zephyr/subsys/bluetooth/host/mesh/access.c

752 lines
16 KiB
C
Raw Normal View History

/* Bluetooth Mesh */
/*
* Copyright (c) 2017 Intel Corporation
*
* SPDX-License-Identifier: Apache-2.0
*/
#include <zephyr.h>
#include <errno.h>
#include <misc/util.h>
#include <misc/byteorder.h>
#include <net/buf.h>
#include <bluetooth/bluetooth.h>
#include <bluetooth/mesh.h>
#define BT_DBG_ENABLED IS_ENABLED(CONFIG_BT_MESH_DEBUG_ACCESS)
#define LOG_MODULE_NAME bt_mesh_access
#include "common/log.h"
#include "mesh.h"
#include "adv.h"
#include "net.h"
#include "lpn.h"
#include "transport.h"
#include "access.h"
#include "foundation.h"
static const struct bt_mesh_comp *dev_comp;
static u16_t dev_primary_addr;
static const struct {
const u16_t id;
int (*const init)(struct bt_mesh_model *model, bool primary);
} model_init[] = {
{ BT_MESH_MODEL_ID_CFG_SRV, bt_mesh_cfg_srv_init },
{ BT_MESH_MODEL_ID_HEALTH_SRV, bt_mesh_health_srv_init },
#if defined(CONFIG_BT_MESH_CFG_CLI)
{ BT_MESH_MODEL_ID_CFG_CLI, bt_mesh_cfg_cli_init },
#endif
#if defined(CONFIG_BT_MESH_HEALTH_CLI)
{ BT_MESH_MODEL_ID_HEALTH_CLI, bt_mesh_health_cli_init },
#endif
};
void bt_mesh_model_foreach(void (*func)(struct bt_mesh_model *mod,
struct bt_mesh_elem *elem,
bool vnd, bool primary,
void *user_data),
void *user_data)
{
int i, j;
for (i = 0; i < dev_comp->elem_count; i++) {
struct bt_mesh_elem *elem = &dev_comp->elem[i];
for (j = 0; j < elem->model_count; j++) {
struct bt_mesh_model *model = &elem->models[j];
func(model, elem, false, i == 0, user_data);
}
for (j = 0; j < elem->vnd_model_count; j++) {
struct bt_mesh_model *model = &elem->vnd_models[j];
func(model, elem, true, i == 0, user_data);
}
}
}
s32_t bt_mesh_model_pub_period_get(struct bt_mesh_model *mod)
{
int period;
if (!mod->pub) {
return 0;
}
switch (mod->pub->period >> 6) {
case 0x00:
/* 1 step is 100 ms */
period = K_MSEC((mod->pub->period & BIT_MASK(6)) * 100U);
break;
case 0x01:
/* 1 step is 1 second */
period = K_SECONDS(mod->pub->period & BIT_MASK(6));
break;
case 0x02:
/* 1 step is 10 seconds */
period = K_SECONDS((mod->pub->period & BIT_MASK(6)) * 10U);
break;
case 0x03:
/* 1 step is 10 minutes */
period = K_MINUTES((mod->pub->period & BIT_MASK(6)) * 10U);
break;
default:
CODE_UNREACHABLE;
}
return period >> mod->pub->period_div;
}
static s32_t next_period(struct bt_mesh_model *mod)
{
struct bt_mesh_model_pub *pub = mod->pub;
u32_t elapsed, period;
period = bt_mesh_model_pub_period_get(mod);
if (!period) {
return 0;
}
elapsed = k_uptime_get_32() - pub->period_start;
BT_DBG("Publishing took %ums", elapsed);
if (elapsed > period) {
BT_WARN("Publication sending took longer than the period");
/* Return smallest positive number since 0 means disabled */
return K_MSEC(1);
}
return period - elapsed;
}
Bluetooth: Mesh: Fix model publication Model publication was broken in a couple of ways: - The Publish Retransmit State was not taken into account at all - Health Server used a single publish state for all elements To implement Publish Retransmit properly, one has to use a callback to track when the message has been sent. The problem with the transport layer sending APIs was that giving a callback would cause the transport layer to assume that segmentation (with acks) is desired, which is not the case for Model Publication (unless the message itself is too large, of course). Because of this, the message sending context receives a new send_rel ("Send Reliable") boolean member that an app can use to force reliable sending. Another challenge with the Publish Retransmit state is that a buffer is needed for storing the AppKey-encrypted SDU once it has been sent out for the first time.To solve this, a new new net_buf_simple member is added to the model publication context. The separate 'msg' input parameter of the bt_mesh_model_publish() API is removed, since the application is now expected to pre-fill pub->msg instead. To help with the publishing API change, the Health Server model gets a new helper macro for initializing the publishing context with a right-sized publishing message. The API for creating Health Server instances is also redesigned since it was so far using a single model publishing state, which would result in erratic behavior in case of multiple elements with the Health Server Model. Now, the application needs to provide a unique publishing context for each Health Server instance. The changes are heavily intertwined, so it's not easily possible to split them into multiple patches, hence the large(ish) patch. Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
2017-11-18 10:24:18 +02:00
static void publish_sent(int err, void *user_data)
{
struct bt_mesh_model *mod = user_data;
s32_t delay;
BT_DBG("err %d", err);
if (mod->pub->count) {
delay = BT_MESH_PUB_TRANSMIT_INT(mod->pub->retransmit);
} else {
delay = next_period(mod);
Bluetooth: Mesh: Fix model publication Model publication was broken in a couple of ways: - The Publish Retransmit State was not taken into account at all - Health Server used a single publish state for all elements To implement Publish Retransmit properly, one has to use a callback to track when the message has been sent. The problem with the transport layer sending APIs was that giving a callback would cause the transport layer to assume that segmentation (with acks) is desired, which is not the case for Model Publication (unless the message itself is too large, of course). Because of this, the message sending context receives a new send_rel ("Send Reliable") boolean member that an app can use to force reliable sending. Another challenge with the Publish Retransmit state is that a buffer is needed for storing the AppKey-encrypted SDU once it has been sent out for the first time.To solve this, a new new net_buf_simple member is added to the model publication context. The separate 'msg' input parameter of the bt_mesh_model_publish() API is removed, since the application is now expected to pre-fill pub->msg instead. To help with the publishing API change, the Health Server model gets a new helper macro for initializing the publishing context with a right-sized publishing message. The API for creating Health Server instances is also redesigned since it was so far using a single model publishing state, which would result in erratic behavior in case of multiple elements with the Health Server Model. Now, the application needs to provide a unique publishing context for each Health Server instance. The changes are heavily intertwined, so it's not easily possible to split them into multiple patches, hence the large(ish) patch. Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
2017-11-18 10:24:18 +02:00
}
if (delay) {
BT_DBG("Publishing next time in %dms", delay);
k_delayed_work_submit(&mod->pub->timer, delay);
}
Bluetooth: Mesh: Fix model publication Model publication was broken in a couple of ways: - The Publish Retransmit State was not taken into account at all - Health Server used a single publish state for all elements To implement Publish Retransmit properly, one has to use a callback to track when the message has been sent. The problem with the transport layer sending APIs was that giving a callback would cause the transport layer to assume that segmentation (with acks) is desired, which is not the case for Model Publication (unless the message itself is too large, of course). Because of this, the message sending context receives a new send_rel ("Send Reliable") boolean member that an app can use to force reliable sending. Another challenge with the Publish Retransmit state is that a buffer is needed for storing the AppKey-encrypted SDU once it has been sent out for the first time.To solve this, a new new net_buf_simple member is added to the model publication context. The separate 'msg' input parameter of the bt_mesh_model_publish() API is removed, since the application is now expected to pre-fill pub->msg instead. To help with the publishing API change, the Health Server model gets a new helper macro for initializing the publishing context with a right-sized publishing message. The API for creating Health Server instances is also redesigned since it was so far using a single model publishing state, which would result in erratic behavior in case of multiple elements with the Health Server Model. Now, the application needs to provide a unique publishing context for each Health Server instance. The changes are heavily intertwined, so it's not easily possible to split them into multiple patches, hence the large(ish) patch. Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
