Quoting Srivatsa Vaddagiri (vatsa@in.ibm.com):
> Heavily based on Paul Menage's (inturn cpuset) work. The big difference
> is that the patch uses task->nsproxy to group tasks for resource control
> purpose (instead of task->containers).

Hi,

we've already had some trouble with nsproxy holding things with
different lifetimes.  As it happens the solution this time was to put
the pid namespace where it belongs - not in nsproxy - so maybe moving
this info into nsproxies will be fine, it just rings a warning bell.

This patch would be all the more reason to do
http://lkml.org/lkml/2007/2/21/217
first though, as 'exit_task_namespaces' becomes even more of a lie.

-serge

> The patch retains the same user interface as Paul Menage's patches. In
> particular, you can have multiple hierarchies, each hierarchy giving a 
> different composition/view of task-groups.
> 
> (Ideally this patch should have been split into 2 or 3 sub-patches, but
> will do that on a subsequent version post)
> 
> Signed-off-by : Srivatsa Vaddagiri <vatsa@in.ibm.com>
> Signed-off-by : Paul Menage <menage@google.com>
> 
> 
> ---
> 
>  linux-2.6.20-vatsa/include/linux/init_task.h |    4 
>  linux-2.6.20-vatsa/include/linux/nsproxy.h   |    5 
>  linux-2.6.20-vatsa/init/Kconfig              |   22 
>  linux-2.6.20-vatsa/init/main.c               |    1 
>  linux-2.6.20-vatsa/kernel/Makefile           |    1 
> 
> 
> ---
> 
> 
> diff -puN include/linux/init_task.h~rcfs include/linux/init_task.h
> --- linux-2.6.20/include/linux/init_task.h~rcfs	2007-03-01 14:20:47.000000000 +0530
> +++ linux-2.6.20-vatsa/include/linux/init_task.h	2007-03-01 14:20:47.000000000 +0530
> @@ -71,6 +71,16 @@
>  }
> 
>  extern struct nsproxy init_nsproxy;
> +
> +#ifdef CONFIG_RCFS
> +#define INIT_RCFS(nsproxy) 						\
> +	.list 		= LIST_HEAD_INIT(nsproxy.list),			\
> +	.ctlr_data 	= {[ 0 ... CONFIG_MAX_RC_SUBSYS-1 ] = NULL },
> +#else
> +#define INIT_RCFS(nsproxy)
> +#endif
> +
> +
>  #define INIT_NSPROXY(nsproxy) {						\
>  	.pid_ns		= &init_pid_ns,					\
>  	.count		= ATOMIC_INIT(1),				\
> @@ -78,6 +88,7 @@ extern struct nsproxy init_nsproxy;
>  	.uts_ns		= &init_uts_ns,					\
>  	.mnt_ns		= NULL,						\
>  	INIT_IPC_NS(ipc_ns)						\
> +	INIT_RCFS(nsproxy)						\
>  }
> 
>  #define INIT_SIGHAND(sighand) {						\
> diff -puN include/linux/nsproxy.h~rcfs include/linux/nsproxy.h
> --- linux-2.6.20/include/linux/nsproxy.h~rcfs	2007-03-01 14:20:47.000000000 +0530
> +++ linux-2.6.20-vatsa/include/linux/nsproxy.h	2007-03-01 14:20:47.000000000 +0530
> @@ -28,6 +28,10 @@ struct nsproxy {
>  	struct ipc_namespace *ipc_ns;
>  	struct mnt_namespace *mnt_ns;
>  	struct pid_namespace *pid_ns;
> +#ifdef CONFIG_RCFS
> +	struct list_head list;
> +	void *ctlr_data[CONFIG_MAX_RC_SUBSYS];
> +#endif
>  };
>  extern struct nsproxy init_nsproxy;
> 
> @@ -35,6 +39,12 @@ struct nsproxy *dup_namespaces(struct ns
>  int copy_namespaces(int flags, struct task_struct *tsk);
>  void get_task_namespaces(struct task_struct *tsk);
>  void free_nsproxy(struct nsproxy *ns);
