Message-ID: <20150217145924.GE11861@treble.redhat.com>
Date:	Tue, 17 Feb 2015 08:59:24 -0600
From:	Josh Poimboeuf <jpoimboe@...hat.com>
To:	Jiri Slaby <jslaby@...e.cz>
Cc:	Seth Jennings <sjenning@...hat.com>, Jiri Kosina <jkosina@...e.cz>,
	Vojtech Pavlik <vojtech@...e.cz>,
	Masami Hiramatsu <masami.hiramatsu.pt@...achi.com>,
	live-patching@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [RFC PATCH 6/9] livepatch: create per-task consistency model

On Sat, Feb 14, 2015 at 12:40:01PM +0100, Jiri Slaby wrote:
> On 02/09/2015, 06:31 PM, Josh Poimboeuf wrote:
> > Add a basic per-task consistency model.  This is the foundation which
> > will eventually enable us to patch those ~10% of security patches which
> > change function prototypes and/or data semantics.
> > 
> > When a patch is enabled, livepatch enters into a transition state where
> > tasks are converging from the old universe to the new universe.  If a
> > given task isn't using any of the patched functions, it's switched to
> > the new universe.  Once all the tasks have been converged to the new
> > universe, patching is complete.
> > 
> > The same sequence occurs when a patch is disabled, except the tasks
> > converge from the new universe to the old universe.
> > 
> > The /sys/kernel/livepatch/<patch>/transition file shows whether a patch
> > is in transition.  Only a single patch (the topmost patch on the stack)
> > can be in transition at a given time.  A patch can remain in the
> > transition state indefinitely, if any of the tasks are stuck in the
> > previous universe.
> > 
> > A transition can be reversed and effectively canceled by writing the
> > opposite value to the /sys/kernel/livepatch/<patch>/enabled file while
> > the transition is in progress.  Then all the tasks will attempt to
> > converge back to the original universe.
> > 
> > Signed-off-by: Josh Poimboeuf <jpoimboe@...hat.com>
> > ---
> >  include/linux/livepatch.h     |  18 ++-
> >  include/linux/sched.h         |   3 +
> >  kernel/fork.c                 |   2 +
> >  kernel/livepatch/Makefile     |   2 +-
> >  kernel/livepatch/core.c       |  71 ++++++----
> >  kernel/livepatch/patch.c      |  34 ++++-
> >  kernel/livepatch/patch.h      |   1 +
> >  kernel/livepatch/transition.c | 300 ++++++++++++++++++++++++++++++++++++++++++
> >  kernel/livepatch/transition.h |  16 +++
> >  kernel/sched/core.c           |   2 +
> >  10 files changed, 423 insertions(+), 26 deletions(-)
> >  create mode 100644 kernel/livepatch/transition.c
> >  create mode 100644 kernel/livepatch/transition.h
> > 
> > diff --git a/include/linux/livepatch.h b/include/linux/livepatch.h
> > index 0e65b4d..b8c2f15 100644
> > --- a/include/linux/livepatch.h
> > +++ b/include/linux/livepatch.h
> > @@ -40,6 +40,7 @@
> >   * @old_size:	size of the old function
> >   * @new_size:	size of the new function
> >   * @patched:	the func has been added to the klp_ops list
> > + * @transition:	the func is currently being applied or reverted
