp->children)
> - return next;
> - return NULL;
> + if (&next->sibling == &cgrp->children)
> + return NULL;
> +
> + if (parent_css->ss)
> + return cgroup_css(next, parent_css->ss->subsys_id);
> + else
> + return &next->dummy_css;
> }
> -EXPORT_SYMBOL_GPL(cgroup_next_child);
> +EXPORT_SYMBOL_GPL(css_next_child);
>
> /**
> - * cgroup_next_descendant_pre - find the next descendant for pre-order walk
> + * css_next_descendant_pre - find the next descendant for pre-order walk
> * @pos: the current position (%NULL to initiate traversal)
> - * @cgroup: cgroup whose descendants to walk
> + * @root: css whose descendants to walk
> *
> - * To be used by cgroup_for_each_descendant_pre(). Find the next
> - * descendant to visit for pre-order traversal of @cgroup's descendants.
> + * To be used by css_for_each_descendant_pre(). Find the next descendant
> + * to visit for pre-order traversal of @root's descendants.
> *
> * While this function requires RCU read locking, it doesn't require the
> * whole traversal to be contained in a single RCU critical section. This
> * function will return the correct next descendant as long as both @pos
> - * and @cgroup are accessible and @pos is a descendant of @cgroup.
> + * and @root are accessible and @pos is a descendant of @root.
> */
> -struct cgroup *cgroup_next_descendant_pre(struct cgroup *pos,
> - struct cgroup *cgroup)
> +struct cgroup_subsys_state *
> +css_next_descendant_pre(struct cgroup_subsys_state *pos,
> + struct cgroup_subsys_state *root)
> {
> - struct cgroup *next;
> + struct cgroup_subsys_state *next;
>
> WARN_ON_ONCE(!rcu_read_lock_held());
>
> - /* if first iteration, pretend we just visited @cgroup */
> + /* if first iteration, pretend we just visited @root */
> if (!pos)
> - pos = cgroup;
> + pos = root;
>
> /* visit the first child if exists */
> - next = cgroup_next_child(NULL, pos);
> + next = css_next_child(NULL, pos);
> if (next)
> return next;
>
> /* no child, visit my or the closest ancestor's next sibling */
> - while (pos != cgroup) {
> - next = cgroup_next_child(pos, pos->parent);
> + while (pos != root) {
> + next = css_next_child(pos, css_parent(pos));
> if (next)
> return next;
> - pos = pos->parent;
> + pos = css_parent(pos);
> }
>
> return NULL;
> }
> -EXPORT_SYMBOL_GPL(cgroup_next_descendant_pre);
> +EXPORT_SYMBOL_GPL(css_next_descendant_pre);
>
> /**
> - * cgroup_rightmost_descendant - return the rightmost descendant of a cgroup
> - * @pos: cgroup of interest
> + * css_rightmost_descendant - return the rightmost descendant of a css
> + * @pos: css of interest
> *
> - * Return the rightmost descendant of @pos. If there's no descendant,
> - * @pos is returned. This can be used during pre-order traversal to skip
> + * Return the rightmost descendant of @pos. If there's no descendant, @pos
> + * is returned. This can be used during pre-order traversal to skip
> * subtree of @pos.
> *
> * While this function requires RCU read locking, it doesn't require the
> @@ -3134,9 +3146,10 @@ EXPORT_SYMBOL_GPL(cgroup_next_descendant_pre);
> * function will return the correct rightmost descendant as long as @pos is
> * accessible.
