Revision eca6be566d47029f945a5f8e1c94d374e31df2ca authored by Sean Christopherson on 15 February 2019, 20:48:40 UTC, committed by Paolo Bonzini on 15 March 2019, 18:24:33 UTC
The series to add memcg accounting to KVM allocations[1] states: There are many KVM kernel memory allocations which are tied to the life of the VM process and should be charged to the VM process's cgroup. While it is correct to account KVM kernel allocations to the cgroup of the process that created the VM, it's technically incorrect to state that the KVM kernel memory allocations are tied to the life of the VM process. This is because the VM itself, i.e. struct kvm, is not tied to the life of the process which created it, rather it is tied to the life of its associated file descriptor. In other words, kvm_destroy_vm() is not invoked until fput() decrements its associated file's refcount to zero. A simple example is to fork() in Qemu and have the child sleep indefinitely; kvm_destroy_vm() isn't called until Qemu closes its file descriptor *and* the rogue child is killed. The allocations are guaranteed to be *accounted* to the process which created the VM, but only because KVM's per-{VM,vCPU} ioctls reject the ioctl() with -EIO if kvm->mm != current->mm. I.e. the child can keep the VM "alive" but can't do anything useful with its reference. Note that because 'struct kvm' also holds a reference to the mm_struct of its owner, the above behavior also applies to userspace allocations. Given that mucking with a VM's file descriptor can lead to subtle and undesirable behavior, e.g. memcg charges persisting after a VM is shut down, explicitly document a VM's lifecycle and its impact on the VM's resources. Alternatively, KVM could aggressively free resources when the creating process exits, e.g. via mmu_notifier->release(). However, mmu_notifier isn't guaranteed to be available, and freeing resources when the creator exits is likely to be error prone and fragile as KVM would need to ensure that it only freed resources that are truly out of reach. In practice, the existing behavior shouldn't be problematic as a properly configured system will prevent a child process from being moved out of the appropriate cgroup hierarchy, i.e. prevent hiding the process from the OOM killer, and will prevent an unprivileged user from being able to to hold a reference to struct kvm via another method, e.g. debugfs. [1]https://patchwork.kernel.org/patch/10806707/ Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
1 parent c7a0e83
klist.c
/*
* klist.c - Routines for manipulating klists.
*
* Copyright (C) 2005 Patrick Mochel
*
* This file is released under the GPL v2.
*
* This klist interface provides a couple of structures that wrap around
* struct list_head to provide explicit list "head" (struct klist) and list
* "node" (struct klist_node) objects. For struct klist, a spinlock is
* included that protects access to the actual list itself. struct
* klist_node provides a pointer to the klist that owns it and a kref
* reference count that indicates the number of current users of that node
* in the list.
*
* The entire point is to provide an interface for iterating over a list
* that is safe and allows for modification of the list during the
* iteration (e.g. insertion and removal), including modification of the
* current node on the list.
*
* It works using a 3rd object type - struct klist_iter - that is declared
* and initialized before an iteration. klist_next() is used to acquire the
* next element in the list. It returns NULL if there are no more items.
* Internally, that routine takes the klist's lock, decrements the
* reference count of the previous klist_node and increments the count of
* the next klist_node. It then drops the lock and returns.
*
* There are primitives for adding and removing nodes to/from a klist.
* When deleting, klist_del() will simply decrement the reference count.
* Only when the count goes to 0 is the node removed from the list.
* klist_remove() will try to delete the node from the list and block until
* it is actually removed. This is useful for objects (like devices) that
* have been removed from the system and must be freed (but must wait until
* all accessors have finished).
*/
#include <linux/klist.h>
#include <linux/export.h>
#include <linux/sched.h>
/*
* Use the lowest bit of n_klist to mark deleted nodes and exclude
* dead ones from iteration.
*/
#define KNODE_DEAD 1LU
#define KNODE_KLIST_MASK ~KNODE_DEAD
static struct klist *knode_klist(struct klist_node *knode)
{
return (struct klist *)
((unsigned long)knode->n_klist & KNODE_KLIST_MASK);
}
static bool knode_dead(struct klist_node *knode)
{
return (unsigned long)knode->n_klist & KNODE_DEAD;
}
static void knode_set_klist(struct klist_node *knode, struct klist *klist)
{
knode->n_klist = klist;
/* no knode deserves to start its life dead */
WARN_ON(knode_dead(knode));
}
static void knode_kill(struct klist_node *knode)
{
/* and no knode should die twice ever either, see we're very humane */
WARN_ON(knode_dead(knode));
*(unsigned long *)&knode->n_klist |= KNODE_DEAD;
}
/**
* klist_init - Initialize a klist structure.
* @k: The klist we're initializing.