> +#ifdef CONFIG_RCFS
> +struct nsproxy *find_nsproxy(struct nsproxy *ns);
> +int namespaces_init(void);
> +#else
> +static inline int namespaces_init(void) { return 0;}
> +#endif
> 
>  static inline void put_nsproxy(struct nsproxy *ns)
>  {
> diff -puN /dev/null include/linux/rcfs.h
> --- /dev/null	2006-02-25 03:06:56.000000000 +0530
> +++ linux-2.6.20-vatsa/include/linux/rcfs.h	2007-03-01 14:20:47.000000000 +0530
> @@ -0,0 +1,72 @@
> +#ifndef _LINUX_RCFS_H
> +#define _LINUX_RCFS_H
> +
> +#ifdef CONFIG_RCFS
> +
> +/* struct cftype:
> + *
> + * The files in the container filesystem mostly have a very simple read/write
> + * handling, some common function will take care of it. Nevertheless some cases
> + * (read tasks) are special and therefore I define this structure for every
> + * kind of file.
> + *
> + *
> + * When reading/writing to a file:
> + *	- the container to use in file->f_dentry->d_parent->d_fsdata
> + *	- the 'cftype' of the file is file->f_dentry->d_fsdata
> + */
> +
> +struct inode;
> +#define MAX_CFTYPE_NAME 64
> +struct cftype {
> +	/* By convention, the name should begin with the name of the
> +	 * subsystem, followed by a period */
> +	char name[MAX_CFTYPE_NAME];
> +	int private;
> +	int (*open) (struct inode *inode, struct file *file);
> +	ssize_t (*read) (struct nsproxy *ns, struct cftype *cft,
> +			 struct file *file,
> +			 char __user *buf, size_t nbytes, loff_t *ppos);
> +	ssize_t (*write) (struct nsproxy *ns, struct cftype *cft,
> +			  struct file *file,
> +			  const char __user *buf, size_t nbytes, loff_t *ppos);
> +	int (*release) (struct inode *inode, struct file *file);
> +};
> +
> +/* resource control subsystem type. See Documentation/rcfs.txt for details */
> +
> +struct rc_subsys {
> +	int (*create)(struct rc_subsys *ss, struct nsproxy *ns,
> +			 struct nsproxy *parent);
> +	void (*destroy)(struct rc_subsys *ss, struct nsproxy *ns);
> +	int (*can_attach)(struct rc_subsys *ss, struct nsproxy *ns,
> +				 struct task_struct *tsk);
> +	void (*attach)(struct rc_subsys *ss, void *new, void *old,
> +				 struct task_struct *tsk);
> +	int (*populate)(struct rc_subsys *ss, struct dentry *d);
> +	int subsys_id;
> +	int active;
> +
> +#define MAX_CONTAINER_TYPE_NAMELEN 32
> +	const char *name;
> +
> +	/* Protected by RCU */
> +	int hierarchy;
> +
> +	struct list_head sibling;
> +};
> +
> +int rc_register_subsys(struct rc_subsys *subsys);
> +/* Add a new file to the given container directory. Should only be
> + * called by subsystems from within a populate() method */
> +int rcfs_add_file(struct dentry *d, const struct cftype *cft);
> +extern int rcfs_init(void);
> +
> +#else
> +
> +static inline int rcfs_init(void) { return 0; }
> +
> +#endif
> +
> +
> +#endif
> diff -puN init/Kconfig~rcfs init/Kconfig
> --- linux-2.6.20/init/Kconfig~rcfs	2007-03-01 14:20:47.000000000 +0530
> +++ linux-2.6.20-vatsa/init/Kconfig	2007-03-01 16:52:50.000000000 +0530
> @@ -238,6 +238,28 @@ config IKCONFIG_PROC
>  	  This option enables access to the kernel configuration file