> */
> -struct cgroup *cgroup_rightmost_descendant(struct cgroup *pos)
> +struct cgroup_subsys_state *
> +css_rightmost_descendant(struct cgroup_subsys_state *pos)
> {
> - struct cgroup *last, *tmp;
> + struct cgroup_subsys_state *last, *tmp;
>
> WARN_ON_ONCE(!rcu_read_lock_held());
>
> @@ -3144,62 +3157,64 @@ struct cgroup *cgroup_rightmost_descendant(struct cgroup *pos)
> last = pos;
> /* ->prev isn't RCU safe, walk ->next till the end */
> pos = NULL;
> - cgroup_for_each_child(tmp, last)
> + css_for_each_child(tmp, last)
> pos = tmp;
> } while (pos);
>
> return last;
> }
> -EXPORT_SYMBOL_GPL(cgroup_rightmost_descendant);
> +EXPORT_SYMBOL_GPL(css_rightmost_descendant);
>
> -static struct cgroup *cgroup_leftmost_descendant(struct cgroup *pos)
> +static struct cgroup_subsys_state *
> +css_leftmost_descendant(struct cgroup_subsys_state *pos)
> {
> - struct cgroup *last;
> + struct cgroup_subsys_state *last;
>
> do {
> last = pos;
> - pos = cgroup_next_child(NULL, pos);
> + pos = css_next_child(NULL, pos);
> } while (pos);
>
> return last;
> }
>
> /**
> - * cgroup_next_descendant_post - find the next descendant for post-order walk
> + * css_next_descendant_post - find the next descendant for post-order walk
> * @pos: the current position (%NULL to initiate traversal)
> - * @cgroup: cgroup whose descendants to walk
> + * @root: css whose descendants to walk
> *
> - * To be used by cgroup_for_each_descendant_post(). Find the next
> - * descendant to visit for post-order traversal of @cgroup's descendants.
> + * To be used by css_for_each_descendant_post(). Find the next descendant
> + * to visit for post-order traversal of @root's descendants.
> *
> * While this function requires RCU read locking, it doesn't require the
> * whole traversal to be contained in a single RCU critical section. This
> * function will return the correct next descendant as long as both @pos
> * and @cgroup are accessible and @pos is a descendant of @cgroup.
> */
> -struct cgroup *cgroup_next_descendant_post(struct cgroup *pos,
> - struct cgroup *cgroup)
> +struct cgroup_subsys_state *
> +css_next_descendant_post(struct cgroup_subsys_state *pos,
> + struct cgroup_subsys_state *root)
> {
> - struct cgroup *next;
> + struct cgroup_subsys_state *next;
>
> WARN_ON_ONCE(!rcu_read_lock_held());
>
> /* if first iteration, visit the leftmost descendant */
> if (!pos) {
> - next = cgroup_leftmost_descendant(cgroup);
> - return next != cgroup ? next : NULL;
> + next = css_leftmost_descendant(root);
> + return next != root ? next : NULL;
> }
>
> /* if there's an unvisited sibling, visit its leftmost descendant */
> - next = cgroup_next_child(pos, pos->parent);
> + next = css_next_child(pos, css_parent(pos));
> if (next)
> - return cgroup_leftmost_descendant(next);
> + return css_leftmost_descendant(next);
>
> /* no sibling left, visit parent */
> - next = pos->parent;
> - return next != cgroup ? next : NULL;
> + next = css_parent(pos);
> + return next != root ? next : NULL;
> }
> -EXPORT_SYMBOL_GPL(cgroup_next_descendant_post);
> +EXPORT_SYMBOL_GPL(css_next_descendant_post);
>
> void cgroup_iter_start(struct cgroup *cgrp, struct cgroup_iter *it)
> __acquires(css_set_lock)
> @@ -4540,9 +4555,9 @@ static int cgroup_destroy_locked(struct cgroup *cgrp)
> /*
> * Mark @cgrp dead. This prevents further task migration and child
> * creation by disabling cgroup_lock_live_group(). Note that
> - * CGRP_DEAD assertion is depended upon by cgroup_next_child() to
> + * CGRP_DEAD assertion is depended upon by css_next_child() to
> * resume iteration after dropping RCU read lock. See
> - * cgroup_next_child() for details.
> + * css_next_child() for details.