* @get: The get function for the embedding object (NULL if none)
* @put: The put function for the embedding object (NULL if none)
*
* Initialises the klist structure. If the klist_node structures are
* going to be embedded in refcounted objects (necessary for safe
* deletion) then the get/put arguments are used to initialise
* functions that take and release references on the embedding
* objects.
*/
void klist_init(struct klist *k, void (*get)(struct klist_node *),
void (*put)(struct klist_node *))
{
INIT_LIST_HEAD(&k->k_list);
spin_lock_init(&k->k_lock);
k->get = get;
k->put = put;
}
EXPORT_SYMBOL_GPL(klist_init);
static void add_head(struct klist *k, struct klist_node *n)
{
spin_lock(&k->k_lock);
list_add(&n->n_node, &k->k_list);
spin_unlock(&k->k_lock);
}
static void add_tail(struct klist *k, struct klist_node *n)
{
spin_lock(&k->k_lock);
list_add_tail(&n->n_node, &k->k_list);
spin_unlock(&k->k_lock);
}
static void klist_node_init(struct klist *k, struct klist_node *n)
{
INIT_LIST_HEAD(&n->n_node);
kref_init(&n->n_ref);
knode_set_klist(n, k);
if (k->get)
k->get(n);
}
/**
* klist_add_head - Initialize a klist_node and add it to front.
* @n: node we're adding.
* @k: klist it's going on.
*/
void klist_add_head(struct klist_node *n, struct klist *k)
{
klist_node_init(k, n);
add_head(k, n);
}
EXPORT_SYMBOL_GPL(klist_add_head);
/**
* klist_add_tail - Initialize a klist_node and add it to back.
* @n: node we're adding.
* @k: klist it's going on.
*/
void klist_add_tail(struct klist_node *n, struct klist *k)
{
klist_node_init(k, n);
add_tail(k, n);
}
EXPORT_SYMBOL_GPL(klist_add_tail);
/**
* klist_add_behind - Init a klist_node and add it after an existing node
* @n: node we're adding.
* @pos: node to put @n after
*/
void klist_add_behind(struct klist_node *n, struct klist_node *pos)
{
struct klist *k = knode_klist(pos);
klist_node_init(k, n);
spin_lock(&k->k_lock);
list_add(&n->n_node, &pos->n_node);
spin_unlock(&k->k_lock);
}
EXPORT_SYMBOL_GPL(klist_add_behind);
/**
* klist_add_before - Init a klist_node and add it before an existing node
* @n: node we're adding.
* @pos: node to put @n after
*/
void klist_add_before(struct klist_node *n, struct klist_node *pos)
{
struct klist *k = knode_klist(pos);
klist_node_init(k, n);
spin_lock(&k->k_lock);
list_add_tail(&n->n_node, &pos->n_node);
spin_unlock(&k->k_lock);
}
EXPORT_SYMBOL_GPL(klist_add_before);
struct klist_waiter {
struct list_head list;
struct klist_node *node;
struct task_struct *process;
int woken;
};
static DEFINE_SPINLOCK(klist_remove_lock);
static LIST_HEAD(klist_remove_waiters);
static void klist_release(struct kref *kref)
{
struct klist_waiter *waiter, *tmp;
struct klist_node *n = container_of(kref, struct klist_node, n_ref);
WARN_ON(!knode_dead(n));
list_del(&n->n_node);
spin_lock(&klist_remove_lock);
list_for_each_entry_safe(waiter, tmp, &klist_remove_waiters, list) {
if (waiter->node != n)
continue;
list_del(&waiter->list);
waiter->woken = 1;
mb();
wake_up_process(waiter->process);
}
spin_unlock(&klist_remove_lock);
knode_set_klist(n, NULL);
}
static int klist_dec_and_del(struct klist_node *n)
{
return kref_put(&n->n_ref, klist_release);
}
static void klist_put(struct klist_node *n, bool kill)
{
struct klist *k = knode_klist(n);
void (*put)(struct klist_node *) = k->put;
spin_lock(&k->k_lock);
if (kill)
knode_kill(n);
if (!klist_dec_and_del(n))
put = NULL;
spin_unlock(&k->k_lock);
if (put)
put(n);
}
/**
* klist_del - Decrement the reference count of node and try to remove.
* @n: node we're deleting.
*/
void klist_del(struct klist_node *n)
{
klist_put(n, true);
}
EXPORT_SYMBOL_GPL(klist_del);
/**
* klist_remove - Decrement the refcount of node and wait for it to go away.
* @n: node we're removing.