2017-11-18 10:24:18 +02:00
}
static const struct bt_mesh_send_cb pub_sent_cb = {
.end = publish_sent,
};
static int publish_retransmit(struct bt_mesh_model *mod)
{
NET_BUF_SIMPLE_DEFINE(sdu, BT_MESH_TX_SDU_MAX);
Bluetooth: Mesh: Fix model publication Model publication was broken in a couple of ways: - The Publish Retransmit State was not taken into account at all - Health Server used a single publish state for all elements To implement Publish Retransmit properly, one has to use a callback to track when the message has been sent. The problem with the transport layer sending APIs was that giving a callback would cause the transport layer to assume that segmentation (with acks) is desired, which is not the case for Model Publication (unless the message itself is too large, of course). Because of this, the message sending context receives a new send_rel ("Send Reliable") boolean member that an app can use to force reliable sending. Another challenge with the Publish Retransmit state is that a buffer is needed for storing the AppKey-encrypted SDU once it has been sent out for the first time.To solve this, a new new net_buf_simple member is added to the model publication context. The separate 'msg' input parameter of the bt_mesh_model_publish() API is removed, since the application is now expected to pre-fill pub->msg instead. To help with the publishing API change, the Health Server model gets a new helper macro for initializing the publishing context with a right-sized publishing message. The API for creating Health Server instances is also redesigned since it was so far using a single model publishing state, which would result in erratic behavior in case of multiple elements with the Health Server Model. Now, the application needs to provide a unique publishing context for each Health Server instance. The changes are heavily intertwined, so it's not easily possible to split them into multiple patches, hence the large(ish) patch. Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
2017-11-18 10:24:18 +02:00
struct bt_mesh_model_pub *pub = mod->pub;
struct bt_mesh_app_key *key;
struct bt_mesh_msg_ctx ctx = {
.addr = pub->addr,
.send_ttl = pub->ttl,
};
struct bt_mesh_net_tx tx = {
.ctx = &ctx,
.src = bt_mesh_model_elem(mod)->addr,
Bluetooth: Mesh: Fix model publication Model publication was broken in a couple of ways: - The Publish Retransmit State was not taken into account at all - Health Server used a single publish state for all elements To implement Publish Retransmit properly, one has to use a callback to track when the message has been sent. The problem with the transport layer sending APIs was that giving a callback would cause the transport layer to assume that segmentation (with acks) is desired, which is not the case for Model Publication (unless the message itself is too large, of course). Because of this, the message sending context receives a new send_rel ("Send Reliable") boolean member that an app can use to force reliable sending. Another challenge with the Publish Retransmit state is that a buffer is needed for storing the AppKey-encrypted SDU once it has been sent out for the first time.To solve this, a new new net_buf_simple member is added to the model publication context. The separate 'msg' input parameter of the bt_mesh_model_publish() API is removed, since the application is now expected to pre-fill pub->msg instead. To help with the publishing API change, the Health Server model gets a new helper macro for initializing the publishing context with a right-sized publishing message. The API for creating Health Server instances is also redesigned since it was so far using a single model publishing state, which would result in erratic behavior in case of multiple elements with the Health Server Model. Now, the application needs to provide a unique publishing context for each Health Server instance. The changes are heavily intertwined, so it's not easily possible to split them into multiple patches, hence the large(ish) patch. Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
2017-11-18 10:24:18 +02:00
.xmit = bt_mesh_net_transmit_get(),
.friend_cred = pub->cred,
};
key = bt_mesh_app_key_find(pub->key);
if (!key) {
return -EADDRNOTAVAIL;
}
tx.sub = bt_mesh_subnet_get(key->net_idx);
ctx.net_idx = key->net_idx;
ctx.app_idx = key->app_idx;
net_buf_simple_add_mem(&sdu, pub->msg->data, pub->msg->len);
Bluetooth: Mesh: Fix model publication Model publication was broken in a couple of ways: - The Publish Retransmit State was not taken into account at all - Health Server used a single publish state for all elements To implement Publish Retransmit properly, one has to use a callback to track when the message has been sent. The problem with the transport layer sending APIs was that giving a callback would cause the transport layer to assume that segmentation (with acks) is desired, which is not the case for Model Publication (unless the message itself is too large, of course). Because of this, the message sending context receives a new send_rel ("Send Reliable") boolean member that an app can use to force reliable sending. Another challenge with the Publish Retransmit state is that a buffer is needed for storing the AppKey-encrypted SDU once it has been sent out for the first time.To solve this, a new new net_buf_simple member is added to the model publication context. The separate 'msg' input parameter of the bt_mesh_model_publish() API is removed, since the application is now expected to pre-fill pub->msg instead. To help with the publishing API change, the Health Server model gets a new helper macro for initializing the publishing context with a right-sized publishing message. The API for creating Health Server instances is also redesigned since it was so far using a single model publishing state, which would result in erratic behavior in case of multiple elements with the Health Server Model. Now, the application needs to provide a unique publishing context for each Health Server instance. The changes are heavily intertwined, so it's not easily possible to split them into multiple patches, hence the large(ish) patch. Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
2017-11-18 10:24:18 +02:00
pub->count--;
return bt_mesh_trans_send(&tx, &sdu, &pub_sent_cb, mod);
Bluetooth: Mesh: Fix model publication Model publication was broken in a couple of ways: - The Publish Retransmit State was not taken into account at all - Health Server used a single publish state for all elements To implement Publish Retransmit properly, one has to use a callback to track when the message has been sent. The problem with the transport layer sending APIs was that giving a callback would cause the transport layer to assume that segmentation (with acks) is desired, which is not the case for Model Publication (unless the message itself is too large, of course). Because of this, the message sending context receives a new send_rel ("Send Reliable") boolean member that an app can use to force reliable sending. Another challenge with the Publish Retransmit state is that a buffer is needed for storing the AppKey-encrypted SDU once it has been sent out for the first time.To solve this, a new new net_buf_simple member is added to the model publication context. The separate 'msg' input parameter of the bt_mesh_model_publish() API is removed, since the application is now expected to pre-fill pub->msg instead. To help with the publishing API change, the Health Server model gets a new helper macro for initializing the publishing context with a right-sized publishing message. The API for creating Health Server instances is also redesigned since it was so far using a single model publishing state, which would result in erratic behavior in case of multiple elements with the Health Server Model. Now, the application needs to provide a unique publishing context for each Health Server instance. The changes are heavily intertwined, so it's not easily possible to split them into multiple patches, hence the large(ish) patch. Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
2017-11-18 10:24:18 +02:00
}
static void mod_publish(struct k_work *work)
{
struct bt_mesh_model_pub *pub = CONTAINER_OF(work,
struct bt_mesh_model_pub,
timer.work);
s32_t period_ms;
int err;
BT_DBG("");
period_ms = bt_mesh_model_pub_period_get(pub->mod);
BT_DBG("period %u ms", period_ms);