> */
> set_bit(CGRP_DEAD, &cgrp->flags);
>
> diff --git a/kernel/cgroup_freezer.c b/kernel/cgroup_freezer.c
> index 19613ba..98ca48d 100644
> --- a/kernel/cgroup_freezer.c
> +++ b/kernel/cgroup_freezer.c
> @@ -50,11 +50,6 @@ static inline struct freezer *css_freezer(struct cgroup_subsys_state *css)
> return css ? container_of(css, struct freezer, css) : NULL;
> }
>
> -static inline struct freezer *cgroup_freezer(struct cgroup *cgroup)
> -{
> - return css_freezer(cgroup_css(cgroup, freezer_subsys_id));
> -}
> -
> static inline struct freezer *task_freezer(struct task_struct *task)
> {
> return css_freezer(task_css(task, freezer_subsys_id));
> @@ -120,7 +115,7 @@ static int freezer_css_online(struct cgroup_subsys_state *css)
> /*
> * The following double locking and freezing state inheritance
> * guarantee that @cgroup can never escape ancestors' freezing
> - * states. See cgroup_for_each_descendant_pre() for details.
> + * states. See css_for_each_descendant_pre() for details.
> */
> if (parent)
> spin_lock_irq(&parent->lock);
> @@ -262,7 +257,7 @@ out:
> static void update_if_frozen(struct cgroup_subsys_state *css)
> {
> struct freezer *freezer = css_freezer(css);
> - struct cgroup *pos;
> + struct cgroup_subsys_state *pos;
> struct cgroup_iter it;
> struct task_struct *task;
>
> @@ -275,8 +270,8 @@ static void update_if_frozen(struct cgroup_subsys_state *css)
> goto out_unlock;
>
> /* are all (live) children frozen? */
> - cgroup_for_each_child(pos, css->cgroup) {
> - struct freezer *child = cgroup_freezer(pos);
> + css_for_each_child(pos, css) {
> + struct freezer *child = css_freezer(pos);
>
> if ((child->state & CGROUP_FREEZER_ONLINE) &&
> !(child->state & CGROUP_FROZEN))
> @@ -309,13 +304,13 @@ out_unlock:
> static int freezer_read(struct cgroup_subsys_state *css, struct cftype *cft,
> struct seq_file *m)
> {
> - struct cgroup *pos;
> + struct cgroup_subsys_state *pos;
>
> rcu_read_lock();
>
> /* update states bottom-up */
> - cgroup_for_each_descendant_post(pos, css->cgroup)
> - update_if_frozen(cgroup_css(pos, freezer_subsys_id));
> + css_for_each_descendant_post(pos, css)
> + update_if_frozen(pos);
> update_if_frozen(css);
>
> rcu_read_unlock();
> @@ -396,7 +391,7 @@ static void freezer_apply_state(struct freezer *freezer, bool freeze,
> */
> static void freezer_change_state(struct freezer *freezer, bool freeze)
> {
> - struct cgroup *pos;
> + struct cgroup_subsys_state *pos;
>
> /* update @freezer */
> spin_lock_irq(&freezer->lock);
> @@ -409,8 +404,8 @@ static void freezer_change_state(struct freezer *freezer, bool freeze)
> * CGROUP_FREEZING_PARENT.
> */
> rcu_read_lock();
> - cgroup_for_each_descendant_pre(pos, freezer->css.cgroup) {
> - struct freezer *pos_f = cgroup_freezer(pos);
> + css_for_each_descendant_pre(pos, &freezer->css) {
> + struct freezer *pos_f = css_freezer(pos);
> struct freezer *parent = parent_freezer(pos_f);
>
> /*
> diff --git a/kernel/cpuset.c b/kernel/cpuset.c
> index 89b76e1..be4f503 100644
> --- a/kernel/cpuset.c
> +++ b/kernel/cpuset.c
> @@ -210,29 +210,29 @@ static struct cpuset top_cpuset = {
> /**
> * cpuset_for_each_child - traverse online children of a cpuset
> * @child_cs: loop cursor pointing to the current child
> - * @pos_cgrp: used for iteration
> + * @pos_css: used for iteration
> * @parent_cs: target cpuset to walk children of
> *
> * Walk @child_cs through the online children of @parent_cs. Must be used
> * with RCU read locked.