> >   */
> >  struct klp_func {
> >  	/* external */
> > @@ -60,6 +61,7 @@ struct klp_func {
> >  	struct list_head stack_node;
> >  	unsigned long old_size, new_size;
> >  	int patched;
> > +	int transition;
> >  };
> >  
> >  /**
> > @@ -128,6 +130,20 @@ extern int klp_unregister_patch(struct klp_patch *);
> >  extern int klp_enable_patch(struct klp_patch *);
> >  extern int klp_disable_patch(struct klp_patch *);
> >  
> > -#endif /* CONFIG_LIVEPATCH */
> > +extern int klp_universe_goal;
> > +
> > +static inline void klp_update_task_universe(struct task_struct *t)
> > +{
> > +	/* corresponding smp_wmb() is in klp_set_universe_goal() */
> > +	smp_rmb();
> > +
> > +	t->klp_universe = klp_universe_goal;
> > +}
> > +
> > +#else /* !CONFIG_LIVEPATCH */
> > +
> > +static inline void klp_update_task_universe(struct task_struct *t) {}
> > +
> > +#endif /* !CONFIG_LIVEPATCH */
> >  
> >  #endif /* _LINUX_LIVEPATCH_H_ */
> > diff --git a/include/linux/sched.h b/include/linux/sched.h
> > index 8db31ef..a95e59a 100644
> > --- a/include/linux/sched.h
> > +++ b/include/linux/sched.h
> > @@ -1701,6 +1701,9 @@ struct task_struct {
> >  #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
> >  	unsigned long	task_state_change;
> >  #endif
> > +#ifdef CONFIG_LIVEPATCH
> > +	int klp_universe;
> > +#endif
> >  };
> >  
> >  /* Future-safe accessor for struct task_struct's cpus_allowed. */
> > diff --git a/kernel/fork.c b/kernel/fork.c
> > index 4dc2dda..1dcbebe 100644
> > --- a/kernel/fork.c
> > +++ b/kernel/fork.c
> > @@ -74,6 +74,7 @@
> >  #include <linux/uprobes.h>
> >  #include <linux/aio.h>
> >  #include <linux/compiler.h>
> > +#include <linux/livepatch.h>
> >  
> >  #include <asm/pgtable.h>
> >  #include <asm/pgalloc.h>
> > @@ -1538,6 +1539,7 @@ static struct task_struct *copy_process(unsigned long clone_flags,
> >  	total_forks++;
> >  	spin_unlock(&current->sighand->siglock);
> >  	syscall_tracepoint_update(p);
> > +	klp_update_task_universe(p);
> >  	write_unlock_irq(&tasklist_lock);
> >  
> >  	proc_fork_connector(p);
> > diff --git a/kernel/livepatch/Makefile b/kernel/livepatch/Makefile
> > index e136dad..2b8bdb1 100644
> > --- a/kernel/livepatch/Makefile
> > +++ b/kernel/livepatch/Makefile
> > @@ -1,3 +1,3 @@
> >  obj-$(CONFIG_LIVEPATCH) += livepatch.o
> >  
> > -livepatch-objs := core.o patch.o
> > +livepatch-objs := core.o patch.o transition.o
> > diff --git a/kernel/livepatch/core.c b/kernel/livepatch/core.c
> > index 85d4ef7..790dc10 100644
> > --- a/kernel/livepatch/core.c
> > +++ b/kernel/livepatch/core.c
> > @@ -28,14 +28,17 @@
> >  #include <linux/kallsyms.h>
> >  
> >  #include "patch.h"
> > +#include "transition.h"
> >  
> >  /*
> > - * The klp_mutex protects the global lists and state transitions of any
> > - * structure reachable from them.  References to any structure must be obtained
> > - * under mutex protection (except in klp_ftrace_handler(), which uses RCU to
> > - * ensure it gets consistent data).