*/
void klist_remove(struct klist_node *n)
{
struct klist_waiter waiter;
waiter.node = n;
waiter.process = current;
waiter.woken = 0;
spin_lock(&klist_remove_lock);
list_add(&waiter.list, &klist_remove_waiters);
spin_unlock(&klist_remove_lock);
klist_del(n);
for (;;) {
set_current_state(TASK_UNINTERRUPTIBLE);
if (waiter.woken)
break;
schedule();
}
__set_current_state(TASK_RUNNING);
}
EXPORT_SYMBOL_GPL(klist_remove);
/**
* klist_node_attached - Say whether a node is bound to a list or not.
* @n: Node that we're testing.
*/
int klist_node_attached(struct klist_node *n)
{
return (n->n_klist != NULL);
}
EXPORT_SYMBOL_GPL(klist_node_attached);
/**
* klist_iter_init_node - Initialize a klist_iter structure.
* @k: klist we're iterating.
* @i: klist_iter we're filling.
* @n: node to start with.
*
* Similar to klist_iter_init(), but starts the action off with @n,
* instead of with the list head.
*/
void klist_iter_init_node(struct klist *k, struct klist_iter *i,
struct klist_node *n)
{
i->i_klist = k;
i->i_cur = NULL;
if (n && kref_get_unless_zero(&n->n_ref))
i->i_cur = n;
}
EXPORT_SYMBOL_GPL(klist_iter_init_node);
/**
* klist_iter_init - Iniitalize a klist_iter structure.
* @k: klist we're iterating.
* @i: klist_iter structure we're filling.
*
* Similar to klist_iter_init_node(), but start with the list head.
*/
void klist_iter_init(struct klist *k, struct klist_iter *i)
{
klist_iter_init_node(k, i, NULL);
}
EXPORT_SYMBOL_GPL(klist_iter_init);
/**
* klist_iter_exit - Finish a list iteration.
* @i: Iterator structure.
*
* Must be called when done iterating over list, as it decrements the
* refcount of the current node. Necessary in case iteration exited before
* the end of the list was reached, and always good form.
*/
void klist_iter_exit(struct klist_iter *i)
{
if (i->i_cur) {
klist_put(i->i_cur, false);
i->i_cur = NULL;
}
}
EXPORT_SYMBOL_GPL(klist_iter_exit);
static struct klist_node *to_klist_node(struct list_head *n)
{
return container_of(n, struct klist_node, n_node);
}
/**
* klist_prev - Ante up prev node in list.
* @i: Iterator structure.
*
* First grab list lock. Decrement the reference count of the previous
* node, if there was one. Grab the prev node, increment its reference
* count, drop the lock, and return that prev node.
*/
struct klist_node *klist_prev(struct klist_iter *i)
{
void (*put)(struct klist_node *) = i->i_klist->put;
struct klist_node *last = i->i_cur;
struct klist_node *prev;
unsigned long flags;
spin_lock_irqsave(&i->i_klist->k_lock, flags);
if (last) {
prev = to_klist_node(last->n_node.prev);
if (!klist_dec_and_del(last))
put = NULL;
} else
prev = to_klist_node(i->i_klist->k_list.prev);
i->i_cur = NULL;
while (prev != to_klist_node(&i->i_klist->k_list)) {
if (likely(!knode_dead(prev))) {
kref_get(&prev->n_ref);
i->i_cur = prev;
break;
}
prev = to_klist_node(prev->n_node.prev);
}
spin_unlock_irqrestore(&i->i_klist->k_lock, flags);
if (put && last)
put(last);
return i->i_cur;
}
EXPORT_SYMBOL_GPL(klist_prev);
/**
* klist_next - Ante up next node in list.
* @i: Iterator structure.
*
* First grab list lock. Decrement the reference count of the previous
* node, if there was one. Grab the next node, increment its reference
* count, drop the lock, and return that next node.
*/
struct klist_node *klist_next(struct klist_iter *i)
{
void (*put)(struct klist_node *) = i->i_klist->put;
struct klist_node *last = i->i_cur;
struct klist_node *next;
unsigned long flags;
spin_lock_irqsave(&i->i_klist->k_lock, flags);
if (last) {
next = to_klist_node(last->n_node.next);
if (!klist_dec_and_del(last))
put = NULL;
} else
next = to_klist_node(i->i_klist->k_list.next);
i->i_cur = NULL;
while (next != to_klist_node(&i->i_klist->k_list)) {
if (likely(!knode_dead(next))) {
kref_get(&next->n_ref);
i->i_cur = next;
break;
}
next = to_klist_node(next->n_node.next);
}
spin_unlock_irqrestore(&i->i_klist->k_lock, flags);
if (put && last)
put(last);
return i->i_cur;
}
EXPORT_SYMBOL_GPL(klist_next);
![swh spinner](/static/img/swh-spinner.gif)
Computing file changes ...