From: Tejun Heo tj@kernel.org

commit 0d2b5955b36250a9428c832664f2079cbf723bec upstream.

of->priv is currently used by each interface file implementation to store private information. This patch collects the current two private data usages into struct cgroup_file_ctx which is allocated and freed by the common path. This allows generic private data which applies to multiple files, which will be used to in the following patch.

Note that cgroup_procs iterator is now embedded as procs.iter in the new cgroup_file_ctx so that it doesn't need to be allocated and freed separately.

v2: union dropped from cgroup_file_ctx and the procs iterator is embedded in cgroup_file_ctx as suggested by Linus.

v3: Michal pointed out that cgroup1's procs pidlist uses of->priv too. Converted. Didn't change to embedded allocation as cgroup1 pidlists get stored for caching.

Signed-off-by: Tejun Heo tj@kernel.org Cc: Linus Torvalds torvalds@linux-foundation.org Reviewed-by: Michal Koutný mkoutny@suse.com [mkoutny: v5.10: modify cgroup.pressure handlers, adjust context] Signed-off-by: Michal Koutný mkoutny@suse.com --- kernel/cgroup/cgroup-internal.h | 17 +++++++++++ kernel/cgroup/cgroup-v1.c | 26 ++++++++-------- kernel/cgroup/cgroup.c | 54 +++++++++++++++++++++------------ 3 files changed, 65 insertions(+), 32 deletions(-)

diff --git a/kernel/cgroup/cgroup-internal.h b/kernel/cgroup/cgroup-internal.h index bfbeabc17a9d..cf637bc4ab45 100644 --- a/kernel/cgroup/cgroup-internal.h +++ b/kernel/cgroup/cgroup-internal.h @@ -65,6 +65,23 @@ static inline struct cgroup_fs_context *cgroup_fc2context(struct fs_context *fc) return container_of(kfc, struct cgroup_fs_context, kfc); }

+struct cgroup_pidlist; + +struct cgroup_file_ctx { + struct { + void *trigger; + } psi; + + struct { + bool started; + struct css_task_iter iter; + } procs; + + struct { + struct cgroup_pidlist *pidlist; + } procs1; +}; + /* * A cgroup can be associated with multiple css_sets as different tasks may * belong to different cgroups on different hierarchies. In the other diff --git a/kernel/cgroup/cgroup-v1.c b/kernel/cgroup/cgroup-v1.c index 69fba563c810..40b8d4e4a810 100644 --- a/kernel/cgroup/cgroup-v1.c +++ b/kernel/cgroup/cgroup-v1.c @@ -393,6 +393,7 @@ static void *cgroup_pidlist_start(struct seq_file *s, loff_t *pos) * next pid to display, if any */ struct kernfs_open_file *of = s->private; + struct cgroup_file_ctx *ctx = of->priv; struct cgroup *cgrp = seq_css(s)->cgroup; struct cgroup_pidlist *l; enum cgroup_filetype type = seq_cft(s)->private; @@ -402,25 +403,24 @@ static void *cgroup_pidlist_start(struct seq_file *s, loff_t *pos) mutex_lock(&cgrp->pidlist_mutex);

/* - * !NULL @of->priv indicates that this isn't the first start() - * after open. If the matching pidlist is around, we can use that. - * Look for it. Note that @of->priv can't be used directly. It - * could already have been destroyed. + * !NULL @ctx->procs1.pidlist indicates that this isn't the first + * start() after open. If the matching pidlist is around, we can use + * that. Look for it. Note that @ctx->procs1.pidlist can't be used + * directly. It could already have been destroyed. */ - if (of->priv) - of->priv = cgroup_pidlist_find(cgrp, type); + if (ctx->procs1.pidlist) + ctx->procs1.pidlist = cgroup_pidlist_find(cgrp, type);