Bluetooth: Mesh: Fix model publication Model publication was broken in a couple of ways: - The Publish Retransmit State was not taken into account at all - Health Server used a single publish state for all elements To implement Publish Retransmit properly, one has to use a callback to track when the message has been sent. The problem with the transport layer sending APIs was that giving a callback would cause the transport layer to assume that segmentation (with acks) is desired, which is not the case for Model Publication (unless the message itself is too large, of course). Because of this, the message sending context receives a new send_rel ("Send Reliable") boolean member that an app can use to force reliable sending. Another challenge with the Publish Retransmit state is that a buffer is needed for storing the AppKey-encrypted SDU once it has been sent out for the first time.To solve this, a new new net_buf_simple member is added to the model publication context. The separate 'msg' input parameter of the bt_mesh_model_publish() API is removed, since the application is now expected to pre-fill pub->msg instead. To help with the publishing API change, the Health Server model gets a new helper macro for initializing the publishing context with a right-sized publishing message. The API for creating Health Server instances is also redesigned since it was so far using a single model publishing state, which would result in erratic behavior in case of multiple elements with the Health Server Model. Now, the application needs to provide a unique publishing context for each Health Server instance. The changes are heavily intertwined, so it's not easily possible to split them into multiple patches, hence the large(ish) patch. Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
2017-11-18 10:24:18 +02:00
if (pub->count) {
err = publish_retransmit(pub->mod);
if (err) {
BT_ERR("Failed to retransmit (err %d)", err);
pub->count = 0U;
Bluetooth: Mesh: Fix model publication Model publication was broken in a couple of ways: - The Publish Retransmit State was not taken into account at all - Health Server used a single publish state for all elements To implement Publish Retransmit properly, one has to use a callback to track when the message has been sent. The problem with the transport layer sending APIs was that giving a callback would cause the transport layer to assume that segmentation (with acks) is desired, which is not the case for Model Publication (unless the message itself is too large, of course). Because of this, the message sending context receives a new send_rel ("Send Reliable") boolean member that an app can use to force reliable sending. Another challenge with the Publish Retransmit state is that a buffer is needed for storing the AppKey-encrypted SDU once it has been sent out for the first time.To solve this, a new new net_buf_simple member is added to the model publication context. The separate 'msg' input parameter of the bt_mesh_model_publish() API is removed, since the application is now expected to pre-fill pub->msg instead. To help with the publishing API change, the Health Server model gets a new helper macro for initializing the publishing context with a right-sized publishing message. The API for creating Health Server instances is also redesigned since it was so far using a single model publishing state, which would result in erratic behavior in case of multiple elements with the Health Server Model. Now, the application needs to provide a unique publishing context for each Health Server instance. The changes are heavily intertwined, so it's not easily possible to split them into multiple patches, hence the large(ish) patch. Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
2017-11-18 10:24:18 +02:00
/* Continue with normal publication */
if (period_ms) {
k_delayed_work_submit(&pub->timer, period_ms);
}
}
return;
}
if (!period_ms) {
return;
}
__ASSERT_NO_MSG(pub->update != NULL);
pub->period_start = k_uptime_get_32();
err = pub->update(pub->mod);
if (err) {
BT_ERR("Failed to update publication message");
return;
}
err = bt_mesh_model_publish(pub->mod);
if (err) {
BT_ERR("Publishing failed (err %d)", err);
}
if (pub->count) {
/* Retransmissions also control the timer */
k_delayed_work_cancel(&pub->timer);
}
}
struct bt_mesh_elem *bt_mesh_model_elem(struct bt_mesh_model *mod)
{
return &dev_comp->elem[mod->elem_idx];
}
struct bt_mesh_model *bt_mesh_model_get(bool vnd, u8_t elem_idx, u8_t mod_idx)
{
struct bt_mesh_elem *elem;
if (elem_idx >= dev_comp->elem_count) {
BT_ERR("Invalid element index %u", elem_idx);
return NULL;
}
elem = &dev_comp->elem[elem_idx];
if (vnd) {
if (mod_idx >= elem->vnd_model_count) {
BT_ERR("Invalid vendor model index %u", mod_idx);
return NULL;
}
return &elem->vnd_models[mod_idx];
} else {
if (mod_idx >= elem->model_count) {
BT_ERR("Invalid SIG model index %u", mod_idx);
return NULL;
}
return &elem->models[mod_idx];
}
}
static void mod_init(struct bt_mesh_model *mod, struct bt_mesh_elem *elem,
bool vnd, bool primary, void *user_data)
{
int i;
if (mod->pub) {
mod->pub->mod = mod;
k_delayed_work_init(&mod->pub->timer, mod_publish);
}
for (i = 0; i < ARRAY_SIZE(mod->keys); i++) {
mod->keys[i] = BT_MESH_KEY_UNUSED;
}
mod->elem_idx = elem - dev_comp->elem;
if (vnd) {
mod->mod_idx = mod - elem->vnd_models;
} else {
mod->mod_idx = mod - elem->models;
}
if (vnd) {
return;
}
for (i = 0; i < ARRAY_SIZE(model_init); i++) {
if (model_init[i].id == mod->id) {
model_init[i].init(mod, primary);
}
}
}
int bt_mesh_comp_register(const struct bt_mesh_comp *comp)
{
/* There must be at least one element */
if (!comp->elem_count) {
return -EINVAL;
}
dev_comp = comp;
bt_mesh_model_foreach(mod_init, NULL);
return 0;
}
void bt_mesh_comp_provision(u16_t addr)
{
int i;
dev_primary_addr = addr;
BT_DBG("addr 0x%04x elem_count %zu", addr, dev_comp->elem_count);
for (i = 0; i < dev_comp->elem_count; i++) {
struct bt_mesh_elem *elem = &dev_comp->elem[i];
elem->addr = addr++;
BT_DBG("addr 0x%04x mod_count %u vnd_mod_count %u",
elem->addr, elem->model_count, elem->vnd_model_count);
}
}
void bt_mesh_comp_unprovision(void)
{
BT_DBG("");
dev_primary_addr = BT_MESH_ADDR_UNASSIGNED;
bt_mesh_model_foreach(mod_init, NULL);
}
u16_t bt_mesh_primary_addr(void)
{
return dev_primary_addr;
}
u16_t *bt_mesh_model_find_group(struct bt_mesh_model *mod, u16_t addr)
{
int i;
for (i = 0; i < ARRAY_SIZE(mod->groups); i++) {
if (mod->groups[i] == addr) {
return &mod->groups[i];
}
}
return NULL;
}
static struct bt_mesh_model *bt_mesh_elem_find_group(struct bt_mesh_elem *elem,
u16_t group_addr)
{
struct bt_mesh_model *model;
u16_t *match;
int i;
for (i = 0; i < elem->model_count; i++) {
model = &elem->models[i];
match = bt_mesh_model_find_group(model, group_addr);
if (match) {
return model;
}
}
for (i = 0; i < elem->vnd_model_count; i++) {
model = &elem->vnd_models[i];
match = bt_mesh_model_find_group(model, group_addr);
if (match) {
return model;
}
}
return NULL;
}
struct bt_mesh_elem *bt_mesh_elem_find(u16_t addr)
{
int i;
for (i = 0; i < dev_comp->elem_count; i++) {
struct bt_mesh_elem *elem = &dev_comp->elem[i];
if (BT_MESH_ADDR_IS_GROUP(addr) ||
BT_MESH_ADDR_IS_VIRTUAL(addr)) {
if (bt_mesh_elem_find_group(elem, addr)) {
return elem;
}
} else if (elem->addr == addr) {
return elem;
}
}
return NULL;
}
u8_t bt_mesh_elem_count(void)
{
return dev_comp->elem_count;
}
static bool model_has_key(struct bt_mesh_model *mod, u16_t key)
{
int i;
for (i = 0; i < ARRAY_SIZE(mod->keys); i++) {
if (mod->keys[i] == key) {
return true;
}
}
return false;
}
static const struct bt_mesh_model_op *find_op(struct bt_mesh_model *models,
u8_t model_count, u16_t dst,
u16_t app_idx, u32_t opcode,
struct bt_mesh_model **model)
{
u8_t i;
for (i = 0U; i < model_count; i++) {
const struct bt_mesh_model_op *op;
*model = &models[i];
if (BT_MESH_ADDR_IS_GROUP(dst) ||
BT_MESH_ADDR_IS_VIRTUAL(dst)) {
if (!bt_mesh_model_find_group(*model, dst)) {
continue;
}
}
if (!model_has_key(*model, app_idx)) {
continue;
}
for (op = (*model)->op; op->func; op++) {
if (op->opcode == opcode) {
return op;
}
}
}
*model = NULL;
return NULL;
}
static int get_opcode(struct net_buf_simple *buf, u32_t *opcode)
{
switch (buf->data[0] >> 6) {
case 0x00:
case 0x01:
if (buf->data[0] == 0x7f) {
BT_ERR("Ignoring RFU OpCode");
return -EINVAL;
}
*opcode = net_buf_simple_pull_u8(buf);
return 0;
case 0x02:
if (buf->len < 2) {
BT_ERR("Too short payload for 2-octet OpCode");
return -EINVAL;
}
*opcode = net_buf_simple_pull_be16(buf);
return 0;
case 0x03:
if (buf->len < 3) {
BT_ERR("Too short payload for 3-octet OpCode");
return -EINVAL;
}
*opcode = net_buf_simple_pull_u8(buf) << 16;
*opcode |= net_buf_simple_pull_le16(buf);
return 0;
}
CODE_UNREACHABLE;
}
bool bt_mesh_fixed_group_match(u16_t addr)
{
/* Check for fixed group addresses */
switch (addr) {
case BT_MESH_ADDR_ALL_NODES:
return true;
case BT_MESH_ADDR_PROXIES:
/* TODO: Proxy not yet supported */
return false;
case BT_MESH_ADDR_FRIENDS:
return (bt_mesh_friend_get() == BT_MESH_FRIEND_ENABLED);
case BT_MESH_ADDR_RELAYS:
return (bt_mesh_relay_get() == BT_MESH_RELAY_ENABLED);
default:
return false;
}
}
void bt_mesh_model_recv(struct bt_mesh_net_rx *rx, struct net_buf_simple *buf)
{
struct bt_mesh_model *models, *model;
const struct bt_mesh_model_op *op;
u32_t opcode;
u8_t count;
int i;
BT_DBG("app_idx 0x%04x src 0x%04x dst 0x%04x", rx->ctx.app_idx,
rx->ctx.addr, rx->ctx.recv_dst);
BT_DBG("len %u: %s", buf->len, bt_hex(buf->data, buf->len));
if (get_opcode(buf, &opcode) < 0) {
BT_WARN("Unable to decode OpCode");
return;
}
BT_DBG("OpCode 0x%08x", opcode);
for (i = 0; i < dev_comp->elem_count; i++) {
struct bt_mesh_elem *elem = &dev_comp->elem[i];
if (BT_MESH_ADDR_IS_UNICAST(rx->ctx.recv_dst)) {
if (elem->addr != rx->ctx.recv_dst) {
continue;
}
} else if (BT_MESH_ADDR_IS_GROUP(rx->ctx.recv_dst) ||
BT_MESH_ADDR_IS_VIRTUAL(rx->ctx.recv_dst)) {
/* find_op() will do proper model/group matching */
} else if (i != 0 ||
!bt_mesh_fixed_group_match(rx->ctx.recv_dst)) {
continue;
}
/* SIG models cannot contain 3-byte (vendor) OpCodes, and
* vendor models cannot contain SIG (1- or 2-byte) OpCodes, so
* we only need to do the lookup in one of the model lists.
*/
if (opcode < 0x10000) {
models = elem->models;
count = elem->model_count;
} else {
models = elem->vnd_models;
count = elem->vnd_model_count;
}
op = find_op(models, count, rx->ctx.recv_dst, rx->ctx.app_idx,
opcode, &model);
if (op) {
struct net_buf_simple_state state;
if (buf->len < op->min_len) {
BT_ERR("Too short message for OpCode 0x%08x",
opcode);
continue;
}
/* The callback will likely parse the buffer, so
* store the parsing state in case multiple models
* receive the message.
*/
net_buf_simple_save(buf, &state);
op->func(model, &rx->ctx, buf);
net_buf_simple_restore(buf, &state);
} else {
BT_DBG("No OpCode 0x%08x for elem %d", opcode, i);
}
}
}
void bt_mesh_model_msg_init(struct net_buf_simple *msg, u32_t opcode)
{
net_buf_simple_init(msg, 0);
if (opcode < 0x100) {
/* 1-byte OpCode */
net_buf_simple_add_u8(msg, opcode);
return;
}
if (opcode < 0x10000) {
/* 2-byte OpCode */
net_buf_simple_add_be16(msg, opcode);
return;
}
/* 3-byte OpCode */
net_buf_simple_add_u8(msg, ((opcode >> 16) & 0xff));
net_buf_simple_add_le16(msg, opcode & 0xffff);
}
static int model_send(struct bt_mesh_model *model,
Bluetooth: Mesh: Fix model publication Model publication was broken in a couple of ways: - The Publish Retransmit State was not taken into account at all - Health Server used a single publish state for all elements To implement Publish Retransmit properly, one has to use a callback to track when the message has been sent. The problem with the transport layer sending APIs was that giving a callback would cause the transport layer to assume that segmentation (with acks) is desired, which is not the case for Model Publication (unless the message itself is too large, of course). Because of this, the message sending context receives a new send_rel ("Send Reliable") boolean member that an app can use to force reliable sending. Another challenge with the Publish Retransmit state is that a buffer is needed for storing the AppKey-encrypted SDU once it has been sent out for the first time.To solve this, a new new net_buf_simple member is added to the model publication context. The separate 'msg' input parameter of the bt_mesh_model_publish() API is removed, since the application is now expected to pre-fill pub->msg instead. To help with the publishing API change, the Health Server model gets a new helper macro for initializing the publishing context with a right-sized publishing message. The API for creating Health Server instances is also redesigned since it was so far using a single model publishing state, which would result in erratic behavior in case of multiple elements with the Health Server Model. Now, the application needs to provide a unique publishing context for each Health Server instance. The changes are heavily intertwined, so it's not easily possible to split them into multiple patches, hence the large(ish) patch. Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
2017-11-18 10:24:18 +02:00
struct bt_mesh_net_tx *tx, bool implicit_bind,
struct net_buf_simple *msg,
const struct bt_mesh_send_cb *cb, void *cb_data)
{
Bluetooth: Mesh: Fix model publication Model publication was broken in a couple of ways: - The Publish Retransmit State was not taken into account at all - Health Server used a single publish state for all elements To implement Publish Retransmit properly, one has to use a callback to track when the message has been sent. The problem with the transport layer sending APIs was that giving a callback would cause the transport layer to assume that segmentation (with acks) is desired, which is not the case for Model Publication (unless the message itself is too large, of course). Because of this, the message sending context receives a new send_rel ("Send Reliable") boolean member that an app can use to force reliable sending. Another challenge with the Publish Retransmit state is that a buffer is needed for storing the AppKey-encrypted SDU once it has been sent out for the first time.To solve this, a new new net_buf_simple member is added to the model publication context. The separate 'msg' input parameter of the bt_mesh_model_publish() API is removed, since the application is now expected to pre-fill pub->msg instead. To help with the publishing API change, the Health Server model gets a new helper macro for initializing the publishing context with a right-sized publishing message. The API for creating Health Server instances is also redesigned since it was so far using a single model publishing state, which would result in erratic behavior in case of multiple elements with the Health Server Model. Now, the application needs to provide a unique publishing context for each Health Server instance. The changes are heavily intertwined, so it's not easily possible to split them into multiple patches, hence the large(ish) patch. Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
2017-11-18 10:24:18 +02:00
BT_DBG("net_idx 0x%04x app_idx 0x%04x dst 0x%04x", tx->ctx->net_idx,
tx->ctx->app_idx, tx->ctx->addr);
BT_DBG("len %u: %s", msg->len, bt_hex(msg->data, msg->len));
if (!bt_mesh_is_provisioned()) {
BT_ERR("Local node is not yet provisioned");
return -EAGAIN;
}
if (net_buf_simple_tailroom(msg) < 4) {
BT_ERR("Not enough tailroom for TransMIC");
return -EINVAL;
}
if (msg->len > BT_MESH_TX_SDU_MAX - 4) {
BT_ERR("Too big message");
return -EMSGSIZE;
}
Bluetooth: Mesh: Fix model publication Model publication was broken in a couple of ways: - The Publish Retransmit State was not taken into account at all - Health Server used a single publish state for all elements To implement Publish Retransmit properly, one has to use a callback to track when the message has been sent. The problem with the transport layer sending APIs was that giving a callback would cause the transport layer to assume that segmentation (with acks) is desired, which is not the case for Model Publication (unless the message itself is too large, of course). Because of this, the message sending context receives a new send_rel ("Send Reliable") boolean member that an app can use to force reliable sending. Another challenge with the Publish Retransmit state is that a buffer is needed for storing the AppKey-encrypted SDU once it has been sent out for the first time.To solve this, a new new net_buf_simple member is added to the model publication context. The separate 'msg' input parameter of the bt_mesh_model_publish() API is removed, since the application is now expected to pre-fill pub->msg instead. To help with the publishing API change, the Health Server model gets a new helper macro for initializing the publishing context with a right-sized publishing message. The API for creating Health Server instances is also redesigned since it was so far using a single model publishing state, which would result in erratic behavior in case of multiple elements with the Health Server Model. Now, the application needs to provide a unique publishing context for each Health Server instance. The changes are heavily intertwined, so it's not easily possible to split them into multiple patches, hence the large(ish) patch. Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
