marker.c

Kernel code of linux kernel

{
	struct hlist_head *head;
	struct hlist_node *node;
	struct marker_entry *e;
	int found = 0;
	size_t len = strlen(name) + 1;
	u32 hash = jhash(name, len-1, 0);

	head = &marker_table[hash & ((1 << MARKER_HASH_BITS)-1)];
	hlist_for_each_entry(e, node, head, hlist) {
		if (!strcmp(name, e->name)) {
			found = 1;
			break;
		}
	}
	if (!found)
		return -ENOENT;
	if (e->single.func != __mark_empty_function)
		return -EBUSY;
	hlist_del(&e->hlist);
	/* Make sure the call_rcu has been executed */
	if (e->rcu_pending)
		rcu_barrier_sched();
	kfree(e);
	return 0;
}

/*
 * Set the mark_entry format to the format found in the element.
 */
static int marker_set_format(struct marker_entry **entry, const char *format)
{
	struct marker_entry *e;
	size_t name_len = strlen((*entry)->name) + 1;
	size_t format_len = strlen(format) + 1;


	e = kmalloc(sizeof(struct marker_entry) + name_len + format_len,
			GFP_KERNEL);
	if (!e)
		return -ENOMEM;
	memcpy(&e->name[0], (*entry)->name, name_len);
	e->format = &e->name[name_len];
	memcpy(e->format, format, format_len);
	if (strcmp(e->format, MARK_NOARGS) == 0)
		e->call = marker_probe_cb_noarg;
	else
		e->call = marker_probe_cb;
	e->single = (*entry)->single;
	e->multi = (*entry)->multi;
	e->ptype = (*entry)->ptype;
	e->refcount = (*entry)->refcount;
	e->rcu_pending = 0;
	hlist_add_before(&e->hlist, &(*entry)->hlist);
	hlist_del(&(*entry)->hlist);
	/* Make sure the call_rcu has been executed */
	if ((*entry)->rcu_pending)
		rcu_barrier_sched();
	kfree(*entry);
	*entry = e;
	trace_mark(core_marker_format, "name %s format %s",
			e->name, e->format);
	return 0;
}

/*
 * Sets the probe callback corresponding to one marker.
 */
static int set_marker(struct marker_entry **entry, struct marker *elem,
		int active)
{
	int ret;
	WARN_ON(strcmp((*entry)->name, elem->name) != 0);

	if ((*entry)->format) {
		if (strcmp((*entry)->format, elem->format) != 0) {
			printk(KERN_NOTICE
				"Format mismatch for probe %s "
				"(%s), marker (%s)\n",
				(*entry)->name,
				(*entry)->format,
				elem->format);
			return -EPERM;
		}
	} else {
		ret = marker_set_format(entry, elem->format);
		if (ret)
			return ret;
	}

	/*
	 * probe_cb setup (statically known) is done here. It is
	 * asynchronous with the rest of execution, therefore we only
	 * pass from a "safe" callback (with argument) to an "unsafe"
	 * callback (does not set arguments).
	 */
	elem->call = (*entry)->call;
	/*
	 * Sanity check :
	 * We only update the single probe private data when the ptr is
	 * set to a _non_ single probe! (0 -> 1 and N -> 1, N != 1)
	 */
	WARN_ON(elem->single.func != __mark_empty_function
		&& elem->single.probe_private
		!= (*entry)->single.probe_private &&
		!elem->ptype);
	elem->single.probe_private = (*entry)->single.probe_private;
	/*
	 * Make sure the private data is valid when we update the
	 * single probe ptr.
	 */
	smp_wmb();
	elem->single.func = (*entry)->single.func;
	/*
	 * We also make sure that the new probe callbacks array is consistent
	 * before setting a pointer to it.
	 */
	rcu_assign_pointer(elem->multi, (*entry)->multi);
	/*
	 * Update the function or multi probe array pointer before setting the
	 * ptype.
	 */
	smp_wmb();
	elem->ptype = (*entry)->ptype;
	elem->state = active;

	return 0;
}

/*
 * Disable a marker and its probe callback.
 * Note: only waiting an RCU period after setting elem->call to the empty
 * function insures that the original callback is not used anymore. This insured
 * by preempt_disable around the call site.
 */
static void disable_marker(struct marker *elem)
{
	/* leave "call" as is. It is known statically. */
	elem->state = 0;
	elem->single.func = __mark_empty_function;
	/* Update the function before setting the ptype */
	smp_wmb();
	elem->ptype = 0;	/* single probe */
	/*
	 * Leave the private data and id there, because removal is racy and
	 * should be done only after an RCU period. These are never used until
	 * the next initialization anyway.
	 */
}