> > + * The klp_mutex is a coarse lock which serializes access to klp data.  All
> > + * accesses to klp-related variables and structures must have mutex protection,
> > + * except within the following functions which carefully avoid the need for it:
> > + *
> > + * - klp_ftrace_handler()
> > + * - klp_update_task_universe()
> >   */
> > -static DEFINE_MUTEX(klp_mutex);
> > +DEFINE_MUTEX(klp_mutex);
> >  
> >  static LIST_HEAD(klp_patches);
> >  
> > @@ -67,7 +70,6 @@ static void klp_find_object_module(struct klp_object *obj)
> >  	mutex_unlock(&module_mutex);
> >  }
> >  
> > -/* klp_mutex must be held by caller */
> >  static bool klp_is_patch_registered(struct klp_patch *patch)
> >  {
> >  	struct klp_patch *mypatch;
> > @@ -285,18 +287,17 @@ static int klp_write_object_relocations(struct module *pmod,
> >  
> >  static int __klp_disable_patch(struct klp_patch *patch)
> >  {
> > -	struct klp_object *obj;
> > +	if (klp_transition_patch)
> > +		return -EBUSY;
> >  
> >  	/* enforce stacking: only the last enabled patch can be disabled */
> >  	if (!list_is_last(&patch->list, &klp_patches) &&
> >  	    list_next_entry(patch, list)->enabled)
> >  		return -EBUSY;
> >  
> > -	pr_notice("disabling patch '%s'\n", patch->mod->name);
> > -
> > -	for (obj = patch->objs; obj->funcs; obj++)
> > -		if (obj->patched)
> > -			klp_unpatch_object(obj);
> > +	klp_init_transition(patch, KLP_UNIVERSE_NEW);
> > +	klp_start_transition(KLP_UNIVERSE_OLD);
> > +	klp_try_complete_transition();
> >  
> >  	patch->enabled = 0;
> >  
> > @@ -340,6 +341,9 @@ static int __klp_enable_patch(struct klp_patch *patch)
> >  	struct klp_object *obj;
> >  	int ret;
> >  
> > +	if (klp_transition_patch)
> > +		return -EBUSY;
> > +
> >  	if (WARN_ON(patch->enabled))
> >  		return -EINVAL;
> >  
> > @@ -351,7 +355,7 @@ static int __klp_enable_patch(struct klp_patch *patch)
> >  	pr_notice_once("tainting kernel with TAINT_LIVEPATCH\n");
> >  	add_taint(TAINT_LIVEPATCH, LOCKDEP_STILL_OK);
> >  
> > -	pr_notice("enabling patch '%s'\n", patch->mod->name);
> > +	klp_init_transition(patch, KLP_UNIVERSE_OLD);
> >  
> >  	for (obj = patch->objs; obj->funcs; obj++) {
> >  		klp_find_object_module(obj);
> > @@ -360,17 +364,24 @@ static int __klp_enable_patch(struct klp_patch *patch)
> >  			continue;
> >  
> >  		ret = klp_patch_object(obj);
> > -		if (ret)
> > -			goto unregister;
> > +		if (ret) {
> > +			pr_warn("failed to enable patch '%s'\n",
> > +				patch->mod->name);
> > +
> > +			klp_unpatch_objects(patch);
> > +			klp_complete_transition();
> > +
> > +			return ret;
> > +		}
> >  	}
> >  
> > +	klp_start_transition(KLP_UNIVERSE_NEW);
> > +
> > +	klp_try_complete_transition();
> > +
> >  	patch->enabled = 1;
> >  
> >  	return 0;
> > -
> > -unregister:
> > -	WARN_ON(__klp_disable_patch(patch));
> > -	return ret;
> >  }
> >  
> >  /**
> > @@ -407,6 +418,7 @@ EXPORT_SYMBOL_GPL(klp_enable_patch);
> >   * /sys/kernel/livepatch
> >   * /sys/kernel/livepatch/<patch>
> >   * /sys/kernel/livepatch/<patch>/enabled
> > + * /sys/kernel/livepatch/<patch>/transition
> >   * /sys/kernel/livepatch/<patch>/<object>