> */
> -#define cpuset_for_each_child(child_cs, pos_cgrp, parent_cs) \
> - cgroup_for_each_child((pos_cgrp), (parent_cs)->css.cgroup) \
> - if (is_cpuset_online(((child_cs) = cgroup_cs((pos_cgrp)))))
> +#define cpuset_for_each_child(child_cs, pos_css, parent_cs) \
> + css_for_each_child((pos_css), &(parent_cs)->css) \
> + if (is_cpuset_online(((child_cs) = css_cs((pos_css)))))
>
> /**
> * cpuset_for_each_descendant_pre - pre-order walk of a cpuset's descendants
> * @des_cs: loop cursor pointing to the current descendant
> - * @pos_cgrp: used for iteration
> + * @pos_css: used for iteration
> * @root_cs: target cpuset to walk ancestor of
> *
> * Walk @des_cs through the online descendants of @root_cs. Must be used
> - * with RCU read locked. The caller may modify @pos_cgrp by calling
> - * cgroup_rightmost_descendant() to skip subtree.
> + * with RCU read locked. The caller may modify @pos_css by calling
> + * css_rightmost_descendant() to skip subtree.
> */
> -#define cpuset_for_each_descendant_pre(des_cs, pos_cgrp, root_cs) \
> - cgroup_for_each_descendant_pre((pos_cgrp), (root_cs)->css.cgroup) \
> - if (is_cpuset_online(((des_cs) = cgroup_cs((pos_cgrp)))))
> +#define cpuset_for_each_descendant_pre(des_cs, pos_css, root_cs) \
> + css_for_each_descendant_pre((pos_css), &(root_cs)->css) \
> + if (is_cpuset_online(((des_cs) = css_cs((pos_css)))))
>
> /*
> * There are two global mutexes guarding cpuset structures - cpuset_mutex
> @@ -430,7 +430,7 @@ static void free_trial_cpuset(struct cpuset *trial)
>
> static int validate_change(struct cpuset *cur, struct cpuset *trial)
> {
> - struct cgroup *cgrp;
> + struct cgroup_subsys_state *css;
> struct cpuset *c, *par;
> int ret;
>
> @@ -438,7 +438,7 @@ static int validate_change(struct cpuset *cur, struct cpuset *trial)
>
> /* Each of our child cpusets must be a subset of us */
> ret = -EBUSY;
> - cpuset_for_each_child(c, cgrp, cur)
> + cpuset_for_each_child(c, css, cur)
> if (!is_cpuset_subset(c, trial))
> goto out;
>
> @@ -459,7 +459,7 @@ static int validate_change(struct cpuset *cur, struct cpuset *trial)
> * overlap
> */
> ret = -EINVAL;
> - cpuset_for_each_child(c, cgrp, par) {
> + cpuset_for_each_child(c, css, par) {
> if ((is_cpu_exclusive(trial) || is_cpu_exclusive(c)) &&
> c != cur &&
> cpumask_intersects(trial->cpus_allowed, c->cpus_allowed))
> @@ -508,13 +508,13 @@ static void update_domain_attr_tree(struct sched_domain_attr *dattr,
> struct cpuset *root_cs)
> {
> struct cpuset *cp;
> - struct cgroup *pos_cgrp;
> + struct cgroup_subsys_state *pos_css;
>
> rcu_read_lock();
> - cpuset_for_each_descendant_pre(cp, pos_cgrp, root_cs) {
> + cpuset_for_each_descendant_pre(cp, pos_css, root_cs) {
> /* skip the whole subtree if @cp doesn't have any CPU */
> if (cpumask_empty(cp->cpus_allowed)) {
> - pos_cgrp = cgroup_rightmost_descendant(pos_cgrp);