/* * Either this is the first start() after open or the matching * pidlist has been destroyed inbetween. Create a new one. */ - if (!of->priv) { - ret = pidlist_array_load(cgrp, type, - (struct cgroup_pidlist **)&of->priv); + if (!ctx->procs1.pidlist) { + ret = pidlist_array_load(cgrp, type, &ctx->procs1.pidlist); if (ret) return ERR_PTR(ret); } - l = of->priv; + l = ctx->procs1.pidlist;

if (pid) { int end = l->length; @@ -448,7 +448,8 @@ static void *cgroup_pidlist_start(struct seq_file *s, loff_t *pos) static void cgroup_pidlist_stop(struct seq_file *s, void *v) { struct kernfs_open_file *of = s->private; - struct cgroup_pidlist *l = of->priv; + struct cgroup_file_ctx *ctx = of->priv; + struct cgroup_pidlist *l = ctx->procs1.pidlist;

if (l) mod_delayed_work(cgroup_pidlist_destroy_wq, &l->destroy_dwork, @@ -459,7 +460,8 @@ static void cgroup_pidlist_stop(struct seq_file *s, void *v) static void *cgroup_pidlist_next(struct seq_file *s, void *v, loff_t *pos) { struct kernfs_open_file *of = s->private; - struct cgroup_pidlist *l = of->priv; + struct cgroup_file_ctx *ctx = of->priv; + struct cgroup_pidlist *l = ctx->procs1.pidlist; pid_t *p = v; pid_t *end = l->list + l->length; /* diff --git a/kernel/cgroup/cgroup.c b/kernel/cgroup/cgroup.c index 4927289a91a9..ddfe6983ea7c 100644 --- a/kernel/cgroup/cgroup.c +++ b/kernel/cgroup/cgroup.c @@ -3590,6 +3590,7 @@ static int cgroup_cpu_pressure_show(struct seq_file *seq, void *v) static ssize_t cgroup_pressure_write(struct kernfs_open_file *of, char *buf, size_t nbytes, enum psi_res res) { + struct cgroup_file_ctx *ctx = of->priv; struct psi_trigger *new; struct cgroup *cgrp; struct psi_group *psi; @@ -3602,7 +3603,7 @@ static ssize_t cgroup_pressure_write(struct kernfs_open_file *of, char *buf, cgroup_kn_unlock(of->kn);

/* Allow only one trigger per file descriptor */ - if (of->priv) { + if (ctx->psi.trigger) { cgroup_put(cgrp); return -EBUSY; } @@ -3614,7 +3615,7 @@ static ssize_t cgroup_pressure_write(struct kernfs_open_file *of, char *buf, return PTR_ERR(new); }

- smp_store_release(&of->priv, new); + smp_store_release(&ctx->psi.trigger, new); cgroup_put(cgrp);

return nbytes; @@ -3644,12 +3645,15 @@ static ssize_t cgroup_cpu_pressure_write(struct kernfs_open_file *of, static __poll_t cgroup_pressure_poll(struct kernfs_open_file *of, poll_table *pt) { - return psi_trigger_poll(&of->priv, of->file, pt); + struct cgroup_file_ctx *ctx = of->priv; + return psi_trigger_poll(&ctx->psi.trigger, of->file, pt); }

static void cgroup_pressure_release(struct kernfs_open_file *of) { - psi_trigger_destroy(of->priv); + struct cgroup_file_ctx *ctx = of->priv; + + psi_trigger_destroy(ctx->psi.trigger); } #endif /* CONFIG_PSI */

@@ -3690,18 +3694,31 @@ static ssize_t cgroup_freeze_write(struct kernfs_open_file *of, static int cgroup_file_open(struct kernfs_open_file *of) { struct cftype *cft = of->kn->priv; + struct cgroup_file_ctx *ctx; + int ret;