> >   * /sys/kernel/livepatch/<patch>/<object>/<func>
> >   */
> > @@ -435,7 +447,9 @@ static ssize_t enabled_store(struct kobject *kobj, struct kobj_attribute *attr,
> >  		goto err;
> >  	}
> >  
> > -	if (val) {
> > +	if (klp_transition_patch == patch) {
> > +		klp_reverse_transition();
> > +	} else if (val) {
> >  		ret = __klp_enable_patch(patch);
> >  		if (ret)
> >  			goto err;
> > @@ -463,9 +477,21 @@ static ssize_t enabled_show(struct kobject *kobj,
> >  	return snprintf(buf, PAGE_SIZE-1, "%d\n", patch->enabled);
> >  }
> >  
> > +static ssize_t transition_show(struct kobject *kobj,
> > +			       struct kobj_attribute *attr, char *buf)
> > +{
> > +	struct klp_patch *patch;
> > +
> > +	patch = container_of(kobj, struct klp_patch, kobj);
> > +	return snprintf(buf, PAGE_SIZE-1, "%d\n",
> > +			klp_transition_patch == patch);
> > +}
> > +
> >  static struct kobj_attribute enabled_kobj_attr = __ATTR_RW(enabled);
> > +static struct kobj_attribute transition_kobj_attr = __ATTR_RO(transition);
> >  static struct attribute *klp_patch_attrs[] = {
> >  	&enabled_kobj_attr.attr,
> > +	&transition_kobj_attr.attr,
> >  	NULL
> >  };
> >  
> > @@ -543,6 +569,7 @@ static int klp_init_func(struct klp_object *obj, struct klp_func *func)
> >  {
> >  	INIT_LIST_HEAD(&func->stack_node);
> >  	func->patched = 0;
> > +	func->transition = 0;
> >  
> >  	return kobject_init_and_add(&func->kobj, &klp_ktype_func,
> >  				    obj->kobj, func->old_name);
> > @@ -725,7 +752,7 @@ static void klp_module_notify_coming(struct klp_patch *patch,
> >  	if (ret)
> >  		goto err;
> >  
> > -	if (!patch->enabled)
> > +	if (!patch->enabled && klp_transition_patch != patch)
> >  		return;
> >  
> >  	pr_notice("applying patch '%s' to loading module '%s'\n",
> > @@ -746,7 +773,7 @@ static void klp_module_notify_going(struct klp_patch *patch,
> >  	struct module *pmod = patch->mod;
> >  	struct module *mod = obj->mod;
> >  
> > -	if (!patch->enabled)
> > +	if (!patch->enabled && klp_transition_patch != patch)
> >  		goto free;
> >  
> >  	pr_notice("reverting patch '%s' on unloading module '%s'\n",
> > diff --git a/kernel/livepatch/patch.c b/kernel/livepatch/patch.c
> > index 281fbca..f12256b 100644
> > --- a/kernel/livepatch/patch.c
> > +++ b/kernel/livepatch/patch.c
> > @@ -24,6 +24,7 @@
> >  #include <linux/slab.h>
> >  
> >  #include "patch.h"
> > +#include "transition.h"
> >  
> >  static LIST_HEAD(klp_ops);
> >  
> > @@ -38,14 +39,34 @@ static void notrace klp_ftrace_handler(unsigned long ip,
> >  	ops = container_of(fops, struct klp_ops, fops);
> >  
> >  	rcu_read_lock();
> > +
> >  	func = list_first_or_null_rcu(&ops->func_stack, struct klp_func,
> >  				      stack_node);
> > -	rcu_read_unlock();
> >  
> >  	if (WARN_ON_ONCE(!func))
> > -		return;
> > +		goto unlock;
> > +
> > +	if (unlikely(func->transition)) {
> > +		/* corresponding smp_wmb() is in klp_init_transition() */
> > +		smp_rmb();
> > +
> > +		if (current->klp_universe == KLP_UNIVERSE_OLD) {
> > +			/*
> > +			 * Use the previously patched version of the function.
> > +			 * If no previous patches exist, use the original
> > +			 * function.