2017-11-18 10:24:18 +02:00
if (!implicit_bind && !model_has_key(model, tx->ctx->app_idx)) {
BT_ERR("Model not bound to AppKey 0x%04x", tx->ctx->app_idx);
return -EINVAL;
}
Bluetooth: Mesh: Fix model publication Model publication was broken in a couple of ways: - The Publish Retransmit State was not taken into account at all - Health Server used a single publish state for all elements To implement Publish Retransmit properly, one has to use a callback to track when the message has been sent. The problem with the transport layer sending APIs was that giving a callback would cause the transport layer to assume that segmentation (with acks) is desired, which is not the case for Model Publication (unless the message itself is too large, of course). Because of this, the message sending context receives a new send_rel ("Send Reliable") boolean member that an app can use to force reliable sending. Another challenge with the Publish Retransmit state is that a buffer is needed for storing the AppKey-encrypted SDU once it has been sent out for the first time.To solve this, a new new net_buf_simple member is added to the model publication context. The separate 'msg' input parameter of the bt_mesh_model_publish() API is removed, since the application is now expected to pre-fill pub->msg instead. To help with the publishing API change, the Health Server model gets a new helper macro for initializing the publishing context with a right-sized publishing message. The API for creating Health Server instances is also redesigned since it was so far using a single model publishing state, which would result in erratic behavior in case of multiple elements with the Health Server Model. Now, the application needs to provide a unique publishing context for each Health Server instance. The changes are heavily intertwined, so it's not easily possible to split them into multiple patches, hence the large(ish) patch. Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
2017-11-18 10:24:18 +02:00
return bt_mesh_trans_send(tx, msg, cb, cb_data);
}
int bt_mesh_model_send(struct bt_mesh_model *model,
struct bt_mesh_msg_ctx *ctx,
struct net_buf_simple *msg,
const struct bt_mesh_send_cb *cb, void *cb_data)
{
Bluetooth: Mesh: Fix model publication Model publication was broken in a couple of ways: - The Publish Retransmit State was not taken into account at all - Health Server used a single publish state for all elements To implement Publish Retransmit properly, one has to use a callback to track when the message has been sent. The problem with the transport layer sending APIs was that giving a callback would cause the transport layer to assume that segmentation (with acks) is desired, which is not the case for Model Publication (unless the message itself is too large, of course). Because of this, the message sending context receives a new send_rel ("Send Reliable") boolean member that an app can use to force reliable sending. Another challenge with the Publish Retransmit state is that a buffer is needed for storing the AppKey-encrypted SDU once it has been sent out for the first time.To solve this, a new new net_buf_simple member is added to the model publication context. The separate 'msg' input parameter of the bt_mesh_model_publish() API is removed, since the application is now expected to pre-fill pub->msg instead. To help with the publishing API change, the Health Server model gets a new helper macro for initializing the publishing context with a right-sized publishing message. The API for creating Health Server instances is also redesigned since it was so far using a single model publishing state, which would result in erratic behavior in case of multiple elements with the Health Server Model. Now, the application needs to provide a unique publishing context for each Health Server instance. The changes are heavily intertwined, so it's not easily possible to split them into multiple patches, hence the large(ish) patch. Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
2017-11-18 10:24:18 +02:00
struct bt_mesh_net_tx tx = {
.sub = bt_mesh_subnet_get(ctx->net_idx),
.ctx = ctx,
.src = bt_mesh_model_elem(model)->addr,
Bluetooth: Mesh: Fix model publication Model publication was broken in a couple of ways: - The Publish Retransmit State was not taken into account at all - Health Server used a single publish state for all elements To implement Publish Retransmit properly, one has to use a callback to track when the message has been sent. The problem with the transport layer sending APIs was that giving a callback would cause the transport layer to assume that segmentation (with acks) is desired, which is not the case for Model Publication (unless the message itself is too large, of course). Because of this, the message sending context receives a new send_rel ("Send Reliable") boolean member that an app can use to force reliable sending. Another challenge with the Publish Retransmit state is that a buffer is needed for storing the AppKey-encrypted SDU once it has been sent out for the first time.To solve this, a new new net_buf_simple member is added to the model publication context. The separate 'msg' input parameter of the bt_mesh_model_publish() API is removed, since the application is now expected to pre-fill pub->msg instead. To help with the publishing API change, the Health Server model gets a new helper macro for initializing the publishing context with a right-sized publishing message. The API for creating Health Server instances is also redesigned since it was so far using a single model publishing state, which would result in erratic behavior in case of multiple elements with the Health Server Model. Now, the application needs to provide a unique publishing context for each Health Server instance. The changes are heavily intertwined, so it's not easily possible to split them into multiple patches, hence the large(ish) patch. Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
2017-11-18 10:24:18 +02:00
.xmit = bt_mesh_net_transmit_get(),
.friend_cred = 0,
};
return model_send(model, &tx, false, msg, cb, cb_data);
}
Bluetooth: Mesh: Fix model publication Model publication was broken in a couple of ways: - The Publish Retransmit State was not taken into account at all - Health Server used a single publish state for all elements To implement Publish Retransmit properly, one has to use a callback to track when the message has been sent. The problem with the transport layer sending APIs was that giving a callback would cause the transport layer to assume that segmentation (with acks) is desired, which is not the case for Model Publication (unless the message itself is too large, of course). Because of this, the message sending context receives a new send_rel ("Send Reliable") boolean member that an app can use to force reliable sending. Another challenge with the Publish Retransmit state is that a buffer is needed for storing the AppKey-encrypted SDU once it has been sent out for the first time.To solve this, a new new net_buf_simple member is added to the model publication context. The separate 'msg' input parameter of the bt_mesh_model_publish() API is removed, since the application is now expected to pre-fill pub->msg instead. To help with the publishing API change, the Health Server model gets a new helper macro for initializing the publishing context with a right-sized publishing message. The API for creating Health Server instances is also redesigned since it was so far using a single model publishing state, which would result in erratic behavior in case of multiple elements with the Health Server Model. Now, the application needs to provide a unique publishing context for each Health Server instance. The changes are heavily intertwined, so it's not easily possible to split them into multiple patches, hence the large(ish) patch. Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
2017-11-18 10:24:18 +02:00
int bt_mesh_model_publish(struct bt_mesh_model *model)
{
NET_BUF_SIMPLE_DEFINE(sdu, BT_MESH_TX_SDU_MAX);
