whiterose

linux unikernel
Log | Files | Refs | README | LICENSE | git clone https://git.ne02ptzero.me/git/whiterose

commit 596766102a3a37ba20c4d9a4130a8a4b42b34646
parent 9022ada8ab6f1f1a932a3c93815061042e6548a5
Author: Linus Torvalds <torvalds@linux-foundation.org>
Date:   Fri, 24 Aug 2018 13:19:27 -0700

Merge branch 'for-4.19' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/cgroup

Pull cgroup updates from Tejun Heo:
 "Just one commit from Steven to take out spin lock from trace event
  handlers"

* 'for-4.19' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/cgroup:
  cgroup/tracing: Move taking of spin lock out of trace event handlers

Diffstat:
Minclude/trace/events/cgroup.h | 47+++++++++++++++++++++++------------------------
Mkernel/cgroup/cgroup-internal.h | 26++++++++++++++++++++++++++
Mkernel/cgroup/cgroup-v1.c | 4++--
Mkernel/cgroup/cgroup.c | 12+++++++-----
4 files changed, 58 insertions(+), 31 deletions(-)

diff --git a/include/trace/events/cgroup.h b/include/trace/events/cgroup.h @@ -53,24 +53,22 @@ DEFINE_EVENT(cgroup_root, cgroup_remount, DECLARE_EVENT_CLASS(cgroup, - TP_PROTO(struct cgroup *cgrp), + TP_PROTO(struct cgroup *cgrp, const char *path), - TP_ARGS(cgrp), + TP_ARGS(cgrp, path), TP_STRUCT__entry( __field( int, root ) __field( int, id ) __field( int, level ) - __dynamic_array(char, path, - cgroup_path(cgrp, NULL, 0) + 1) + __string( path, path ) ), TP_fast_assign( __entry->root = cgrp->root->hierarchy_id; __entry->id = cgrp->id; __entry->level = cgrp->level; - cgroup_path(cgrp, __get_dynamic_array(path), - __get_dynamic_array_len(path)); + __assign_str(path, path); ), TP_printk("root=%d id=%d level=%d path=%s", @@ -79,45 +77,45 @@ DECLARE_EVENT_CLASS(cgroup, DEFINE_EVENT(cgroup, cgroup_mkdir, - TP_PROTO(struct cgroup *cgroup), + TP_PROTO(struct cgroup *cgrp, const char *path), - TP_ARGS(cgroup) + TP_ARGS(cgrp, path) ); DEFINE_EVENT(cgroup, cgroup_rmdir, - TP_PROTO(struct cgroup *cgroup), + TP_PROTO(struct cgroup *cgrp, const char *path), - TP_ARGS(cgroup) + TP_ARGS(cgrp, path) ); DEFINE_EVENT(cgroup, cgroup_release, - TP_PROTO(struct cgroup *cgroup), + TP_PROTO(struct cgroup *cgrp, const char *path), - TP_ARGS(cgroup) + TP_ARGS(cgrp, path) ); DEFINE_EVENT(cgroup, cgroup_rename, - TP_PROTO(struct cgroup *cgroup), + TP_PROTO(struct cgroup *cgrp, const char *path), - TP_ARGS(cgroup) + TP_ARGS(cgrp, path) ); DECLARE_EVENT_CLASS(cgroup_migrate, - TP_PROTO(struct cgroup *dst_cgrp, struct task_struct *task, bool threadgroup), + TP_PROTO(struct cgroup *dst_cgrp, const char *path, + struct task_struct *task, bool threadgroup), - TP_ARGS(dst_cgrp, task, threadgroup), + TP_ARGS(dst_cgrp, path, task, threadgroup), TP_STRUCT__entry( __field( int, dst_root ) __field( int, dst_id ) __field( int, dst_level ) - __dynamic_array(char, dst_path, - cgroup_path(dst_cgrp, NULL, 0) + 1) __field( int, pid ) + __string( dst_path, path ) __string( comm, task->comm ) ), @@ -125,8 +123,7 @@ DECLARE_EVENT_CLASS(cgroup_migrate, __entry->dst_root = dst_cgrp->root->hierarchy_id; __entry->dst_id = dst_cgrp->id; __entry->dst_level = dst_cgrp->level; - cgroup_path(dst_cgrp, __get_dynamic_array(dst_path), - __get_dynamic_array_len(dst_path)); + __assign_str(dst_path, path); __entry->pid = task->pid; __assign_str(comm, task->comm); ), @@ -138,16 +135,18 @@ DECLARE_EVENT_CLASS(cgroup_migrate, DEFINE_EVENT(cgroup_migrate, cgroup_attach_task, - TP_PROTO(struct cgroup *dst_cgrp, struct task_struct *task, bool threadgroup), + TP_PROTO(struct cgroup *dst_cgrp, const char *path, + struct task_struct *task, bool threadgroup), - TP_ARGS(dst_cgrp, task, threadgroup) + TP_ARGS(dst_cgrp, path, task, threadgroup) ); DEFINE_EVENT(cgroup_migrate, cgroup_transfer_tasks, - TP_PROTO(struct cgroup *dst_cgrp, struct task_struct *task, bool threadgroup), + TP_PROTO(struct cgroup *dst_cgrp, const char *path, + struct task_struct *task, bool threadgroup), - TP_ARGS(dst_cgrp, task, threadgroup) + TP_ARGS(dst_cgrp, path, task, threadgroup) ); #endif /* _TRACE_CGROUP_H */ diff --git a/kernel/cgroup/cgroup-internal.h b/kernel/cgroup/cgroup-internal.h @@ -8,6 +8,32 @@ #include <linux/list.h> #include <linux/refcount.h> +#define TRACE_CGROUP_PATH_LEN 1024 +extern spinlock_t trace_cgroup_path_lock; +extern char trace_cgroup_path[TRACE_CGROUP_PATH_LEN]; + +/* + * cgroup_path() takes a spin lock. It is good practice not to take + * spin locks within trace point handlers, as they are mostly hidden + * from normal view. As cgroup_path() can take the kernfs_rename_lock + * spin lock, it is best to not call that function from the trace event + * handler. + * + * Note: trace_cgroup_##type##_enabled() is a static branch that will only + * be set when the trace event is enabled. + */ +#define TRACE_CGROUP_PATH(type, cgrp, ...) \ + do { \ + if (trace_cgroup_##type##_enabled()) { \ + spin_lock(&trace_cgroup_path_lock); \ + cgroup_path(cgrp, trace_cgroup_path, \ + TRACE_CGROUP_PATH_LEN); \ + trace_cgroup_##type(cgrp, trace_cgroup_path, \ + ##__VA_ARGS__); \ + spin_unlock(&trace_cgroup_path_lock); \ + } \ + } while (0) + /* * 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 @@ -135,7 +135,7 @@ int cgroup_transfer_tasks(struct cgroup *to, struct cgroup *from) if (task) { ret = cgroup_migrate(task, false, &mgctx); if (!ret) - trace_cgroup_transfer_tasks(to, task, false); + TRACE_CGROUP_PATH(transfer_tasks, to, task, false); put_task_struct(task); } } while (task && !ret); @@ -865,7 +865,7 @@ static int cgroup1_rename(struct kernfs_node *kn, struct kernfs_node *new_parent ret = kernfs_rename(kn, new_parent, new_name_str); if (!ret) - trace_cgroup_rename(cgrp); + TRACE_CGROUP_PATH(rename, cgrp); mutex_unlock(&cgroup_mutex); diff --git a/kernel/cgroup/cgroup.c b/kernel/cgroup/cgroup.c @@ -83,6 +83,9 @@ EXPORT_SYMBOL_GPL(cgroup_mutex); EXPORT_SYMBOL_GPL(css_set_lock); #endif +DEFINE_SPINLOCK(trace_cgroup_path_lock); +char trace_cgroup_path[TRACE_CGROUP_PATH_LEN]; + /* * Protects cgroup_idr and css_idr so that IDs can be released without * grabbing cgroup_mutex. @@ -2638,7 +2641,7 @@ int cgroup_attach_task(struct cgroup *dst_cgrp, struct task_struct *leader, cgroup_migrate_finish(&mgctx); if (!ret) - trace_cgroup_attach_task(dst_cgrp, leader, threadgroup); + TRACE_CGROUP_PATH(attach_task, dst_cgrp, leader, threadgroup); return ret; } @@ -4636,7 +4639,7 @@ static void css_release_work_fn(struct work_struct *work) struct cgroup *tcgrp; /* cgroup release path */ - trace_cgroup_release(cgrp); + TRACE_CGROUP_PATH(release, cgrp); if (cgroup_on_dfl(cgrp)) cgroup_rstat_flush(cgrp); @@ -4979,7 +4982,7 @@ int cgroup_mkdir(struct kernfs_node *parent_kn, const char *name, umode_t mode) if (ret) goto out_destroy; - trace_cgroup_mkdir(cgrp); + TRACE_CGROUP_PATH(mkdir, cgrp); /* let's create and online css's */ kernfs_activate(kn); @@ -5167,9 +5170,8 @@ int cgroup_rmdir(struct kernfs_node *kn) return 0; ret = cgroup_destroy_locked(cgrp); - if (!ret) - trace_cgroup_rmdir(cgrp); + TRACE_CGROUP_PATH(rmdir, cgrp); cgroup_kn_unlock(kn); return ret;