> > +			 */
> > +			func = list_entry_rcu(func->stack_node.next,
> > +					      struct klp_func, stack_node);
> > +
> > +			if (&func->stack_node == &ops->func_stack)
> > +				goto unlock;
> > +		}
> > +	}
> >  
> >  	klp_arch_set_pc(regs, (unsigned long)func->new_func);
> > +unlock:
> > +	rcu_read_unlock();
> >  }
> >  
> >  struct klp_ops *klp_find_ops(unsigned long old_addr)
> > @@ -174,3 +195,12 @@ int klp_patch_object(struct klp_object *obj)
> >  
> >  	return 0;
> >  }
> > +
> > +void klp_unpatch_objects(struct klp_patch *patch)
> > +{
> > +	struct klp_object *obj;
> > +
> > +	for (obj = patch->objs; obj->funcs; obj++)
> > +		if (obj->patched)
> > +			klp_unpatch_object(obj);
> > +}
> > diff --git a/kernel/livepatch/patch.h b/kernel/livepatch/patch.h
> > index bb34bd3..1648259 100644
> > --- a/kernel/livepatch/patch.h
> > +++ b/kernel/livepatch/patch.h
> > @@ -23,3 +23,4 @@ struct klp_ops *klp_find_ops(unsigned long old_addr);
> >  
> >  extern int klp_patch_object(struct klp_object *obj);
> >  extern void klp_unpatch_object(struct klp_object *obj);
> > +extern void klp_unpatch_objects(struct klp_patch *patch);
> > diff --git a/kernel/livepatch/transition.c b/kernel/livepatch/transition.c
> > new file mode 100644
> > index 0000000..2630296
> > --- /dev/null
> > +++ b/kernel/livepatch/transition.c
> > @@ -0,0 +1,300 @@
> > +/*
> > + * transition.c - Kernel Live Patching transition functions
> > + *
> > + * Copyright (C) 2015 Josh Poimboeuf <jpoimboe@...hat.com>
> > + *
> > + * This program is free software; you can redistribute it and/or
> > + * modify it under the terms of the GNU General Public License
> > + * as published by the Free Software Foundation; either version 2
> > + * of the License, or (at your option) any later version.
> > + *
> > + * This program is distributed in the hope that it will be useful,
> > + * but WITHOUT ANY WARRANTY; without even the implied warranty of
> > + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
> > + * GNU General Public License for more details.
> > + *
> > + * You should have received a copy of the GNU General Public License
> > + * along with this program; if not, see <http://www.gnu.org/licenses/>.
> > + */
> > +
> > +#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
> > +
> > +#include <linux/cpu.h>
> > +#include <asm/stacktrace.h>
> > +#include "../sched/sched.h"
> > +
> > +#include "patch.h"
> > +#include "transition.h"
> > +
> > +static void klp_transition_work_fn(struct work_struct *);
> > +static DECLARE_DELAYED_WORK(klp_transition_work, klp_transition_work_fn);
> > +
> > +struct klp_patch *klp_transition_patch;
> > +
> > +int klp_universe_goal = KLP_UNIVERSE_UNDEFINED;
> > +
> > +static void klp_set_universe_goal(int universe)
> > +{
> > +	klp_universe_goal = universe;
> > +
> > +	/* corresponding smp_rmb() is in klp_update_task_universe() */
> > +	smp_wmb();
> > +}
> > +
> > +/*
> > + * The transition to the universe goal is complete.  Clean up the data
> > + * structures.
> > + */
> > +void klp_complete_transition(void)
> > +{
> > +	struct klp_object *obj;
> > +	struct klp_func *func;
> > +
> > +	for (obj = klp_transition_patch->objs; obj->funcs; obj++)
> > +		for (func = obj->funcs; func->old_name; func++)
> > +			func->transition = 0;
> > +
> > +	klp_transition_patch = NULL;
> > +}
> > +
> > +static int klp_stacktrace_address_verify_func(struct klp_func *func,
> > +					      unsigned long address)
> > +{
> > +	unsigned long func_addr, func_size;
> > +
> > +	if (klp_universe_goal == KLP_UNIVERSE_OLD) {
> > +		 /* check the to-be-unpatched function (the func itself) */
> > +		func_addr = (unsigned long)func->new_func;
> > +		func_size = func->new_size;
> > +	} else {
> > +		/* check the to-be-patched function (previous func) */
> > +		struct klp_ops *ops;
> > +
> > +		ops = klp_find_ops(func->old_addr);
> > +
> > +		if (list_is_singular(&ops->func_stack)) {
> > +			/* original function */
> > +			func_addr = func->old_addr;
> > +			func_size = func->old_size;
> > +		} else {
> > +			/* previously patched function */
> > +			struct klp_func *prev;
> > +
> > +			prev = list_next_entry(func, stack_node);
> > +			func_addr = (unsigned long)prev->new_func;
> > +			func_size = prev->new_size;
> > +		}
> > +	}
> > +
> > +	if (address >= func_addr && address < func_addr + func_size)
> > +		return -1;
> > +
> > +	return 0;
> > +}
> > +
> > +/*
> > + * Determine whether the given return address on the stack is within a
> > + * to-be-patched or to-be-unpatched function.