Bluetooth: Mesh: Fix model publication Model publication was broken in a couple of ways: - The Publish Retransmit State was not taken into account at all - Health Server used a single publish state for all elements To implement Publish Retransmit properly, one has to use a callback to track when the message has been sent. The problem with the transport layer sending APIs was that giving a callback would cause the transport layer to assume that segmentation (with acks) is desired, which is not the case for Model Publication (unless the message itself is too large, of course). Because of this, the message sending context receives a new send_rel ("Send Reliable") boolean member that an app can use to force reliable sending. Another challenge with the Publish Retransmit state is that a buffer is needed for storing the AppKey-encrypted SDU once it has been sent out for the first time.To solve this, a new new net_buf_simple member is added to the model publication context. The separate 'msg' input parameter of the bt_mesh_model_publish() API is removed, since the application is now expected to pre-fill pub->msg instead. To help with the publishing API change, the Health Server model gets a new helper macro for initializing the publishing context with a right-sized publishing message. The API for creating Health Server instances is also redesigned since it was so far using a single model publishing state, which would result in erratic behavior in case of multiple elements with the Health Server Model. Now, the application needs to provide a unique publishing context for each Health Server instance. The changes are heavily intertwined, so it's not easily possible to split them into multiple patches, hence the large(ish) patch. Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
2017-11-18 10:24:18 +02:00
struct bt_mesh_model_pub *pub = model->pub;
struct bt_mesh_app_key *key;
Bluetooth: Mesh: Fix model publication Model publication was broken in a couple of ways: - The Publish Retransmit State was not taken into account at all - Health Server used a single publish state for all elements To implement Publish Retransmit properly, one has to use a callback to track when the message has been sent. The problem with the transport layer sending APIs was that giving a callback would cause the transport layer to assume that segmentation (with acks) is desired, which is not the case for Model Publication (unless the message itself is too large, of course). Because of this, the message sending context receives a new send_rel ("Send Reliable") boolean member that an app can use to force reliable sending. Another challenge with the Publish Retransmit state is that a buffer is needed for storing the AppKey-encrypted SDU once it has been sent out for the first time.To solve this, a new new net_buf_simple member is added to the model publication context. The separate 'msg' input parameter of the bt_mesh_model_publish() API is removed, since the application is now expected to pre-fill pub->msg instead. To help with the publishing API change, the Health Server model gets a new helper macro for initializing the publishing context with a right-sized publishing message. The API for creating Health Server instances is also redesigned since it was so far using a single model publishing state, which would result in erratic behavior in case of multiple elements with the Health Server Model. Now, the application needs to provide a unique publishing context for each Health Server instance. The changes are heavily intertwined, so it's not easily possible to split them into multiple patches, hence the large(ish) patch. Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
2017-11-18 10:24:18 +02:00
struct bt_mesh_msg_ctx ctx = {
};
struct bt_mesh_net_tx tx = {
.ctx = &ctx,
.src = bt_mesh_model_elem(model)->addr,
Bluetooth: Mesh: Fix model publication Model publication was broken in a couple of ways: - The Publish Retransmit State was not taken into account at all - Health Server used a single publish state for all elements To implement Publish Retransmit properly, one has to use a callback to track when the message has been sent. The problem with the transport layer sending APIs was that giving a callback would cause the transport layer to assume that segmentation (with acks) is desired, which is not the case for Model Publication (unless the message itself is too large, of course). Because of this, the message sending context receives a new send_rel ("Send Reliable") boolean member that an app can use to force reliable sending. Another challenge with the Publish Retransmit state is that a buffer is needed for storing the AppKey-encrypted SDU once it has been sent out for the first time.To solve this, a new new net_buf_simple member is added to the model publication context. The separate 'msg' input parameter of the bt_mesh_model_publish() API is removed, since the application is now expected to pre-fill pub->msg instead. To help with the publishing API change, the Health Server model gets a new helper macro for initializing the publishing context with a right-sized publishing message. The API for creating Health Server instances is also redesigned since it was so far using a single model publishing state, which would result in erratic behavior in case of multiple elements with the Health Server Model. Now, the application needs to provide a unique publishing context for each Health Server instance. The changes are heavily intertwined, so it's not easily possible to split them into multiple patches, hence the large(ish) patch. Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
2017-11-18 10:24:18 +02:00
.xmit = bt_mesh_net_transmit_get(),
};
int err;
BT_DBG("");
Bluetooth: Mesh: Fix model publication Model publication was broken in a couple of ways: - The Publish Retransmit State was not taken into account at all - Health Server used a single publish state for all elements To implement Publish Retransmit properly, one has to use a callback to track when the message has been sent. The problem with the transport layer sending APIs was that giving a callback would cause the transport layer to assume that segmentation (with acks) is desired, which is not the case for Model Publication (unless the message itself is too large, of course). Because of this, the message sending context receives a new send_rel ("Send Reliable") boolean member that an app can use to force reliable sending. Another challenge with the Publish Retransmit state is that a buffer is needed for storing the AppKey-encrypted SDU once it has been sent out for the first time.To solve this, a new new net_buf_simple member is added to the model publication context. The separate 'msg' input parameter of the bt_mesh_model_publish() API is removed, since the application is now expected to pre-fill pub->msg instead. To help with the publishing API change, the Health Server model gets a new helper macro for initializing the publishing context with a right-sized publishing message. The API for creating Health Server instances is also redesigned since it was so far using a single model publishing state, which would result in erratic behavior in case of multiple elements with the Health Server Model. Now, the application needs to provide a unique publishing context for each Health Server instance. The changes are heavily intertwined, so it's not easily possible to split them into multiple patches, hence the large(ish) patch. Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
2017-11-18 10:24:18 +02:00
if (!pub) {
return -ENOTSUP;
}
Bluetooth: Mesh: Fix model publication Model publication was broken in a couple of ways: - The Publish Retransmit State was not taken into account at all - Health Server used a single publish state for all elements To implement Publish Retransmit properly, one has to use a callback to track when the message has been sent. The problem with the transport layer sending APIs was that giving a callback would cause the transport layer to assume that segmentation (with acks) is desired, which is not the case for Model Publication (unless the message itself is too large, of course). Because of this, the message sending context receives a new send_rel ("Send Reliable") boolean member that an app can use to force reliable sending. Another challenge with the Publish Retransmit state is that a buffer is needed for storing the AppKey-encrypted SDU once it has been sent out for the first time.To solve this, a new new net_buf_simple member is added to the model publication context. The separate 'msg' input parameter of the bt_mesh_model_publish() API is removed, since the application is now expected to pre-fill pub->msg instead. To help with the publishing API change, the Health Server model gets a new helper macro for initializing the publishing context with a right-sized publishing message. The API for creating Health Server instances is also redesigned since it was so far using a single model publishing state, which would result in erratic behavior in case of multiple elements with the Health Server Model. Now, the application needs to provide a unique publishing context for each Health Server instance. The changes are heavily intertwined, so it's not easily possible to split them into multiple patches, hence the large(ish) patch. Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