/**
 * marker_update_probe_range - Update a probe range
 * @begin: beginning of the range
 * @end: end of the range
 *
 * Updates the probe callback corresponding to a range of markers.
 */
void marker_update_probe_range(struct marker *begin,
	struct marker *end)
{
	struct marker *iter;
	struct marker_entry *mark_entry;

	mutex_lock(&markers_mutex);
	for (iter = begin; iter < end; iter++) {
		mark_entry = get_marker(iter->name);
		if (mark_entry) {
			set_marker(&mark_entry, iter,
					!!mark_entry->refcount);
			/*
			 * ignore error, continue
			 */
		} else {
			disable_marker(iter);
		}
	}
	mutex_unlock(&markers_mutex);
}

/*
 * Update probes, removing the faulty probes.
 *
 * Internal callback only changed before the first probe is connected to it.
 * Single probe private data can only be changed on 0 -> 1 and 2 -> 1
 * transitions.  All other transitions will leave the old private data valid.
 * This makes the non-atomicity of the callback/private data updates valid.
 *
 * "special case" updates :
 * 0 -> 1 callback
 * 1 -> 0 callback
 * 1 -> 2 callbacks
 * 2 -> 1 callbacks
 * Other updates all behave the same, just like the 2 -> 3 or 3 -> 2 updates.
 * Site effect : marker_set_format may delete the marker entry (creating a
 * replacement).
 */
static void marker_update_probes(void)
{
	/* Core kernel markers */
	marker_update_probe_range(__start___markers, __stop___markers);
	/* Markers in modules. */
	module_update_markers();
}

/**
 * marker_probe_register -  Connect a probe to a marker
 * @name: marker name
 * @format: format string
 * @probe: probe handler
 * @probe_private: probe private data
 *
 * private data must be a valid allocated memory address, or NULL.
 * Returns 0 if ok, error value on error.
 * The probe address must at least be aligned on the architecture pointer size.
 */
int marker_probe_register(const char *name, const char *format,
			marker_probe_func *probe, void *probe_private)
{
	struct marker_entry *entry;
	int ret = 0;
	struct marker_probe_closure *old;

	mutex_lock(&markers_mutex);
	entry = get_marker(name);
	if (!entry) {
		entry = add_marker(name, format);
		if (IS_ERR(entry)) {
			ret = PTR_ERR(entry);
			goto end;
		}
	}
	/*
	 * If we detect that a call_rcu is pending for this marker,
	 * make sure it's executed now.
	 */
	if (entry->rcu_pending)
		rcu_barrier_sched();
	old = marker_entry_add_probe(entry, probe, probe_private);
	if (IS_ERR(old)) {
		ret = PTR_ERR(old);
		goto end;
	}
	mutex_unlock(&markers_mutex);
	marker_update_probes();		/* may update entry */
	mutex_lock(&markers_mutex);
	entry = get_marker(name);
	WARN_ON(!entry);
	entry->oldptr = old;
	entry->rcu_pending = 1;
	/* write rcu_pending before calling the RCU callback */
	smp_wmb();
	call_rcu_sched(&entry->rcu, free_old_closure);
end:
	mutex_unlock(&markers_mutex);
	return ret;
}
EXPORT_SYMBOL_GPL(marker_probe_register);

/**
 * marker_probe_unregister -  Disconnect a probe from a marker
 * @name: marker name
 * @probe: probe function pointer
 * @probe_private: probe private data
 *
 * Returns the private data given to marker_probe_register, or an ERR_PTR().
 * We do not need to call a synchronize_sched to make sure the probes have
 * finished running before doing a module unload, because the module unload
 * itself uses stop_machine(), which insures that every preempt disabled section
 * have finished.
 */
int marker_probe_unregister(const char *name,
	marker_probe_func *probe, void *probe_private)
{
	struct marker_entry *entry;
	struct marker_probe_closure *old;
	int ret = -ENOENT;