> > + */
> > +static void klp_stacktrace_address_verify(void *data, unsigned long address,
> > +					  int reliable)
> > +{
> > +	struct klp_object *obj;
> > +	struct klp_func *func;
> > +	int *ret = data;
> > +
> > +	if (*ret)
> > +		return;
> > +
> > +	for (obj = klp_transition_patch->objs; obj->funcs; obj++) {
> > +		if (!obj->patched)
> > +			continue;
> > +		for (func = obj->funcs; func->old_name; func++) {
> > +			if (klp_stacktrace_address_verify_func(func, address)) {
> > +				*ret = -1;
> > +				return;
> > +			}
> > +		}
> > +	}
> > +}
> > +
> > +static int klp_stacktrace_stack(void *data, char *name)
> > +{
> > +	return 0;
> > +}
> > +
> > +static const struct stacktrace_ops klp_stacktrace_ops = {
> > +	.address = klp_stacktrace_address_verify,
> > +	.stack = klp_stacktrace_stack,
> > +	.walk_stack = print_context_stack_bp,
> > +};
> > +
> > +/*
> > + * Try to safely transition a task to the universe goal.  If the task is
> > + * currently running or is sleeping on a to-be-patched or to-be-unpatched
> > + * function, return false.
> > + */
> > +static bool klp_transition_task(struct task_struct *t)
> > +{
> > +	struct rq *rq;
> > +	unsigned long flags;
> > +	int ret;
> > +	bool success = false;
> > +
> > +	if (t->klp_universe == klp_universe_goal)
> > +		return true;
> > +
> > +	rq = task_rq_lock(t, &flags);
> > +
> > +	if (task_running(rq, t) && t != current) {
> > +		pr_debug("%s: pid %d (%s) is running\n", __func__, t->pid,
> > +			 t->comm);
> > +		goto done;
> > +	}
> > +
> > +	ret = 0;
> > +	dump_trace(t, NULL, NULL, 0, &klp_stacktrace_ops, &ret);
> > +	if (ret) {
> > +		pr_debug("%s: pid %d (%s) is sleeping on a patched function\n",
> > +			 __func__, t->pid, t->comm);
> > +		goto done;
> > +	}
> > +
> > +	klp_update_task_universe(t);
> > +
> > +	success = true;
> > +done:
> > +	task_rq_unlock(rq, t, &flags);
> > +	return success;
> > +}
> > +
> > +/*
> > + * Try to transition all tasks to the universe goal.  If any tasks are still
> > + * stuck in the original universe, schedule a retry.
> > + */
> > +void klp_try_complete_transition(void)
> > +{
> > +	unsigned int cpu;
> > +	struct task_struct *g, *t;
> > +	bool complete = true;
> > +
> > +	/* try to transition all normal tasks */
> > +	read_lock(&tasklist_lock);
> > +	for_each_process_thread(g, t)
> > +		if (!klp_transition_task(t))
> > +			complete = false;
> > +	read_unlock(&tasklist_lock);
> > +
> > +	/* try to transition the idle "swapper" tasks */
> > +	get_online_cpus();
> > +	for_each_online_cpu(cpu)
> > +		if (!klp_transition_task(idle_task(cpu)))
> > +			complete = false;
> > +	put_online_cpus();
> > +
> > +	/* if not complete, try again later */
> > +	if (!complete) {
> > +		schedule_delayed_work(&klp_transition_work,
> > +				      round_jiffies_relative(HZ));
> > +		return;
> > +	}
> > +
> > +	/* success! unpatch obsolete functions and do some cleanup */
> > +
> > +	if (klp_universe_goal == KLP_UNIVERSE_OLD) {
> > +		klp_unpatch_objects(klp_transition_patch);
> > +
> > +		/* prevent ftrace handler from reading old func->transition */
> > +		synchronize_rcu();
> > +	}
> > +
> > +	pr_notice("'%s': %s complete\n", klp_transition_patch->mod->name,
> > +		  klp_universe_goal == KLP_UNIVERSE_NEW ? "patching" :
> > +							  "unpatching");
> > +
> > +	klp_complete_transition();
> > +}
> > +
> > +static void klp_transition_work_fn(struct work_struct *work)
> > +{
> > +	mutex_lock(&klp_mutex);
> > +
> > +	if (klp_transition_patch)
> > +		klp_try_complete_transition();
> > +
> > +	mutex_unlock(&klp_mutex);
> > +}
> > +
> > +/*
> > + * Start the transition to the specified universe so tasks can begin switching
> > + * to it.