> + pos_css = css_rightmost_descendant(pos_css);
> continue;
> }
>
> @@ -589,7 +589,7 @@ static int generate_sched_domains(cpumask_var_t **domains,
> struct sched_domain_attr *dattr; /* attributes for custom domains */
> int ndoms = 0; /* number of sched domains in result */
> int nslot; /* next empty doms[] struct cpumask slot */
> - struct cgroup *pos_cgrp;
> + struct cgroup_subsys_state *pos_css;
>
> doms = NULL;
> dattr = NULL;
> @@ -618,7 +618,7 @@ static int generate_sched_domains(cpumask_var_t **domains,
> csn = 0;
>
> rcu_read_lock();
> - cpuset_for_each_descendant_pre(cp, pos_cgrp, &top_cpuset) {
> + cpuset_for_each_descendant_pre(cp, pos_css, &top_cpuset) {
> /*
> * Continue traversing beyond @cp iff @cp has some CPUs and
> * isn't load balancing. The former is obvious. The
> @@ -635,7 +635,7 @@ static int generate_sched_domains(cpumask_var_t **domains,
> csa[csn++] = cp;
>
> /* skip @cp's subtree */
> - pos_cgrp = cgroup_rightmost_descendant(pos_cgrp);
> + pos_css = css_rightmost_descendant(pos_css);
> }
> rcu_read_unlock();
>
> @@ -886,16 +886,16 @@ static void update_tasks_cpumask_hier(struct cpuset *root_cs,
> bool update_root, struct ptr_heap *heap)
> {
> struct cpuset *cp;
> - struct cgroup *pos_cgrp;
> + struct cgroup_subsys_state *pos_css;
>
> if (update_root)
> update_tasks_cpumask(root_cs, heap);
>
> rcu_read_lock();
> - cpuset_for_each_descendant_pre(cp, pos_cgrp, root_cs) {
> + cpuset_for_each_descendant_pre(cp, pos_css, root_cs) {
> /* skip the whole subtree if @cp have some CPU */
> if (!cpumask_empty(cp->cpus_allowed)) {
> - pos_cgrp = cgroup_rightmost_descendant(pos_cgrp);
> + pos_css = css_rightmost_descendant(pos_css);
> continue;
> }
> if (!css_tryget(&cp->css))
> @@ -1143,16 +1143,16 @@ static void update_tasks_nodemask_hier(struct cpuset *root_cs,
> bool update_root, struct ptr_heap *heap)
> {
> struct cpuset *cp;
> - struct cgroup *pos_cgrp;
> + struct cgroup_subsys_state *pos_css;
>
> if (update_root)
> update_tasks_nodemask(root_cs, heap);
>
> rcu_read_lock();
> - cpuset_for_each_descendant_pre(cp, pos_cgrp, root_cs) {
> + cpuset_for_each_descendant_pre(cp, pos_css, root_cs) {
> /* skip the whole subtree if @cp have some CPU */
> if (!nodes_empty(cp->mems_allowed)) {
> - pos_cgrp = cgroup_rightmost_descendant(pos_cgrp);
> + pos_css = css_rightmost_descendant(pos_css);
> continue;
> }
> if (!css_tryget(&cp->css))
> @@ -1973,7 +1973,7 @@ static int cpuset_css_online(struct cgroup_subsys_state *css)
> struct cpuset *cs = css_cs(css);
> struct cpuset *parent = parent_cs(cs);
> struct cpuset *tmp_cs;
> - struct cgroup *pos_cgrp;
> + struct cgroup_subsys_state *pos_css;
>
> if (!parent)
> return 0;
> @@ -2005,7 +2005,7 @@ static int cpuset_css_online(struct cgroup_subsys_state *css)
> * (and likewise for mems) to the new cgroup.
