[PATCH 07/12] cgroup: make all functions of all cftypes be invoked

From: Paolo Valente
Date: Mon Nov 12 2018 - 04:57:27 EST


From: Angelo Ruocco <angeloruocco90@xxxxxxxxx>

When two or more policies share a file their respective cftypes are
linked together. The allowed operations on those files are: open,
release, write and show, mapped to the functions defined in the
cftypes.

This commit makes the cgroup core invoke, whenever one of those
operations is requested, the respective function of all the cftypes
linked together.

Signed-off-by: Angelo Ruocco <angeloruocco90@xxxxxxxxx>
Signed-off-by: Paolo Valente <paolo.valente@xxxxxxxxxx>
---
kernel/cgroup/cgroup.c | 181 ++++++++++++++++++++++++++++++++++++-------------
1 file changed, 132 insertions(+), 49 deletions(-)

diff --git a/kernel/cgroup/cgroup.c b/kernel/cgroup/cgroup.c
index e3cc437669a8..6d4cfd6395ec 100644
--- a/kernel/cgroup/cgroup.c
+++ b/kernel/cgroup/cgroup.c
@@ -3481,66 +3481,107 @@ static int cgroup_cpu_pressure_show(struct seq_file *seq, void *v)
static int cgroup_file_open(struct kernfs_open_file *of)
{
struct cftype *cft = of->kn->priv;
+ struct cftype *n;
+ int ret = 0;

+ list_for_each_cft(cft, n) {
+ if (cft->open)
+ ret = cft->open(of);
+ /*
+ * If there has been an error with the open function of one of
+ * the cft associated with the file, we call the release
+ * function of all the cftype associated to cft whose open
+ * function succeded.
+ */
+ if (ret) {
+ struct cftype *c = of->kn->priv;
+ struct cftype *n;
+
+ list_for_each_cft(c, n) {
+ if (cft == c)
+ break;
+ if (c->release)
+ c->release(of);
+ }
+ break;
+ }
+ }

- if (cft->open)
- return cft->open(of);
- return 0;
+ return ret;
}

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

- if (cft->release)
- cft->release(of);
+ list_for_each_cft(cft, n)
+ if (cft->release)
+ cft->release(of);
}

+/*
+ * Call all the write functions of the cftypes associated with the file.
+ *
+ * When a write fails, don't keep trying to write into the file via the write
+ * functions of the other cftypes associated with it.
+ */
static ssize_t cgroup_file_write(struct kernfs_open_file *of, char *buf,
size_t nbytes, loff_t off)
{
struct cgroup_namespace *ns = current->nsproxy->cgroup_ns;
struct cgroup *cgrp = of->kn->parent->priv;
struct cftype *cft = of->kn->priv;
+ struct cftype *n;
struct cgroup_subsys_state *css;
- int ret;
+ int ret = 0;

- /*
- * If namespaces are delegation boundaries, disallow writes to
- * files in an non-init namespace root from inside the namespace
- * except for the files explicitly marked delegatable -
- * cgroup.procs and cgroup.subtree_control.
- */
- if ((cgrp->root->flags & CGRP_ROOT_NS_DELEGATE) &&
- !(cft->flags & CFTYPE_NS_DELEGATABLE) &&
- ns != &init_cgroup_ns && ns->root_cset->dfl_cgrp == cgrp)
- return -EPERM;
+ list_for_each_cft(cft, n) {
+ /*
+ * If namespaces are delegation boundaries, disallow writes to
+ * files in an non-init namespace root from inside the
+ * namespace except for the files explicitly marked
+ * delegatable - cgroup.procs and cgroup.subtree_control.
+ */
+ if ((cgrp->root->flags & CGRP_ROOT_NS_DELEGATE) &&
+ !(cft->flags & CFTYPE_NS_DELEGATABLE) &&
+ ns != &init_cgroup_ns && ns->root_cset->dfl_cgrp == cgrp)
+ return -EPERM;

- if (cft->write)
- return cft->write(of, buf, nbytes, off);
+ if (cft->write) {
+ ret = cft->write(of, buf, nbytes, off);