	mutex_lock(&markers_mutex);
	entry = get_marker(name);
	if (!entry)
		goto end;
	if (entry->rcu_pending)
		rcu_barrier_sched();
	old = marker_entry_remove_probe(entry, probe, probe_private);
	mutex_unlock(&markers_mutex);
	marker_update_probes();		/* may update entry */
	mutex_lock(&markers_mutex);
	entry = get_marker(name);
	if (!entry)
		goto end;
	entry->oldptr = old;
	entry->rcu_pending = 1;
	/* write rcu_pending before calling the RCU callback */
	smp_wmb();
	call_rcu_sched(&entry->rcu, free_old_closure);
	remove_marker(name);	/* Ignore busy error message */
	ret = 0;
end:
	mutex_unlock(&markers_mutex);
	return ret;
}
EXPORT_SYMBOL_GPL(marker_probe_unregister);

static struct marker_entry *
get_marker_from_private_data(marker_probe_func *probe, void *probe_private)
{
	struct marker_entry *entry;
	unsigned int i;
	struct hlist_head *head;
	struct hlist_node *node;

	for (i = 0; i < MARKER_TABLE_SIZE; i++) {
		head = &marker_table[i];
		hlist_for_each_entry(entry, node, head, hlist) {
			if (!entry->ptype) {
				if (entry->single.func == probe
						&& entry->single.probe_private
						== probe_private)
					return entry;
			} else {
				struct marker_probe_closure *closure;
				closure = entry->multi;
				for (i = 0; closure[i].func; i++) {
					if (closure[i].func == probe &&
							closure[i].probe_private
							== probe_private)
						return entry;
				}
			}
		}
	}
	return NULL;
}

/**
 * marker_probe_unregister_private_data -  Disconnect a probe from a marker
 * @probe: probe function
 * @probe_private: probe private data
 *
 * Unregister a probe by providing the registered private data.
 * Only removes the first marker found in hash table.
 * Return 0 on success or error value.
 * We do not need to call a synchronize_sched to make sure the probes have
 * finished running before doing a module unload, because the module unload
 * itself uses stop_machine(), which insures that every preempt disabled section
 * have finished.
 */
int marker_probe_unregister_private_data(marker_probe_func *probe,
		void *probe_private)
{
	struct marker_entry *entry;
	int ret = 0;
	struct marker_probe_closure *old;

	mutex_lock(&markers_mutex);
	entry = get_marker_from_private_data(probe, probe_private);
	if (!entry) {
		ret = -ENOENT;
		goto end;
	}
	if (entry->rcu_pending)
		rcu_barrier_sched();
	old = marker_entry_remove_probe(entry, NULL, probe_private);
	mutex_unlock(&markers_mutex);
	marker_update_probes();		/* may update entry */
	mutex_lock(&markers_mutex);
	entry = get_marker_from_private_data(probe, probe_private);
	WARN_ON(!entry);
	entry->oldptr = old;
	entry->rcu_pending = 1;
	/* write rcu_pending before calling the RCU callback */
	smp_wmb();
	call_rcu_sched(&entry->rcu, free_old_closure);
	remove_marker(entry->name);	/* Ignore busy error message */
end:
	mutex_unlock(&markers_mutex);
	return ret;
}
EXPORT_SYMBOL_GPL(marker_probe_unregister_private_data);

/**
 * marker_get_private_data - Get a marker's probe private data
 * @name: marker name
 * @probe: probe to match
 * @num: get the nth matching probe's private data
 *
 * Returns the nth private data pointer (starting from 0) matching, or an
 * ERR_PTR.
 * Returns the private data pointer, or an ERR_PTR.
 * The private data pointer should _only_ be dereferenced if the caller is the
 * owner of the data, or its content could vanish. This is mostly used to
 * confirm that a caller is the owner of a registered probe.
 */
void *marker_get_private_data(const char *name, marker_probe_func *probe,
		int num)
{
	struct hlist_head *head;
	struct hlist_node *node;
	struct marker_entry *e;
	size_t name_len = strlen(name) + 1;
	u32 hash = jhash(name, name_len-1, 0);
	int i;

	head = &marker_table[hash & ((1 << MARKER_HASH_BITS)-1)];
	hlist_for_each_entry(e, node, head, hlist) {
		if (!strcmp(name, e->name)) {
			if (!e->ptype) {
				if (num == 0 && e->single.func == probe)
					return e->single.probe_private;
				else
					break;
			} else {
				struct marker_probe_closure *closure;
				int match = 0;
				closure = e->multi;
				for (i = 0; closure[i].func; i++) {
					if (closure[i].func != probe)
						continue;
					if (match++ == num)
						return closure[i].probe_private;
				}
			}
		}
	}
	return ERR_PTR(-ENOENT);
}
EXPORT_SYMBOL_GPL(marker_get_private_data);