2017-11-18 10:24:18 +02:00
if (pub->addr == BT_MESH_ADDR_UNASSIGNED) {
return -EADDRNOTAVAIL;
}
Bluetooth: Mesh: Fix model publication Model publication was broken in a couple of ways: - The Publish Retransmit State was not taken into account at all - Health Server used a single publish state for all elements To implement Publish Retransmit properly, one has to use a callback to track when the message has been sent. The problem with the transport layer sending APIs was that giving a callback would cause the transport layer to assume that segmentation (with acks) is desired, which is not the case for Model Publication (unless the message itself is too large, of course). Because of this, the message sending context receives a new send_rel ("Send Reliable") boolean member that an app can use to force reliable sending. Another challenge with the Publish Retransmit state is that a buffer is needed for storing the AppKey-encrypted SDU once it has been sent out for the first time.To solve this, a new new net_buf_simple member is added to the model publication context. The separate 'msg' input parameter of the bt_mesh_model_publish() API is removed, since the application is now expected to pre-fill pub->msg instead. To help with the publishing API change, the Health Server model gets a new helper macro for initializing the publishing context with a right-sized publishing message. The API for creating Health Server instances is also redesigned since it was so far using a single model publishing state, which would result in erratic behavior in case of multiple elements with the Health Server Model. Now, the application needs to provide a unique publishing context for each Health Server instance. The changes are heavily intertwined, so it's not easily possible to split them into multiple patches, hence the large(ish) patch. Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
2017-11-18 10:24:18 +02:00
key = bt_mesh_app_key_find(pub->key);
if (!key) {
return -EADDRNOTAVAIL;
}
Bluetooth: Mesh: Fix model publication Model publication was broken in a couple of ways: - The Publish Retransmit State was not taken into account at all - Health Server used a single publish state for all elements To implement Publish Retransmit properly, one has to use a callback to track when the message has been sent. The problem with the transport layer sending APIs was that giving a callback would cause the transport layer to assume that segmentation (with acks) is desired, which is not the case for Model Publication (unless the message itself is too large, of course). Because of this, the message sending context receives a new send_rel ("Send Reliable") boolean member that an app can use to force reliable sending. Another challenge with the Publish Retransmit state is that a buffer is needed for storing the AppKey-encrypted SDU once it has been sent out for the first time.To solve this, a new new net_buf_simple member is added to the model publication context. The separate 'msg' input parameter of the bt_mesh_model_publish() API is removed, since the application is now expected to pre-fill pub->msg instead. To help with the publishing API change, the Health Server model gets a new helper macro for initializing the publishing context with a right-sized publishing message. The API for creating Health Server instances is also redesigned since it was so far using a single model publishing state, which would result in erratic behavior in case of multiple elements with the Health Server Model. Now, the application needs to provide a unique publishing context for each Health Server instance. The changes are heavily intertwined, so it's not easily possible to split them into multiple patches, hence the large(ish) patch. Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
2017-11-18 10:24:18 +02:00
if (pub->msg->len + 4 > BT_MESH_TX_SDU_MAX) {
BT_ERR("Message does not fit maximum SDU size");
return -EMSGSIZE;
}
if (pub->count) {
BT_WARN("Clearing publish retransmit timer");
k_delayed_work_cancel(&pub->timer);
}
net_buf_simple_add_mem(&sdu, pub->msg->data, pub->msg->len);
Bluetooth: Mesh: Fix model publication Model publication was broken in a couple of ways: - The Publish Retransmit State was not taken into account at all - Health Server used a single publish state for all elements To implement Publish Retransmit properly, one has to use a callback to track when the message has been sent. The problem with the transport layer sending APIs was that giving a callback would cause the transport layer to assume that segmentation (with acks) is desired, which is not the case for Model Publication (unless the message itself is too large, of course). Because of this, the message sending context receives a new send_rel ("Send Reliable") boolean member that an app can use to force reliable sending. Another challenge with the Publish Retransmit state is that a buffer is needed for storing the AppKey-encrypted SDU once it has been sent out for the first time.To solve this, a new new net_buf_simple member is added to the model publication context. The separate 'msg' input parameter of the bt_mesh_model_publish() API is removed, since the application is now expected to pre-fill pub->msg instead. To help with the publishing API change, the Health Server model gets a new helper macro for initializing the publishing context with a right-sized publishing message. The API for creating Health Server instances is also redesigned since it was so far using a single model publishing state, which would result in erratic behavior in case of multiple elements with the Health Server Model. Now, the application needs to provide a unique publishing context for each Health Server instance. The changes are heavily intertwined, so it's not easily possible to split them into multiple patches, hence the large(ish) patch. Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
2017-11-18 10:24:18 +02:00
ctx.addr = pub->addr;
ctx.send_ttl = pub->ttl;
ctx.net_idx = key->net_idx;
ctx.app_idx = key->app_idx;
tx.friend_cred = pub->cred;
Bluetooth: Mesh: Fix model publication Model publication was broken in a couple of ways: - The Publish Retransmit State was not taken into account at all - Health Server used a single publish state for all elements To implement Publish Retransmit properly, one has to use a callback to track when the message has been sent. The problem with the transport layer sending APIs was that giving a callback would cause the transport layer to assume that segmentation (with acks) is desired, which is not the case for Model Publication (unless the message itself is too large, of course). Because of this, the message sending context receives a new send_rel ("Send Reliable") boolean member that an app can use to force reliable sending. Another challenge with the Publish Retransmit state is that a buffer is needed for storing the AppKey-encrypted SDU once it has been sent out for the first time.To solve this, a new new net_buf_simple member is added to the model publication context. The separate 'msg' input parameter of the bt_mesh_model_publish() API is removed, since the application is now expected to pre-fill pub->msg instead. To help with the publishing API change, the Health Server model gets a new helper macro for initializing the publishing context with a right-sized publishing message. The API for creating Health Server instances is also redesigned since it was so far using a single model publishing state, which would result in erratic behavior in case of multiple elements with the Health Server Model. Now, the application needs to provide a unique publishing context for each Health Server instance. The changes are heavily intertwined, so it's not easily possible to split them into multiple patches, hence the large(ish) patch. Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