> > + */
> > +void klp_start_transition(int universe)
> > +{
> > +	if (WARN_ON(klp_universe_goal == universe))
> > +		return;
> > +
> > +	pr_notice("'%s': %s...\n", klp_transition_patch->mod->name,
> > +		  universe == KLP_UNIVERSE_NEW ? "patching" : "unpatching");
> > +
> > +	klp_set_universe_goal(universe);
> > +}
> > +
> > +/*
> > + * Can be called in the middle of an existing transition to reverse the
> > + * direction of the universe goal.  This can be done to effectively cancel an
> > + * existing enable or disable operation if there are any tasks which are stuck
> > + * in the original universe.
> > + */
> > +void klp_reverse_transition(void)
> > +{
> > +	struct klp_patch *patch = klp_transition_patch;
> > +
> > +	klp_start_transition(!klp_universe_goal);
> > +	klp_try_complete_transition();
> > +
> > +	patch->enabled = !patch->enabled;
> > +}
> > +
> > +/*
> > + * Reset the universe goal and all tasks to the starting universe, and set all
> > + * func->transition's to 1 to prepare for patching.
> > + */
> > +void klp_init_transition(struct klp_patch *patch, int universe)
> > +{
> > +	struct task_struct *g, *t;
> > +	unsigned int cpu;
> > +	struct klp_object *obj;
> > +	struct klp_func *func;
> > +
> > +	klp_transition_patch = patch;
> > +
> > +	/*
> > +	 * If the previous transition was in the opposite direction, we may
> > +	 * already be in the requested initial universe.
> > +	 */
> > +	if (klp_universe_goal == universe)
> > +		goto init_funcs;
> > +
> > +	klp_set_universe_goal(universe);
> > +
> > +	/* init all normal task universes */
> > +	read_lock(&tasklist_lock);
> > +	for_each_process_thread(g, t)
> > +		klp_update_task_universe(t);
> > +	read_unlock(&tasklist_lock);
> > +
> > +	/* init all idle "swapper" task universes */
> > +	get_online_cpus();
> > +	for_each_online_cpu(cpu)
> > +		klp_update_task_universe(idle_task(cpu));
> > +	put_online_cpus();
> > +
> > +init_funcs:
> > +	/* corresponding smp_rmb() is in klp_ftrace_handler() */
> > +	smp_wmb();
> > +
> > +	for (obj = patch->objs; obj->funcs; obj++)
> > +		for (func = obj->funcs; func->old_name; func++)
> > +			func->transition = 1;
> 
> So I finally got to review of this one. I have only two concerns:
> 1) it removes the ability for the user to use 'no consistency model'.
> But you don't need to worry about this, I plan to implement this as soon
> as you send v2 of these.

Ok, sounds good.

> 2) How is this 'transition = 1' store above guaranteed to reach other
> CPUs before you start registering ftrace handlers? The CPUs need not see
> the update when some handler is already invoked before start_transition
> AFAICS.

Yeah, I think the order of the 'transition = 1' and adding the func to
the ops stack list should be enforced.  Also I'll probably rework the
barriers a little bit in v2 so that they're more explicit and better
commented.

-- 
Josh
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

Powered by blists - more mailing lists