- /*
- * kernfs guarantees that a file isn't deleted with operations in
- * flight, which means that the matching css is and stays alive and
- * doesn't need to be pinned. The RCU locking is not necessary
- * either. It's just for the convenience of using cgroup_css().
- */
- rcu_read_lock();
- css = cgroup_css(cgrp, cft->ss);
- rcu_read_unlock();
+ if (ret)
+ break;
+ continue;
+ }

- if (cft->write_u64) {
- unsigned long long v;
- ret = kstrtoull(buf, 0, &v);
- if (!ret)
- ret = cft->write_u64(css, cft, v);
- } else if (cft->write_s64) {
- long long v;
- ret = kstrtoll(buf, 0, &v);
- if (!ret)
- ret = cft->write_s64(css, cft, v);
- } else {
- ret = -EINVAL;
+ /*
+ * kernfs guarantees that a file isn't deleted with operations
+ * in flight, which means that the matching css is and stays
+ * alive and doesn't need to be pinned. The RCU locking is not
+ * necessary either. It's just for the convenience of using
+ * cgroup_css().
+ */
+ rcu_read_lock();
+ css = cgroup_css(cgrp, cft->ss);
+ rcu_read_unlock();
+
+ if (cft->write_u64) {
+ unsigned long long v;
+
+ ret = kstrtoull(buf, 0, &v);
+ if (!ret)
+ ret = cft->write_u64(css, cft, v);
+ } else if (cft->write_s64) {
+ long long v;
+
+ ret = kstrtoll(buf, 0, &v);
+ if (!ret)
+ ret = cft->write_s64(css, cft, v);
+ } else {
+ return -EINVAL;
+ }
}

return ret ?: nbytes;
@@ -3562,22 +3603,64 @@ static void cgroup_seqfile_stop(struct seq_file *seq, void *v)
seq_cft(seq)->seq_stop(seq, v);
}

+/*
+ * A file shared by more cftypes may be showing different values. In that case
+ * call all the show functions and print the name of the owner that defined
+ * them.
+ */
static int cgroup_seqfile_show(struct seq_file *m, void *arg)
{
struct cftype *cft = seq_cft(m);
+ struct cftype *n;
struct cgroup_subsys_state *css = seq_css(m);
+ char *first_seqshow_str = NULL;
+ size_t first_str_size = 0;
+ size_t current_str_size = 0;
int ret = 0;

- if (cft->seq_show)
- ret = cft->seq_show(m, arg);
- else if (cft->seq_show_cft)
- ret = cft->seq_show_cft(m, cft, arg);
- else if (cft->read_u64)
- seq_printf(m, "%llu\n", cft->read_u64(css, cft));
- else if (cft->read_s64)
- seq_printf(m, "%lld\n", cft->read_s64(css, cft));
- else
- ret = -EINVAL;
+ list_for_each_cft(cft, n) {
+ if (cft->seq_show) {
+ ret = cft->seq_show(m, arg);
+ if (ret)
+ break;
+ } else if (cft->seq_show_cft) {
+ ret = cft->seq_show_cft(m, cft, arg);
+ if (ret)
+ break;
+ } else if (cft->read_u64) {
+ seq_printf(m, "%llu\n", cft->read_u64(css, cft));
+ } else if (cft->read_s64) {
+ seq_printf(m, "%lld\n", cft->read_s64(css, cft));
+ } else {
+ ret = -EINVAL;
+ break;
+ }
+ current_str_size = m->count - current_str_size;
+
+ if (first_seqshow_str == NULL) {
+ first_seqshow_str = kmalloc(m->size, GFP_KERNEL);
+ first_str_size = m->count;
+ strcpy(first_seqshow_str, m->buf);
+ first_str_size = m->count;
+ } else if (strcmp(first_seqshow_str,
+ m->buf + m->count - current_str_size)) {
+ first_str_size = -1;
+ }
+
+ if (current_str_size) {
+ seq_printf(m, " - %s\n", cft->owner_name);
+ current_str_size = m->count;
+ }
+ }
+
+ /*
+ * If all the cft->seqfile_show/read are equal, truncate the
+ * output of the seqfile to the length of the first string.
+ */
+ if (first_str_size != -1)
+ m->count = first_str_size;
+
+ kfree(first_seqshow_str);
return ret;
}

--
2.16.1