> */
> rcu_read_lock();
> - cpuset_for_each_child(tmp_cs, pos_cgrp, parent) {
> + cpuset_for_each_child(tmp_cs, pos_css, parent) {
> if (is_mem_exclusive(tmp_cs) || is_cpu_exclusive(tmp_cs)) {
> rcu_read_unlock();
> goto out_unlock;
> @@ -2252,10 +2252,10 @@ static void cpuset_hotplug_workfn(struct work_struct *work)
> /* if cpus or mems changed, we need to propagate to descendants */
> if (cpus_updated || mems_updated) {
> struct cpuset *cs;
> - struct cgroup *pos_cgrp;
> + struct cgroup_subsys_state *pos_css;
>
> rcu_read_lock();
> - cpuset_for_each_descendant_pre(cs, pos_cgrp, &top_cpuset) {
> + cpuset_for_each_descendant_pre(cs, pos_css, &top_cpuset) {
> if (!css_tryget(&cs->css))
> continue;
> rcu_read_unlock();
> diff --git a/mm/memcontrol.c b/mm/memcontrol.c
> index ab64dfc..2285319 100644
> --- a/mm/memcontrol.c
> +++ b/mm/memcontrol.c
> @@ -1082,7 +1082,7 @@ struct mem_cgroup *try_get_mem_cgroup_from_mm(struct mm_struct *mm)
> static struct mem_cgroup *__mem_cgroup_iter_next(struct mem_cgroup *root,
> struct mem_cgroup *last_visited)
> {
> - struct cgroup *prev_cgroup, *next_cgroup;
> + struct cgroup_subsys_state *prev_css, *next_css;
>
> /*
> * Root is not visited by cgroup iterators so it needs an
> @@ -1091,11 +1091,9 @@ static struct mem_cgroup *__mem_cgroup_iter_next(struct mem_cgroup *root,
> if (!last_visited)
> return root;
>
> - prev_cgroup = (last_visited == root) ? NULL
> - : last_visited->css.cgroup;
> + prev_css = (last_visited == root) ? NULL : &last_visited->css;
> skip_node:
> - next_cgroup = cgroup_next_descendant_pre(
> - prev_cgroup, root->css.cgroup);
> + next_css = css_next_descendant_pre(prev_css, &root->css);
>
> /*
> * Even if we found a group we have to make sure it is
> @@ -1104,13 +1102,13 @@ skip_node:
> * last_visited css is safe to use because it is
> * protected by css_get and the tree walk is rcu safe.
> */
> - if (next_cgroup) {
> - struct mem_cgroup *mem = mem_cgroup_from_cont(
> - next_cgroup);
> + if (next_css) {
> + struct mem_cgroup *mem = mem_cgroup_from_css(next_css);
> +
> if (css_tryget(&mem->css))
> return mem;
> else {
> - prev_cgroup = next_cgroup;
> + prev_css = next_css;
> goto skip_node;
> }
> }
> @@ -4939,10 +4937,10 @@ static void mem_cgroup_reparent_charges(struct mem_cgroup *memcg)
> */
> static inline bool __memcg_has_children(struct mem_cgroup *memcg)
> {
> - struct cgroup *pos;
> + struct cgroup_subsys_state *pos;
>
> /* bounce at first found */
> - cgroup_for_each_child(pos, memcg->css.cgroup)
> + css_for_each_child(pos, &memcg->css)
> return true;
> return false;
> }
> diff --git a/security/device_cgroup.c b/security/device_cgroup.c
> index e0ca464..9bf230a 100644
> --- a/security/device_cgroup.c
> +++ b/security/device_cgroup.c
> @@ -56,11 +56,6 @@ static inline struct dev_cgroup *css_to_devcgroup(struct cgroup_subsys_state *s)
> return s ? container_of(s, struct dev_cgroup, css) : NULL;
> }
>
> -static inline struct dev_cgroup *cgroup_to_devcgroup(struct cgroup *cgroup)
> -{
> - return css_to_devcgroup(cgroup_css(cgroup, devices_subsys_id));
> -}
> -
> static inline struct dev_cgroup *task_devcgroup(struct task_struct *task)
> {
> return css_to_devcgroup(task_css(task, devices_subsys_id));
> @@ -447,13 +442,13 @@ static void revalidate_active_exceptions(struct dev_cgroup *devcg)
> static int propagate_exception(struct dev_cgroup *devcg_root,
> struct dev_exception_item *ex)
> {
> - struct cgroup *root = devcg_root->css.cgroup, *pos;
> + struct cgroup_subsys_state *pos;
> int rc = 0;
>
> rcu_read_lock();
>
> - cgroup_for_each_descendant_pre(pos, root) {
> - struct dev_cgroup *devcg = cgroup_to_devcgroup(pos);
> + css_for_each_descendant_pre(pos, &devcg_root->css) {
> + struct dev_cgroup *devcg = css_to_devcgroup(pos);
>
> /*
> * Because devcgroup_mutex is held, no devcg will become
> --
> 1.8.3.1
>
--
Michal Hocko
SUSE Labs
--
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/