2017-11-18 10:24:18 +02:00
tx.sub = bt_mesh_subnet_get(ctx.net_idx),
pub->count = BT_MESH_PUB_TRANSMIT_COUNT(pub->retransmit);
Bluetooth: Mesh: Fix model publication Model publication was broken in a couple of ways: - The Publish Retransmit State was not taken into account at all - Health Server used a single publish state for all elements To implement Publish Retransmit properly, one has to use a callback to track when the message has been sent. The problem with the transport layer sending APIs was that giving a callback would cause the transport layer to assume that segmentation (with acks) is desired, which is not the case for Model Publication (unless the message itself is too large, of course). Because of this, the message sending context receives a new send_rel ("Send Reliable") boolean member that an app can use to force reliable sending. Another challenge with the Publish Retransmit state is that a buffer is needed for storing the AppKey-encrypted SDU once it has been sent out for the first time.To solve this, a new new net_buf_simple member is added to the model publication context. The separate 'msg' input parameter of the bt_mesh_model_publish() API is removed, since the application is now expected to pre-fill pub->msg instead. To help with the publishing API change, the Health Server model gets a new helper macro for initializing the publishing context with a right-sized publishing message. The API for creating Health Server instances is also redesigned since it was so far using a single model publishing state, which would result in erratic behavior in case of multiple elements with the Health Server Model. Now, the application needs to provide a unique publishing context for each Health Server instance. The changes are heavily intertwined, so it's not easily possible to split them into multiple patches, hence the large(ish) patch. Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
2017-11-18 10:24:18 +02:00
BT_DBG("Publish Retransmit Count %u Interval %ums", pub->count,
BT_MESH_PUB_TRANSMIT_INT(pub->retransmit));
err = model_send(model, &tx, true, &sdu, &pub_sent_cb, model);
Bluetooth: Mesh: Fix model publication Model publication was broken in a couple of ways: - The Publish Retransmit State was not taken into account at all - Health Server used a single publish state for all elements To implement Publish Retransmit properly, one has to use a callback to track when the message has been sent. The problem with the transport layer sending APIs was that giving a callback would cause the transport layer to assume that segmentation (with acks) is desired, which is not the case for Model Publication (unless the message itself is too large, of course). Because of this, the message sending context receives a new send_rel ("Send Reliable") boolean member that an app can use to force reliable sending. Another challenge with the Publish Retransmit state is that a buffer is needed for storing the AppKey-encrypted SDU once it has been sent out for the first time.To solve this, a new new net_buf_simple member is added to the model publication context. The separate 'msg' input parameter of the bt_mesh_model_publish() API is removed, since the application is now expected to pre-fill pub->msg instead. To help with the publishing API change, the Health Server model gets a new helper macro for initializing the publishing context with a right-sized publishing message. The API for creating Health Server instances is also redesigned since it was so far using a single model publishing state, which would result in erratic behavior in case of multiple elements with the Health Server Model. Now, the application needs to provide a unique publishing context for each Health Server instance. The changes are heavily intertwined, so it's not easily possible to split them into multiple patches, hence the large(ish) patch. Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
2017-11-18 10:24:18 +02:00
if (err) {
/* Don't try retransmissions for this publish attempt */
pub->count = 0U;
/* Make sure the publish timer gets reset */
publish_sent(err, model);
Bluetooth: Mesh: Fix model publication Model publication was broken in a couple of ways: - The Publish Retransmit State was not taken into account at all - Health Server used a single publish state for all elements To implement Publish Retransmit properly, one has to use a callback to track when the message has been sent. The problem with the transport layer sending APIs was that giving a callback would cause the transport layer to assume that segmentation (with acks) is desired, which is not the case for Model Publication (unless the message itself is too large, of course). Because of this, the message sending context receives a new send_rel ("Send Reliable") boolean member that an app can use to force reliable sending. Another challenge with the Publish Retransmit state is that a buffer is needed for storing the AppKey-encrypted SDU once it has been sent out for the first time.To solve this, a new new net_buf_simple member is added to the model publication context. The separate 'msg' input parameter of the bt_mesh_model_publish() API is removed, since the application is now expected to pre-fill pub->msg instead. To help with the publishing API change, the Health Server model gets a new helper macro for initializing the publishing context with a right-sized publishing message. The API for creating Health Server instances is also redesigned since it was so far using a single model publishing state, which would result in erratic behavior in case of multiple elements with the Health Server Model. Now, the application needs to provide a unique publishing context for each Health Server instance. The changes are heavily intertwined, so it's not easily possible to split them into multiple patches, hence the large(ish) patch. Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
2017-11-18 10:24:18 +02:00
return err;
}
Bluetooth: Mesh: Fix model publication Model publication was broken in a couple of ways: - The Publish Retransmit State was not taken into account at all - Health Server used a single publish state for all elements To implement Publish Retransmit properly, one has to use a callback to track when the message has been sent. The problem with the transport layer sending APIs was that giving a callback would cause the transport layer to assume that segmentation (with acks) is desired, which is not the case for Model Publication (unless the message itself is too large, of course). Because of this, the message sending context receives a new send_rel ("Send Reliable") boolean member that an app can use to force reliable sending. Another challenge with the Publish Retransmit state is that a buffer is needed for storing the AppKey-encrypted SDU once it has been sent out for the first time.To solve this, a new new net_buf_simple member is added to the model publication context. The separate 'msg' input parameter of the bt_mesh_model_publish() API is removed, since the application is now expected to pre-fill pub->msg instead. To help with the publishing API change, the Health Server model gets a new helper macro for initializing the publishing context with a right-sized publishing message. The API for creating Health Server instances is also redesigned since it was so far using a single model publishing state, which would result in erratic behavior in case of multiple elements with the Health Server Model. Now, the application needs to provide a unique publishing context for each Health Server instance. The changes are heavily intertwined, so it's not easily possible to split them into multiple patches, hence the large(ish) patch. Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
2017-11-18 10:24:18 +02:00
return 0;
}
struct bt_mesh_model *bt_mesh_model_find_vnd(struct bt_mesh_elem *elem,
u16_t company, u16_t id)
{
u8_t i;
for (i = 0U; i < elem->vnd_model_count; i++) {
if (elem->vnd_models[i].vnd.company == company &&
elem->vnd_models[i].vnd.id == id) {
return &elem->vnd_models[i];
}
}
return NULL;
}
struct bt_mesh_model *bt_mesh_model_find(struct bt_mesh_elem *elem,
u16_t id)
{
u8_t i;
for (i = 0U; i < elem->model_count; i++) {
if (elem->models[i].id == id) {
return &elem->models[i];
}
}
return NULL;
}
const struct bt_mesh_comp *bt_mesh_comp_get(void)
{
return dev_comp;
}