>  	  through /proc/config.gz.
> 
> +config RCFS
> +	bool "Resource control file system support"
> +	default n
> +	help
> +	  This option will let you create and manage resource containers,
> +	  which can be used to aggregate multiple processes, e.g. for
> +	  the purposes of resource tracking.
> +
> +	  Say N if unsure
> +
> +config MAX_RC_SUBSYS
> +       int "Number of resource control subsystems to support"
> +       depends on RCFS
> +       range 1 255
> +       default 8
> +
> +config MAX_RC_HIERARCHIES
> +       int "Number of rcfs hierarchies to support"
> +       depends on RCFS
> +       range 2 255
> +       default 4
> +
>  config CPUSETS
>  	bool "Cpuset support"
>  	depends on SMP
> diff -puN init/main.c~rcfs init/main.c
> --- linux-2.6.20/init/main.c~rcfs	2007-03-01 14:20:47.000000000 +0530
> +++ linux-2.6.20-vatsa/init/main.c	2007-03-01 14:20:47.000000000 +0530
> @@ -52,6 +52,7 @@
>  #include <linux/lockdep.h>
>  #include <linux/pid_namespace.h>
>  #include <linux/device.h>
> +#include <linux/rcfs.h>
> 
>  #include <asm/io.h>
>  #include <asm/bugs.h>
> @@ -512,6 +513,7 @@ asmlinkage void __init start_kernel(void
>  	setup_per_cpu_areas();
>  	smp_prepare_boot_cpu();	/* arch-specific boot-cpu hooks */
> 
> +	namespaces_init();
>  	/*
>  	 * Set up the scheduler prior starting any interrupts (such as the
>  	 * timer interrupt). Full topology setup happens at smp_init()
> @@ -608,6 +610,7 @@ asmlinkage void __init start_kernel(void
>  #ifdef CONFIG_PROC_FS
>  	proc_root_init();
>  #endif
> +	rcfs_init();
>  	cpuset_init();
>  	taskstats_init_early();
>  	delayacct_init();
> diff -puN kernel/Makefile~rcfs kernel/Makefile
> --- linux-2.6.20/kernel/Makefile~rcfs	2007-03-01 14:20:47.000000000 +0530
> +++ linux-2.6.20-vatsa/kernel/Makefile	2007-03-01 16:52:50.000000000 +0530
> @@ -50,6 +50,7 @@ obj-$(CONFIG_RELAY) += relay.o
>  obj-$(CONFIG_UTS_NS) += utsname.o
>  obj-$(CONFIG_TASK_DELAY_ACCT) += delayacct.o
>  obj-$(CONFIG_TASKSTATS) += taskstats.o tsacct.o
> +obj-$(CONFIG_RCFS) += rcfs.o
> 
>  ifneq ($(CONFIG_SCHED_NO_NO_OMIT_FRAME_POINTER),y)
>  # According to Alan Modra <alan@linuxcare.com.au>, the -fno-omit-frame-pointer is
> diff -puN kernel/nsproxy.c~rcfs kernel/nsproxy.c
> --- linux-2.6.20/kernel/nsproxy.c~rcfs	2007-03-01 14:20:47.000000000 +0530
> +++ linux-2.6.20-vatsa/kernel/nsproxy.c	2007-03-01 14:20:47.000000000 +0530
> @@ -23,6 +23,11 @@
> 
>  struct nsproxy init_nsproxy = INIT_NSPROXY(init_nsproxy);
> 
> +#ifdef CONFIG_RCFS
> +static LIST_HEAD(nslisthead);
> +static DEFINE_SPINLOCK(nslistlock);
> +#endif
> +
>  static inline void get_nsproxy(struct nsproxy *ns)
>  {
>  	atomic_inc(&ns->count);
> @@ -71,6 +76,12 @@ struct nsproxy *dup_namespaces(struct ns
>  			get_pid_ns(ns->pid_ns);
>  	}
> 
> +#ifdef CONFIG_RCFS
> +	spin_lock(&nslistlock);
> +	list_add(&ns->list, &nslisthead);
> +	spin_unlock(&nslistlock);
> +#endif
> +
>  	return ns;
>  }
> 
> @@ -145,5 +156,44 @@ void free_nsproxy(struct nsproxy *ns)
>  		put_ipc_ns(ns-&