- if (cft->open) - return cft->open(of); - return 0; + ctx = kzalloc(sizeof(*ctx), GFP_KERNEL); + if (!ctx) + return -ENOMEM; + of->priv = ctx; + + if (!cft->open) + return 0; + + ret = cft->open(of); + if (ret) + kfree(ctx); + return ret; }

static void cgroup_file_release(struct kernfs_open_file *of) { struct cftype *cft = of->kn->priv; + struct cgroup_file_ctx *ctx = of->priv;

if (cft->release) cft->release(of); + kfree(ctx); }

static ssize_t cgroup_file_write(struct kernfs_open_file *of, char *buf, @@ -4625,21 +4642,21 @@ void css_task_iter_end(struct css_task_iter *it)

static void cgroup_procs_release(struct kernfs_open_file *of) { - if (of->priv) { - css_task_iter_end(of->priv); - kfree(of->priv); - } + struct cgroup_file_ctx *ctx = of->priv; + + if (ctx->procs.started) + css_task_iter_end(&ctx->procs.iter); }

static void *cgroup_procs_next(struct seq_file *s, void *v, loff_t *pos) { struct kernfs_open_file *of = s->private; - struct css_task_iter *it = of->priv; + struct cgroup_file_ctx *ctx = of->priv;

if (pos) (*pos)++;

- return css_task_iter_next(it); + return css_task_iter_next(&ctx->procs.iter); }

static void *__cgroup_procs_start(struct seq_file *s, loff_t *pos, @@ -4647,21 +4664,18 @@ static void *__cgroup_procs_start(struct seq_file *s, loff_t *pos, { struct kernfs_open_file *of = s->private; struct cgroup *cgrp = seq_css(s)->cgroup; - struct css_task_iter *it = of->priv; + struct cgroup_file_ctx *ctx = of->priv; + struct css_task_iter *it = &ctx->procs.iter;

/* * When a seq_file is seeked, it's always traversed sequentially * from position 0, so we can simply keep iterating on !0 *pos. */ - if (!it) { + if (!ctx->procs.started) { if (WARN_ON_ONCE((*pos))) return ERR_PTR(-EINVAL); - - it = kzalloc(sizeof(*it), GFP_KERNEL); - if (!it) - return ERR_PTR(-ENOMEM); - of->priv = it; css_task_iter_start(&cgrp->self, iter_flags, it); + ctx->procs.started = true; } else if (!(*pos)) { css_task_iter_end(it); css_task_iter_start(&cgrp->self, iter_flags, it);

commit 467a726b754f474936980da793b4ff2ec3e382a7 upstream.

The idea is to check: a) the owning user_ns of cgroup_ns, b) capabilities in init_user_ns.

The commit 24f600856418 ("cgroup-v1: Require capabilities to set release_agent") got this wrong in the write handler of release_agent since it checked user_ns of the opener (may be different from the owning user_ns of cgroup_ns). Secondly, to avoid possibly confused deputy, the capability of the opener must be checked.

Fixes: 24f600856418 ("cgroup-v1: Require capabilities to set release_agent") Cc: stable@vger.kernel.org Link: https://lore.kernel.org/stable/20220216121142.GB30035@blackbody.suse.cz/ Signed-off-by: Michal Koutný mkoutny@suse.com Reviewed-by: Masami Ichikawa(CIP) masami.ichikawa@cybertrust.co.jp Signed-off-by: Tejun Heo tj@kernel.org --- kernel/cgroup/cgroup-v1.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/kernel/cgroup/cgroup-v1.c b/kernel/cgroup/cgroup-v1.c index 40b8d4e4a810..8f0ea12d7cee 100644 --- a/kernel/cgroup/cgroup-v1.c +++ b/kernel/cgroup/cgroup-v1.c @@ -544,6 +544,7 @@ static ssize_t cgroup_release_agent_write(struct kernfs_open_file *of, char *buf, size_t nbytes, loff_t off) { struct cgroup *cgrp; + struct cgroup_file_ctx *ctx;

BUILD_BUG_ON(sizeof(cgrp->root->release_agent_path) < PATH_MAX);

@@ -551,8 +552,9 @@ static ssize_t cgroup_release_agent_write(struct kernfs_open_file *of, * Release agent gets called with all capabilities, * require capabilities to set release agent. */ - if ((of->file->f_cred->user_ns != &init_user_ns) || - !capable(CAP_SYS_ADMIN)) + ctx = of->priv; + if ((ctx->ns->user_ns != &init_user_ns) || + !file_ns_capable(of->file, &init_user_ns, CAP_SYS_ADMIN)) return -EPERM;

cgrp = cgroup_kn_lock_live(of->kn, false);